Rally post-terrorism

Chen, Shuai
A icle — Published Ve sion
Rally pos - e o ism
Jou nal o Popula ion Economics
P o ided in Coope a ion wi h:
Sp inge Na u e
Sugges ed Ci a ion: Chen, Shuai (2025) : Rally pos - e o ism, Jou nal o Popula ion Economics, ISSN
1432-1475, Sp inge , Be lin, Heidelbe g, Vol. 38, Iss. 2,
h ps://doi.o g/10.1007/s00148-025-01106-2
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/323186
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h p://c ea i ecommons.o g/licenses/by/4.0/
Jou nal o Popula ion Economics (2025) 38:50
h ps://doi.o g/10.1007/s00148-025-01106-2
ORIGINAL PAPER
Rally pos - e o ism
Shuai Chen1,2,3
Recei ed: 13 July 2022 / Accep ed: 16 May 2025 / Published online: 2 June 2025
© The Au ho (s) 2025
Abs ac
This s udy examines whe he he “ ally ’ ound he lag” phenomenon is p esen
a e e o is a acks, and in es iga es explana ions o his inc ease in con idence
in na ional poli ical ins i u ions and app o al o he coun y leade ’s job pe o mance.
I exploi a ia ions in e o is occu ences and esul s (success o ailu e) ac oss sub-
na ional EU egions whe e a leas one a ack ook place du ing he da a pe iod. I
show empi ically ha bo h e o is occu ences and esul s a e plausibly exogenous
o he p io poli ical and economic clima e. Conduc ing a di e ence-in-di e ences
analysis, I compa e changes in poli ical con idence and app o al among indi iduals
who we e exposed o an a ack in hei egion wi h hose who we e no . U ilizing
ano he mo e sophis ica ed iden i ica ion, I also compa e such poli ical changes a e
success ul a acks wi h hose a e ailed a acks. I ind ha pos - e o ism, indi idual
poli ical con idence and suppo signi ican ly inc eased by mo e han 10 pe cen age
poin s, and ha his poli ical inc emen was o e 5 pe cen age poin s a e success-
ul a acks ela i e o ailed ones. Such ally e ec s we e empo a y and aded away
wi hin a yea . Fu he mo e, I explo e a ious po en ial channels sugges ing pa io ism
and ci ic engagemen as mechanisms while ejec ing pe cei ed economic cap u e and
poli ical acquisi ion as al e na i e explana ions.
Keywo ds Rally ’ ound he lag e ec ·Te o ism ·Con idence in ins i u ions ·
Poli ical app o al ·Pa io ism ·Ci ic engagemen ·Economic cap u e ·
Poli ical acquisi ion
JEL Classi ica ion D74 ·H12 ·P16
Responsible edi o : Al onso Flo es-Lagunes
BShuai Chen
sc956@leices e .ac.uk
1Uni e si y o Leices e , Uni e si y Road, Leices e LE1 7RH, UK
2Ins i u e o Labo Economics (IZA), Bonn, Ge many
3Global Labo O ganiza ion (GLO), Essen, Ge many
123
50 Page 2 o 29 S. Chen
1 In oduc ion
In he poli ical sciences li e a u e, he “ ally ’ ound he lag” e ec e e s o he phe-
nomenono a iseinpopula suppo o hena ionalleade ship ollowingin e na ional
c ises (Muelle 1970; He he ing on and Nelson 2003). A e he 9/11 a acks and
despi e he inc eased coun e e o ism spending wo ldwide, bo h ansna ional and
domes ic e o is inciden s ha e esu ged, and he consequences o ansna ional e -
o ism ha e become mo e disas ous han hose o domes ic e o ism, especially
since 2008 (see Sec ion 2 o Gaibulloe and Sandle 2019). Thus, e o ism is likely
o ha e become an impo an po en ial sou ce o he ally phenomenon. Such a ally
e ec ollowing e o is a acks has impo an economic implica ions. Fo immedia e
c isis aid o managemen a e e o ism, a go e nmen o en needs o ealloca e eco-
nomic esou ces omp oduc i ein es men s owa dgo e nmen spending(Blombe g
e al. 2004). Mo eo e , making coun e insu gency e o s, launching econs uc ion
and communi y p ojec s, p o iding social se ices and aking la ge numbe s o indis-
c imina e concilia o y ac s a e e ec i e measu es he go e nmen can ake in he
sho - o mid- e m o educe e o ism (Be man 2009;Be mane al.2011; Dugan and
Chenowe h 2012). The success o hese measu es elies on he ci ilian ally: popu-
la suppo o he na ional leade ship and/o ci ic pa icipa ion in he p ojec s and
se ices.
The cu en s udy examines he possibili y o he ally e ec by in es iga ing he
in luences o localized e o ism a he subna ional le el on indi idual con idence in
na ional poli ical ins i u ions and poli ical suppo o he coun y’s leade in he Eu o-
pean Union. Mo eo e , I explo e a ious po en ial mechanisms, including pa io ism,
ci ic engagemen , pe cei ed economic cap u e and pe cei ed poli ical acquisi ion by
he go e nmen . This in es iga ion may acili a e unde s anding whe he such poli ical
solida i y is mo e likely o be olun a y o o be elici ed by he poli ical incumben .
Speci ically, I manually assemble a comp ehensi e su ey a he indi idual le el
wi h he uni e se o e o is a acks in he EU a he i s -o de subna ional (i.e., NUTS
1) egional le el. This is possible h ough u ilizing su ey in o ma ion abou in e iew
da es and loca ions. In his way, I connec he expe ience o e o ism wi h indi idual
con idence in poli ical ins i u ions and poli ical app o al. I examine bo h con idence
in he na ional go e nmen and app o al o he coun y leade ’s pe o mance.
I apply a di e ence-in-di e ences (DID) app oach o compa e changes in poli ical
con idence and app o al among indi iduals who we e exposed o a e o is a ack
in hei egion wi h hose who we e no . I es ic he sample o indi iduals esiding
in egions ha we e hi by a leas one a ack du ing he da a pe iod o 2008–2016.
I impose his es ic ion because a ge ed egions may di e om non- a ge ed ones
in a ious aspec s, including he poli ical landscape. Such geog aphic he e ogenei y
migh also in e wine wi h local poli ical shocks o igge e o is occu ences.
Inspi ed by B odeu (2018), I ake wo s eps o e i y my iden i ica ion s a -
egy. Fi s , I show ha he occu ence o e o ism is o hogonal o a ious po en ial
con ounding a i udes o, and pe cep ions o , law and o de , poli ics and economic
condi ions o 1 yea , 6 mon hs, o 3 mon hs be o e a e o is inciden . Thus, I a gue
ha he a ack is plausibly exogenous o he p io poli ical and economic clima e ha
may be co ela ed wi h indi idual poli ical con idence and app o al.
123
Rally pos - e o ism Page 3 o 29 50
Second, in he same manne , I again p o ide e idence ha he success o ailu e o
e o ism (condi ional on i s occu ence) is also plausibly exogenous o he pe cei ed
poli ical and economic si ua ion be o e an a ack. The success o ailu e o e o is
a acks is de ined based on he ype o a ack ha ook place, and hus is dependen on
he la e ac o .1The iden i ica ion o e o ism e ec s hen comes om he “inhe en
andomness” o he success o ailu e o an a ack (B odeu 2018). Wi h his mo e
sophis ica ed s a egy, I hen compa e changes in poli ical con idence and app o al
a e success ul a acks wi h hose a e ailed ones. By hese wo means, I documen
a consis en ally e ec a e e o ism.
I ind ha in he subsequen mon hs up o 1 yea pos -a acks, he likelihood o
ha ing con idence in he na ional go e nmen inc eased by 11 pe cen age poin s. The
likelihood o app o al o he coun y leade ’s pe o mance ose by mo e han 13
pe cen age poin s a e a e o is inciden . Bo h e ec s a e sizable, in compa ison o
a ound 40% a e o poli ical con idence and app o al p io o e o ism. Mo eo e , he
inc emen was 6.5 pe cen age poin s o con idence in he na ional go e nmen , and
mo e han 5 pe cen age poin s o app o al o he coun y leade ’s pe o mance a e
success ul a acks ela i e o ailed ones. These inc emen s a e no i ial, compa ed
wi h abou 50% a e o con idence and 45% a e o app o al in he case o ailed
e o ism. The ally e ec is hus exhibi ed pos - e o ism. Howe e , such an e ec
is sho - e m and ades away wo qua e s a e he e o is inciden , and one qua e
a e he success ul a ack in compa ison o he ailed a ack.
In a second s ep, I explo e po en ial channels h ough which e o ism a ec s indi-
idual poli ical con idence and app o al. Fi s , I p o ide sugges i e e idence o he
mechanisms o pa io ism and ci ic engagemen . In egions hi by e o is a acks,
s onge pa io ism may p e ail among ci ilians. Wi h p oxies o pa io ism, I doc-
umen a signi ican imp o emen in na ionalis sen imen ollowing e o ism. I also
ind ha he inc eases in poli ical con idence and app o al a e a acks launched by
domes ic pe pe a o s we e only hal as la ge as hose a e a acks launched by non-
domes ic e o is s. This e idence may also e lec an elemen o na ionalism. Ano he
mechanism o ci ic engagemen may come in o e ec i esiden s pa icipa e in ac i -
i ies o communi y ebuilding and need aid o suppo om he go e nmen . Thei
imp ession o he go e nmen may imp o e h ough mu ual coope a ion wi h hese
ac i i ies. I ind e idence o enhanced ci ic engagemen and educed dissen om
o icials a e e o is inciden s, which implies his channel o ci ic engagemen .
I hen in es iga e o he al e na i e explana ions ela ed o pe cei ed economic
cap u eandpoli icalacquisi ionby hego e nmen .A e a e o is a ack, hego e n-
men may alloca e economic esou ces o he a ec ed egion o paci y local ci ilians
and main ain hei poli ical suppo . Wi hou di ec measu emen s o esou ce allo-
ca ion, I ins ead analyze he be o e/a e e o ism a ia ion in a composi e index o
sa is ac ion wi h a ious aspec s o he esiden s’ a ea, including in as uc u e and
housing, as well as a ia ions in con idence in he police and pe cei ed sa e y in he
local egion. I do no ind any signi ican e ec s on he subjec i e o pe cei ed eal-
loca ion o economic esou ces in he a ge ed a eas. Gi en he subjec i e na u e o
1Appendix C epo s he ypes o e o is a acks and hei co esponding de ini ions o success in he
Global Te o ism Da abase. I an a ack is no conside ed success ul, i is ega ded as ailed.
123
50 Page 4 o 29 S. Chen
poli ical con idence and app o al, pe cei ed economic cap u e does no seem o play
an impo an ole in explaining he poli ical e ec s o e o ism. In addi ion, a go e n-
men may g an mo e poli ical eedom in o de o acqui e local suppo . Examining
he pe cei ed eedom o exp essing poli ical iews and he pe cei ed eedom o
he media, I do no ind e idence o any co esponding imp o emen a e e o ism.
Gi en ha he Eu opean Union is al eady an ad anced democ acy, he e may no be
much oom in e ms o anchise ex ension o he go e nmen o manipula e.
This s udy add esses se e al s ands o li e a u e. P incipally, i adds o he g owing
academic discussion on he poli ical consequences o e o ism. Gode oid (2023)
conduc ed a me a-analysis on his li e a u e, concluding ha e o ism was associa ed
wi h ou -g oup hos ili y, poli ical conse a ism and he ally phenomenon, hough he
e ec sizes depended on con ex s. The i s ocus o he discussion is on suppo o
and a i udes owa d he incumben . Gassebne e al. (2011) analyzed mo e han wo
housand cabine s in o e 150 coun ies be ween 1970 and 2002, and ound ha on
a e age, e o ism sho ened cabine du a ion. Using he iming o he e o is a acks
in Mad id on 11 Ma ch 2004, only 3 days be o e he cong essional elec ion, Mon-
al o (2011) compa ed indi iduals o ing be o e he e o is ac s wi h hose o ing
a e wa ds. He documen ed a nega i e impac o he a acks on elec o al suppo o
he uling conse a i e pa y. Kib is (2011) used he secu i y o ce casual ies in he
igh agains e o is s as a measu emen o he e o is in ensi y in Tu key, and ound
ha Tu kish o e s we e sensi i e o hese inciden s; hey blamed he go e nmen o
he losses. Chowanie z (2011) analyzed he esponses o poli ical eli es and pa ies
o e o is ac s, and ound a ally e ec among hem. Chanley (2002) and Pe in and
Smolek (2009) in es iga ed he a e ma h o he 9/11 a acks in he USA and docu-
men ed an inc ease in us in he na ional go e nmen . Simila ly, Dinesen and Jæge
(2013) examined changes in us in ins i u ions a e he 3/11 Mad id e o ism, and
expanded he ypes o ins i u ions o poli ics, media and jus ice. Nussio (2020) s ud-
ied he a i udinal and emo ional consequences o he 2016 Be lin Ch is mas ma ke
Islamis e o ism, and documen ed unchanged poli ical a i udes bu escala ed sad-
ness and ange . Somewha less ela ed is he wo k o Bozzoli and Mülle (2011), who
used he London e o is a acks in 2005 as a shock o examine indi idual isk pe -
cep ions. They ound ha e o is -induced isk assessmen enhanced he willingness
o ade o secu i y o libe ies. Linking changes in indi idual democ a ic a i udes o
egional a ia ions in e o is a acks in Pakis an, Rehman and Vanin (2017) iden i ied
a posi i e associa ion be ween pe sis en exposu e o e o ism and lowe suppo o
democ a ic alues.
Closes o my wo k a e he pape s o Pe i e al. (2020,2021) and Ama asinghe
(2023). Pe i e al. (2020,2021) independen ly s udied he link be ween e o ism
and poli ical opinions using Eu opean Social Su ey da a. The au ho s documen ed
inc eased us in he coun y’s pa liamen and sa is ac ion wi h he na ional go e n-
men , bu li le e idence o changes in a i udes o immig a ion. Ama asinghe (2023)
examined how a cons uc ed composi e index o public discon en changes a e e -
o ism, wi h da a a he coun y le el o 135 coun ies. The au ho ound a sho - e m
inc ease in public discon en ollowing success ul a acks, and p o ided sugges i e
e idence ha such discon en is aimed a he go e nmen . Compa ed wi h hese s ud-
ies and o he p io wo k, my s udy adds o he li e a u e in he ollowing aspec s: i s ,
123

Rally pos - e o ism Page 5 o 29 50
he ich Gallup su ey da a in my s udy help o be e delinea e po en ial mechanisms
o he e o ism-induced ally e ec . I am able o simul aneously explo e di e en
channels by analyzing p oxy a iables o pa io ism, ci ic engagemen , pe cei ed
economic cap u e and poli ical acquisi ion by he go e nmen . Thus, his s udy o e s
he i s a emp o empi ically disen angle olun a y solida i y om economically
o poli ically elici ed solida i y in he con ex o e o ism. Second, in he empi i-
cal analysis, I aim o combine mic o-based da a a he indi idual le el wi h speci ic
cha ac e is ics o e o ism such as a ack ypes, logis ics and salience. This app oach
acili a es he compa ison o a acks o he same class. Gi en he empi ical e idence
p o ided in he pape o plausible exogenei y o he occu ence and esul o e o -
ism, his app oach can hence imp o e he iden i ica ion o causal e ec s. Mo eo e , I
exploi wo iden i ica ion s a egies using e o is occu ence and esul , espec i ely,
o e i y he obus ness o my esul s and conclusions. Finally, le e aging e en da a
a he mo e p ecise subna ional adminis a i e (i.e., NUTS 1) egional le el, I es ab-
lish he pos -localized- e o ism ally e ec a he indi idual le el, in e ms o bo h
na ional poli ical ins i u ions and he coun y’s leade . Following localized e o ism
in he EU, no only did indi idual con idence in he na ional go e nmen inc ease, bu
indi idual suppo o he coun y leade ’s pe o mance also imp o ed in he sho un.
Ano he ocus o he poli ical discussion is on he co esponding shi o he poli i-
cal landscape pos - e o ism. Gould and Klo (2010) elied on geog aphic a ia ions in
e o ism om 1988 o 2006 in Is ael o in es iga e poli ical esponses o Is aelis. They
ound ha Is aelis became mo e willing o make e i o ial concessions and shi ed
hei poli ical landscape o he le ollowing e o is inciden s. E en hough Is aelis
ended o o e o igh -wing pa ies, hese pa ies mo ed o he le a e e o ism.2
Using a Ge man indi idual panel om 1999 o 2003, Schülle (2015) s udied he e ec
o he 9/11 e o is a acks on poli ical mobiliza ion and de ec ed a esul ing shi in
a conse a i e di ec ion. As o he same inciden , Kaushal e al. (2007) ound a ise
in ha e c imes and a educ ion in in as a e mig a ion o A ab and Muslim men, and
Åslund and Roo h (2005) documen ed de e io a ing a i udes owa d ce ain e hnic
mino i ies in Sweden a e he 9/11 a acks in he U.S. Alesina e al. (2019)de el-
oped a model showing ha o ing o pa ies agains c iminal o ganiza ions dec eased
a e iolence conduc ed by hese o ganiza ions. They ound empi ical e idence by
analyzing la ge da a in Sicily and ac oss I alian egions. Combining long- e m indi-
idual panels om se e al Wes e n coun ies wi h global e o ism in o ma ion o e
20 yea s, Akay e al. (2020) ound ha global e o ism educed subjec i e well-being
and aised he in en ion o o e o conse a i e pa ies. Gia azzi e al. (2020) u i-
lized a machine lea ning me hod o analyze ex s o Twi e use s in Ge many. They
disco e ed ha a e e o ism and c iminal e en s, Twi e ex s on a e age became
simila o hose o he a - igh pa y A D. They also ound an associa ion be ween
shi s in ex ual simila i y and changes in he o e sha es in ede al elec ions. The as
majo i y o he abo e s udies p esen ed a o ing inclina ion owa d he igh -wing o
conse a i e pa ies in he mid- o longe - e m. By con as , he cu en pape deals
wi h he p omp poli ical a i udinal esponses o e o is a acks. I documen new
2Be ebi and Klo (2006,2008) and Ge mansky and Zei zo (2014) also documen ed an inc ease in o ing
o ing-wing pa ies in esponse o e o ism in Is ael.
123
50 Page 6 o 29 S. Chen
e idence o he ansi o y a i udinal eac ions conce ning poli ical con idence and
app o al di ec ly a e e o is inciden s. The co esponding ally e ec in e ms o
hese poli ical a i udes became nonsigni ican wi hin a yea ollowing e o ism.
Mo eo e , my pape complemen s he li e a u e on he economic impac o con-
lic s o un es . Wi h ega d o e o ism, Abadie and Ga deazabal (2003) documen ed
a 10 pe cen age poin s educ ion o GDP pe capi a in he Basque Coun y ollowing
he ou b eak o e o ism in he la e 1960s compa ed wi h a syn he ic con ol egion.
Bloom e al. (2007) showed ha du ing pe iods o high unce ain y, such as a e
9/11, he esponsi eness o in es men by i ms o any gi en policy s imulus may be
much weake . Kaushal e al. (2007) disco e ed an associa ion be ween 9/11 and a 10
pe cen age poin s empo a y decline in he eal wage and weekly ea nings o A ab
and Muslim men. Simila ly, Elsayed and De G ip (2018) ound ha ollowing Islamic
e o is a acks ac oss Eu ope, Muslim immig an s’ a i udes owa d in eg a ion wo s-
ened, and ha hei (pe cei ed) in eg a ion ela ed o seg ega ion and unemploymen
also de e io a ed in he Ne he lands. Analyzing Is aelis’ eac ions o e o ism du -
ing he “Al-Aqsa” In i ada, Becke and Rubins ein (2011) ound ha occasional use s
educed hei demand o goods and se ices subjec o e o is a acks as a esponse o
ea s o u u e dange s. B odeu (2018) no ed ha success ul e o is a acks educed
jobs and o al ea nings in a ec ed egions ela i e o ailed a acks. The success ul
e o is a acks especially diminished housing p ices and aised consume pessimism
abou business and buying condi ions. Te o ism leads o a ealloca ion o economic
esou ces and an inc ease in coun e e o ism expendi u e. Fo ins ance, Di Tella and
Scha g odsky (2004) and D aca e al. (2011) bo h documen ed a no able inc ease
in police deploymen immedia ely a e e o is inciden s in a ge ed egions, and a
esul ing educ ion in local c ime a es. Acco ding o he calcula ion by Muelle and
S ewa (2014), domes ic coun e e o ism spending pe yea was 25 billion US dolla s
in 2010 be o e he 9/11 e o is a acks, whe e he expendi u e inc eased h ee- old
in he subsequen decade. In addi ion o coun e e o ism measu es, an incumben
may also y o accommoda e he c isis in a poli ical way by ex ending he anchise
(Acemoglu and Robinson 2000a,b,2001; Aid and F anck 2015; Besley and Pe s-
son 2011), o may a emp o acqui e suppo in an economic way by ealloca ing
esou ces o a ec ed egions (Ma ioli 2020;Zhu2018).3In he cu en s udy, I no
only in es iga e bo h channels o pe cei ed economic cap u e and pe cei ed poli ical
acquisi ion by he go e nmen pos -c isis, bu also analyze whe he hese s a egies
wo k. I documen no signi ican imp o emen s in con idence in he police o pe cei ed
sa e y, no signi ican pe cei ed ealloca ion o economic esou ces and no inc ease in
pe cei ed poli ical eedom in he a ge ed egions om he pe spec i e o ci izens.
Las bu no leas , he cu en s udy is closely ela ed o esea ch on changes in
poli ical us and ci izen coope a ion co esponding o changes in ins i u ional pe -
o mance. Some s udies in poli ical science, o example, Mishle and Rose (2001),
Ro hs ein and S olle (2008) and Zme li and New on (2008), a gued ha poli ical us
was endogenous o he pe o mance o poli ical ins i u ions; ha is, con idence in
poli ical ins i u ions was a consequence a he han a cause o hei pe o mance. This
3Poli ical connec ions wi h i ms u he acili a e such esou ce alloca ions (Faccio 2006; Fisman 2001;
Hodle and Raschky 2014; Shlei e and Vishny 1994).
123
Rally pos - e o ism Page 7 o 29 50
idea was modeled as “consensually s ong s a e equilib ium” by Acemoglu (2005).
Among empi ical s udies in economics, Aksoy e al. (2020) ound a pe sis en nega-
i e impac o epidemic exposu e in people’s “imp essionable yea s” on con idence in
poli ical leade ship and ins i u ions. Such esul s we e mainly d i en by indi iduals
in weak s a es incapable o ac ing agains epidemics, he eby disappoin ing ci izens.
Dell e al. (2015) documen ed ha in No he n Vie nam, whe e he e was a s onge
bu eauc a ic capaci y, ci izens we e mo e able o o ganize and coope a e wi h he local
go e nmen . Fu he mo e, using high-s akes lab-in- he- ield expe imen s in Pakis an,
Acemoglu e al. (2020) ound ha ecei ing in o ma ion on he imp o ed se ice
deli e y o s a e cou s inc eased ci izens’ willingness o in e ac wi h, and us in,
he cou s. People wi h upda ed in o ma ion also educed hei in e ac ions wi h non-
s a eac o s.Mys udy, ins ead,documen s inc eased con idence in poli ical ins i u ions
and poli ical app o al among ci izens in ad e sely a ec ed egions, who did no e en
pe cei e imp o emen s in local communi y basics ollowing e o is inciden s. This
was e en he case i a go e nmen ailed o p e en e o ism (i.e., an a ack was suc-
cess ul). Howe e , addi ional analyses in he cu en s udy also sugges ha he ally
e ec s a e no independen o a go e nmen ’s pe o mance agains e o ism: indeed,
an inc ease in poli ical us and suppo diminished i e o is ac s occu ed epea -
edly in a egion o e ime. Addi ionally, I p o ide sugges i e e idence o ein o ced
ci izens’ ci ic engagemen , as well as educed dissen om o icials a e e o is
inciden s.
2 Da a
In my analysis, I combine he Gallup Wo ldwide Resea ch su ey da a wi h he Global
Te o ism Da abase (GTD). The Gallup Wo ldwide Resea ch d aws a na ionally ep-
esen a i e andom sample o indi iduals annually om mo e han 150 coun ies
a ound he wo ld. E e y yea i ypically su eys a leas 1000 indi iduals in e e y
coun y, wi h ei he a1-h ace- o- acein e iewo a 30-min elephonein e iew.These
c oss sec ional da a include esponden s’ pe sonal cha ac e is ics, such as gende , age,
highes comple ed le el o educa ion, ma i al s a us, numbe o child en li ing in he
household, whe he he esponden was bo n in he su eyed coun y and employmen
s a us. In addi ion, he su ey asks a se o co e ques ions co e ing opics including, o
example, well-being, poli ical opinions and eligious opinions. Impo an ly, he da a
eco d he da e and he i s -o de subna ional adminis a i e (i.e., NUTS 1) egion o
when and whe e he indi idual in e iew was conduc ed.4I exploi his in o ma ion
la e o me ge he Gallup da a wi h he e o ism da a. Da a a e a ailable o he pe iod
2008–2016.
The Global Te o ism Da abase (GTD) documen s o e 190,000 in e na ional and
domes ic e o is a acks ha ook place a ound he wo ld om 1970 o 2018. The da a
de ine e o is a acks as “ he h ea ened o ac ual use o illegal o ce and iolence by
a non-s a e ac o o a ain a poli ical, economic, eligious, o social goal h ough ea ,
coe cion, o in imida ion.” To be included in he GTD, an inciden has o p esen h ee
4Please e e o Appendix C2 o mo e de ailed in o ma ion abou he Gallup Wo ldwide Resea ch Su ey.
123
50 Page 8 o 29 S. Chen
Table 1 Means o e o ism cha ac e is ics by a ack ype, weapon ype and logis ics
Success (%) Inju y Dea h No. obs (%)
A ack ype
Assassina ion 23.08 1.38 0.92 13 (2.41)
A med assaul 90.67 7.75 1.75 75 (13.89)
Bombing/explosion 45.91 1.54 0.19 257 (47.59)
Facul y/in as uc u e 94.58 0.18 0.01 166 (30.74)
Una med assaul 80.95 2.90 0.14 21 (3.89)
O he and unknown 100.00 1.00 0.88 8 (1.48)
Weapon ype
Fi ea ms 86.67 11.56 2.89 45 (8.33)
Explosi es/bombs/dynami e 45.45 1.50 0.19 264 (48.89)
Incendia y 93.14 0.32 0.01 175 (32.41)
Melee 93.02 2.02 0.33 43 (7.96)
O he and unknown 69.23 2.83 0.54 13 (2.41)
In e na ional logis ics 77.14 10.44 1.54 35 (6.48)
Domes ic logis ics 80.52 0.77 0.04 77 (14.26)
Unknown 65.89 1.63 0.34 428 (79.26)
To al 68.70 2.04 0.38 540 (100)
a ibu es: i mus be in en ional, i mus en ail some le el o iolence o immedia e
h ea o iolence, and he pe pe a o s o he inciden mus be subna ional ac o s. In
addi ion, he inciden mus mee a leas wo o he h ee ollowing c i e ia: i s , he
ac mus be aimed a a aining a poli ical, economic, eligious, o social goal; second,
he e mus be e idence o an in en ion o coe ce, in imida e, o con ey some o he
message o a la ge audience (o audiences) han he immedia e ic ims; and hi d, he
ac ion mus be ou side he con ex o legi ima e wa a e ac i i ies. In he cu en s udy,
I ollow he abo e de ini ion and ules.
Fo he as majo i y o e o is inciden s iled, he da a include in o ma ion abou
heda eandloca ion, he ypeo hea ack,whe he hea ackwassuccess ulo no , he
weapons used, he na u e o he a ge , he numbe o casual ies, and he iden i y o he
pe pe a o (s). The success o ailu e o e o is a acks is de ined based on he a ack
ype ha ook place, hus is dependen on he la e ac o .5Fo ins ance, a bombing
is ega ded as success ul i he explosi e de ice de ona es, and as ailed i i does
no de ona e. In o de o alle ia e conce ns abou he co ela ion be ween e o ism
esul s and di e ences in coun e e o ism capaci ies, I include coun y×yea ixed
e ec s and egion-speci ic ime ends, in addi ion o egion ixed e ec s, o accoun
o bo h ime- a ying and ime-in a ian geog aphical a ia ions in coun e e o ism
capaci ies.
Table 1displays he means o some impo an cha ac e is ics o he a acks in my
sample. Among he 540 e o is inciden s occu ing in EU coun ies om 2008 o
5See Appendix C o he ypes o e o is a acks and hei co esponding de ini ions o success in he
Global Te o ism Da abase.
123
Rally pos - e o ism Page 15 o 29 50
po en ial con ounde s a e epo ed in Table B2 o Appendix B. None o hem a e
signi ican e en ma ginally a he 10% le el, whe he o he e o ism occu ence o
o he e o ism success. The e o e, acco ding o he co ec ed indi idual coe icien
es s, bo h he e o ism occu ence and he e o ism success a e plausibly exogenous
o he poli ical and economic clima e p io o e o ism.
The es ima es sugges ha his la ge se o po en ial con ounde s is conside ably
balanced be ween indi iduals be o e hey we e con on ed wi h a e o is inciden and
indi iduals who we e no exposed o e o ism, as well as be ween indi iduals be o e
hey expe ienced a success ul e o is a ack and indi iduals be o e hey aced a ailed
one. The e o e, nei he he occu ence no he esul o e o ism can be p edic ed by
he pe cei ed poli ical and economic condi ions be o e he inciden . I exploi bo h
s a egies based on he occu ence and esul o e o is inciden s in he empi ical
analyses. The conclusions d awn om he es ima es wi h he wo app oaches a e in
acco d, hough he ally e ec s iden i ied based on e o is esul s appea mo e ansi-
o y. Acco ding o he abo e s a is ical es s, e o is occu ences seem mo e plausibly
exogenous han e o is esul s o he poli ico-economic clima e p io o a acks. Fu -
he mo e, Eq.1based on e o ism occu ences is pa simonious, and he coe icien
o success ul e o ism ela i e o ailed e o ism in Eq.2based on e o ism esul s
unde es ima es he e ec o e o ism which is o in e es ( epo ed la e in Table 2),
es ima es based on Eq.1a e p e e ed and hence ega ded as he baseline.
Table 2 E ec s o e o ism on con idence in go e nmen and app o al o coun y leade ’s pe o mance
(1) (2) (3) (4)
A. Con idence in
go e nmen
Pos -a ack 0.081 (0.032)** 0.081 (0.031)*** 0.110 (0.013)*** 0.045 (0.030)
Success×pos 0.065 (0.026)**
B. App o al o
leade ’s pe o -
mance
Pos -a ack 0.057 (0.031)* 0.057 (0.030)* 0.134 (0.021)*** 0.084 (0.037)**
Success×pos 0.051 (0.030)*
Ind. co a ia es Y Y Y
Te o . co a ia es Y Y
Region FE Y Y Y Y
Yea ×mon h FE Y Y Y Y
No e: Based on 61,718 obse a ions. Indi idual co a ia es con ain a male dummy a iable, he quad a ic
o age, dummy a iables o comple ed educa ional le els, dummy a iables o ma i al s a us, numbe
o child en li ing in he household, a dummy a iable o being bo n in he coun y su eyed and dummy
a iables o unemploymen and ou o he labo o ce. Te o ism co a ia es include a ack- ype ixed e ec s,
a dummy a iable o logis ically in e na ional a ack, a dummy a iable wi h he alue o 1 i he a ack is
unknown o be logis ically in e na ional o no and ano he dummy a iable o he majo a ack. A cons an
is also included in e e y model, bu no shown o pa simony. S anda d e o s clus e ed a he egional le el
in pa en heses; *p<0.1; **p<0.05; ***p<0.01
123

50 Page 16 o 29 S. Chen
Fig. 4 Con idence in go e nmen and app o al o coun y leade ’s pe o mance ac oss ime a ound he
e o ism occu ence. No e: Scale in he ho izon al axis is in qua e s. The segmen s deno e 90% (da k)
and 95% (ligh ) con idence in e als, espec i ely, o es ima ed coe icien s o in e ac ions o e o ism and
ime pe iods a ound he occu ence o e o ism ( he e ical dashed line)
4 Pa ame e es ima es
4.1 Pa allel p e- e o ism ends
To examine he ends o he ou come a iables o indi iduals be o e hey we e
exposed o e o ism e sus hose who did no expe ience such e en s, I add a se o
dummy a iables o one qua e , wo qua e s and h ee qua e s o 1 yea be o e an
a ack, espec i ely, in Eq.1. I he coe icien s o hese qua e ly dummy a iables
p e- e o ism a e no signi ican ly dis inguishable om ze o, i is e idence o pa allel
p e- ends. The co esponding es ima es a e isualized in Fig.4.16 The igu e shows
ha all es ima es p e-a acks a e no signi ican ly di e en om ze o a he 5% le el.
Mo eo e , he coe icien s o 3 mon hs and 6 mon hs a e a e o is inciden a e
sizable and signi ican o bo h ou come a iables, while es ima es 9 mon hs a e
a acks sh ink and al eady lose signi icance a he 5% le el. Hence, he esul s imply
ha he e ec s o e o ism oncon idence in poli icalins i u ions and poli icalapp o al
a e me ely sho e m.
Simila ly, Figs.4and 5display he coe icien es ima es in he same ype o e en
s udy, by sepa a ing success ul e o is a acks om ailed ones.17 Following he
occu ence o e o is inciden s ( he e ical dashed line), he ed ci cle es ima es
wi h hick segmen s deno e success ul a acks and he g een squa e es ima es wi h
hin segmen s e e o ailed ones. The ally phenomenon was e iden a e success ul
e o is a acks wi hin one qua e , and disappea ed by he second qua e . I ma e i-
alized mo e slowly a e ailed e o ism: he ally e ec g adually eme ged, and hen
became signi ican in he second qua e a e he inciden s, bu anished in he hi d
qua e as well. Hence, i espec i e o he a ack esul , e o ism igge ed a empo a y
ally phenomenon which aded away wi hin a yea , hough i ope a ed signi ican ly
o one qua e a a di e en ime depending on he esul .
16 The es ima es a e epo ed in Table B3 o Appendix B2.
17 Likewise, he ele an es ima es a e also epo ed in Table B4 o Appendix B2.
123
Rally pos - e o ism Page 17 o 29 50
Fig. 5 Con idence in go e nmen and app o al o coun y leade ’s pe o mance ac oss ime; sepa a ing
success ul e o ism om ailed a acks. No e: A e e o is occu ences ( he e ical dashed line), he ed
ci cle es ima es wi h hick segmen s e e o success ul e o ism and he g een squa e es ima es wi h hin
segmen s deno e ailed e o ism; see also no e unde Fig.4
4.2 E ec s o e o ism on poli ical con idence and app o al
Table 2p o ides an o e iew o he main pa ame e es ima es o he e ec s o e o -
ism on con idence in he na ional go e nmen and on poli ical app o al o he coun y
leade ’s pe o mance.18 The i s column includes egion ixed e ec s and ime pe iod
ixed e ec s, wi hou o he con ols. Column 2 adds indi idual cha ac e is ics, and
column 3 u he includes e o ism co a ia es. Column 4 displays he esul s o he
mo e sophis ica ed model— he e ec s o success ul e o is a acks in compa ison o
ailed ones.
Panel A epo s es ima es o con idence in he na ional go e nmen . In he Eu o-
pean Union a e e o is a acks, he p obabili y o ha ing con idence in he na ional
go e nmen signi ican ly inc eased by 8 o 11 pe cen age poin s ac oss di e en model
speci ica ions in columns 1 o 3. Despi e hei quali a i e simila i y, he di e gence in
he sizes o he e ec indica es he impo ance o he e o ism co a ia es. Wi hou
aking hem in o accoun , he poli ical e ec s could be neu alized by di e en classes
o a acks. In column 4, he coe icien in he second ow shows he addi i e e ec o
success ul e o ism o ha o ailed e o ism. Thus, compa ed wi h ailed a acks,
success ul e o is inciden s signi ican ly inc eased his p obabili y by 6.5 pe cen age
poin s; he occu ence o e o ism (also he ailed e en s compa ed wi h no e o ism
exposu e) inc eased he p obabili y by 4.5 pe cen age poin s, hough no signi ican ly.
I we sum up he wo coe icien s in column 4, we ob ain he e ec o success ul e -
o is a acks ela i e o no e o ism exposu e. The magni ude o he sum is he same
as he es ima e in column 3 since success ul a acks domina e no only he numbe bu
also he impac o e o ism (see he desc ip i es in Table 1). Impo an ly, I ind ha
he e ec s and hei sizes a e consis en be ween speci ica ions in columns 3 and 4.
Hence, bo h iden i ica ion s a egies in column 3 based on e o ism occu ences and
in column 4 based on hei esul s yield he same conclusions. Es ima es in columns 3
and 4 may be bo h in e p e ed as causal e ec s. As discussed in he p e ious sec ion,
18 Ex ended baseline es ima es a e epo ed in Appendix B3. I no speci ied o he wise, he analyses la e
a e all based on he same es ima ion sample as in Table 2.
123
50 Page 18 o 29 S. Chen
e o ism occu ences seem mo e plausibly exogenous han e o ism esul s o he
poli ico-economic clima e p io o e o is a acks. Addi ionally, he model speci ica-
ion in column 3 is pa simonious, and he coe icien o success ul e o ism ela i e o
ailed e o ism in column 4 unde es ima es he e ec o e o ism, i.e., he e ec el-
a i e o no e o ism exposu e, es ima es in column 3 a e p e e ed and hus ega ded
as he baseline.
Likewise, panel B p esen s he e ec s on app o al o he coun y leade ’s pe -
o mance. The occu ence o e o is a acks signi ican ly aised he p obabili y o
app o al o he coun y leade ’s pe o manceby6 o13.5 pe cen age poin s, depending
on he model speci ica ion. Success ul a acks signi ican ly inc eased he pe o mance
app o al by abou 5 pe cen age poin s ela i e o ailed ones (column 4).
The e ec s o e o ism occu ences in column 3 o bo h panels a e sizable, in
compa ison o a ound 40% a e o poli ical con idence and app o al p io o e o ism.
The ela i e e ec s o success ul a acks o ailed ones, displayed in column 4 o
bo h panels, a e also no i ial, compa ed wi h abou 50% a e o con idence and
45% a e o app o al in he case o ailed e o ism, espec i ely. Mo eo e , he ally
e ec s o success ul a acks in compa ison o ailed a acks, ha is, he coe icien s o
Success ×pos , a e smalle and less s a is ically signi ican han hei coun e pa s
ollowing (gene al/any) e o is inciden s, namely he coe icien s in column 3. This
makessensein ha ailed e o ismd o ea(milde ) allyphenomenon( hecoe icien s
o Pos -a ack in column 4). A he end o Sec ion 3, I ha e p esen ed empi ical
e idence ha bo h he occu ence o e o ism and he success o a acks may be
plausibly exogenous. In his sec ion, I show ha bo h app oaches based on e o ism
occu encesand esul sdeli e consis en conclusions,e en hough he allye ec s a e
mo e ansi o y when dis inguishing success ul a acks om ailed ones (see Fig.5).
4.3 Po en ial mechanisms
In hissubsec ion,Iexplo epo en ialmechanisms h oughwhich e o is a ackscould
ha e a ec ed con idence in na ional poli ical ins i u ions and poli ical app o al o he
coun y’s leade . I p o ide sugges i e e idence ha e o ism could exe in luence by
igge ing s onge pa io ism and ci ic engagemen among ci izens. Howe e , I do no
ind e idence o he al e na i e channels o pe cei ed economic cap u e o poli ical
acquisi ion by he go e nmen .19
4.3.1 Pa io ism
Pa io ismisonemechanism ha mayd i e he iseinpoli icalcon idenceandapp o al
pos - e o ism. This p oposed mechanism is inspi ed by he li e a u e dealing wi h
he ally ’ ound he lag e ec , in which he Pa io ism School a gues ha c isis-
induced na ionalis sen imen ansla es in o poli ical suppo o he leade ship a e
(in e na ional) c ises (B ody and Shapi o 1989).
19 Mo e discussion on al e na i e explana ions o he mechanisms explo ed in his sec ion can be ound
in Appendix B4.
123
Rally pos - e o ism Page 19 o 29 50
Table 3 Mechanisms o pa io ism, ci ic engagemen , economic cap u e and poli ical acquisi ion
(1) (2) (3) (4)
A. Pa io ism
Mo e ab oad Mo e wi hin yea
Pos -a ack −0.087 (0.012)*** −0.074 (0.019)***
No. obs. 61,718 7333
B. Ci ic engage-
men
Dona ion Volun ee Help s ange s Voice opinion
Pos -a ack 0.031 (0.016)* 0.039 (0.034) 0.064 (0.030)** −0.041 (0.015)***
No. obs. 59,078 43,650
C. Economic cap-
u e
Index Con . police Feel. sa e
Pos -a ack 0.017 (0.012) 0.022 (0.015) 0.004 (0.015)
No. obs. 59,078
D. Poli ical acqui-
si ion
Poli. iew ee. Media ee.
Pos -a ack −0.189 (0.011)*** −0.086 (0.016)***
No. obs. 36,451 57,565
No e: Indi idual co a ia es con ain a male dummy a iable, he quad a ic o age, dummy a iables o
comple ed educa ional le els, dummy a iables o ma i al s a us, numbe o child en li ing in he household,
a dummy a iable o being bo n in he coun y su eyed and dummy a iables o unemploymen and ou
o he labo o ce. Te o ism co a ia es include a ack- ype ixed e ec s, a dummy a iable o logis ically
in e na ional a ack, a dummy a iable wi h he alue o 1 i he a ack is unknown o be logis ically
in e na ional o no and ano he dummy a iable o he majo a ack. A cons an is also included in e e y
model, bu no shown o pa simony. S anda d e o s clus e ed a he egional le el in pa en heses; *p<
0.1; **p<0.05; ***p<0.01
In panel A o Table 3, I use unwillingness o mo e pe manen ly ab oad ideally
(a e a e o is a ack) as one p oxy o pa io ism.20 D ehe e al. (2011) showed ha
e o is inciden s boos ed skilled emig a ion due o de e io a ion in local li ing and
wo king condi ions. I , e en in an ideal si ua ion, indi iduals would be less likely o
mo e ab oad a e a e o is a ack, i may e lec a s ong na ionalis ically a ached
sen imen . A e a acks, he p obabili y o being willing o pe manen ly mo e ab oad
in an ideal si ua ion was signi ican ly educed by 8.7 pe cen age poin s. Among he
po en ial emig an s, he likelihood o planning o lea e hei coun y wi hin he ollow-
ing yea signi ican ly dec eased by 7.4 pe cen age poin s pos -a acks. These esul s
a e consis en wi h he pa io ism mechanism: e o is inciden s happening in a pe -
son’s egion o esidence imp o ed hei coun y a achmen and hence dec eased hei
willingness o mo e ab oad and o mo e soon. La e , as a he e ogenei y analysis, I
show ha poli ical con idence and app o al imp o ed wo- old a e a acks launched
20 The able no e means ha he same sample and he same co a ia es as o he baseline model speci ica ion
in column 3 o Table 2a e used o es ima ion.
123
50 Page 20 o 29 S. Chen
by non-domes ic e o is s compa ed wi h a e a acks launched by domes ic pe pe-
a o s. This di e ence is in acco d wi h he idea o he na ionalis a achmen .
4.3.2 Ci ic engagemen
Nex , I s udy he po en ial channel o ci ic engagemen . A e he occu ence o e -
o is inciden s in hei egion, esiden s may be mo i a ed o pa icipa e in ebuilding
and enhancing hei own local communi y. Needing suppo om he go e nmen
and/o coope a ing wi h he go e nmen in econs uc ion ac i i ies, indi iduals may
ha e g ea e poli ical con idence and app o al. I examine di e en dimensions o ci ic
engagemen : cha i y dona ion, olun ee pa icipa ion in an o ganiza ion, p o ision o
help o a s ange and opinion oiced o a public o icial.
Panel B o Table 3 epo s he esul s. Pos - e o ism, he p obabili ies o cha i y
dona ion and helping s ange s signi ican ly ose by a ound 3 pe cen age poin s and
6.5 pe cen age poin s, espec i ely. Howe e , he likelihood o people oicing hei
opinion o a public o icial was signi ican ly educed by abou 4 pe cen age poin s.
This las esul makes sense, in ha a e a e o is a ack, ci izens engage in ci ic
e en s and coope a e wi h ins i u ions, and hence a e less likely o oice dissen ing
opinions o public o icials.21 The es ima es in panel B p o ide e idence o enhanced
ci ic engagemen among esiden s in ad e sely a ec ed egions, and hus suppo he
channel o ci ic engagemen .
4.3.3 Economic cap u e
A e a e o is inciden occu s, in addi ion o immedia e esponses such as escuing
ic ims and hun ing o and a es ing pe pe a o s, he go e nmen may ake s a egic
measu es opaci yci ilians,especiallyin heimpac ed egion,ino de op ese e hei
con idence and suppo . By alloca ing economic esou ces o he a ge ed egion and
imp o ing (o a leas main aining he p e- e o is inciden ) pe cei ed local economic
clima e, he go e nmen will gain app o al om ci ilians a ec ed by e o ism. I e m
his s a egy “economic cap u e.” The economic cap u e in his case goes beyond he
necessa y measu es o c isis aid o managemen , such as econs uc ion o he acili ies
damaged in an a ack.
InpanelCo Table3,Ip esen hees ima es o hepossiblemechanismo pe cei ed
economic cap u e by he go e nmen . The dependen a iable in column 1 is an eco-
nomiccap u eindexo a eabasics.Thisindexiscons uc edbasedondummy a iables
o sa is ac ion wi h a se ies o elemen s in he esponden s’ a ea o esidence. These
elemen s, o “basics,” include in as uc u e, such as oads and highways, he pub-
lic anspo a ion sys em, housing a ailabili y, he educa ional sys em o schools and
heal hca e a ailabili y.22 I he go e nmen alloca es economic esou ces o a ge ed
21 Ano he piece o e idence o lesspoli ical dissen ollowing e o ismisillus a ed inFig.4:immedia ely
a e a acks (i.e., in he i s qua e pos - e o ism), no only did he a e age p obabili ies o poli ical
con idence and app o al imp o e, bu hei s anda d e o s and hence a iances also sh ank. This sugges s
ha people became mo e uni ed wi h poli ical ins i u ions.
22 De ails o speci ic ques ions abou hese elemen s and how he index is c ea ed can be ound in Appendix
C.
123

Rally pos - e o ism Page 21 o 29 50
egions, hese basics may imp o e. Lacking an objec i e measu emen o imp o e-
men o hese elemen s, he index cons uc ed o sa is ac ion wi h basics p o ides a
pe cei ed composi e al e na i e.
Column 1 in panel C displays he e ec s o e o ism on he economic cap u e
index. The e was a nonsigni ican inc ease in he index o 0.017 (in a ange o 0–1)
a e a e o is inciden . Tha is o say, on a e age, ci ilians’ sa is ac ion wi h he
o e all a ea basics imp o ed by a ound 1.7 pe cen age poin s a e hei egion o
esidence was hi by an a ack; howe e , he imp o emen was no signi ican . Thus,
he pe cei ed economic cap u e mechanism does no seem impo an in explaining
he e ec s o e o ism on con idence in he go e nmen and app o al o he coun y
leade ’s pe o mance.
Thego e nmen may edeploymo epoliceo ice s oimp o ep o ec ionin a ge ed
egions a e he occu ence o e o ism (Di Tella and Scha g odsky 2004; D aca e al.
2011), and his could be ega ded as pa o a ealloca ion o economic esou ces.
Hence, in columns 2 and 3 o panel C, I examine he in luence o e o ism on he
dummy a iables o ha ing con idence in he police and eeling sa e while walking
alone a nigh . Howe e , I ind nei he signi ican no sizable e ec s on con idence in
he police o pe cei ed sa e y.
In his subsec ion, I do no a emp o a gue ha a e o ism-besieged go e nmen
hasno ied oeconomicallycap u epoli icalsuppo o imp o e hesecu i ymeasu es
in a ge ed a eas ollowing e o ism. Ins ead, I documen ha ci izens in impac ed
egions did no pe cei e signi ican imp o emen s in hese communi y basics. Consid-
e ing he subjec i e and pe cei ed cha ac e is ics o poli ical con idence and app o al,
I do no ind e idence ha he inc eased le els o poli ical us and suppo pos -
e o ism can be a ibu ed o he pe cei ed ealloca ion o economic esou ces o he
ele an a eas.
4.3.4 Poli ical acquisi ion
Along wi h economic esou ce alloca ion, a go e nmen may ex end he anchise o
acqui e local poli ical suppo a e e o is a acks. In au ho i a ian egimes, democ-
a iza ion can become a concessiona y op ion unde he h ea o widesp ead un es
(Acemoglu and Robinson 2000b; Aid and F anck 2015). Ne e heless, he oom o
ex ending he anchise is limi ed in ad anced democ acies such as he EU. Thus, I
examine he changes in he (pe cei ed) eedom o exp essing poli ical iews and he
(pe cei ed) eedom o he media, a he han changes ela ed o o ing igh s.
In he las panel, he nega i e coe icien es ima es o Pos -a ack a e signi ican
o bo h (pe cei ed) eedom indica o s. I he go e nmen ’s s a egy o anchise
ex ension wo ked, we would see a signi ican ise in hese measu es o poli ical ee-
dom pos - e o ism. Hence, poli ical acquisi ion does no seem o be a mechanism.23
The signi ican nega i e impac s o e o ism occu ences on pe cei ed eedom could
23 One po en ial conce n is ha i implemen ed, he anchise would be ex ended a he na ional le el
a he han a he egional le el. Thus, I eplica e his analysis a he coun y le el and ob ain a ma ginally
signi ican nega i e e ec (−0.048) on he eedom o exp essing poli ical iews, and a nonsigni ican e ec
(0.026) on he eedom o he media.
123
50 Page 22 o 29 S. Chen
be in e p e ed as ei he he es ic ions imposed by he au ho i ies (especially in less-
libe al EU membe s a es) o he ea s o u u e e enge and u he a acks in he
a ec ed egions. Nei he o he explana ions alida es he channel o poli ical acqui-
si ion.
The e o e, he explo a ion o po en ial mechanisms sugges s ha pa io ism and
ci icengagemen o ci iliansin a ge ed egionswe elikely obuildlinks om e o is
inciden s o poli ical con idence and app o al. None heless, I do no ind e idence
ha he inc ease in poli ical us and suppo a e e o ism was due o he pe cei ed
alloca ion o economic esou ces o pe cei ed poli ical acquisi ion by he go e nmen .
4.4 He e ogenei y in es iga ion
In his subsec ion, I examine he e ogeneous e ec s in e ms o a ious indi idual cha -
ac e is ics such as gende , employmen s a us, educa ional a ainmen and age coho
and e o ism cha ac e is ics including logis ics and geog aphy/ins i u ion wi hin he
EU. The aim is o explo e which speci ic subg oups o people and coun ies and which
classes o a acks d o e his pos - e o ism ally e ec .
Fo ease o compa ison, in panel A o Table 4, I ep oduce he baseline es ima es
om column 3 o Table 2. Fi s , I explo e whe he he e a e di e en ials be ween
men and women. Pe in and Smolek (2009) documen ed a s onge ally phenomenon
among men han among women; howe e , my es ima es in panel B do no acco d wi h
hei indings. Consis en ly, inc eases in bo h poli ical con idence and app o al a e
e o is inciden s we e signi ican o bo h gende s and o a sligh ly la ge magni ude
among women han men. In he las ow o he panel, I es whe he he gende di e -
ence is signi ican o no . The co esponding p- alues indica e a nonsigni ican gap
be ween men and women in he ally e ec .24
Second, e o ism ad e sely a ec s business condi ions and leads o job losses
(B odeu 2018). People laid o a e a e o is inciden may blame he go e nmen
o a lack o capaci y o deal wi h he c isis, and hence hei own unemploymen .
None heless, hey may also be ang y wi h e o is s because o hei own job loss, and
hus suppo he go e nmen ’s coun e e o ism measu es. The esul s in panel C seem
o suppo he second hypo hesis. Bo h he employed and he non-employed p esen
signi ican ally synd ome a e e o ism. Among he non-employed, imp o emen s
in poli ical con idence and app o al we e mo e sizable han hose o he employed.
Howe e , such a di e gence in magni ude is onlysigni ican o app o alo hecoun y
leade ’s pe o mance, as shown by he p- alue o he di e ence es .
Thi d, in panel D o Table 4, I examine he di e ences be ween college deg ee
holde s and indi iduals wi hou such a deg ee. Su p isingly, he e ec s o e o ism on
poli ical con idence and app o al o college deg ee holde s a e signi ican ly nega i e
24 He e ogeneous e ec s in e ms o gende a e signi ican o nei he poli ical con idence no poli ical
app o al, and he e ogeneous e ec s ega ding labo ma ke s a us and age coho a e only pa ially signi i-
can . This migh no necessa ily exclude he possibili y o he e ogenei ies conce ning hese cha ac e is ics.
S a is ical non-signi icance o hese he e ogenei ies could also be due o ewe obse a ions wi h e o ism
exposu e a e sample di isions and hence dec eased a ia ions ha can be le e aged o compa e es ima ed
e ec s be ween sub-samples. Un o una ely, i seems in easible o e i y his possibili y by inc easing he
p opo ion o obse a ions wi h e o ism exposu e.
123
Rally pos - e o ism Page 23 o 29 50
Table 4 E ec s o e o ism on poli ical con idence and app o al–he e ogenei y in es iga ion
(1) Con idence go . (2) App o al leade No. obs
A. Baseline 0.110 (0.013)*** 0.134 (0.021)*** 61,718
B. Gende
Men (βm) 0.108 (0.024)*** 0.140 (0.025)*** 26,801
Women (βw) 0.125 (0.022)*** 0.155 (0.022)*** 34,917
p alue (βm=βw) 0.677 0.438
C. Labo ma ke s a us
Employed (βep) 0.106 (0.023)*** 0.107 (0.017)*** 33,635
Non-employed (βnep) 0.125 (0.025)*** 0.211 (0.036)*** 28,083
p alue (βep =βnep) 0.645 0.000***
D. Educa ional a ainmen
College deg ee (βcg) -0.247 (0.035)*** -0.300 (0.027)*** 15,930
No college deg ee (βncg) 0.135 (0.014)*** 0.165 (0.022)*** 45,788
p alue (βcg =βncg) 0.000*** 0.000***
E. Age coho
50 & younge (β50−) 0.082 (0.022)*** 0.123 (0.018)*** 32,636
Olde han 50 (β50+) 0.270 (0.052)*** 0.187 (0.054)*** 29,082
p alue (β50−=β50+) 0.006*** 0.160
F. Logis ics
Domes ic (βdm) 0.111 (0.013)*** 0.135 (0.021)*** 60,742
Non-domes ic (βndm) 0.265 (0.039)*** 0.246 (0.030)*** 61,563
p alue (βdm =βndm) 0.000*** 0.002***
G. Geog aphy/ins i u ion
Cen al and Eas e n Eu ope (βceu) 0.184 (0.022)*** 0.072 (0.039)* 19,782
Wes e n Eu ope (βweu) 0.114 (0.012)*** 0.138 (0.022)*** 41,936
p alue (βceu =βweu) 0.030** 0.016**
No e: Only he coe icien s o Pos -a ack a e p esen ed. Indi idual co a ia es con ain a male dummy
a iable, he quad a ic o age, dummy a iables o comple ed educa ional le els, dummy a iables o
ma i al s a us, numbe o child en li ing in he household, a dummy a iable o being bo n in he coun y
su eyed and dummy a iables o unemploymen and ou o he labo o ce. Te o ism co a ia es include
a ack- ype ixed e ec s, a dummy a iable o logis ically in e na ional a ack, a dummy a iable wi h he
alue o 1 i he a ack is unknown o be logis ically in e na ional o no and ano he dummy a iable o
he majo a ack. A cons an is also included in e e y model, bu no shown o pa simony. S anda d e o s
clus e ed a he egional le el in pa en heses;
*p<0.1; **p<0.05; ***p<0.01
and e y la ge. Appa en ly, he ally phenomenon a e e o ism was d i en by lowe
educa ed people wi hou a college deg ee while such poli ical us and suppo among
highe educa ed indi iduals declined subs an ially. The di e ence es s in he las ow
o panel D con i m such a signi ican di e gence.
Fu he mo e, I sepa a e he sample in o wo age g oups: one g oup o 50 yea s old
and below, and he o he olde han 50. Fo bo h ou come a iables, he coe icien s
a e signi ican o he wo g oups. Howe e , he e o ism-induced ally synd ome was
123
50 Page 24 o 29 S. Chen
mo e sizable among he olde coho han he younge one. The p- alues show ha
such age coho di e en ials a e signi ican o con idence in he na ional go e nmen ,
especially.
In addi ion o indi idual cha ac e is ics, I also examine he e ogenei y ega ding
e o ism logis ics. In panel F, I dis inguish a acks wi h domes ic logis ics om hose
wi h in e na ional o unknown logis ics. The coe icien s a e signi ican o bo h ypes
o a acks, bu oughly wice as la ge o non-domes ic e o ism as o domes ic e -
o ism. Hence, esiden s in a ec ed egions seemed o exhibi a conside ably s onge
ally synd ome a e a acks ha we e launched by non-domes ic pe pe a o s. This
subs an ial gap is u he con i med by he es conduc ed in he las ow. Such a di -
e ence is consis en wi h he pa io ism mechanism: people inc eased hei suppo
o he go e nmen and app o al o hei coun y leade ’s pe o mance, pa icula ly
when a acked by o eign e o is g oups.
Finally, I in es iga e di e en ials be ween EU membe s a es in Wes e n Eu ope,
and hose in Cen al and Eas e n Eu ope, as he la e a e adi ionally pe cei ed as less
libe al o less ma u ely democ a ic.25 The es ima es in panel G show ha con idence in
he na ional go e nmen imp o ed ollowing e o is a acks mo e no ably in Cen al
and Eas e n Eu ope, while app o al o he coun y leade ’s pe o mance in Wes e n
Eu ope inc eased almos wice as much as in Cen al and Eas e n Eu ope. The es s in
he las ow con i m such signi ican geog aphic and ins i u ional gaps. These esul s
imply ha he ally e ec s seemed o be exhibi ed mo e no ably in he o m o poli i-
cal ins i u ions— he na ional go e nmen in he cu en con ex — han he indi idual
leade ship in Cen al and Eas e n Eu ope. Ne e heless, he ally phenomenon was
signi ican ly p esen ega ding bo h poli ical ins i u ions and indi idual leade ship in
Wes e n Eu ope.
In his subsec ion, I s udy he e ogeneous ally e ec s in di e en subg oups and
ypes o e o ism. I ind ha a e a acks, poli ical con idence and app o al inc eased
mo e subs an ially among he non-employed, indi iduals wi hou a college deg ee
and olde people. Howe e , poli ical con idence and app o al educed signi ican ly,
bo h economically and s a is ically, among college deg ee holde s. Mo eo e , a acks
launchedby domes ic pe pe a o s hadonlyhal heex en o poli ical e ec sasa acks
launched by non-domes ic e o is s. This may e lec a na ionalis ically a ached sen-
imen ; a mechanism o he ally synd ome. The ally phenomenon was exhibi ed
signi ican ly in e ms o bo h poli ical ins i u ions and indi idual leade ship in Wes -
e n Eu ope, while in Cen al and Eas e n Eu ope, i was p esen ed mo e s ongly in
he o m o poli ical ins i u ions.
4.5 Sensi i i y analyses
To check he obus ness o my es ima es, I conduc a ange o sensi i i y analyses. I
ex end he pos - e o ism window pe iod o explo e he pe sis ence o he ally e ec .
I also include coun y×yea ixed e ec s and egion-speci ic ime ends.
25 Acco ding o he de ini ion om he OECD, Cen al and Eas e n Eu opean membe s a es o he EU
include Bulga ia, C oa ia, he Czech Republic, Hunga y, Poland, Romania, he Slo ak Republic, Slo enia
and he h ee Bal ic S a es: Es onia, La ia and Li huania.
123