Corruption and political accountability in good and bad economic times

Ba inas-Fo e o, And és; Sca ascini, Ca los G.
Wo king Pape
Co up ion and poli ical accoun abili y in good and bad
economic imes
IDB Wo king Pape Se ies, No. IDB-WP-1690
P o ided in Coope a ion wi h:
In e -Ame ican De elopmen Bank (IDB), Washing on, DC
Sugges ed Ci a ion: Ba inas-Fo e o, And és; Sca ascini, Ca los G. (2025) : Co up ion and poli ical
accoun abili y in good and bad economic imes, IDB Wo king Pape Se ies, No. IDB-WP-1690, In e -
Ame ican De elopmen Bank (IDB), Washing on, DC,
h ps://doi.o g/10.18235/0013597
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/324837
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 ps://c ea i ecommons.o g/licenses/by/3.0/igo/
Co up ion and Poli ical
A
ccoun abili y
in Good and Bad Economic Times
A
nd és Ba inas-Fo e o
Ca los Sca ascini
WORKING PAPER No IDB-WP-1690
In e -
A
me ican De elopmen Bank
Depa men o Resea ch and Chie Economis
July 2025
* In e -
A
me ican De elopmen Bank
Co up ion and Poli ical Accoun abili y
in Good and Bad Economic Times
A
nd és Ba inas-Fo e o*
Ca los Sca ascini*
In e -
A
me ican De elopmen Bank
Depa men o Resea ch and Chie Economis
July 2025
Ca aloging-in-Publica ion da a p o ided by he
In e -Ame ican De elopmen Bank
Felipe He e a Lib a y
Ba inas-Fo e o, And es.
Co up ion and poli ical accoun abili y in good and bad economic imes /
And es Ba inas-Fo e o, Ca los Sca ascini.
p. cm. — (IDB Wo king Pape Se ies ; 1690)
Includes bibliog aphical e e ences.
1. Poli ical co up ion-Ma hema ical models. 2. Public goods-Co up
p ac ices-Ma hema ical models. I. Sca ascini, Ca los G., 19 71- II. In e -
Ame ican De elopmen Bank. Depa men o Resea ch and Chie Economis .
III. Ti le. IV. Se ies.
IDB-WP-1690
h p://www.iadb.o g
Copy igh © 2025 In e -Ame ican De elopmen Bank ("IDB"). This wo k is subjec o a C ea i e
Commons license CC BY 3.0 IGO (h ps://c ea i ecommons.o g/licenses/by/3.0/igo/legalcode). The
e ms and condi ions indica ed in he URL link mus be me and he espec i e ecogni ion mus be
g an ed o he IDB.
Fu he o sec ion 8 o he abo e license, any media ion ela ing o dispu es a ising unde such license
shall be conduc ed in acco dance wi h he WIPO Media ion Rules. Any dispu e ela ed o he use o
he wo ks o he IDB ha canno be se led amicably shall be submi ed o a bi a ion pu suan o he
Uni ed Na ions Commission on In e na ional T ade Law (UNCITRAL) ules. The use o he IDB's name
o any pu pose o he han o a ibu ion, and he use o IDB's logo shall be subjec o a sepa a e
w i en license ag eemen be ween he IDB and he use and is no au ho ized as pa o his license.
No e ha he URL link includes e ms and condi ions ha a e an in eg al pa o his license.
The opinions exp essed in his wo k a e hose o he au ho s and do no necessa ily e lec he iews o
he In e -Ame ican De elopmen Bank, i s Boa d o Di ec o s, o he coun ies hey ep esen .
Abs ac
While he li e a u e ex ensi ely explo es he s uc u al enable s o co up ion and i s
ad e se e ec s on economic pe o mance, less is known abou how he s a e o he
economy in luences co up ion and poli ical accoun abili y. To add ess his gap, we
de elop a heo e ical model in which poli icians may di e esou ces om public
goods and ci izens can espond by punishing co up ion. In ou model, pe iods o
posi i e economic condi ions inc ease co up ion while weakening accoun abili y. We
alida e hese p edic ions h ough a labo a o y expe imen , inding ha co up ion
a es signi ican ly ise when economic condi ions a e good. Howe e , ci izens’ willing-
ness o punish co up poli icians emains s able ac oss he business cycle. Punishmen
decisions a e d i en by obse ed public good alloca ions; low alloca ions p omp
signi ican ly highe punishmen a es han high alloca ions, e en esul ing in he pun-
ishmen o hones poli icians du ing bad economic imes. Addi ionally, we assess he
ole o co up ion expec a ions in shaping esponses: ci izens wi h p io belie s ha
poli icians a e co up a e less likely o punish han hose who belie e poli icians a e
hones when public good p o ision is low. Accoun abili y becomes mo e challenging
when ci izens s uggle o clea ly iden i y co up ion, and ci izens a e mo e o gi ing
o co up ion du ing good economic imes and i hey al eady mis us poli icians.
These indings unde sco e he impo ance o obus anspa ency and accoun abili y
mechanisms in upholding go e nance s anda ds, pa icula ly in he ace o economic
luc ua ions and public mis us .
JEL classi ica ions: D72, D73, H41, C91
Keywo ds: Ren seeking, Economic Booms, Co up ion, Punishmen , Labo a o y ex-
pe imen , Down u ns
The au ho s hank pa icipan s a REPAL 2024, LACEA-LAMES 2024, MPSA 2025, he Uni e si y o Vi ginia semina , Ma ín
A danaz, He nán Beja ano, Ma iana Blanco, E nes o Cal o, Dan Ginge ich, Phil Kee e , And ea Lopez-Luzu iaga, Julia Sei he , and
an anonymous e iewe o hei commen s and sugges ions. The esea ch desc ibed in his a icle has Uni e sidad del Rosa io IRB
App o al CS474 (Ma ch 15, 2024) and was p e egis e ed a he OSF Regis ies (
www.os .io/ m 92
, May 8, 2024). This esea ch was
inanced by he In e -Ame ican De elopmen Bank (IDB). IDB managemen had no in ol emen in he s udy design, analysis, o
in e p e a ion o he da a, in he w i ing o he epo , o in he decision o submi he a icle o publica ion. The opinions exp essed
he e a e hose o he au ho s and do no necessa ily e lec he iews o he In e -Ame ican De elopmen Bank, i s Boa d o Di ec o s,
o he coun ies hey ep esen . The au ho s decla e ha hey ha e no known compe ing inancial 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 .

1 In oduc ion
Poli ical co up ion dis up s g ow h (Mau o,1995;Ba dhan,1997), deepens po e y (Gup a
e al.,1998), exace ba es inequali y (Olken,2006), e odes us be ween ci izens and he s a e
(Kee e and Sca ascini,2022;Khli and Ama a,2019), and leads o signi ican economic
ine iciencies (Dal Bó and Rossi,2007;Fisman and S ensson,2007). Economic and poli ical
cycles can ampli y o mi iga e hese ha m ul e ec s (Saha and Sen,2023) and may also
in luence he likelihood o co up ion i sel .
Du ing an economic bonanza, inc eased esou ces may p o ide mo e oppo uni ies o
co up p ac ices (Galb ai h,1997;Lei e and Weidmann,1999). Con e sely, du ing imes
o economic ha dship, co up ion isks can in ensi y as he alue o en s g ows, and
public o icials may exploi he popula ion’s inc eased ulne abili y (I le s and Hinks,
2015;To sello,2010). Economic cycles also shape pe cep ions o co up ion, as ci izens
may a ibu e down u ns o poli ical mismanagemen o become mo e a uned o he issue
du ing pe iods o ha dship (Má quez Romo and Rome o-Vidal,2023;Zechmeis e and
Zizumbo-Colunga,2013).
Fo ins ance, da a om a na ionally ep esen a i e su ey conduc ed ac oss 26 La in
Ame ican and Ca ibbean coun ies (LAPOP,2024), om 2016 o 2023, e eal a no able
end: esponden s’ pe cep ions o poli ical co up ion a e in e sely ela ed o hei iews
on economic condi ions o e he pas yea (Figu e 1.)
1
This co ela ion sugges s ha he
s a e o he economy could a ec suspicions o co up ion o ac ual co up ac i i ies.
I co up ion belie s a e in luenced by he cycle, he e may be inc eased incen i es o
co up ion, as poli ical accoun abili y dec eases.
In his pape , we examine how economic good and bad economic imes impac he beha io
o poli icians when i comes o local co up ion and ci izens when i comes o poli ical
accoun abili y.
2
To analyze hese dynamics, we de elop a heo e ical model o poli ical
co up ion and poli ical accoun abili y, whe e ci izens pay axes and can punish poli icians,
and poli icians p o ide public goods (o pocke pa o he e enues). We ope a ionalize
he economic cycle h ough an exogenous economic shock ha modi ies he cos o public
1
This pa e n is consis en ac oss all coun ies in he egion and ac oss suppo o he cu en p esiden , as
de ailed in he Online Appendix (see Figu es D1,D2 and D3).
2
In ou model, poli icians who ac as “ en -seeke s” a he han "o ice-seeke s" a e he ones who ame local
co up ion (see Pe sson and Tabellini,2002). Ou concep o accoun abili y di e ges om he adi ional
unde s anding ound in seminal poli ical accoun abili y li e a u e (see Ba o,1973;P zewo ski e al.,1999).
Ins ead, we cha ac e ize accoun abili y as ci izen’s willingness o punish poli icians while hey a e in o ice
(Klašnja e al.,2021).
1
goods cons uc ion, a ec ing poli icians’ incen i es, he quan i y o public goods p o ided,
and ci izens’ subsequen choices o hold poli icians accoun able.
Figu e 1: Pe cep ion o he Numbe o Poli icians In ol ed in Co up ion ac oss Na ional
Economic Si ua ion Pe cep ion
β
= -.25***
p < .00001
3.2
3.4
3.6
3.8
4
Numbe o poli icians in ol ed in co up ion (1 None - 5 All)
Wo se han
be o e Same as
be o e
Be e han
be o e
Economic si ua ion o he coun y in las 12 mon hs (1 Wo se - 3 Be e )
Linea i 95% CIs
No es: The igu e depic s he sca e plo binned be ween he pe cep ion o he numbe o poli icians in ol ed in co up ion and he
pe cep ion o he coun y’s economic si ua ion o e he pas 12 mon hs. Sca e binned es ima ion o all esponden s o he LAPOP
ques ion in 26 coun ies be ween 2016 and 2023. The da k line indica es he linea i om a eg ession model wi h
n= 79,787
. The
g ay a ea ep esen s he 95% clus e ed con idence in e als a he coun y le el. The analysis includes coun y and yea ixed e ec s.
The eg ession analysis is p esen ed in he Online Appendix D1. S a is ical signi icance le els:
∗∗∗ p < 0.01,∗∗ p < 0.05,∗p < 0.1
, n.s
p > 0.1.
We model economic condi ions as changes in he cos o public goods o wo main
easons. Fi s , i allows us o in oduce andom shocks o ci izens’ ma e ial well-being and
obse e how hese luc ua ions shape bo h co up ion and poli ical accoun abili y. Since
ci izens de i e u ili y om public goods (Pe sson and Tabellini,2002), inc eases in cos
educe p o ision and consequen ly lowe ci izens’ wel a e, p omp ing hem o po en ially
punish incumben s e en in he absence o co up ion (Ben on,2005;Kayse and Pe ess,
2012;Lewis-Beck and S egmaie ,2000).
3
Second, in many de eloping coun ies, public
inances usually employed in he cons uc ion o public goods a e closely ied o ola ile
commodi y e enues (Ma ínez,2023). As such, andom luc ua ions in p oduc ion cos s
p o ide an ex e nally alid analogue o eal-wo ld economic shocks ha a e independen
3
We de ia e om o he s udies ha employ inc emen s in p oduc i i y as he main sou ce o he posi i e
economic shock (see Bai e al.,2017).
2
o incumben s’ decisions and ac ions.4
In ou model, a local poli ician, a na ional au ho i y, and a ci izen in e ac s a egically. In
he i s s age o he game, an economic shock occu s ( he cons uc ion cos o he public
goods is de ined by na u e), which is common knowledge o bo h he ci izen and he
poli ician. The ep esen a i e ci izen pays income axes ha a e ans e ed o he poli ician.
The poli ician ecei es he ans e and decides how much o in es in public goods.
5
A e
he poli ician’s alloca ion decision, he na ional au ho i y may p o ide addi ional unds
o public goods cons uc ion (an exogenous ac ion). In he inal s age, he ci izen obse es
he ac ual alloca ion o public goods, conside ing he economic condi ions, and decides
whe he o punish (o disapp o e) he poli ician–—a choice ha incu s a cos .
In his model, ci izens and poli icians can exhibi di e en in insic mo i a ions. Poli icians
may a y in hei mo al commi men o in eg i y (Ma inelli,2022;Tanne e al.,2022), and
ci izens may di e in hei mo i a ion o hold poli icians accoun able (Adse à e al.,2003;
Leigh e al.,2020;Nannicini e al.,2013).
The p esence o a na ional au ho i y in ou model allows us o inco po a e h ee common
dimensions: i) in many coun ies, public goods p o ision is a sha ed esponsibili y be ween
local and na ional au ho i ies;
6
ii) co up ion is o en acili a ed by impe ec in o ma ion
abou poli icians’ ac ions.
7
;iii) i enables us o simula e an en i onmen in which co up ion
is di icul o de ec , he eby in oducing eal incen i es o pa icipan s o engage in
co up beha io .
In ou heo e ical amewo k, we iden i y a signaling equilib ium shaped by wo p ima y
dynamics. Fi s , poli icians who de i e g ea e u ili y om hones y a e less inclined o
4
Fo ins ance, oyal ies om na u al esou ce ex ac ion se e as a majo income sou ce o municipali ies
in Colombia and one o he main sou ces o inancing o public goods (Ma ínez,2023). Thus, he shi in
cons uc ion cos s we assume in he model o simplici y mimics an economic shock simila o hose d i en
by luc ua ions in commodi y p ices. Mo eo e , he ele ance o economic shocks o commodi y p ices
ex ends beyond public goods p o ision, as hey ha e also been shown o in luence c iminal beha io (Dube
and Va gas,2013) and e en he e hical conduc o poli icians (Ashe and No osad,2023).
5
By allowing he poli ician o decide how much o alloca e, we concen a e on “assignmen ” p oblems,
which a e common app oaches when hinking abou co up ion (see Bane jee e al.,2012).
6
Fo example, in Colombia, since he 1990s, local and na ional au ho i ies ha e join ly managed essen ial
public se ices, including d inking wa e , sani a ion, heal h, and educa ion (Alesina e al.,2005). In
de eloping coun ies, i is also he case ha in e na ional agencies and o ganiza ions p o ide local public
goods ha complemen local p o ision. Ou case could hen be used o hink abou he ole o o eign aid
and co up ion.
7
De ails on who unds speci ic se ices and he amoun s alloca ed a e a ely common knowledge (Bu man
and Slem od,2020;López-Labo da e al.,2023), c ea ing unce ain y ha weakens accoun abili y; co up ion
ypically occu s unde condi ions o sec ecy (Hu and Oak,2023;Shlei e and Vishny,1993)
3
engage in co up ion, while ci izens who place a lowe alue on poli ical accoun abili y
a e less likely o incu he cos o disapp o ing o co up poli icians. Second, ollowing
a posi i e economic shock ha lowe s he cos o public goods, poli icians may be mo e
p one o co up ion, an icipa ing educed ci izen igilance in holding hem accoun able.
This expec a ion o lowe disapp o al inc eases he expec ed bene i s o co up ac ions.
Thus, ou model sugges s ha economic booms should no only ele a e local co up ion
bu also weaken poli ical accoun abili y.8
Building on hese heo e ical insigh s, we designed and implemen ed a labo a o y expe i-
men o empi ically es ou hypo heses. In he expe imen , pa icipan s we e assigned
o one o wo oles: ci izens o poli icians.
9
Pa icipan s we e ma ched using a pe ec
s ange ma ching p o ocol, in o med o he game s uc u e wi h emphasis on he im-
po ance o public goods, and epea ed he same ac i i y ac oss mul iple ounds. The
design inco po a ed wi hin-subjec a ia ion in public goods cos , se o ei he high cos
o low cos , simula ing exogenous a ia ion in he economic imes. This s uc u e allows
us o in es iga e he e ec o economic booms on poli ical beha io , p o iding empi ical
suppo o he heo e ical amewo k ou lined ea lie . De ailed expe imen al p ocedu es
a e p esen ed in Sec ion 3.
Ou indings indica e ha co up ion a es in high-cos en i onmen s (economic down-
u ns) a e app oxima ely 51%. In con as , in low-cos en i onmen s, such as hose ollow-
ing good economic imes, co up ion ises o 58%, ma king a 7-pe cen age poin inc ease.
Rega ding poli ical accoun abili y, we obse e ha ci izens’ willingness o disapp o e o
poli icians is 16% in high-cos en i onmen s and sligh ly lowe , a 15%, du ing economic
booms. These esul s ca y signi ican economic implica ions: good economic imes a e
associa ed wi h oughly a 14% ise in local poli ical co up ion, unde sco ing he di icul y
o main aining poli ical accoun abili y du ing pe iods o economic expansion.
Gi en he s uc u e o he game, he inal alloca ion o public goods also se es as a quali y
signal o he poli ician’s ac ions (Mani and Mukand,2007). Consequen ly, we examine
how he obse ed public goods alloca ion a ec s he ci izen’s willingness o disapp o e
o he poli ician (Fe ejohn,1986). Ou esul s indica e ha ci izens a e willing o incu
pe sonal cos s o punish co up poli icians, wi h a disapp o al a e o 36% o hose
8
This idea is consis en wi h he heo e ical esul s de eloped by Bha acha yya and Hodle (2010), and
wi h empi ical esul s linking economic condi ions and pe cep ions o co up ion in he popula ion
(Má quez Romo and Rome o-Vidal,2023;Zechmeis e and Zizumbo-Colunga,2013).
9
The expe imen used loaded aming, wi h pa icipan s explici ly in o med o hei ole i les. P io esea ch
indica es ha aming e ec s a e ypically minimal in co up ion games (Abbink and Hennig-Schmid ,
2006;Ba and Se a,2009;Aleksee , Cha ness and Gneezy,2017;Bane jee,2016).
4
wi h δηw being he mone a y punishmen ecei ed by he disapp o al om he ci izen.
On he o he hand, he ci izen de i es u ili y om he public good p o ision
(Θ)
and om
he non- axed income
y(1 −τ)
. Fo simplici y, he alua ion o each uni o he public good
is ixed a
κ∈R+
. Addi ionally, no e ha he ma ginal bene i o each uni is equal o he
ixed cos o disapp o al, meaning ha choosing o disapp o e o he poli ician ep esen s
a loss o one public good uni .21 We cha ac e ize he ci izen’s payo as:
Uc=y(1 −τ) + κΘ+δ((αc−Θ)λ−κ),(3)
whe e
λ∈[0,1]
deno es he ci izen’s pos e io belie abou he poli ician’s ype once she
obse es he signal p o ided by he public goods alloca ion (
Θ
). I is wo h men ioning
ha he ci izen upda es he belie s in a Bayesian manne ,
22
and hey a e consis en wi h
he p io belie s o co up ion (β∈(0,1)).
In gene al, ci izens may pe cei e highe e u ns om p o es when public good p o ision
is sca ce (Alme e al.,2017;G ossman,1991). The e o e, highe alloca ions o public
goods ac as a ese e u ili y o ci izens. In ou model, his ese e u ili y is ep esen ed
by
(αc−Θ)
, whe e lowe alloca ions o public goods inc ease he ma ginal bene i o
p o es . Addi ionally, ci izens beha e s a egically. The e o e, he ma ginal bene i s o
non-app o al a e media ed by he pos e io belie s o co up ion (λ).
While he model sugges s ha poli icians possess deepe insigh s in o he public goods
inancing scheme, i aises he ques ion o why his ype o in o ma ion asymme y is
deemed mo e c i ical han he asymme y conce ning economic ac o s ha in luence
cons uc ion cos s. No ably, he cos s o public goods end o align closely wi h hose
o p i a e goods, and gi en ha ci izens egula ly pu chase hese p i a e goods, i is
plausible o assume hey can easonably es ima e public cos s. In con as , he in icacies
o budge a y p ocesses emain la gely inaccessible o he gene al public, unde sco ing he
a ionale o ocusing on budge a y a he han economic in o ma ional asymme ies.
Figu e 3summa izes he model’s imeline. The pa icipan s’ ypes
(αc, ϕp)
, as well as he
21
Acco ding o O iz e al. (2022), social discon en can lead o he de as a ion o in as uc u e and weal h.
Images o bu ning ca s o buildings equen ly appea in media co e age abou p o es s in many na ions.
Consequen ly, i makes sense o exp ess he cos o p o es s as he loss o one uni o public goods.
22Fo a speci ic alloca ion o public goods, he belie s upda ing p ocess ollows:
λ≡P (σ= 0|Θ(π) = Θm(π))=P (Θ(π) = Θm(π)|σ= 0)×P (σ= 0)
P (Θ(π) = Θm(π))∀m={h, a, l}
11

cos s o public goods
π={π, π}
a e exogenously de e mined a he beginning o he game.
Then, he poli ician makes he policy decision, he na ional au ho i y andomly decides
whe he o p o ide esou ces o he municipali y, and he public goods alloca ion is shown
o he ci izen. Finally, once he ci izen obse es he inal alloca ion o public goods, she
decides whe he o app o e o disapp o e he poli ician, and payo s a e ealized.
Figu e 3: Timeline o he Poli ical Co up ion and Accoun abili y Model
Na u e Poli ician Na ional
Au ho i y Ci izen
Pa icipan s Cos o Policy Resou ces and App o al
ypes public goods choice public goods alloca ion choice
ϕp,αcπhγ,Θδ
No es: Pa icipan ypes a e d awn om hei co esponding dis ibu ions
(ϕp∼U[0,b
ϕ])
,
(αc∼ [0, α]
. The cos o public goods
(π)
ollows a Be noulli dis ibu ion, and
π∈ {π, π}
. Finally, he na ional au ho i y decides o alloca e addi ional esou ces
(γ= 1)
wi h a
p obabili y o µ∈(0,1).
2.3 Signaling Equilib ium
Now, we de ine s a egies, belie s, and equilib ium o he ex ensi e game o incomple e
in o ma ion played by he ci izen and he poli ician. The in o ma ion se a ailable o he
poli ician be o e making he policy choice is cha ac e ized by hei own ype and he public
good cos . Thus, we de ine a s a egy o poli ician as a mapping σ,
σ:Ip→[0,1] whe e Ip≡(ϕp, π)∈[0,1] × {π, π},
which speci ies he p obabili y o adop ing he policy decision
h
as a unc ion o he
in o ma ion ecei ed by he poli ician a he beginning o he game.
The ci izen knows he public good cos , obse es he alloca ion o public goods gi en ha
cos , and makes a decision gi en he in insic mo i a ion o disapp o al
αc
. Thus, we
de ine a s a egy o he ci izen as a mapping ν,
ν:{Θh(π),Θa(π),Θl(π)} × [0, α]→[0,1],
ha goes om he se o possible public goods alloca ions gi en a speci ic cos and he
dis ibu ion o possible disapp o al a es
[0, α]
o a ange be ween
[0,1]
ha deno es he
p obabili y o en o cing he punishmen agains he poli ician.
Finally, we de ine a belie sys em o he ci izen as a mapping λ,
12
λ:{Θh,Θa,Θl} × {0,1} → [0,1]
ha goes om he obse ed alloca ion o public goods
Θ
and p io belie s
β∈(0,1)
o
co up ion o a ange o upda ed belie s o co up ion in [0,1].
De ini ion 1. (Signaling Equilib ium) A signaling Bayesian equilib ium is gi en by
(i) a s a egy o he poli ician (σ)
(ii) a s a egy o he ci izen (ν)
(iii)
a belie sys em o he ci izen
(λ)
such ha : he ci izen s a egy
(ν)
is op imal gi en he
belie s, he ci izen belie s a e consis en wi h he poli ician’s s a egy—meaning ha
λ
is
de i ed om p io belie s
(β)
and he poli ician’s s a egy using Bayes’ ule o e e y signal,
and a poli ician s a egy (σ) ha is op imal gi en he ci izen s a egy.
2.4 Equilib ium Analysis
Gi en he s uc u e and iming o he game, in he inal s age o he in e ac ion, he ci izen
obse es he inal alloca ion o public goods and upda es he pos e io belie s abou
co up ion in a Bayesian manne . As illus a ed in Figu e 3, he sequen ial na u e o he
game allows he poli ician o an icipa e he s a egic beha io o he ci izen in he la e
s age, enabling he o choose he op imal s a egy in he ea ly s age o he game. The e o e,
o cha ac e ize he equilib ium, we i s de e mine he ci izen’s op imal s a egy
ν
and
hen he poli ician’s op imal s a egy σ.
Lemma 1. In e e y signal equilib ium, ega dless o he cos
(π)
, i he ci izen obse es a high-
alloca ion signal Θh, he ci izen plays a s a egy o no punishmen .
ν(αc,Θh)=0
P oo . See Online Appendix B.1.
F om Lemma 1, we can es ablish ha ega dless o he cos , a high alloca ion o public
goods consis en ly signals an hones poli ician, and consequen ly, e en ci izens who
p io i ize poli ical accoun abili y ha e no incen i e o disapp o e o he incumben . This
esul aligns wi h he idea ha ci izens ewa d good poli icians o keep hem in o ice and
punish bad poli icians o emo e hem om o ice.
De ini ion 2. (Cu o s a egy). A s a egy
ν
is a cu o s a egy i he e is some
α∗(Θm(π)) ∈
13
(0, α], wi h m={a, l}such ha :
ν(αc,Θm(π)) = 




1i αc≥α∗(Θm(π))
0i αc< α∗(Θm(π))
Lemma 2. In e e y signaling equilib ium, i he ci izen obse es an ambiguous
(Θa(π))
o
low-alloca ion signal (Θl(π)), he ci izen plays a cu o s a egy ollowing De ini ion 2.
P oo . See Online Appendix B.2.
F om Lemma 2, we conclude ha o each possible alloca ion
Θm(π)
he e exis s an
α∗(Θm(π))
cu o ha de ines he ci izen’s s a egy
ν
. This means ha he ci izen obse es
he p i a e in insic ewa d o poli ical accoun abili y, he cons uc ion cos , and he
inal alloca ion o public goods, and i he in insic ewa d o disapp o al is highe
han
α∗(Θm(π))
, she would disapp o e o he poli ician. Mo eo e , as ci izens ecei e
di e en in o ma ion abou he public goods alloca ion, he cu o may shi acco ding o
he cons uc ion cos (π)and he obse ed alloca ion.
Lemma 3. In e e y signaling equilib ium, π∈ {π, π},α∗(Θl(π)) ≤α∗(Θa(π))
P oo . See Online Appendix B.3.
F om Lemma 3, he deciding cu o unde an ambiguous alloca ion is always g ea e han
o equal o he deciding cu o unde a low alloca ion. Thus, ci izens who a e willing o
punish unde ambiguous alloca ions mus ha e a highe in insic ewa d o disapp o al
han ci izens who punish unde low alloca ions. Fu he mo e, a ci izen who punishes a
poli ician unde an ambiguous alloca ion—–whe e he e is unce ain y o he poli ician’s
co up ion—–will also punish unde a low alloca ion, whe e he e is ce ain y ha he
poli ician was co up .
Gi en ha we ha e al eady cha ac e ized he ci izen’s op imal s a egy
(ν)
, which ollows
a consis en belie sys em
(λ)
, i emains o cha ac e ize he poli ician’s op imal s a egy
consis en wi h he ci izen’s s a egy o de ine a signaling equilib ium, as p esen ed in
De ini ion 3.
De ini ion 3. (Th eshold s a egy). A s a egy
σ
is a h eshold s a egy i he e is some
ϕ∗(π)≤b
ϕ
,
14
and o e e y π∈ {π, π}such ha :
σ(ϕp) = 




1i ϕp≥ϕ∗(π)
0i ϕp< ϕ∗(π)
Lemma 4. In e e y signaling equilib ium, he poli ician en o ces a h eshold s a egy ollowing
De ini ion 3.
P oo . See Online Appendix B.4.
Lemma 4es ablishes ha he e mus exis a h eshold
ϕ∗(π)
which de ines he poli ician’s
s a egy
σ
. This means ha he poli ician obse es he p i a e in insic ewa d o doing
he “ igh hing”
(ϕp)
, he cons uc ion cos , and hen she compa es he expec ed payo o
being co up e sus being hones . I he in insic ewa d o ac ing wi h in eg i y exceeds
ϕ∗(π), she will choose o be hones and alloca e he en i e budge o public goods.
2.5
Impac s o Economic Times on Poli ical Co up ion and Accoun -
abili y
Rega ding he impac s o good economic imes on poli ical co up ion and poli ical ac-
coun abili y, we i s de ine good economic imes as a shi in cons uc ion cos s om a
high cos
(π)
o a low cos
(π)
. Mo eo e , hese cos changes a e andomly de e mined and
announced o bo h ci izens and poli icians a he beginning o he game. Wi h his in mind,
Theo em 1. In a signaling equilib ium, and o all m={a, l}, i π≥π, hen
α∗(Θm(π)) ≥α∗(Θm(π))
P oo . See Online Appendix B.5.
The main esul o ou heo e ical model is ha good economic imes, cap u ed by a decline
in he cos o public goods p o ision om a high le el
(π)
o a low le el
(π)
, educe ci izens’
willingness o punish poli icians by aising he op imal punishmen h eshold. This shi
is d i en by a highe expec ed ma e ial payo , as he expec ed public goods alloca ion
is highe du ing good economic imes. Impo an ly, he inc ease in he cu o applies
ega dless o he obse ed alloca ion, e lec ing a gene al endency owa d mo e lenien
beha io in bo h low and ambiguous alloca ion scena ios.
15
This leads o he ollowing co olla y:
Co olla y 1. In a signaling equilib ium, and o all Θmwi h m={a, l}, i π≥π, hen
Zα∗(Θm(π))
0 (α)dα ≥Zα∗(Θm(π))
0 (α)dα
P oo . See Online Appendix B.6.
As a di ec implica ion o Theo em 1, Co olla y 1demons a es ha , gi en an inc ease
in he op imal cu o poin s and a gi en signal
α∗(Θm)
, a poli ician is mo e likely o be
punished in high-cos en i onmen s han in low-cos se ings,
Figu e 4 isually illus a es he changes in he p obabili y o punishmen . Speci ically, i
compa es he ini ial se o cu o s,
α∗(π) = {α∗(Θl(π)), α∗(Θa(π))}
wi h he new pai o
cu o s α∗(π) = {α∗(Θl(π)), α∗(Θa(π))}, ollowing he cos educ ion.
Figu e 4: Ci izen Dis ibu ion o Types (α)
αlow(π)
αlow(π)
αamb(π)
αamb(π)
(α)
No
Punish Punish i Θ=ΘlPunish i Θ=Θl
and also i Θ=Θa
α
No es: The igu e depic s he dis ibu ion o he ypes
(αc∼ [0, α]
o he ci izen. In da k g ay, he mass o ci izens ha a e no willing
o punish. In medium g ay, he mass o ci izens who a e willing o punish unde low-alloca ions o public goods. In ligh g ay, he mass
o ci izens ha a e willing o punish unde low-alloca ions and ambiguous-alloca ions o public goods. In black (solid,
ϕ∗(π)
), we p esen
he op imal cu o unde a high-cos se ing. In blue (dashed, ϕ∗(π)), we p esen he op imal cu o unde a low-cos se ing.
Finally, we examine how a o able economic condi ions in luence a poli ician’s ma ginal
p opensi y o engage in co up ion.
Theo em 2. In a signaling equilib ium, i π≥π
ϕ∗(π)≤ϕ∗(π)
P oo . See Online Appendix B.7.
16

As es ablished in Co olla y 1, an economic boom educes he likelihood ha ci izens will
punish co up poli icians. This decline in pe cei ed accoun abili y lowe s he expec ed
cos o co up beha io , he eby inc easing i s ne expec ed bene i . Theo em 2 o malizes
his esul , showing ha he equilib ium p opensi y o engage in co up ion ises as he
cos o public goods p o ision alls.
Co olla y 2. In a signaling equilib ium wi h ϕ∗(π)∈R+, i π≥π
Zϕ∗(π)
0U(ϕ)dϕ ≤Zϕ∗(π)
0U(ϕ)dϕ
P oo .
This esul ollows di ec ly om he p ocedu e desc ibed in Online Appendix B.6,
by changing α∗(Θm(π)) o ϕ∗(π).
Finally, Co olla y 2p esen s a di ec implica ion o Theo em 2, showing ha he p obabili y
o a poli ician engaging in co up ac i i ies is g ea e when cons uc ion cos s a e low
compa ed o when hey a e high. These heo e ical esul s a e u he illus a ed in Figu e
5. Taken oge he , ou indings highligh he subs an ial in luence o p e ailing economic
condi ions on he incidence o co up ion and he mechanisms o poli ical accoun abili y.
Figu e 5: Poli ician Dis ibu ion o Types (ϕ)a e he Economic Shock
ϕ
(ϕ)
1
b
ϕ
0b
ϕ
ϕ∗(π)ϕ∗(π)
Co up Hones
No es: The igu e depic s he dis ibu ion o he ypes
(ϕp∼U[0,b
ϕ])
o he poli ician. In medium g ay, he mass o poli icians who a e
co up . In ligh g ay, he mass o poli icians who a e hones . In black (solid,
ϕ∗(π)
), we p esen he op imal cu o unde a high-cos
se ing. In blue (dashed, ϕ∗(π)), we p esen he op imal cu o unde a low-cos se ing.
3 Expe imen al Design
Building on hese heo e ical esul s, we conduc ed a se ies o labo a o y expe imen s using
oT ee (Chen e al.,2016). The expe imen s in ol ed nea ly 800 college s uden s ec ui ed
17
h ough he Rosa io Expe imen al and Beha io al Economics Lab (RE
β
EL, G eine (2015))
a Uni e sidad del Rosa io in Bogo a, Colombia. We collec ed da a om 24 independen
sessions, each las ing an a e age o one and a hal hou s.
3.1 Poli ical Co up ion Game (PCG)
The main ac i i y o he expe imen was a poli ical co up ion game (PCG) designed o
eplica e ou heo e ical model. Pa icipan s we e andomly assigned o one o wo oles:
poli ician (P) o ci izen (C), and his ole emained ixed o he en i e session. Each ci izen
ecei ed an endowmen o 100 expe imen al poin s (EP),
23
while each poli ician had a
po en ial sala y o 120 EP. Pa icipan s we e pai ed wi h a membe o he opposi e ole
unde a pe ec s ange -ma ching design.
Once pai ed, he ci izen paid 40 EP in axes, which we e ans e ed o he poli ician.
The poli ician’s main ask was o decide how much o his budge o alloca e owa d
cons uc ing public goods: ei he he en i e amoun (40 EP) o hal (20 EP), keeping he
emaining hal as pe sonal income. In his se up, public goods ep esen an addi ional
sou ce o income o ci izens. A he beginning o he expe imen , bo h pa icipan s we e
in o med ha each uni o public goods cons uc ed would p o ide he ci izen wi h an
addi ional 30 EP, highligh ing he social bene i s o highe public goods alloca ions and
ein o cing he poli ician’s du y o ac in he ci izen’s in e es .
In his se up, each uni o public goods has a cos . Pa icipan s a e awa e ha he con-
s uc ion cos is ei he 20 EP (high cos ) o 10 EP (low cos ), wi h each op ion ha ing an
equal p obabili y. Bo h pa icipan s we e in o med abou he public goods cos (
π
) a
he beginning o each ound, ensu ing no in o ma ion asymme y abou he s a e o he
economy.
A e obse ing he public goods cos (
π
), and once he poli ician made he alloca ion
decision, he compu e simula ed a na ional au ho i y by andomly deciding whe he
o alloca e addi ional unds. I he compu e alloca ed esou ces, 20 EP we e added o
he cons uc ion o public goods; o he wise, only he poli ician’s con ibu ion was used.
Impo an ly, he compu e ’s decision emained p i a e, meaning ci izens obse ed only
he inal public goods alloca ion wi hou knowing he indi idual con ibu ions om he
poli ician o he highe -le el go e nmen (in his case, he compu e ’s andom alloca ion).
23
Each EP was alued a USD 0.08. Pa icipan s ecei ed a USD 3 show-up ee. Ci izens (Poli icians) ea ned
an a e age o USD 13.1 (USD 12.9). Gi en ha he minimum wage in Colombia is USD 1.42/hou , he
ea nings ob ained a e abou nine imes he hou ly minimum wage.
18
Public goods p o ision ollows equa ion (1).
Once he ci izen obse ed he public goods p o ided, she decided whe he o educe he
poli ician’s sala y (“punish”) o 80 EP
24
by paying 30 EP, o o “app o e” he poli ician
and allow he o ecei e he ull sala y. The in e ac ion ended once he ci izen made he
choice, and no eedback was p o ided wi hin he ound.
3.1.1 Addi ional De ails
This in e ac ion epea ed o e i e decision ounds.
25
In each ound, pa icipan s e-
po ed hei i s -o de belie s abou hei pa ne ’s beha io : poli icians p edic ed he
ci izen’s decision, while ci izens p edic ed he poli ician’s policy choice. These belie s
we e incen i e-compa ible, as pa icipan s ea ned 5 EP each ime hey co ec ly p edic ed
hei pa ne ’s decision. To ensu e expec a ions we e eco ded be o e obse ing o he s’
beha io , each pa icipan ollowed a sepa a e expe imen al imeline.26
3.1.2 T ea men s
We exploi a wi hin-subjec s design ac oss all i e ounds. Gi en ha he cos o public
goods
(π )
is andomly selec ed by he compu e a he beginning o he ound and was
de ined a he pai le el, we deno e he ollowing wo s a es o he economy:
•
High-cos en i onmen -o bad economic imes- (Con ol): Unde his se ing, he
cons uc ion cos is se o 20EP. The e o e, he maximum amoun o public goods
ha can be p oduced is h ee, and he lowes amoun o public goods ha can be
p oduced is one.
•
Low-cos en i onmen -o good economic imes- (T ea men ): Unde his se ing,
he cons uc ion cos is se o 10EP. The e o e, he maximum amoun o public goods
ha can be p oduced is six, and he lowes amoun o public goods ha can be
p oduced is wo.
24Gi en he s uc u e o he expe imen , he u ili y unc ions unde he PCG a e:
Up= 120 + (1 −h)20 −δ(0.33)120 + hϕp
Uc= 100(1 −0.4) + 30Θ+δ((αc−Θ)λ−30)
25
Pa icipan s knew ha only one andomly selec ed ound would coun owa d hei inal expe imen al
paymen .
26The expe imen al imeline o each ype o pa icipan is displayed in he Online Appendix C1
19
3.2 Coo dina ion Game (Social No ms)
Finally, pa icipan s engaged in a coo dina ion game based on K upka and Webe (2013),
used o elici incen i e-compa ible no ma i e expec a ions om ci izens and poli icians
du ing he PCG.
27
Each pa icipan a ed he social app op ia eness o he poli ician’s be-
ha io unde each possible cos scena io. Speci ically, subjec s indica ed whe he alloca ing
he en i e budge was e y socially inapp op ia e, socially inapp op ia e, socially app o-
p ia e, o e y socially app op ia e. Pa icipan s ea ned 5 EP each ime hei no ma i e
expec a ion aligned wi h he majo i y iew among pa icipan s o hei own ype. A he
end o he ac i i y, hey comple ed a non-incen i ized socio-demog aphic ques ionnai e
and p o ided paymen de ails. Table D2 in he Appendix p esen s desc ip i e s a is ics.
4 Resul s
4.1 Poli ical Co up ion
A e poli icians co up ? Pooling pa icipan s ac oss all i e ounds, we ind ha hey a e
co up 55% o he ime. No ably, 22% o he poli icians we e always co up , while 12% o
hem we e always hones .
28
Thei willingness o engage in co up ion is no independen
o he economic cycle. Figu e 6, which illus a es he pe cen age o co up poli icians
segmen ed by cos ype, shows ha he uncondi ional a e age co up ion a e is 51% when
he cos is high, ising o nea ly 59% when he cons uc ion cos is low—a s a is ically
signi ican inc ease a he 1% le el. This inc ease ep esen s a 14% ise in he p obabili y o
engaging in co up ion compa ed o he high-cos en i onmen .
To measu e he causal ea men e ec s on poli ical co up ion, we es ima e he ollowing
model:
Co up p, ,s =α0+α1Cos p, +x′
pγ+ Λ +ηs+εp, ,s,(4)
whe e
Co up i, ,s
is a dummy a iable ha akes he alue o 1 i he poli ician was co up
and 0 o he wise.
Cos p,
, he ea men a iable, is a dummy a iable ha akes he alue
o 1 i he cons uc ion cos is low and 0 i he cons uc ion cos is high. Addi ionally, we
include a se o con ols
(x′
p)
ha conside sociodemog aphic a iables, a se ies o ound
27
The coo dina ion game was implemen ed a he end o he expe imen , wi h pa icipan iden i ies emaining
anonymous.
28
The dis ibu ion o he numbe o imes poli icians we e co up is p esen ed in Figu e D5 in he Online
Appendix.
20
include he same se o sociodemog aphic con ols p esen ed in he es ima ion o equa ion
(4). Finally, we es ima e each model conside ing i he poli ician was hones
(σ= 1)
o
co up (σ= 0).
Figu e 9: Poli ician Expec ed Punishmen Segmen ed by Cos s (π)and Policy Choice
0.112
0.065
***
0
0.10
0.20
0.30
0.40
0.50
Belie s o punishmen (%)
Hones (σ = 1)
0.346
0.283
**
0
0.10
0.20
0.30
0.40
0.50
Belie s o punishmen (%)
Co up (σ = 0)
High cos Low cos
No es: The igu e depic s he a e age expec ed disapp o al a e by he poli ician ac oss all ounds, segmen ed by p oduc ion cos s and
policy choices. High cos = 20 EP. Low cos = 10 EP. Da k lines show 95% clus e ed obus con idence in e als. No con ols included.
∗∗∗ p < 0.01,∗∗ p < 0.05,∗p < 0.1
, n.s
p > 0.1
. Repo ed signi icance le els a e de i ed om hypo hesis es ing o a ully sa u a ed
linea model, whe e he dependen a iable is he ci izen’s choice o disapp o e o he incumben , and he eg esso s a e dummy
a iables associa ed wi h he ea men s.
The OLS es ima es a e shown in Table 4. Columns (1) o (3) p esen es ima es o pun-
ishmen expec a ions when he poli ician was hones . Hones poli icians expec o be
punished 11% o he ime, consis en wi h he unce ain y su ounding he inal public
goods alloca ion, which may gene a e an ambiguous alloca ion signal. The coe icien s
u he indica e ha in good economic imes, poli icians expec a 4 o 5 pe cen age poin
educ ion in he likelihood o punishmen .
Simila ly, Columns (4) o (6) p esen es ima es o cases whe e he poli ician was co up .
The expec ed disapp o al a e is app oxima ely 35%, signi ican ly highe han ha o
hones poli icians (p- alue = 0.000). In good economic imes, co up poli icians an icipa e
a 6 o 7 pe cen age poin educ ion in he likelihood o punishmen . The poin es ima es
also e eal signi ican economic implica ions, wi h economic booms leading o a educ ion
27

Table 4: Poli icians Expec ed Punishmen ac oss cos (π), Segmen ed by Policy Choice
Dep Va :Belie o punishmen Poli ician’s decision (σ)
Hones Co up
(1) (2) (3) (4) (5) (6)
Low cos (T ea men ) -0.048*** -0.044** -0.041** -0.063** -0.068** -0.064**
(0.02) (0.02) (0.02) (0.03) (0.03) (0.03)
Cons an 0.112*** 0.380*** 0.411*** 0.346*** 0.574*** 0.560***
(0.02) (0.13) (0.13) (0.02) (0.17) (0.19)
Mean Dep. Va iable 0.11 0.11 0.11 0.35 0.35 0.35
E ec size (%) -42.3 -39.0 -36.7 -18.1 -19.7 -18.5
Con ols ✓ ✓ ✓ ✓
Round FE ✓ ✓
Session FE ✓ ✓
Obse a ions 899 899 899 1101 1101 1101
Clus e s 309 309 309 351 351 351
Adjus ed R-squa ed 0.01 0.09 0.16 0.00 0.02 0.08
No es: The se o con ols included in Con ols a e: age, sex, s a um, us measu e, isk measu e, pa ience
measu e, and p e ious expe ience pa icipa ing in expe imen s. The Mean Dep. Va iable exp esses he uncondi ional
con ol mean o he dependen a iable. The E ec size (%) deno es he size o he ea men e ec as a pe cen age
compa ed wi h he uncondi ional mean o he con ol g oup. S anda d e o s clus e ed a he indi idual le el a e
p esen ed in pa en heses. ∗∗∗p < 0.01,∗∗ p < 0.05,∗p < 0.1
in he expec ed disapp o al a e anging om -42% o -20% compa ed o he con ol g oup.
5.2 He e ogeneous Punishmen Beha io ac oss Co up ion Belie s
Rega ding poli ical accoun abili y beha io , we le e age ci izens’ incen i e-compa ible
empi ical expec a ions o es ima e how p io belie s abou co up ion in luence punish-
men . Figu e 10 displays he uncondi ional disapp o al a e ac oss he inal public goods
alloca ion o public goods segmen ed by p io belie s o co up ion.
F om Figu e 10, when ci izens obse e a low public goods alloca ion and expec he
poli ician o be hones , he disapp o al a e is 48%; howe e , o hose expec ing he
poli ician o be co up , he a e d ops o 26%, e lec ing a 22pp educ ion. Addi ionally,
when ci izens obse e an ambiguous alloca ion, p io belie s ha e minimal e ec , wi h
disapp o al a es consis en ly a ound 9%. Las ly, o high alloca ions o public goods,
disapp o al is 3% among ci izens expec ing hones y and 6% among hose expec ing
dishones y, hough his di e ence is no s a is ically signi ican .
28
Figu e 10: Ci izen Punishmen Segmen ed by P io belie s (β)ac oss Alloca ions (Θ).
0.48
0.09
0.03
0.26
0.10
0.06
*** n.s. n.s.
0
0.10
0.20
0.30
0.40
0.50
0.60
% o ci izens ha disapp o e
Low alloca ion
θl
Ambiguous alloca ion
θa
High alloca ion
θh
Belie poli ician was hones Belie poli ician was co up
No es: The igu e depic s he a e age disapp o al a e o he ci izen in all ounds, segmen ed by p io belie s abou co up ion and he
inal alloca ion o public goods. High cos = 20 EP. Low cos = 10 EP. The da k lines show obus con idence in e als clus e ed wi h
95%. No con ols included.
∗∗∗ p < 0.01,∗∗ p < 0.05,∗p < 0.1
, n.s
p > 0.1
. Repo ed signi icance le els come om hypo hesis es ing
om a ully sa u a ed linea model whe e he dependen a iable is he ci izen’s choice o disapp o e o he incumben , and eg esso s
a e dummy a iables associa ed wi h ea men s.
Gi en ha punishmen seems o be he e ogeneous ac oss p io belie s o co up ion, and
in o de o o mally es his mechanism, we es ima e he ollowing model:
Punishc, ,s =α0+α1Θa
c+α2Θl
c+α3BC +β1(Θa
c×BC) + β2(Θl
c×BC) + Λ +ηs+εc, ,s,(8)
whe e BC deno es he ci izen’s p io belie s o co up ion. Table 5p esen s he esul s on
he he e ogeneous e ec o p io belie s ac oss obse ed alloca ions.
Column (1) indica es ha p io belie s o co up ion do no enhance poli ical accoun abil-
i y; ins ead, hey educe he disapp o al a e. While ci izens a e mo e likely o punish
poli icians when hey obse e a low alloca ion o public goods, he disapp o al a e is
signi ican ly highe i hey expec he poli ician o be hones (48%) compa ed o when hey
expec he poli ician o be co up (26%). This beha io sugges s ha when ci izens an ici-
pa e poo pe o mance, hei mo i a ion o punish dec eases, aligning wi h he indings o
De la Cues a e al. (2022).
29
Table 5: Ci izens Disapp o al Ra e ac oss Alloca ions, by P io Belie s o Co up ion (β)
Dep Va :Ci izen punishmen OLS
(1) (2) (3) (4)
Ambiguous alloca ion (θa) 0.048*** 0.055*** 0.047** 0.045**
(0.01) (0.02) (0.02) (0.02)
Low alloca ion (θl) 0.314*** 0.449*** 0.445*** 0.442***
(0.03) (0.04) (0.04) (0.04)
Belie Co up -0.051*** 0.022 0.016 0.017
(0.02) (0.02) (0.02) (0.02)
Ambiguous alloca ion (θa)×Belie Co up -0.011 -0.001 0.003
(0.03) (0.03) (0.03)
Low alloca ion (θl)×Belie Co up -0.245*** -0.236*** -0.227***
(0.05) (0.05) (0.05)
Cons an 0.073*** 0.034*** 0.356* 0.227
(0.01) (0.01) (0.20) (0.20)
Mean Dep. Va iable 0.05 0.05 0.05 0.05
E ec size (%) 104.8 119.3 102.8 97.1
Con ols ✓ ✓
Round FE ✓
Session FE ✓
Obse a ions 1995 1995 1995 1995
Clus e s 399 399 399 399
Adjus ed R-squa ed 0.13 0.15 0.17 0.18
p- alue H0:θa=θl0.000 0.000 0.000 0.000
No es: The se o con ols included in Con ols a e: age, sex, s a um, us measu e, isk measu e, pa ience
measu e, p e ious expe ience pa icipa ing in expe imen s, and he willingness- o-p o es measu e om a
6-i em ques ionnai e. The Mean Dep. Va iable exp esses he uncondi ional con ol mean o he dependen
a iable. The p- alue
H0:θa=θl
deno es he s a is ical es o di e ences ac oss alloca ions. S anda d e o s
clus e ed a he indi idual le el a e p esen ed in pa en heses. ∗∗∗p < 0.01,∗∗ p < 0.05,∗p < 0.1
5.3 Does Cos A ec P io Belie s o Co up ion?
Gi en ha p io belie s o co up ion in luence ci izen’s willingness o disapp o e, we
now es ima e he e ec o good economic imes on hose p io belie s. Table 6p esen s
OLS es ima es o he a e age ea men e ec o low cos s on p io belie s o co up ion. In
low-cos en i onmen s, ci izens expec poli icians o be 8pp less co up han in high-cos
en i onmen s. Fu he mo e, he poin es ima es indica e subs an ial economic impac , wi h
educ ions in incen i e-compa ible empi ical expec a ions nea ing 14%. These indings
sugges ha ins ead o an icipa ing inc eased co up ion in good economic imes, ci izens’
us in poli icians ac ually ises–—a esul consis en wi h obse ed beha io in he La in
Ame ica egion (Figu e 1).
The ull punishmen mechanism un olds as ollows: a o able economic condi ions in-
c ease he incen i es o co up ion, leading o mo e equen ins ances o inadequa e
public goods p o ision. Ini ially, ci izens a e mo e inclined o us poli icians, as good
30
Table 6: Ci izens P io Belie s o Co up ion ac oss Cos (π)
Dep Va :Belie o co up ion P io belie s (β)
(1) (2) (3)
Low cos (T ea men ) -0.084*** -0.077*** -0.073***
(0.03) (0.03) (0.03)
Cons an 0.574*** 0.356** 0.309**
(0.02) (0.14) (0.14)
Mean Dep. Va iable 0.57 0.57 0.57
E ec size (%) -14.6 -13.5 -12.8
Con ols ✓ ✓
Round FE ✓
Session FE ✓
Obse a ions 1995 1995 1995
Clus e s 399 399 399
Adjus ed R-squa ed 0.01 0.04 0.07
No es: The se o con ols included in Con ols a e: age, sex, s a um, us
measu e, isk measu e, pa ience measu e, p e ious expe ience pa icipa ing in
expe imen s, and he willingness- o-p o es measu e om a 6-i em ques ion-
nai e. The Mean Dep. Va iable exp esses he uncondi ional con ol mean o he
dependen a iable. The E ec size (%) deno es he size o he ea men e ec
as a pe cen age compa ed wi h he uncondi ional mean o he con ol g oup.
S anda d e o s clus e ed a he indi idual le el a e p esen ed in pa en heses.
∗∗∗p < 0.01,∗∗ p < 0.05,∗p < 0.1
imes educe hei p io belie s abou he likelihood o co up ion. Howe e , when o -
e s obse e a low alloca ion o public goods and subsequen ly lea n ha he incumben
engaged in co up beha io , hey expe ience a sense o be ayal, which leads o a sha p
inc ease in poli ical disapp o al.
5.4 Social No ms
Las ly, we conduc a compa a i e analysis o esponses om he coo dina ion game
ques ionnai e (K upka and Webe ,2013), based on he ollowing ques ion: “In he si ua ion
whe e he cos was (High,Low). I a poli ician decides o keep 20 EP o he collec ed axes, you
conside his decision o be...?” In his ask, pa icipan s a emp ed o ma ch he modal
app op ia eness a ing o o he s wi hin hei pa icipan ype.
As shown in he le panel o Figu e 11, in high-cos scena ios, a consensus eme ges
among poli icians ha pe sonally e aining a po ion o collec ed axes is e y socially
inapp op ia e. Howe e , his beha io shi s d ama ically in a low-cos en i onmen
whe e a signi ican po ion o poli icians iew keeping some o he axes o pe sonal use
31
as mo e accep able, illus a ing a elaxa ion o social no ms (desc ip i e no m) agains
co up ion. Ci izens’ belie s beha e simila ly. High-cos se ings ein o ce he s ong social
inapp op ia eness o co up ion. Ye , in low-cos se ings, ewe ci izens iew co up
beha io as highly inapp op ia e, sugges ing some deg ee o no maliza ion o ole ance.
Figu e 11: Social No ms His og am ac oss Cos s and Pa icipan s Types
χ2 p- alue = 0.000
0
10
20
30
40
50
60
Pe cen age %
Ve y
socially
app op ia e
Socially
app op ia e Socially
innapp op ia e
Ve y
socially
innapp op ia e
Poli ician
χ2 p- alue = 0.000
0
10
20
30
40
50
60
Pe cen age %
Ve y
socially
app op ia e
Socially
app op ia e Socially
innapp op ia e
Ve y
socially
innapp op ia e
Ci izen
You conside his decision o be...
I a poli ician decides o keep 20 EP o he collec ed axes
Low cos (πl) High cos (πh)
No es: The igu e illus a es he dis ibu ion o esponses o he ques ion: “In he si ua ion whe e he cos was (High,Low), i a poli ician
decides o keep 20 EP o he collec ed axes, do you conside his decision o be...?” The his og am shows esponses o low cos (in g ay) and
high cos (in dashed black). The le panel p esen s he dis ibu ion o esponses o poli icians, while he igh panel shows hose o
ci izens. (Le panel: χ2 es p- alue = 0.000; Righ panel: χ2 es p- alue = 0.000.)
6 Discussion
In his s udy, we in es iga e he ela ionship be ween economic condi ions and ci izens’
willingness o hold poli icians accoun able, as well as he impac o economic condi ions
on poli ical co up ion. While he li e a u e ex ensi ely documen s he ad e se e ec s
o co up ion on economic pe o mance, less a en ion has been gi en o how economic
luc ua ions shape co up ion le els and accoun abili y. By add essing his gap, ou
esea ch o e s new insigh s in o poli ical beha io du ing economic luc ua ions and lays
he g oundwo k o mo e e ec i e go e nance s a egies. We achie e his by de eloping a
heo e ical model and empi ically es ing i s p edic ions h ough a labo a o y expe imen .
32

Ou heo e ical model p o ides key insigh s in o he ela ionship be ween economic pe o -
mance and poli ical beha io . We p edic ha a o able economic condi ions, cha ac e ized
by low cos s o p o iding public goods, inc ease poli icians’ inclina ion o engage in co -
up ac i i ies. This occu s la gely because poli icians an icipa e a lowe likelihood o
punishmen , as ci izens may be mo e willing o o e look co up ion in imes o economic
p ospe i y. Addi ionally, ou model p edic s ha good economic imes weaken poli ical ac-
coun abili y, e lec ed in ci izens’ educed willingness o disapp o e o co up poli icians.
This highligh s a b oade end o diminished punishmen and weake en o cemen .
Empi ically, ou expe imen s suppo hese heo e ical indings. Co up ion a es a e
signi ican ly highe in low-cos en i onmen s, wi h an inc ease o app oxima ely 14%
compa ed o he baseline. Addi ionally, in hese low-cos se ings, poli icians an icipa ed
a nea ly 7-pe cen age-poin educ ion in he disapp o al a e, which u he con ibu ed
o he ise in co up ion. In con as , ci izens’ willingness o disapp o e o poli icians
emained ela i ely s able ac oss cos le els, as hey p io i ized punishmen based on he
obse ed alloca ion o public goods (Fe ejohn,1986). Speci ically, ci izens a e willing o
punish co up poli icians nea ly 36% o he ime when he alloca ion is low and be ween
3 and 6 pe cen o he ime when he alloca ion is high.
Poli icians ecognize ha ci izens can iden i y co up ion o some ex en , belie ing hey
a e h ee imes mo e likely o be punished when co up han when hones . Howe e ,
poli icians do no expec o be punished e e y ime hey engage in co up ion, and an-
icipa e occasional punishmen e en when ac ing hones ly. Consequen ly, incen i es o
co up ion inc ease when poli icians pe cei e ci izens as less sophis ica ed. Mo eo e ,
ci izens a e less likely o punish co up ion i hey al eady suspec poli icians a e co up ,
sugges ing ha mis us os e s condi ions ha enable u he co up ion.
While ou indings o e no el insigh s, we acknowledge impo an limi a ions ega ding
ex e nal alidi y. Fi s , as wi h any labo a o y expe imen , ou se ing abs ac s om some
ea u es o eal-wo ld poli ics, mos no ably he s akes in ol ed. In p ac ice, co up ion
can ca y legal consequences such as imp isonmen and may de e eelec ion, whe eas ou
expe imen al con ex imposes no such penal ies. Tha said, al hough he magni ude o he
incen i es di e s, we see no compelling eason o expec ha he di ec ion o beha io al
esponses o he unde lying logic o he accoun abili y mechanism would e e se in
highe -s akes en i onmen s.
A second conce n ela es o he composi ion o ou expe imen al sample, which consis s
p ima ily o uni e si y s uden s. This may be iewed as a limi a ion, pa icula ly in a
33
se ing whe e o he - ega ding p e e ences play a cen al ole. While ea lie e idence
sugges s ha s uden s may exhibi highe le els o sel ish beha io in economic games
ela i e o non-s uden s (Belo e al.,2015), mo e ecen s udies ind b oadly simila
dis ibu ions o social p e e ences, especially wi h espec o al uis ic and maximin (i.e.,
conce n o imp o ing he wel a e o he wo s -o ) ypes ac oss s uden and non-s uden
popula ions, pa icula ly among indi iduals wi h highe le els o educa ional a ainmen
(Eppe e al.,2023). This is pa icula ly ele an gi en ha bo h ou sample and he
popula ion o in e es (e.g., public o icials) a e highly educa ed. E en i one we e o
assume ha sel ish beha io exhibi by s uden s may be d i ing he e ec s on poli ical
co up ion, o he s udies using simila expe imen al samples in Colombia epo ha
app oxima ely 70% o college s uden s beha e in a non-sel ish manne (Ba inas-Fo e o
and Sca ascini,2025), sugges ing ample he e ogenei y in o he - ega ding mo i es du ing
ou expe imen al in e ac ion.33
Based on hese indings, policymake s should emain igilan abou he isk o inc eased
co up ion du ing pe iods o economic g ow h. S eng hening anspa ency and accoun -
abili y mechanisms, pa icula ly du ing pe iods o p ospe i y, is c ucial o main aining
high go e nance s anda ds. Accoun abili y becomes mo e challenging when ci izens s ug-
gle o iden i y co up ion, a di icul y ampli ied by economic cycles. When mul iple le els
o go e nmen sha e esponsibili y o p o iding public goods, holding poli icians ac-
coun able becomes e en mo e complex. Thus, while decen aliza ion may o e a solu ion,
i is e ec i e only i i cla i ies a he han obscu es esponsibili ies.
Fu u e esea ch could expand on ou indings by examining he long- e m e ec s o
good and bad economic imes on co up ion ac oss di e en con ex s and explo ing how
in o ma ion dissemina ion migh mi iga e he nega i e impac o economic booms on
poli ical accoun abili y. Addi ionally, analyzing he in e ac ion be ween a ious ypes
o economic shocks and poli ical sys ems could p o ide aluable insigh s o comba ing
co up ion unde di e se economic condi ions.
33
Ou expe imen was conduc ed a a leading Colombian uni e si y ha has his o ically p oduced a
signi ican sha e o he coun y’s poli ical eli e, including mul iple o me p esiden s. Mo eo e , nea ly 40%
o pa icipan s a e en olled in deg ee p og ams di ec ly ela ed o public adminis a ion and go e nance,
making his sample pa icula ly ele an o s udying he p ospec i e beha io o u u e public o icials.
34
Re e ences
Abadie, Albe o, Susan A hey, Guido W. Imbens, and Je ey M. Woold idge, “When
Should You Adjus S anda d E o s o Clus e ing?,” The Qua e ly Jou nal o Economics,
2023, 138 (1), 1–35.
Abbink, Klaus and Danila Se a, “An i-co up ion Policies: Lessons om he Lab,” in
“New Ad ances in Expe imen al Resea ch on Co up ion,” Eme ald G oup Publishing
Limi ed, 2012.
and Heike Hennig-Schmid , “Neu al e sus loaded ins uc ions in a b ibe y expe i-
men ,” Expe imen al Economics, 2006, 9, 103–121.
, Be nd I lenbusch, and Elke Renne , “An Expe imen al B ibe y Game,” Jou nal o Law,
Economics and O ganiza ion, 2002, 18 (2), 428–454.
Ades, Albe o and Ra ael Di Tella, “Ren s, compe i ion, and co up ion,” Ame ican
Economic Re iew, 1999, 89 (4), 982–993.
Adse à, Alícia, Ca les Boix, and Ma k Payne, “A e You Being Se ed? Poli ical Accoun -
abili y and Quali y o Go e nmen ,” The Jou nal o Law, Economics and O ganiza ion, 2003,
19 (2), 445 –490.
Aleksee , Aleksand , Ga y Cha ness, and U i Gneezy, “Expe imen al me hods: When
and why con ex ual ins uc ions a e impo an ,” Jou nal o Economic Beha io and O gani-
za ion, 2017, 134, 48–59.
Alesina, Albe o, Albe o Ca asquilla, and Juan Eche a ia, “Decen aliza ion in Colom-
bia,” in Albe o Alesina, ed., Ins i u ional Re o ms in Colombia., MIT P ess, 2005.
Alme , Ch is ian, Je emy Lau en -Lucche i, and Manuel Oechslin, “Wa e sca ci y and
io ing: Disagg ega ed e idence om Sub-Saha an A ica,” Jou nal o En i onmen al
Economics and Managemen , 2017, 86 (1), 193–209.
Al unba¸s, Yene and John Tho n on, “Does inancial de elopmen educe co up ion?,”
Economics Le e s, 2012, 114 (2), 221–223.
A danaz, Ma ín and Alejand o Izquie do, “Cu en expendi u e upswings in good imes
and public in es men downswings in bad imes? New e idence om de eloping
coun ies,” Jou nal o Compa a i e Economics, 2022, 50 (1), 118–134.
35
A mand, Alex, Alexande Cou s, Ped o Vicen e, and Ines Vilela, “Measu ing co up ion
in he ield using beha io al games,” Jou nal o Public Economics, 2023, 218 (10), 47–99.
A man ie , Oli ie and Amadou Boly, “A con olled ield expe imen on co up ion,”
Eu opean Economic Re iew, 2011, 55 (8), 1072–1082.
Ashe , Sam and Paul No osad, “Ren -seeking and c iminal poli icians: E idence om
mining booms,” Re iew o Economics and S a is ics, 2023, 105 (1), 20–39.
Bai, Jie, Seema Jayachand an, Edmund J. Malesky, and Benjamin Olken, “Fi m G ow h
and Co up ion: Empi ical E idence om Vie nam,” The Economic Jou nal, 2017, 129
(618), 651–677.
Bane jee, Abhiji , Sendhil Mullaina han, and Rema Hanna, “Co up ion,” Technical
Repo , Na ional Bu eau o Economic Resea ch 2012.
Bane jee, R iwik, “On he in e p e a ion o b ibe y in a labo a o y co up ion game: mo al
ames and social no ms,” Expe imen al Economics, 2016, 19, 240–267.
Ba dhan, P anab, “Co up ion and De elopmen : A Re iew o Issues,” Jou nal o Economic
Li e a u e, 1997, 35 (3), 1320–1346.
Ba inas-Fo e o, And es, “Why should my g oup us you s? Collec i e us and us -
wo hiness unde Economic Shocks,” Documen o CEDE, 2024, (29), 1–48.
and Ca los Sca ascini, “Wage cu s, Social Dis ibu ional P e e ences and Poli ical
Co up ion,” IDB Wo king Pape , 2025, pp. 1–42.
Ba , Abigail and Danila Se a, “The e ec s o ex e nali ies and aming on b ibe y in a
pe y co up ion expe imen ,” Expe imen al Economics, 2009, 12, 488–503.
Ba o, Robe J, “The con ol o poli icians: an economic model,” Public Choice, 1973, 14,
19–42.
Belo , Michele, Raymond Duch, and Luis Mille , “A comp ehensi e compa ison o
s uden s and non-s uden s in classic expe imen al games,” Jou nal o Economic Beha io
& O ganiza ion, 2015, 113, 26–33.
Ben on, Allyson Lucinda, “Dissa is ied Democ a s o Re ospec i e Vo e s?: Economic
Ha dship, Poli ical Ins i u ions, and Vo ing Beha io in La in Ame ica,” Compa a i e
Poli ical S udies, 2005, 38 (4), 417–442.
36
Online Appendix
Co up ion and Poli ical Accoun abili y in Good
and Bad Economic Times
And es Ba inas-Fo e o Ca los Sca ascini
Con en s
A Ex ensi e Rep esen a ion Payo s 2
B P oo s 2
B.1 P oo Lemma 1 ................................... 2
B.2 P oo Lemma 2 ................................... 3
B.3 P oo Lemma 3 ................................... 4
B.4 P oo Lemma 4 ................................... 5
B.5 P oo Theo em 1 ................................... 7
B.6 P oo Co olla y 1 .................................. 8
B.7 P oo Theo em 2 ................................... 9
C Expe imen al Design 9
C.1 Pa icipan s’ Timelines ............................... 9
D Addi ional Figu es and Tables 10
E Expe imen al Sc een Ins uc ions 20
E.1 O iginal Ins uc ions in Spanish ........................... 20
1

A Ex ensi e Rep esen a ion Payo s
Rega ding ou expe imen al se ing, we ha e he ollowing ec o o public goods alloca-
ions: Hig-cos en i onmen :
Θh,Θa,Θl|π= 20EP= (3,2,1)
and Low-cos en i onmen :
Θh,Θa,Θl|π= 10EP= (6,4,2).
Poli ician’s Payo s:






U1
pol.
U2
pol.
U3
pol.
U4
pol.







=


w(1 −η) + ϕp
w+ϕp
w(1 −η) +
w+



Ci izen’s Payo s:









U1
ci i.
U2
ci i.
U3
ci i.
U4
ci i.
U5
ci i.
U6
ci i.









=









y(1 −τ) + κ(Θh)−κ
y(1 −τ) + κ(Θh)
y(1 −τ) + κ(Θa) + ((αc−Θa)λ−κ)
y(1 −τ) + κ(Θa)
y(1 −τ) + κ(Θl) + ((αc−Θl)−κ)
y(1 −τ) + κ(Θl)









B P oo s
B.1 P oo Lemma 1
P oo .
Unde a high-alloca ion
Θh
, he ci izen upda es he pos e io belie s o co up ion
ollowing,
λ≡P σ= 0|Θ(π) = Θh(π)=P Θ(π) = Θh(π)|σ= 0×P (σ= 0)
P (Θ(π) = Θh(π))(B1)
F om Figu e 2, obse e ha a highe alloca ion o public goods is only possible i he
poli ician is hones . The e o e,
P (Θ(π) = Θh(π)|σ= 0) = 0
. Upda ing
(λ)
ollowing
equa ion (B1), we ind ha
λ= 0
. Gi en his signal, he op imiza ion payo p esen ed in
Equa ion (3) can be exp essed as:
max
δ∈{0,1}y(1 −τ) + κΘh(π) + δ(−κ).
2
Finally, no e ha any s a egy
ν > 0
esul s in a lowe ma e ial payo . In ui i ely, his
occu s because, unde a high alloca ion signal, he ci izen has no eason o p o es , as
he poli ician was hones . The e o e, unde his signal, and ega dless o he cos
(π)
, he
ci izen plays:
ν(Θh(π)) = 0 ∀αc∈[0, α]
B.2 P oo Lemma 2
P oo .
Unde a low-alloca ion
Θl
, he ci izen upda es he pos e io belie s o co up ion
ollowing Equa ion (B1). Once again, om Figu e 2obse e ha a low-alloca ion o public
goods is only possible i he poli ician was co up . Mo eo e , he belie -upda ing p ocess
(λ)
esul s in he ollowing:
λ=(1−µ)×β
(1−µ)(β)= 1
. Gi en his signal and he upda ed belie s, he
op imiza ion p oblem o he ci izen can be w i en as:
max
δ∈{0,1}y(1 −τ) + κΘl(π) + δ(αc−Θl(π)) −κ
In ui i ely, he ci izen op imizes by choosing a mixed s a egy
ν
ha maximizes he
payo , which depends en i ely on:
(αc−Θl(π)) −κ
. F om Sec ion 2, ecall ha
Θl(π) =
τy −
π
. Rea anging e ms, we ind ha he op imal s a egy unde a low-alloca ion
signal is:
ν(αc,Θl(π)) =















1i αc≥κ+τy −
π
0i αc< κ +τy −
π(B2)
Mo eo e , ega ding he op imal cu o gi en a low-alloca ion signal, we ind ha :
α∗(Θl(π)) = κ+τy −
π∈(0, α]
Following he same logic, gi en an ambiguous-alloca ion
Θa
, he belie -upda ing p ocess
esul s in:
λ=µβ
µβ+(1−β)(1−µ)
. Gi en his signal, he op imiza ion p oblem o he ci izen
can be exp essed as:
max
δ∈{0,1}y(1 −τ) + κΘa(π) + δ (αc−Θa(π)) µβ
µβ + (1 −β)(1 −µ)!−κ!
3
Simila ly, he ci izen chooses a mixed s a egy
ν
ha maximizes he payo , which depends
on:
(αc−Θa(π))µβ
µβ+(1−β)(1−µ)−κ
. Gi en ha
Θa(π) = τy
π
, and ea anging e ms,
he op imal s a egy unde an ambiguous-alloca ion signal is:
ν(αc,Θa(π)) =















1i αc≥κ µβ + (1 −β)(1 −µ)
µβ !+τy
π
0i αc< κ µβ + (1 −β)(1 −µ)
µβ !+τy
π(B3)
Mo eo e , ega ding he op imal cu o unde an ambiguous alloca ion signal, we ind
ha :
α∗(Θa(π)) = (κ µβ + (1 −β)(1 −µ)
µβ !+τy
π)∈(0, α]
Gi en ha cu o poin s de e mine he ci izen’s op imal s a egy unde bo h signals and
ega dless o he cos , his p o es he lemma.
B.3 P oo Lemma 3
P oo .
Employing Lemma 2, and ocusing on a pa icula
π={π, π}
wi hou loss o
gene ali y, we can de ine he ci izen’s se o op imal cu o s
α∗(Θl,Θa) = κ+τy −
π, κ µβ + (1 −β)(1 −µ)
µβ !+τy
π!
The e o e, we compa ed hese cu o s ollowing:
κ+τy −
π
| {z }
α∗(Θl)
< κ µβ + (1 −β)(1 −µ)
µβ !+τy
π
| {z }
α∗(Θa)
Rea anging e ms,
κ−
π< κ µβ + (1 −β)(1 −µ)
µβ !
Gi en ha κ∈R+, i is possible o ew i e his las exp ession as:
1−
κπ< µβ + (1 −β)(1 −µ)
µβ !
4
Simpli ying, we ob ain
1−
κπ<1 + (1 −β)(1 −µ)
µβ !
Gi en ha
∈(0, τy)
,
κ∈R+
, and
π∈ {π, π} ⊂ R+
, he exp ession

κπ 
is always posi-
i e. On he o he hand, gi en ha
µ∈(0,1)
, and
β∈(0,1)
, he exp ession
µβ+(1−β)(1−µ)
µβ 
is always posi i e.
The e o e, he exp ession
1−
κπ
| {z }
Posi i e
<1 + (1 −β)(1 −µ)
µβ !
| {z }
Posi i e
,
is always ue. This comple es he p oo .
B.4 P oo Lemma 4
P oo .
The poli ician may conside a s a egy
(σ)
ha aligns wi h he ci izen’s s a egy
ν
and he ci izen’s upda ed belie s λ, o ming an equilib ium. Hence, once he public good
cos s a e de e mined, he poli ician compa es he expec ed payo o each possible ac ion,
conside ing he ci izens’ beha io , and chooses he highes .
S a egy 1: Expec ed payo o being Co up
I he poli ician chooses o be co up , he e is a p obabili y
(µ)
ha he na ional au ho i y
ully co e s he decision
(γ= 1)
, esul ing in he ci izen obse ing an ambiguous alloca-
ion signal
Θa
. In his case, he ci izen will punish he poli ician i equa ion (B3) holds.
Con e sely, wi h a p obabili y o
(1 −µ)
, he na ional au ho i y does no p o ide esou ces
(γ= 0)
, leading he ci izen o obse e a low alloca ion signal
Θl
, and he ci izen will punish
i equa ion (B2) holds. The e o e, he expec ed payo o being co up
(h= 0)
is gi en by
E[Up|h= 0, π]=w+ −µ[P(αc> α∗(Θa(π)))(ηw)]−(1 −µ)hPαc> α∗(Θl(π))(ηw)i
Gi en ha αc∼ [0, α], we can ew i e his exp ession as
w+ −µ 1−Zα∗(Θa(π))
0 (α)dα !(ηw)−(1 −µ) 1−Zα∗(Θl(π))
0 (α)dα !(ηw)(B4)
5
S a egy 2: Expec ed payo o being hones
I he poli ician decides o be hones , she knows ha wi h p obabili y
(µ)
, he na ional
au ho i y will p o ide addi ional esou ces
(γ= 1)
, and he ci izen obse es a high
alloca ion signal
Θh
. F om Lemma 1, we know ha unde his signal he ci izen will ne e
punish he poli ician. Con e sely, wi h p obabili y
(1 −µ)
, he na ional au ho i y does
no p o ide esou ces
(γ= 0)
, and he ci izen obse es an ambiguous signal
Θa
o public
goods. Again, unde an ambiguous signal, he ci izen will disapp o e i equa ion (B3)
holds. The e o e, he expec ed payo o being hones (h= 1) is gi en by
E[Up|h= 1, π] = w+ϕp−(1 −µ)[P(αc> α∗(Θa(π)))(ηw)]
Rew i ing his exp ession, we ob ain
E[Up|h= 1, π] = w+ϕp−(1 −µ) 1−Zα∗(Θl(π))
0 (α)dα !(ηw)(B5)
To ind an op imal s a egy
(σ)
, he poli ician conside s bo h expec ed payo s and chooses
he one ha maximizes he u ili y. Unde bo h scena ios, he poli ician ecei es he sala y
(w)
and also aces he possibili y o punishmen unde an ambiguous signal. The e o e,
she compa es he expec ed u ili y o being hones o e being co up , gi en a pa icula
public goods cos (π). This compa ison can be exp essed as
ϕp−(1 −µ) 1−Zα∗(Θa(π))
0
(α)dα !(ηw)
| {z }
Being hones
≥
−µ 1−Zα∗(Θa(π))
0
(α)dα !(ηw)−(1 −µ) 1−Zα∗(Θl(π))
0
(α)dα !(ηw)
| {z }
Being co up
(B6)
Fac o izing, we ob ain
ϕp≥ + (1 −2µ) 1−Zα∗(Θa(π))
0
(α)dα !−(1 −µ) 1−Zα∗(Θl(π))
0
(α)dα !
The e o e,
ϕ∗(π) = + (1 −2µ) 1−Zα∗(Θa(π))
0
(α)dα !−(1 −µ) 1−Zα∗(Θl(π))
0
(α)dα !(B7)
6

Finally, only es s o p o e ha
ϕ∗(π)<b
ϕ
. Fi s , gi en ha
µ∈(0,1)
beha es mono oni-
cally, we can ocus on he ex eme alue cases o µ.
Case 1: limµ→0
ϕ∗(π) = + 1−Zα∗(Θa(π))
0 (α)dα !− 1−Zα∗(Θl(π))
0 (α)dα !
F om Lemma 3, i ollows ha
Rα∗(Θl(π))
0 (α)dα ≤Rα∗(Θa(π))
0 (α)dα
, and gi en ha
< b
ϕ
,
we ob ain
ϕ∗(π) = + 1−Zα∗(Θa(π))
0 (α)dα !− 1−Zα∗(Θl(π))
0 (α)dα !< < b
ϕ
So, when limµ→0,
ϕ∗(π)<b
ϕ
Case 2: limµ→1
ϕ∗(π) = − 1−Zα∗(Θa(π))
0 (α)dα !
Gi en ha α∗(Θa(π)) >0, we ob ain
ϕ∗(π) = − 1−Zα∗(Θa(π))
0 (α)dα !< < b
ϕ
So, when limµ→1,
ϕ∗(π)<b
ϕ
This comple es he p oo .
B.5 P oo Theo em 1
P oo . Case 1: (Low alloca ion):
α∗(Θl(π)) = κ+τy −
π
De i a i e wi h espec o π:
d
dπα∗(Θl(π)) = d
dπ κ+τy −
π=−τy −
π2
7
Since
τy −
is a posi i e cons an and
π2
is always posi i e,
−τy−
π2
is nega i e. The e o e,
α∗(Θl(π))
is a dec easing unc ion o
π
. Since
α∗(Θl(π))
dec eases as
π
inc eases, i ollows
ha :
α∗(Θl(π)) ≤α∗(Θl(π))
whene e π≥π.
Case 2: (Ambiguous alloca ion):
α∗(Θa(π)) = κ µβ + (1 −β)(1 −µ)
µβ !+τy
π
De i a i e wi h espec o π:
d
dπα∗(Θa(π)) = d
dπ κ µβ + (1 −β)(1 −µ)
µβ !+τy
π!=−τy
π2
Since
τy
is a posi i e cons an and
π2
is always posi i e,
−τy
π2
is nega i e. The e o e,
α∗(Θa(π))
is a dec easing unc ion o
π
. Since
α∗(Θa(π))
dec eases as
π
inc eases, i ollows
ha :
α∗(Θa(π)) ≤α∗(Θa(π))
whene e π≥π. This comple es he p oo o Theo em 1.
B.6 P oo Co olla y 1
P oo . F om Theo em 1,α∗(Θm(π)) ≥α∗(Θm(π)), hen ea anging e ms,
Zα∗(Θm(π))
0 (α)dα +Zα∗(Θm(π))
α∗(Θ)m(π) (α)dα ≥Zα∗(Θm(π))
0 (α)dα
Zα∗(Θm(π))
α∗(Θ)m(π) (α)dα ≥0
This comple es he p oo o Co olla y 1.
8
B.7 P oo Theo em 2
P oo . F om Equa ion (B7), we ha e ha
ϕ∗(π) = + (1 −2µ) 1−Zα∗(Θa(π))
0 (α)dα !−(1 −µ) 1−Zα∗(Θl(π))
0 (α)dα !
F om Co olla y 1, we can eadily see ha
1−Zα∗(Θm(π))
0 (α)dα !≤ 1−Zα∗(Θm(π))
0 (α)dα !∀m={a, l}
Then, i ollows ha
ϕ∗(π)≤ϕ∗(π)
C Expe imen al Design
C.1 Pa icipan s’ Timelines
Figu e C1: Expe imen al Timeline in PCG ac oss Pa icipan s Types
Compu e Poli ician Poli ician Compu e
Cos o Policy Belie s Resou ces and
public goods choice o punishmen public goods alloca ion
πhE[δ]γ,Θ
Compu e Ci izen Compu e Ci izen
Cos o Belie s Resou ces an App o al
public goods o co up ion public goods alloca ion choice
πβγ,Θδ
No es: The igu e illus a es he expe imen al imeline ac oss di e en pa icipan ypes. Belie s o punishmen e e s o he poli ician’s
esponse o he ques ion: "Do you hink he ci izen would disapp o e o you?" Belie s o co up ion e e s o he ci izen’s esponse o
he ques ion: "How many esou ces do you hink he poli ician would alloca e?"
9
D Addi ional Figu es and Tables
Figu e D1: Pe cep ion o Numbe o Poli icians In ol ed in Co up ion ac oss Na ional
Economic Si ua ion Pe cep ion.
2
2.5
3
3.5
4
4.5
Numbe o poli icians in ol ed in co up ion (1 None - 5 All)
Wo se han
be o e Same as
be o e
Be e han
be o e
Economic si ua ion o he coun y in las 12 mon hs (1 Wo se - 3 Be e )
A gen ina
Bahamas
Belize
Boli ia
B azil
Canada
Chile
3
3.5
4
4.5
Numbe o poli icians in ol ed in co up ion (1 None - 5 All)
Wo se han
be o e Same as
be o e
Be e han
be o e
Economic si ua ion o he coun y in las 12 mon hs (1 Wo se - 3 Be e )
Hai i
Hondu as
Jamaica
Mexico
Nica agua
Panama
No es: The igu e depic s he sca e plo binned be ween he pe cep ion o he numbe o poli icians in ol ed in co up ion and he
pe cep ion o he coun y’s economic si ua ion o e he pas 12 mon hs. Sca e binned es ima ion o all esponden s o he LAPOP
ques ion in 26 coun ies be ween 2016 and 2023. The da k line indica es he linea i . G ay dashed lines ep esen he 95% obus
con idence in e als. The analysis includes coun y and yea ixed e ec s. S a is ical signi icance le els:
∗∗∗ p < 0.01,∗∗ p < 0.05,∗p <
0.1, n.s p > 0.1.
10
Figu e D7: Dis ibu ion o T ea ed Rounds by Pa icipan Type (PCG)
0
5
10
15
20
25
30
35
Pe cen age %
None 1234All
N° o ounds wi h low cos (by pa icipan )
Ci izens (C) Poli icians (P)
No es: The igu e illus a es he pe cen age dis ibu ion o he numbe o ounds in which ci izens disapp o e o poli icians du ing he
game. The his og am ep esen s di e en ca ego ies o poli ical accoun abili y beha io , anging om "Ne e punish" (0 ounds o
disapp o al) o "Always punish" (5 ounds o poli ical disapp o al).
17

Figu e D8: Dis ibu ion o To al Co up ion in he Poli ical Co up ion Game (PCG)
0.18
0.18
0.27
0.09
0.27
0.10
0.13
0.17
0.23
0.12
0.25
0.11
0.17
0.20
0.18
0.09
0.25
0.16
0.07
0.22
0.21
0.13
0.21
0.12
0.13
0.15
0.32
0.12
0.17
0.000.00
0.30
0.20
0.20
0.30
0
20
40
60
80
100
Pe cen age %
0 1 2 3 4 5
N° o ounds wi h low cos (by pa icipan )
Co up = 0 imes
Co up = 1 imes
Co up = 2 imes
Co up = 3 imes
Co up = 4 imes
Co up = 5 imes
No es: We p esen he dis ibu ion o o al numbe o imes being co up ac oss he o al numbe o ounds ha he poli ician ace a low
cos . No ice ha "Always hones " poli icians ace se e al ounds wi h low cos se ings.
Figu e D9: Dis ibu ion o To al Punishmen in he Poli ical Co up ion Game (PCG)
0.60
0.25
0.05
0.10
0.50
0.23
0.20
0.07
0.50
0.28
0.18
0.03
0.49
0.29
0.15
0.06
0.53
0.27
0.16
0.01
0.70
0.20
0.10
0
20
40
60
80
100
Pe cen age %
012345
N° o ounds wi h low cos (by pa icipan )
Punish = 0 imes
Punish = 1 imes
Punish = 2 imes
Punish = 3 imes
No es: We p esen he dis ibu ion o o al numbe o punishmen ounds ac oss he o al numbe o ounds ha he ci izen aces a low
cos . No ice ha "Ne e punish" ci izens ace se e al ounds wi h low cos se ings.
18
Figu e D10: Willingness o Punish by LAPOP Ques ion, by Pa icipan Type
0
10
20
30
40
Pe cen
0 2 4 6 8 10
Willing o p o es o be e sala ies
0
20
40
60
Pe cen
0 2 4 6 8 10
Willing o p o es o be e heal h
0
10
20
30
40
Pe cen
0 2 4 6 8 10
Willing o p o es agains clima e change
0
10
20
30
40
Pe cen
0 2 4 6 8 10
Willing o p o es o democ acy igh s
0
10
20
30
40
Pe cen
0 2 4 6 8 10
Willing o p o es agains co up ion
0
10
20
30
40
Pe cen
0 2 4 6 8 10
Willing o p o es agains inequali y
Ci izens (C) Poli icians (P)
No es: The igu e illus a es he willingness o ci izens (in g ay) and poli icians (in ed) o p o es o a ious causes, including be e sala ies, heal h, clima e change, democ acy igh s,
co up ion, and inequali y. The his og am ep esen s he willingness o p o es , sco ed om 0 o 10.
19
E Expe imen al Sc een Ins uc ions
E.1 O iginal Ins uc ions in Spanish
Figu e E1: Gene al ins uc ions
20
Figu e E2: In o med Consen
Figu e E3: Welcome
21
Figu e E4: Ini ial Ins uc ions
Figu e E5: Ins uc ions - 1
22

Figu e E6: Ins uc ions - 2
23
Figu e E7: Round s ages
24
Figu e E8: Round s ages - 2
25
Figu e E9: Paymen
Figu e E10: Con ol Ques ion A - 1
26
Figu e E19: End o he ound
Figu e E20: New ound sc een
33

Figu e E21: Ci izen’s Social No ms Sc een
Figu e E22: Poli ician’s Social No ms Sc een
34
Figu e E23: Final Resul s
35
Figu e E24: Final Resul s - 2
36
Figu e E25: Ques ionnai e
37
38

39
40
41