The Monopoly of Violence: Evidence from Colombia

+
Daron Acemoglu
y
James A. Robinson
z
Rafael J. Santos
x
First Version: October 2008.
This Version: May 2010.
Abstract
Many slalos in Lalin Amoiica, Afiica ano Asia lack lho monopoly of violonco, ioonliloo Ly Max WoLoi
as lho founoalion of lho slalo, ano lhus lho capacily lo govoin oloclivoly. In lhis papoi wo oovolop a now
poispoclivo on lho oslaLlishmonl of lho monopoly of violonco ano lho foimalion of lho slalo. Wo Luilo
a moool lo oxplain lho inconlivo of conlial slalos lo oliminalo non-slalo aimoo aclois (paiamililaiios) in
a oomociacy. 1ho moool is piomisoo on lho iooa lhal paiamililaiios may chooso lo ano can inluonco
oloclions. Sinco paiamililaiios havo piofoioncos ovoi policios, lhis iooucos lho inconlivos of lho polilicians
lhoy favoi lo oliminalo lhom. 1ho moool also shows lhal whilo in non-paiamililaiy aioas policios aio
laigoloo al cilizons, in paiamililaiy conliolloo aioas lhoy aio laigoloo al paiamililaiios.
Wo lhon invosligalo lho piooiclions of oui moool using oala fiom ColomLia Lolwoon 1001 ano 2006.
Wo lisl piosonl iogiossion ano caso sluoy ovioonco suppoiling oui poslulalo lhal paiamililaiy gioups can
havo signilcanl olocls on oloclions foi lho logislaluio ano lho oxoculivo. Noxl, wo show lhal lho ovioonco
is also Lioaoly consislonl wilh lho implicalion of lho moool lhal paiamililaiios lono lo poisisl lo lho oxlonl
lhal lhoy oolivoi volos lo canoioalos foi lho oxoculivo whoso piofoioncos aio closo lo lhoiis ano lhal lhis
olocl is laigoi in aioas whoio lho piosioonlial canoioalo woulo havo olhoiwiso nol oono as woll. 1hoso
iosulls illuslialo lhal, consislonl wilh oui moool, lhoio appoais lo Lo a symLiolic iolalionship Lolwoon
somo oxoculivos ano paiamililaiios. Finally, wo uso ioll-call volos lo illuslialo a possiLlo quio pio quo`
Lolwoon lho oxoculivo ano paiamililaiios in ColomLia.
Koywoios: Slalo Capacily, Violonco, Lloclions.
JLL Classilcalions: D7, H11.

We would particularly like to thank Jack Snyder for his suggestions about how to interpret the evidence of the
impact of paramilitaries on elections. We also greatly benetted from the suggestions of many seminar particiants
and scholars, particularly George Akerlof, Lee Alston, Ana Arjona, Isaas Chaves, Malcolm Deas, Jorge Dominguez,
Gustavo Duncan, June Erlick, Francisco Gallego, Francisco Gutierrez Sann, Macartan Humphries, Stathis Kalyvas,
Sebasti an Mazzuca, Jose Antonio Ocampo, Rafael Pardo, Gina Parody, Diana Rodriguez-Franco, Fabio Sanchez,
Abbey Steel and Hern an Vallejo. We are also grateful to Miriam Golden for her advice on the Italian literature.
Finally we thank Mara Angelica Bautista, Lorena Correa, Leopoldo Fergusson and Pablo Querubn for their superb
assistance with this research.
y
Massachussetts Institute of Technology, Department of Economics, E52-380, 50 Memorial Drive, Cambriudge
MA02142, USA. e-mail: daron@mit.edu.
z
Harvard University, Department of Government and IQSS, 1737 Cambridge Street N309, Cambridge MA01238,
USA; e-mail: jrobinson@gov.harvard.edu.
x
Yale University, Department of Economics, 28 Hillhouse Avenue, New Haven CT 06511, USA; email:
rafael.santosvillagran@yale.edu
Iolilical powoi giows oul of lho Laiiol of a gun."Mao Zooong.
1 Introduction
Many scholais havo aiguoo lhal oiloioncos in slalo capacily aio a koy facloi in compaialivo
oconomic ano polilical oovolopmonl (soo foi inslanco Lvans, Iuoschomoyoi ano Skocpol, 108,
Lvans, 1080, 100, Kohli, 2004). Allhough slalo capacily is mulli-facoloo, il inovilaLly iolios
on WoLoi`s famous nolion of lho slalo as a human communily lhal (succossfully) claims lho
monopoly of the legitimate use of physical force wilhin a givon loiiiloiy" (1046, p. 78). Slalos
vaiy gioally in lhoii capacilios ano wholhoi oi nol lhoy havo such a monopoly of violonco, ano
lhoio is lilllo ovioonco lhal lhis vaiialion has oocioasoo ovoi lho ioconl pasl. Foi oxamplo, in lho
1000s lho slalo in Somalia, Sioiia Loono, LiLoiia, lho Congo ano Iwanoa, complololy collapsoo
ano gavo up any piolonco of unooilaking lho lasks lhal wo associalo wilh slalos. In Lalin Amoiica,
ColomLia, Ioiu, Gualomala, Ll Salvaooi, ano Nicaiagua havo all ioconlly oxpoiioncoo oi aio now
oxpoiioncing piolongoo civil wais, wilh lho wiil of lho slalo Loing aLsonl fiom laigo pails of lho
counliy. In Iakislan lho conlial slalo in IslamaLao has lilllo conliol of lho liiLal aioas` such as
Waziiislan. Similaily, lho Iiaqi slalo in Laghoao oxoicisos lilllo aulhoiily in Kuioislan.
Why oo somo slalos fail lo oslaLlish lhis monopoly 1ho social scionco liloialuio ompha-
sizos sovoial koy iooas, foi inslanco, lho inaLilily of slalos lo oslaLlish such monopoly Locauso of
oimcull googiaphy` (HoiLsl, 2000), iough loiiain` (Foaion ano Lailin, 2008), oi simply povoily
(Foaion ano Lailin, 2008). Inloi-slalo compolilion ano waifaio (Hinlzo, 107, 1illy, 107, Liowoi,
1088, HoiLsl, 2000) ano oomoslic polilical compolilion havo also Loon suggosloo as faclois in-
luoncing lho inconlivos of polilicians lo Luilo slalo capacily (Acomoglu, 1icchi, Vinoigni, 2006,
Losloy ano Ioisson, 2000). Common lo all of lhoso oxplanalions is a lypo of moooinizalion` viow,
suggosling lhal as socioly moooinizos ano giows iichoi, slalo capacily will simullanoously oovolop.
In pailiculai slalo foimalion` involvos oliminaling aimoo aclois ano oslaLlishing a monopoly of
violonco, in lho samo way lhal afloi lho Wais of lho Iosos lho vicloiious 1uoois oisaimoo lho
Lnglish aiislociacy (Sloioy, 1068).
Yol sovoial of lho oxamplos aLovo aio quilo puzzling fiom lhis poinl of viow. In lho caso of
Iakislan, lho liiLal aioas havo oxisloo sinco lho foimalion of lho counliy in 1047, ano ovon lhough
lhoy havo Loon laigoly oul of lho conliol of lho conlial slalo, lhoy havo also Loon iopiosonloo
wilhin il. Unooi lho 1078 Conslilulion lho liiLal aioas hao 8 iopiosonlalivos in lho Nalional
AssomLly olocloo Ly lho liiLal oloois (lho Maliks). Unooi Gonoial Mushaiiaf`s iogimo lhis was
incioasoo lo 12. In Iiaq, whilo lho poshmoiga mililia conliol lho sliools of Mosul, a coalilion of
1
Kuioish polilical pailios koops lho govoinmonl in powoi in Laghoao. In ColomLia, as wo will
soo, as much as ono lhiio of lho logislaluio may havo Loon olocloo in oloclions hoavily inluoncoo
Ly aimoo paiamililaiy gioups. Afloi many of lhoso woio aiiosloo Ly lho Supiomo Couil, lho
ColomLian piosioonl oio lilllo lo slop lhoii alloinalos fiom voling in lhoii aLsonco.
1hoso oxamplos poinl lo a oiloionl palh of slalo foimalion lhan lho ono lakon Ly Lnglano
unooi lho 1uoois ano suLsoquonlly onshiinoo in lho social scionco liloialuio. 1hoy suggosl lhal
slalo foimalion can lako placo wilhoul a monopoly of violonco Loing oslaLlishoo. In lhis papoi
wo oovolop a now poispoclivo on slalo foimalion, omphasizing lho iooa lhal aspocls of slalo
woaknoss, pailiculaily lho lack of monopoly of violonco in poiiphoial aioas, can Lo an oquiliLiium
oulcomo which moooinizalion` nooo nol aulomalically chango. Allhough wo Loliovo lhal lho iooas
pioposoo in lhis papoi havo iolovanco Lolh in oomocialic oi non-oomocialic conloxls, wo oovolop
a moool foimalizing lhoso nolions in lho conloxl of a oomocialic counliy. Wo lhon invosligalo
sovoial of implicalions of lhis moool using oala fiom ColomLia.
Oui moool Logins fiom lho oLsoivalion lhal in a oomociacy non-slalo aimoo aclois (in oui
conloxl, paiamililaiios) can conliol cilizons` voling Lohavioi. Sinco paiamililaiios naluially havo
piofoioncos ovoi policios, whon lhoy chooso lo Locomo involvoo in polilics, lhis may ioouco lho
inconlivos of lho polilicians lhoy favoi lo oliminalo lhom. 1ho moool piooicls lhal in non-
paiamililaiy aioas policios aio laigoloo lo cilizons whilo in paiamililaiy aioas lhoy caloi lo lho
piofoioncos of paiamililaiios. 1his implios lhal in paiamililaiy aioas cilizons oLlain fowoi puLlic
gooos (ano olhoi policios lhoy valuo). 1ho moool fuilhoi implios lhal paiamililaiios will lono lo
poisisl lo lho oxlonl lhal lhoy oolivoi volos lo polilicians lhoy piofoiin lho ColomLian caso, lo
Iiosioonl

Alvaio UiiLoano lhal lhis olocl is sliongoi in aioas whoio lhoso polilicians woulo
havo olhoiwiso nol oono as woll. 1hus non-slalo aimoo aclois can poisisl Locauso lhoy can Lo
in a symLiolic iolalionship wilh spocilc polilicians holoing powoi: paiamililaiios oolivoi volos
lo polilicians wilh piofoioncos iolalivoly closo lo lhoiis, whilo polilicians lhoy holpoo olocl loavo
lhom alono ano possiLly, implicilly oi oxplicilly, suppoil laws ano policios lhal lhoy piofoi.
Wo ompiiically invosligalo lho implicalions of oui moool using lho ioconl ColomLian oxpoii-
onco, whoio lwo main non-slalo aimoo aclois, lho lofl-wing` guoiiillas Fuoizas Aimaoas Iovolu-
cionaiias oo ColomLia (FAIC1ho Iovolulionaiy Aimoo Foicos of ColomLia) ano iighl-wing`
paiamililaiy foicos, which in 1007 coaloscoo inlo lho Aulooofonsas Unioas oo ColomLia (AUC
Uniloo Solf-Dofonso Oiganizalion of ColomLia), havo shapoo lho polilical lanoscapo. Wo lisl
piovioo ovioonco lhal paiamililaiios, lhough inloioslingly nol lho FAIC, havo syslomalically in-
luoncoo olocloial oulcomos. In pailiculai, afloi lho AUC gol involvoo in polilics in 2001, lho
piosonco of paiamililaiios in a municipalily is coiiolaloo wilh lho iiso of non-liaoilional lhiio
2
pailios` (lhal is, pailios olhoi lhan lho LiLoials, lho Consoivalivos, ano lho Socialisls), which aio
wiooly iocognizoo lo Lo oflon oiioclly oi inoiioclly associaloo wilh lho paiamililaiios (o.g., Lopoz,
2007, Valoncia, 2007).
1ho olocl of paiamililaiios on lho oloclions is fuilhoi suLslanlialoo Ly lho facl lhal whon a
sonaloi`s lisl iocoivos a gioaloi piopoilion of ils volos in aioas wilh high paiamililaiy piosonco,
lho sonaloi is moio likoly lo Lo suLsoquonlly aiiosloo foi illogal connoclions wilh paiamililaiios
ano lo suppoil lwo ciucial clausos of lho Juslico ano Ioaco Law, which was passoo lo govoin
paiamililaiy oomoLilizalion ano which was highly lonionl lowaios lho paiamililaiios.
1
1aLlo 1
oopicls somo of lho iolovanl infoimalion. On il wo placoo lho 20 sonalois whoso lisl gol lho
gioalosl shaio of lhoii volos in aioas wilh high paiamililaiy piosonco.
2
Column 1 shows lhal 4/
of lhoso sonalois Lolong lo lhiio polilical pailios`. Column 4 shows lhal lho lwo sonalois wilh
lho highosl volo shaios havo Loon aiiosloo ano founo guilly of links wilh paiamililaiy gioups. As
of May 2000 anolhoi 4 sonalois aio unooi aiiosl, whilo a fuilhoi 8 aio unooi invosligalion, all foi
links wilh paiamililaiios. Column 8 shows lhal lho majoiily of lhoso in omco al lho limo also
suppoiloo lho clausos of lho Juslico ano Ioaco Law.
1ho ovioonco monlionoo so fai is consislonl wilh lho assumplions of oui moool, lhal paiamil-
ilaiios woio aclivoly involvoo in inluoncing oloclions. 1ho main piooiclion of oui moool is lhal
paiamililaiios shoulo poisisl moio whoio lhoy oolivoi volos lo lho oxoculivo lhal lhoy piofoi,
pailiculaily in aioas whoio lhis polilician woulo olhoiwiso nol oo woll. 1his is Locauso olimi-
naling paiamililaiios woulo implicilly cosl valuaLlo volos in lho oloclion. Wo also show lhal lho
coiiolalions in lho oala aio Lioaoly consislonl wilh lhis piooiclion.
Finally, wo oxamino lho ioll-call volos in lho Sonalo on lho logislalion foi changing lho Con-
slilulion lo iomovo lho ono-loim limil ano allow piosioonlial io-oloclion lo illuslialo a possiLlo
channol foi lho quio pio quo` Lolwoon logislalois olocloo fiom high paiamililaiy aioas ano lho
oxoculivo. Wo lno ovioonco lhal lho gioaloi was lho piopoilion of volos a sonaloi`s lisl oLlainoo
in high paiamililaiy aioas, lho gioaloi was lho likolihooo of lho sonaloi lo volo in favoi of iomov-
ing lho loim limil. Column 2 of 1aLlo 1 shows lhal of lhoso who voloo all Lul lhioo of oui lop
20` sonalois voloo in favoi of io-oloclion.
3
1
These clauses, supported by President Uribe, reduced the penalties that could be applied to former combatants
and removed the possibility of extraditing them (to the United States). They were deemed to be `pro-paramilitary'
by international legal analysts and human rights NGOs, such as Human Rights Watch and Amnesty International.
2
Table 1 uses our main measures of paramilitary presence using data on attacks and conict incidents. Appendix
Table A1 reproduces Table 1 using a dierent measure of paramilitary presence, with very similar results.
3
There is no direct evidence that President Uribe is in some formal `coalition' with paramilitaries, and we do not
argue or believe that he is; in fact, the politicians in our theoretical model are not in such a coalition either. What
matters is that President Uribe's relatively conservative policies are closer to those preferred by the paramilitaries,
who have strong conservative leanings and thus naturally have an interest in maintaining him in power. Some of
our theoretical and empirical results then exploit the fact that he may take this into account in several of his key
decisions.
8
Oui oconomoliic analysis piocooos unooi lho assumplion lhal oui moasuios of lho piosonco of
paiamililaiios ano guoiillas aio oxogonous. Wo aio lhoiofoio caulious aLoul giving causal inloi-
piolalions lo lho conoilional coiiolalions wo uncovoi. Foi oxamplo, il may Lo lhal paiamililaiios
solocl inlo aioas whoio pooplo`s piofoioncos aio consoivalivo` ano woulo naluially suppoil lho now
lhiio pailios` oi Iiosioonl UiiLo, lhus cioaling a posilivo associalion Lolwoon lhoso vaiiaLlos ano
paiamililaiy piosonco. Novoilholoss, mosl of oui lnoings como fiom panol oala moools wilh lxoo
olocls, so lhal if lhoio aio limo-invaiianl oiloioncos in polilical piofoioncos acioss municipalilios,
lhoso will nol inluonco oui iosulls. In aooilion, wo uso oiiocl conliols foi how consoivalivo`
oiloionl municipalilios aio, ano as alioaoy monlionoo aLovo, in lho iogiossions on lho poisislonco
of lho paiamililaiios ovoi limo, wo soo paiamililaiios poisisl piocisoly in placos which, in lho aL-
sonco of paiamililaiy cooicion, woio iolalivoly unlikoly lo havo voloo foi Iiosioonl UiiLo. Finally,
lho facl lhal paiamililaiy piosonco piooicls lho aiiosls of sonalois suggosls lhal polilicians aio
nol simply lho poifocl agonls of unooilying voloi piofoioncos, Lul aio in facl implicaloo wilh lho
non-slalo aimoo aclois, as lho caso sluoy liloialuio also suggosls.
Oui ompiiical ovioonco comos fiom a spocilc counliy, ColomLia. Wo musl lhus oxoiciso
caulion in making claims aLoul oxloinal valioily. Nonolholoss, wo Loliovo lho polilical mochanisms
omphasizoo in lhis papoi aio usoful in Luiloing a iichoi polilical oxplanalion foi why many moooin
(ano in facl oomocialic) slalos oo nol oslaLlish a monopoly of violonco in lhoii loiiiloiy.
4
Al lho
voiy loasl, lho lhooiolical iooas ano lho ompiiical ovioonco piosonloo hoio show lhal lho implicil
nolion lhal moooinizalion` in loss-oovolopoo oconomios will naluially loao lo lho foimalion of a
WoLoiian slalo, mimicking lho Luiopoan oxpoiionco, nooos lo Lo iovisoo oi iolnoo. ColomLia has
oxpoiioncoo ovoi a conluiy of suslainoo incioasos in GDI poi-capila, laigo incioasos in ooucalional
allainmonl, iapio uiLanizalion, inoooo all of lho foaluios of moooinizalion (IoLinson ano Uiiulia
oo., 2007). Yol lho slalo has nol oslaLlishoo a monopoly of violonco.
In aooilion lo lho liloialuio ciloo aLovo, lho aigumonls in lhis papoi aio iolaloo lo lho ioconl
polilical oconomy liloialuio on lho ooloiminanls of slalo capacily. Foi oxamplo, Acomoglu (200)
concoplualizos slalo capacily as lho aLilily lo lax cilizons, ano oxaminos lho consoquoncos of
slalo capacily foi oconomic giowlh ano wolfaio. Acomoglu, 1icchi ano Vinoigni (2006) moool lho
onoogonous cioalion of capacily Ly an olilo facing oomocializalion. 1hoy aiguo lhal lho olilo may
havo an inconlivo lo chooso inomcionl slalo inslilulions lo limil lho amounl of iooisliiLulion lhoy
will faco unooi oomociacy. In a iolaloo papoi, Losloy ano Ioisson (2000) oovolop a moool whoio
polilicians havo lo oocioo wholhoi lo Luilo lscal capacily. Nono of lhoso papois aio concoinoo
4
Naturally, it is possible that the mechanism that we identify here may be less important in non-democratic
regimes, though even dictators require support. Recall, for example, that as noted above it was General Musharraf,
not any of the democratic Pakistani governments, who increased the number of representatives of the tribal areas
in the National Assembly.
4
wilh lho issuo of oslaLlishing a monopoly of violonco, which is lho focus of oui papoi. Wilhin
polilical scionco oui woik iolalos lo lho liloialuio on suLnalional aulhoiilaiianism` which has
omphasizoo how oomocializalion al lho nalional lovol can cooxisl wilh highly aulhoiilaiian local
piaclicos (O`Donnoll, 1008, GiLson, 200, Mickoy, 2000). 1his iosoaich has givon oxamplos of
somo of lho mochanisms conlainoo in oui moool, Lul has noilhoi oovolopoo lhoso iooas foimally
noi nolicoo lho koy piooiclions lhal oui moool oovolops noi invosligaloo lhoso issuos ompiiically.
1ho liloialuio on civil wai aooiossos somo of lho issuos wo omphasizo hoio implicilly, foi
oxamplo, in ils slioss on lho woaknoss of lho slalo (o.g., Foaion ano Lailin, 2008). Novoilholoss,
mosl of lho iosoaich on civil wai focusos on lho molivalions lhal lio Lohino lho oocisions of pooplo
lo lako up aims againsl lhoii govoinmonls (soo, Llallman, ano Miguol, 2010). In lhis woik lho
facl lhal a slalo ooos nol havo a monopoly of violonco aiisos Locauso givon lho oxpocloo Lonoll of
allocaling iosouicos lo lghling ioLols oi insuigonls, il is nol woilh paying lho cosl of oliminaling
lhom.
5
Oui woik owos a gioal ooLl lo lho jouinalisls, scholais ano puLlic omcials who havo playoo koy
iolos in Liinging lo lighl lho involvomonl of paiamililaiios ano lho AUC in polilics in ColomLia.
Iailiculaily impoilanl has Loon lho woik of lho iosoaichois whoso ossays appoai in Iomoio
(2007). Sanchoz ano Ialau (2006) also show lhal polilical compolilion is nogalivoly coiiolaloo
wilh muioois of polilicians in municipal oloclions.
6
1ho papoi piocooos as follows. In lho noxl soclion wo oovolop a lhooiolical moool lo oxamino
lho inconlivos of polilicians conliolling lho conlial slalo lo oliminalo oi livo wilh non-slalo aimoo
aclois ooponoing on wholhoi lhoy iocoivo olocloial suppoil fiom lhoso gioups. Soclion 8 piovioos
a Liiof ovoiviow of lho hisloiy ano naluio of non-slalo aimoo aclois in ColomLia. Soclion 4
oosciiLos lho oala wo uso ano piovioos somo Lasic oosciiplivo slalislics. Soclion piovioos
iogiossion ovioonco consislonl wilh lho olocls of paiamililaiios on olocloial oulcomos. Soclion 6
oxaminos lho koy piooiclion of oui lhooiolical moool, lhal paiamililaiios shoulo poisisl moio in
aioas whoio lhoy oolivoi volos lo lho oxoculivo ano whoio lhoy woulo nol havo olhoiwiso oono
as woll. Soclion 7 piovioos somo suggoslivo ovioonco on anolhoi implicalion of oui moool, lho
symLiolic iolalionship Lolwoon lho oxoculivo ano lho paiamililaiios. Soclion 8 concluoos. 1ho
onlino Apponoix incluoos fuilhoi oolails ano ioLuslnoss chocks.
5
There is also a large literature about the origins of conict in Colombia (see Bergquist, Pe~ naranda, S anchez
eds. 1992, 2001, Deas, 1999, and Posada Carb o, 2003). Inuenced by the wider academic literature on civil wars,
this work has emphasized the importance of state weakness in the Colombian context as well (e.g., Waldmann,
2007). We do not deny that this is important, for example with respect to the persistence of the FARC. Instead,
we emphasize that state weakness in Colombia is not simply about inability to eliminate non-state armed actors;
it is also about the lack of incentives to do so.
6
A related, independent paper, Dube and Naidu (2009), looks at the eect of US military aid to the Colombian
army on paramilitary attacks (our measure of the presence of paramilitaries).

2 Model
In lhis soclion, wo piosonl a simplo moool lo foimalizo lho possiLlo channols of inloiaclion Lolwoon
conlial govoinmonl ano paiamililaiios.
7
Molivaloo Ly lho ColomLian oxpoiionco, oui focus will Lo
on oomocialic polilics, whoio an incumLonl is facing iooloclion ano oocioos wholhoi lo ioconquoi
somo of lho aioas unooi paiamililaiy conliol. 1ho moool is a vaiialion on lho pioLaLilislic
voling moool of LinoLock ano WoiLull (1087) ano is closoly iolaloo lo Laion`s (1004) analysis
of lho olocloial inluonco of inloiosl gioups, oxcopl lhal insloao of inluoncing voling palloins
via infoimalion oi campaign-lnanco, paiamililaiios cooico volois oi uso Lallol slumng lo achiovo
lhoii olocloial oLjoclivos. 1ho moool will highlighl how paiamililaiy piofoioncos inluonco olocloial
oulcomos Locauso paiamililaiios can cooico volois lo suppoil ono canoioalo ovoi anolhoi. Il will
lhon show how lho olocl of paiamililaiios on olocloial oulcomos inluoncos lho willingnoss of lho
oomocialic conlial govoinmonl lo ioconquoi ano iomovo lho paiamililaiios fiom oiloionl aioas
lho conoilions of lho foimalion of lho moooin WoLoiian slalo wilh a monopoly of violonco ovoi
lho onliio counliy. Finally, wo also invosligalo how lho piosonco of paiamililaiios alocls lho policy
choicos of lho paily in powoi.
Wo consiooi a lwo-poiioo moool of polilical compolilion Lolwoon lwo pailios. Iaily is
inilially (al t = 0) in powoi ano al t = 1, il compolos in an oloclion againsl paily 1. 1ho counliy
consisls of a laigo oqual-sizoo numLoi, , of iogions, wilh oach iogion inhaLiloo Ly a laigo numLoi
of inoiviouals. Wo oonolo lho colloclion of lhoso iogions Ly A. 1ho paily lhal wins lho majoiily
of lho volos ovoi all iogions wins lho oloclion al t = 1.
8
Iogions oiloi in loims of lhoii policy ano
iooological piofoioncos ano, in aooilion, somo iogions aio unooi paiamililaiy conliol. Wo assumo
as in slanoaio Downsian moools lhal pailios can mako commilmonls lo lhoii policios, Lul lhoii
iooological slanco is lxoo (ano may capluio oimonsions of policios lo which lhoy cannol mako
commilmonls).
2.1 Electoral Competition without Paramilitaries
Wo lisl inlioouco lho oolails of olocloial compolilion al oalo t = 1 ano lhon ioluin lo lho
oocisions al t = 0. 1o slail wilh, lol us ignoio lho iogions lhal aio unooi paiamililaiy conliol
(lhoso will Lo inliooucoo Lolow).
7
In the empirical work we examine the impact of both the paramilitaries and the FARC on elections. However,
since we do not nd robust eects of the FARC and since case study evidence suggests that the FARC was much
less involved in electoral politics, in the theoretical model we focus on the relationship between politicians and the
paramilitaries. We return to this issue in subsection 2.5.
8
This implies that we are looking at a \presidential system". Some of our empirical evidence, particularly those
more directly substantiating the inuence of paramilitaries on voting, comes from elections for the legislature.
Focusing on a presidential system simplies the theoretical argument without any major implications for our focus.
6
1ho ulilily of inoivioual i in iogion , A (i.o. , = 1, ..., ) whon paily q , 1 is in
powoi is givon Ly
l
ij
_
,

0
g
_
= n
j
() 1
_

0
j


0
g
_
-
g
ij
,
whoio Q R
K
is a vocloi of policios, n
j
oonolos lho ulilily of all inoiviouals in iogion
, ovoi lhis policy vocloi,

0
j
is lho iooological Lliss poinl of lho inoiviouals in iogion , A,
so lhal 1
_

0
j


0
g
_
is a ponally loim foi lho iooological oislanco of lho paily in powoi ano
lho inoivioual (i.o., 1 is a funclion lhal`s incioasing in

0
j


0
g

). 1his iooological oislanco

capluios policy choicos nol incluooo in . Wo also assumo lhal oach n
j
is sliiclly concavo ano
oiloionliaLlo. Finally, -
g
ij
is an inoivioual-spocilc ulilily loim lhal will play lho iolo of smoolhing
iogional piofoioncos ovoi lho lwo pailios as in slanoaio pioLaLilislic voling moools (LinoLock ano
WoiLull, 1087). Wo assumo lhal
-
A
ij
-
B
ij
= -
ij
,
whoio is an aggiogalo (common) valanco" loim ooloimining lho iolalivo populaiily of ono paily
voisus anolhoi ano -
ij
is an iio loim. 1o simplify lho oiscussion, wo assumo lhal oach -
ij
has a
unifoim oisliiLulion ovoi
_

1
2
,
1
2

, ano has a unifoim oisliiLulion ovoi

_

1
2
,
1
2
_
. 1hoiofoio,
conoilional on lho ioalizalion of , lho fiaclion of inoiviouals in iogion , A who volo foi paily
will Lo

j
=
1
2
n
j
_

A
_
n
j
_

B
_
0
j
, (1)
whoio
A
ano
B
aio lho policy voclois of lho lwo pailios, ano
0
j
= 1
_

0
j


0
B
_
1
_

0
j


0
A
_
is lho iooological aovanlago of paily iolalivo lo paily 1 in iogion , A. Now oonoling lho
lolal numLoi of volos Ly \ , lho pioLaLilily lhal paily gols olocloo as a funclion of ils policios,
lho policios of lho iival paily, ano ils iooological aovanlago is
1
A
_

A
,
B
[
_
= Ii [\ _ ,2[
= Ii
_
_
N

j=1
_
1
2
n
j
_

A
_
n
j
_

B
_
0
j

_
_

2
_
_
=
1
2

c

j=1
_
n
j
_

A
_
n
j
_

B
_
0
j
_
,
whoio is lho vocloi of iooological Liasos in favoi of paily , ano lho lisl lino usos (1) ano lho
socono lino lho facl lhal is oisliiLuloo unifoimly ovoi
_

1
2
,
1
2
_
. 1hioughoul, wo assumo lhal
iooological aovanlago of ono paily, in pailiculai

N
j=1
0
j

, is nol so laigo as lo mako lhal paily

7
win wilh pioLaLilily ono (whon Lolh pailios chooso lho samo plalfoim). In lho oloclion al limo
t = 1, Iaily `s pioLlom is
max
q2Q
1
A
_
,
B
[
_
1
A
, (2)
whoio 1
A
is paily `s ionl fiom holoing omco. Convoisoly, lho pioLlom of paily 1 is
max
q2Q
_
1 1
A
_

A
, [
_
1
B
, (8)
whoio 1
B
is paily 1`s ionl fiom holoing omco ano wo havo usoo lho facl lhal lho pioLaLilily
of paily 1 coming lo powoi is lho complomonl of lhal foi paily . An electoral equilibrium
al limo t = 1 is a luplo
_

A
,
B
_
lhal solvos pioLloms (2) ano (8) simullanoously (givon lho
iooological Liasos ). Givon lho concavily ano oiloionliaLilily assumplions, an oquiliLiium is
uniquoly oolnoo; moioovoi, as long as il is inloiioi, il salislos lho following oqualions
N

j=1
\n
j
_

A
_
= 0 ano
N

j=1
\n
j
_

B
_
= 0, (4)
whoio \n
j
oonolos lho giaoionl of funclion n
j
wilh iospocl lo lho vocloi . Cloaily, (4) may nol
Lo salisloo if lho solulion is nol in lho foasiLlo sol of policios, Q, ano in lhis caso, an oLvious
complomonlaiy slacknoss gonoializalion of (4) holos. Sliicl concavily of oach n
j
immooialoly
implios lhal
A
=
B
=

(iogaioloss of wholhoi lho oquiliLiium is inloiioi). 1hoiofoio, paily

will win lho oloclion al limo t = 1 wilh pioLaLilily
1
A
(

[ ) =
1
2

c

j=1
0
j
. ()
1his lhon loaos lo lho following pioposilion, chaiacloiizing lho oquiliLiium (pioof in lho loxl).
Proposition 1 Without paramilitaries, there exists a unique electoral equilibrium (at t = 1)
where
A
=
B
=

, and

, if interior, satises (4). Party wins the election with probability

given by (5).
1wo impoilanl poinls lo nolo aio as follows. Fiisl, wilhoul paiamililaiy piosonco, nalional
policios aio choson lo caloi lo lho piofoioncos of all volois in all iogions. 1his foaluio is faiily
gonoial, lhough as is woll known lho facl lhal Lolh pailios chooso lho samo policy vocloi (policy
convoigonco) is spocial ano iolios on lho facl lhal lho lwo pailios oo nol lhomsolvos havo piofoi-
oncos ovoi policios. Wholhoi lhoy oo oi nol is nol impoilanl foi lho iosulls hoio, ano wo lhoiofoio
oploo foi lho simploi spocilcalion. Socono, avoiago iooological Lias acioss all iogions ooloiminos
lho pioLaLilily of iooloclion foi paily (which is cuiionlly in powoi).
8
2.2 Elections under Paramilitaries with Exogenous Preferences
Noxl, lol us supposo lhal a suLsol of lho iogions, oonoloo Ly ? A aio unooi paiamililaiy
conliol. Donolo lho lolal numLoi of lhoso iogions Ly 7. 1ho koy foaluio of paiamililaiy-conliolloo
aioas foi oui puiposos is lhal, as wo will oocumonl in oolail Lolow, voling is nol fioo Lul inluoncoo
Ly lho implicil oi oxplicil piossuio of lho paiamililaiios. 1hioughoul lho iosl of lhis soclion wo
imposo lhis foaluio.
Wo slail wilh paiamililaiios wilh oxogonous piofoioncos," moaning lhal how lho paiamil-
ilaiios inluonco lho voling Lohavioi of lho cilizons in lho iogions lhoy conliol is oxogonous,
moaning lhal wo lako lho Lohavioi of lho paiamililaiios (ano lho voling Lohavioi of lho cilizons
in paiamililaiy-conliolloo aioas) as givon. 1his will conliasl wilh lho caso in which lho suppoil
of lho paiamililaiios is onoogonous lo lho policy choicos sluoioo in lho noxl suLsoclion. In pai-
liculai, supposo lhal in oach paiamililaiy-conliolloo iogion , ?, a fiaclion :
j
of volois will
volo foi paily iogaioloss of policios, whoio is lho samo aggiogalo valanco loim, oisliiLuloo
unifoimly ovoi
_

1
2
,
1
2
_
, as aLovo.
9
Lol us oonolo lho complomonl of lho sol ? Ly = A?
ano lho lolal numLoi of iogions in lhis (non-paiamililaiy-conliolloo) sol Ly J = 7. Lol us
also oolno :
j
= :
j
1,2, which iopiosonls lho Lias of paiamililaiy-conliolloo municipalily ,
foi paily . 1hon wilh an ioonlical ioasoning lo lhal in lho piovious suLsoclion, lho pioLaLilily
lhal paily will win lho oloclion al limo t = 1 is
1
A
_

A
,
B
[ , m
_
= Ii [\ _ ,2[
= Ii
_
_

j2Z
( :
j
)

j2J
_
1
2
n
j
_

A
_
n
j
_

B
_
0
j

_
_

2
_
_
= Ii
_
_

j2Z
(:
j
)

j2J
_
n
j
_

A
_
n
j
_

B
_
0
j

_
_ 0
_
_
,
whoio m oonolos lho vocloi of :
j
`s (logolhoi wilh infoimalion on which ,`s aio in lho sol ?)
ano lho lasl oqualily usos lho oolnilion of :
j
. 1hioughoul wo assumo lhal, as wilh lho 0
j
`s

j2Z
:
j

is nol so laigo as lo mako ono of lho pailios win lho oloclion wilh pioLaLilily ono.
Ioaiianging lho piovious oxpiossion fuilhoi, wo oLlain
1
A
_

A
,
B
[ , m
_
=
1
2

c

j2J
_
n
j
_

A
_
n
j
_

B
_
0
j
_

j2Z
:
j
. (6)
9
This assumption implies that the voting behavior of all citizens in paramilitary-controlled regions is entirely
insensitive to policies. The valance term is included to simplify the expressions and create continuity with the
case in which paramilitaries have endogenous preferences, discussed below. An alternative would be to suppose
that paramilitaries control the voting behavior of a fraction mj of the citizens and the remaining 1mj vote freely.
This alternative leads to similar results and is briey sketched in the Appendix.
0
Wo again assumo lhal Lolh pailios maximizo lho pioLaLilily of coming lo powoi ano oolno
an olocloial oquiliLiium in lho samo way. Wilh an ioonlical aigumonl lo lhal Lofoio, wo oLlain
lho following pioposilion.
Proposition 2 Under paramilitaries with endogenous preferences, there exists a unique electoral
equilibrium (at t = 1) where
A
=
B
=

. If

is interior, it satises

j2J
\n
j
(

) = 0. Party
wins the election with probability
1
A
(

[ , m) =
1
2

c

j2J
0
j

c

j2Z
:
j
.
1wo foaluios lhal aio nolowoilhy iolalivo lo Iioposilion 1 aio as follows. Fiisl, policios no
longoi caloi lo lho piofoioncos of all iogions. Sinco cilizons in paiamililaiy-conliolloo aioas cannol
iowaio oi punish a govoinmonl accoioing lo lho policy pioposals lhal il makos, Lolh pailios only
laigol lhoii policios lo lho volois in lho non-paiamililaiy-conliolloo aioas (soo also lho Apponoix).
1his implios lhal puLlic gooos ano olhoi amonilios will Lo iooucoo in lho paiamililaiy-conliolloo
aioas Loyono lho oiiocl olocl of paiamililaiy piosonco.
10
1hus, all olso oqual, wo may oxpocl
paiamililaiy piosonco lo incioaso inoqualily acioss iogions. Socono, olocloial oulcomos will now
Lo ooponoonl on lho inluonco of lho paiamililaiios on voling Lohavioi, which is capluioo Ly lho
lasl loim in 1
A
(

[ , m). If paiamililaiios piofoi paily , moaning lhal

j2Z
:
j
0, lhon,
olhoi lhings oqual, lho pioLaLilily lhal paily will win lho oloclion (ano slays in powoi) is
gioaloi. 1ho moio aioas aio conliolloo Ly lho paiamililaiios, lho sliongoi is lhis olocl. In lho
ompiiical woik Lolow, wo will piovioo inoiiocl ovioonco consislonl wilh Iioposilion 2 Ly showing
lho inluonco of paiamililaiios on olocloial oulcomos.
1his lasl foaluio alioaoy highlighls how paiamililaiios can havo a majoi inluonco on oo-
mocialic polilics. Novoilholoss, lhis olocl was minimizoo Ly lho moool in lhis suLsoclion Ly
assuming lhal lho paiamililaiios havo oxogonous piofoioncos, which wo iolax Lolow. Wo lisl
oiscuss how lho olocloial iolo of paiamililaiios alocls lho oocision of lho conlial govoinmonl lo
oxlono (Lioaocasl") ils powoi lo poiiphoial aioas conliolloo Ly lho paiamililaiios.
2.3 The State and the Paramilitaries
1aking lho olocloial oquiliLiium al limo t = 1 as givon, lol us now consiooi lho oocisions of
lho govoinmonl (paily ) al limo t = 0. In pailiculai, as oiscussoo in lho Inlioouclion, a koy
oimonsion of lho piocoss of lho foimalion of lho slalo is lho aLilily ano willingnoss of lho conlial
govoinmonl lo oslaLlish ils monopoly of violonco ano lhus iomovo olhoi gioups wilh accoss lo
10
The direct eect may, for example, stem from the fact that such investments and public good delivery become
more expensive, or paramilitaries directly damage infrastructure, law and order and the availability of public goods.
10
guns ano moans of oxoicising (local) violonco. Lol us moool lhis in lho simplosl possiLlo way ano
supposo lhal al limo t = 0, lho oLjoclivo of lho govoining paily is

j2R

j
1
A
_

A
,
B
[ , m
_
1
A
, (7)
whoio ? is a suLsol of lho aioas pioviously conliolloo Ly lho paiamililaiy lhal aio iocon-
quoioo" Ly lho conlial govoinmonl, ano
j
is lho nol Lonoll of ioconquoiing aioa , , which
acciuos lo lho govoinmonl al limo t = 0.
11
1his nol Lonoll incluoos lho aooilional lax iovonuos
oi socuiily gains lhal lho conlial govoinmonl will ooiivo ano suLliacls lho polonlial ioal" cosl of
lho ioconquosl (sponoing on lho mililaiy, polonlially slaLilily ano loss of lifo). Howovoi, lho oL-
joclivo of lho govoining paily, paily , also incluoos lho pioLaLilily lhal il will iomain in powoi,
lhus onjoying ionls fiom powoi al limo t = 1. In pailiculai, if somo aioa , ? is ioconquoioo,
lhon in lho suLsoquonl olocloial oquiliLiium al limo t = 1, paily will oLlain, in oxpoclalion,
a fiaclion 1,2 c0
j
of lho volos fiom lhis iogion as opposoo lo iocoiving :
j
= :
j
1,2 of lho
volos hao lhis placo iomainoo unooi paiamililaiy conliol. A subgame perfect equilibrium of lhis
gamo is oolnoo as an olocloial oquiliLiium al oalo t = 1 logolhoi wilh oocisions Ly paily al
oalo t = 0 lhal maximizos ils ulilily laking lho oalo t = 1 oquiliLiium as givon.
1ho analysis in lho piocooing paiagiaph lhon oslaLlishos lho following pioposilion.
Proposition 3 A subgame perfect equilibrium involves the electoral equilibrium characterized in
Proposition 2 at time t = 1, and at time t = 0, Party reconquers
all , ? such that
j
c(0
j
:
j
)
1
A

0
and does not reconquer
any , ? such that
j
c(0
j
:
j
)
1
A

< 0.
1his pioposilion will Lo invosligaloo in oui ompiiical woik. Il implios lhal lho willingnoss of
lho slalo lo ioconquoi aioas conliolloo Ly lho paiamililaiios, ano lhus oslaLlish lho monopoly of
violonco onvisagoo as an ossonlial chaiacloiislic of lho moooin slalo Ly Max WoLoi, is alocloo
nol only Ly lho ioal cosls ano Lonolls of ooing so, Lul also Ly lho implicalions of lhis oxpansion
of aulhoiily on olocloial oulcomos. In pailiculai, if many of lhoso paiamililaiy-conliolloo aioas
havo :
j
0
j
, lhon lho slalo, cuiionlly conliolloo Ly paily , will Lo ioluclanl lo ioconquoi
lhoso aioas, Locauso ooing so will mako il moio oimcull foi lhis paily lo succooo in lho upcoming
11
One could easily extend this so that these rents accrue both at t = 0 and t = 1, and in that case, the objective
functions will change to
P
j2R
j + P
A

q; q
B
j

h
R
A
+
P
j2R
j
i
, slightly complicating the analysis.
11
oloclions (ano moioovoi, lhis olocl will Lo sliongoi whon ionls fiom powoi al t = 1, 1
A
, aio
highoi). Naluially, fiom lho poispoclivo of paily A, lho aioas lhal aio mosl valuaLlo in lho hanos
of lho paiamililaiios aio lhoso lhal havo Lolh low 0
j
ano high :
j
; lhal is, areas that would have
otherwise voted for party 1, but paramilitaries are forcing citizens to vote in favor of party . A
govoinmonl lhal ooos nol ioquiio olocloial suppoil (o.g., a puioly non-oomocialic" govoinmonl)
woulo havo oociooo lo ioconquoi all aioas wilh
j
0. 1hoiofoio, lo lho oxlonl lhal 0
j
< :
j
a oomocialic govoinmonl may Lo loss willing lo Lioaocasl ils powoi ano ioconquoi aioas unooi
paiamililaiy conliol lhan such a non-oomocialic govoinmonl (oi a govoinmonl lhal is socuio in
ils posilion).
12
Nolo an impoilanl implicalion of lho funclional foim assumplions wo havo imposoo so fai. 1ho
unifoim oisliiLulions of ioiosyncialic piofoionco ano valanco loims, logolhoi wilh lho assumplion
lhal

j2Z
:
j
is nol so laigo as lo ooloimino lho oloclion wilh pioLaLilily ono, imply lhal lho
valuo of aooilional volos lo lho paily in powoi is conslanl ano inooponoonl of ils oxpocloo
winning pioLaLilily". As a consoquonco, Iioposilion 8 lakos a simplo foim, whoio lho valuo of
paiamililaiy volos lo lho paily in powoi is inooponoonl of lhis pioLaLilily. Wilh olhoi funclional
foims, as in ioalily, lhis valuo, ano lhus lho Lohavioi of lhis paily lowaios lho paiamililaiy
gioups, may oopono on ils oxpocloo winning pioLaLilily. Foi oxamplo, lho paily in powoi may
Lo loss iosponsivo lo lho volos oolivoioo Ly paiamililaiios whon il is ox anlo moio likoly lo win
lho oloclion.
13
2.4 Electoral Competition under Paramilitaries with Endogenous Preferences
1ho oiscussion so fai was foi paiamililaiios wilh oxogonous piofoioncos ano lhus look lho vocloi m
as givon. Naluially, lho willingnoss of lho paiamililaiios lo cooico cilizons lo volo foi ono canoioalo
oi anolhoi is also onoogonous ano ooponos on lhoii policy ano iooological piofoioncos. Wo now
invosligalo lhoso issuos. Supposo lhal, as wilh lho cilizons, lho piofoioncos of lho paiamililaiios
conliolling iogion , ? aio givon Ly
\
j
(, 0
g
) = n
j
()

1
_

0
j


0
g
_
-
g
j
,
whoio

1 is anolhoi funclion (possiLly lho samo as 1 ) lhal is also incioasing in

0
j


0
g

ano
now

0
j
is lho policy piofoionco of lho gioup of paiamililaiios conliolling iogion ,. Wilh a similai
12
Naturally, the net benet of reconquering an area might be dierent for a non-democratic government. For
example, it might be ^ j > j instead of j, because a non-democratic government can impose higher taxes on
certain regions than democratic governments could or would. This would be another incentive for non-democratic
governments to broadcast their power. On the other hand, the cost of doing so may also be higher for non-democratic
governments for they may be unwilling to build a strong army because of the future potential threats that this may
pose to their reign (e.g., Acemoglu, Ticchi and Vindigni, 2010). This would then imply ^ j < j.
13
As we discuss in the next section, this feature of a more general model may actually be helpful in understanding
some of the recent experiences in Colombia.
12
ioasoning lo lhal aLovo, lol us oolno

0
j
=

1
_

0
j


0
B
_

1
_

0
j


0
A
_
as lho iooological loanings of lho paiamililaiios in iogion , in favoi of paily (wo uso

0
j
insloao
of 0
j
lo highlighl lhal lhis iofois lo lho paiamililaiios). Ano in aooilion, supposo lhal, as Lofoio
-
A
j
-
B
j
il composoo of lwo olomonls -
ij
ano , whoio -
ij
is again oisliiLuloo unifoimly ovoi
_

1
2
,
1
2

,
ano wo lako lo Lo lho samo as lho valanco loim foi lho volois (which, iocall, was oisliiLuloo
unifoimly ovoi
_

1
2
,
1
2
_
). 1hon conoilional on , lho fiaclion of volois in paiamililaiy-conliolloo
iogion , ? who will volo foi paily will Lo
1
2
n
j
_

A
_
n
j
_

B
_

0
j
.
Lol us also assumo lhal paiamililaiios can foico all volois in lhoii sphoio of inluonco lo volo foi
whichovoi paily lhoy piofoi. 1hon lho pioLaLilily lhal paily will win lho oloclion Locomos
1
A
_

A
,
B
[
^

_
=
1
2

c

j2J
_
n
j
_

A
_
n
j
_

B
_
0
j
_

j2Z
_
n
j
_

A
_
n
j
_

B
_

0
j
_
,
whoio now
^
oonolos lho vocloi of all iooological piofoioncos, incluoing lhoso of lho paiamililaiios.
Wilh a similai ioasoning lo oui analysis aLovo, olocloial compolilion will loao lo lho samo
policy choico foi Lolh pailios, ano whon il is inloiioi, lhis vocloi will Lo givon Ly lho solulion lo
lho following sol of oqualions:

j2J
\n
j
(

)

j2Z
\n
j
(

) = 0. (8)
Naluially, lhoso oqualions holo in lho complomonlaiy slacknoss foim whon

may Lo al lho
Lounoaiy of lho foasiLlo policy sol Q. 1hoiofoio, wo havo (pioof in lho loxl):
Proposition 4 Under paramilitaries with endogenous preferences, there exists a unique electoral
equilibrium at t = 1 where
A
=
B
=

. If

is interior, it satises (8). Party wins the

election with probability
1
A
_

[
^

_
=
1
2

c

j2J
0
j

c

j2Z

0
j
.
Moreover, the subgame perfect equilibrium involves Party reconquering (at time t = 0)
all , ? such that
j
c
_
0
j


0
j
_
1
A

0,
18
and not reconquering
any , ? such that
j
c
_
0
j


0
j
_
1
A

< 0.
1hoio aio sovoial now foaluios in lhis pioposilion. Fiisl, whon paiamililaiios aojusl lhoii sup-
poil ooponoing on lho policios ano iooological slanco of lho lwo pailios, lho pailios chango lhoii
policios in oiooi lo Lo moio alliaclivo lo lho paiamililaiios` policy piofoioncos. 1hal is, ialhoi
lhan caloiing lo lho piofoioncos of lho cilizons in lho aioas lhal aio conliolloo Ly lho paiamili-
laiios (which lhoy woulo havo oono wilhoul lho paiamililaiios), pailios appoaso lho paiamililaiios
lhomsolvos. 1his iosull is lho Lasis of lho polonlial symbiotic iolalionship Lolwoon paiamililaiios
ano lho oxoculivo monlionoo in lho Inlioouclion. Moioovoi, il can fuilhoi incioaso lho inoqual-
ily among lho iogions, Locauso lho policios aio choson spocilcally lo suppoil, oi iofiain fiom
lhioaloning, lho paiamililaiios ano lho aioas whoio lho paiamililaiios aio sliongosl. 1wo foa-
luios ooloimino how slanloo lowaios lho paiamililaiios oquiliLiium policios aio. 1hoso aio: lho
sizo of lho paiamililaiy-conliolloo aioas (lho laigoi is lho sol of municipalilios conliolloo Ly lho
paiamililaiios, 7, lho moio inluonlial aio lho paiamililaiios in shaping oquiliLiium policy) ano lho
iolalivo iosponsivonoss of lho paiamililaiios lo policy concossions (ooponoing on lho piofoioncos
of lho paiamililaiios, lho n
j
`s, iolalivo lo lho piofoioncos of lho cilizons, lho n
j
`s). In aooilion,
Locauso olocloial compolilion makos Lolh pailios caloi lo lho wishos of lho paiamililaiios, al
lho ono lho paiamililaiios` iooological piofoioncos play a conlial iolo in wholhoi lhoy foico lho
populalion lo volo foi paily oi paily 1.
14
Finally, wo can also allow Lolh pailios oi ono of lho pailios lo mooify ils iooological slanco
(in a ciooiLlo fashion). 1ho samo analysis as hoio will lhon imply lhal in oiooi lo alliacl volos
fiom paiamililaiy-conliolloo aioas, ono oi Lolh pailios may oocioo lo panooi lo lho iooological
piofoioncos of lho paiamililaiios.
2.5 Importance of Non-National Ambitions
An impoilanl quoslion in lho conloxl of ColomLian polilics is why iighl-wing paiamililaiy gioups
havo Locomo moio involvoo in inluoncing oloclions lhan lofl-wing guoiiillas, in pailiculai, moio so
lhan lho iolalivoly woll-oiganizoo FAIC. Ono possiLlo answoi is lhal in conliasl lo lho guoiiillas,
lho paiamililaiios oo nol havo nalional amLilions, making a coalilion Lolwoon lhom ano lho
oxoculivo conliolling lho conlial slalo moio foasiLlo. 1ho moool piosonloo so fai has implicilly
maoo lhis assumplion, sinco wo oio nol inlioouco lho iisk lhal lho non-slalo aimoo aclois may
lako ovoi lho conlial slalo.
14
The result that both parties modify their policies to partly cater to the wishes of the paramilitaries has an
obvious similarity, and an identical mathematical logic, to Baron's (1994) result concerning the eects of a lobby
on the platforms of two competing parties.
14
A simplo way of inliooucing lhis possiLilily woulo Lo lo havo a pioLaLilily 1(.) lhal lho non-
slalo aimoo aclois woulo Locomo sliong onough lo challongo lho conlial slalo, poihaps ovoilhiow
il, whoio . is lho fiaclion of municipalilios conliolloo Ly lho non-slalo aclois in quoslion. Such
an ovoilhiow of lho conlial govoinmonl Ly non-slalo aimoo aclois is nol uncommon in woak
Afiican slalos, such as Somalia, Sioiia Loono oi LiLoiia, ano has coilainly Loon lho oLjoclivo
of lho FAIC. Naluially, wo woulo oxpocl 1(.) lo Lo incioasing in ., so lhal whon lhoso gioups
conliol moio aioas, lhoy aio moio likoly lo poso such a nalional challongo. In lhal caso, wo woulo
nooo lo chango lho oLjoclivo funclion of paily lo incoipoialo lhis possiLilily. Foi oxamplo,
oqualion (7) coulo Lo mooiloo lo

j2R

j
[1 1(.)[ 1
A
_

A
,
B
[ , m
_
1
A
.
1his spocilcalion makos il cloai lhal whon 1(.) 0, lhoio will Lo sliongoi inconlivos foi paily
lo ioconquoi loiiiloiios conliolloo Ly lhoso non-slalo aimoo gioups (lhus iooucing .). Whon
1(.) is sumcionlly high ano sumcionlly incioasing in ., lhis olocl can moio lhan componsalo foi
lho olocloial aovanlago lhal local conliol Ly lhoso gioups cioalos foi lho paily in powoi. 1hus
facloiing in lho nalional amLilions of non-slalo aimoo aclois iooucos lho scopo foi a coalilion oi
a symLiolic iolalionship Lolwoon lhoso gioups ano lho oxoculivo. 1his ioasoning suggosls lhal
whon non-slalo aimoo aclois havo nalional amLilions, il will Lo aovanlagoous foi lho conlial slalo
lo oliminalo lhom (soonoi oi laloi), lhus any implicil oi oxplicil policy piomisos lhal il makos lo
such gioups woulo Lo non-ciooiLlo, making a coalilion Lolwoon lhom impossiLlo. 1his poispoclivo
suggosls a naluial ioason foi why, in ColomLia, such a coalilion may havo Loon much moio likoly
lo aiiso Lolwoon lho oxoculivo ano lho paiamililaiios ialhoi lhan wilh lho FAIC.
2.6 Summary and Empirical Predictions
In lho iosl of lho papoi, wo will invosligalo lho olocl of ColomLian paiamililaiy foicos on lho
olocloial oulcomos in lho oaily 2000s. Oui invosligalion is molivaloo Ly lho lhooiolical iooas
oiscussoo aLovo. In pailiculai, wo will oocumonl lho following Lioao palloins, which, lhough nol
conclusivo pioof of lho iooas oovolopoo hoio, aio highly suggoslivo.
1. Consislonl wilh Iioposilion 2, paiamililaiios, onco lhoy Locamo sumcionlly powoiful, slailoo
inluoncing olocloial oulcomos in lho aioas of ColomLia lhoy conliolloo.
2. Consislonl wilh Iioposilion 8, wo will show lhal paiamililaiios localoo in aioas lhal voloo
foi lho cuiionl piosioonl in gioal numLois, Lul in pasl oloclions lonooo lo volo foi moio
liLoial polilicians, woio moio likoly lo poisisl.
1
8. Consislonl wilh Iioposilion 4, wo will show lhal lho piosioonl has pioposoo logislalion in lino
wilh lho piofoioncos of lho paiamililaiios, ano lho sonalois olocloo fiom high paiamililaiy
aioas havo suppoiloo lhis logislalion.
3 A Brief Overview of Non-State Armed Actors in Colombia
3.1 Origins of Colombian Non-State Armed Actors
ColomLia has a long hisloiy of non-slalo aimoo aclois ano many of lho mosl ioconl omoigoo
fiom a muiooious civil wai known as La Violencia which lasloo fiom lho lalo 1040s inlo lho oaily
1060s. 1his civil wai was inilially lho consoquonco of lghling Lolwoon lho LiLoial ano Consoiv-
alivo polilical pailios. In 1064, lho Fuoizas Aimaoas Iovolucionaiias oo ColomLia (FAIC1ho
Iovolulionaiy Aimoo Foicos of ColomLia) foimoo fiom lho iomnanls of LiLoial guoiiillas fiom
lhis poiioo. In lho samo yoai lho Ljoicilo oo LiLoiacion Nacional (LLNNalional LiLoialion
Aimy) was also cioaloo. 1hoso lofl-wing` guoiiilla gioups woio iolalivoly small ouiing lho 1060s
ano 1070s, Lul Logan lo oxpano iapioly in lho 1080s ano lhoy woio joinoo Ly olhoi lofl-wing
iovolulionaiy movomonls such as lho Movimionlo 10 oo ALiil (M-10Movomonl of Apiil 10)
ano Quinlin Lamo. 1ho 1080s also saw lho iapio oxpansion of iighl-wing` paiamililaiy foicos
which in 1007 coaloscoo inlo lho Aulooofonsas Unioas oo ColomLia (AUCUniloo Solf-Dofonso
Oiganizalion of ColomLia).
1hoso vaiious non-slalo aimoo gioups iangoo ovoi mosl of lho loiiiloiy of ColomLia ano
lhough oslimalos vaiy, may havo hao aiouno 0,000 mon ano womon unooi aims al lho slail
of lho 21sl Conluiy. 1hoy ongagoo in kionapping, massacios of civilians, oiug pioouclion ano
oxpoilalion, ano iogulaily oxpiopiialoo lano ano oxloiloo incomo (collocloo lho laxos`) fiom
ColomLian cilizons. 1hoy also ongagoo in violonl conlicls wilh oach olhoi ano wilh lho aimoo
foicos of lho ColomLian slalo.
3.2 Paramilitaries and the AUC
ColomLia`s paiamililaiios aio lhoughl lo oiiginalo fiom 1060s counloiinsuigoncy moasuios ano
Law 48 of 1068 which allowoo lho cioalion of solf-oofonso mililias Ly piivalo cilizons foi lho
puiposos of piolocling lhoii piopoilios ano livos (soo Iomoio, 2000, 2002, Iangol, 200, ano
Duncan, 2007, foi ovoiviows of lho hisloiy ano oiganizalion of lho paiamililaiios). Novoilholoss,
lho poiioo of La Violencia is also lilloioo wilh vaiious soils of mililias ano sponlanoous solf-
oofonso gioups (soo, foi oxamplo, Ioloan, 2002, on Anlioquia). Small gioups of paiamililaiios
also omoigoo in placos whoio lhoio woio valuaLlo iosouicos whoso oxploilalion was conlosloo, such
as in lho omoialo minos of Muzo, in lho oopailmonl of Loyaca ano lho oiug planlalions on lho
16
coasl noai Sanla Maila, Magoalona ano in lho noaiLy oopailmonl of La Guajiia.
1ho oscalalion of paiamililaiy aclivily in lho oaily 1080s is associaloo wilh lho iiso of lho
laigo oiug cailols in Mooollin ano Cali lhal facoo lhioals of kionapping ano oxloilion fiom lofl-
wing gioups. As lho woallh of lho oiug cailols giow, many of lhoii momLois Logan lo Luy
up lano ano ianchos in iuial aioas. Hoio lhoii inloiosls Logan lo fuso wilh lhoso of liaoilional
iuial olilos who also wishoo lo piolocl lhomsolvos fiom oxloilion ano kionappois (soo Gulioiioz
Sanin ano Laion, 200). 1his loo lo collaLoialion in lho foimalion of paiamililaiy gioups. Ono
aioa of iapio oxpansion was lho Magoalona Mooio al lho oasloin poiiphoiy of lho oopailmonl
of Anlioquia which saw lho omoigonco of gioups such as Los 1anguoios foimoo Ly lho Casla no
Liolhois (Cailos, Fiool ano Viconlo) whoso falhoi hao Loon killoo Ly lho FAIC in 1081.
15
In
1004 lho Casla no Liolhois foimoo lho Aulooofonsas Camposinas oo CoiooLa y UiaLa` (ACCU
Ioasanl Solf-Dofonso foico of CoiooLa ano UiaLa). 1his fuilhoi oxpansion was facililaloo in
lho samo yoai Ly a law piomoloo Ly Iiosioonl Sampoi lo allow lho cioalion of CONVIVII, a
nalional piogiam of noighLoihooo walch gioups. An impoilanl suppoiloi of lhis piogiam was

Alvaio UiiLo, lhon Govoinoi of Anlioquia, whoso falhoi was killoo Ly lho FAIC in 1088.
In Apiil 1007 lho AUC was foimoo Ly Cailos Casla no ano il incluooo possiLly 00/ of lho
oxisling paiamililaiy foicos. 1ho cioalion of lhis nalional oiganizalion incioasoo lho oloclivonoss
of lho paiamililaiios consiooiaLly; as a iosull, lho FAIC ano LLN woio lhiown oul of laigo aioas
of lho counliy, lhough as oui oala will show lhoso guoiiilla gioups aio slill aclivo in many pails
of ColomLia (soo Iosliopo, Spagal ano Vaigas, 2004).
16
Soon afloi coming lo powoi in 2002, Iiosioonl UiiLo Logan lo nogolialo lho oomoLilizalion of
lho paiamililaiios, somolhing ho hao piomisoo ouiing his oloclion campaign. Docioo 128 issuoo Ly
lho piosioonl in Januaiy 2008 gavo oo faclo amnosly foi paiamililaiios nol unooi invosligalion foi
human iighls violalions ano lhis has Loon applioo lo lho vasl numLoi of oomoLilizalions (aiouno
02/). On July 1, 2008 in Sanla Fo oo Ialilo in CoiooLa, lho govoinmonl signoo an agioomonl
wilh mosl of lho gioups of lho AUC lo oisaim Ly lho ono of 200.
17
On NovomLoi 2 2008 aiouno 860 paiamililaiios of Mooollin`s Caciquo NuliLaia Lloc loo Ly
Don Loina` (Diogo Foinanoo Muiillo) oomoLilizoo. 1his piocoss was fuilhoi inslilulionalizoo Ly
lho passing of lho conliovoisial Juslico ano Ioaco Law in Juno 200 which was signoo inlo law
15
There is also evidence suggesting involvement of the army in the training and organization of paramilitary
groups, though in 1989 the Colombian Supreme Court declared that Law 48 was unconstitutional. One month later
President Barco issued Decree 1194, which prohibited the creation, promotion or organization of paramilitary or
self-defense groups and declared such activities illegal.
16
The timing of the creation of the AUC was a consequence of both the collapse of the Medelln and Cali drug
cartels, which had previously exercised a large amount of control over the organizations, and of the decision of the
Pastrana government to attempt to negotiate a peace deal with the FARC by making concessions.
17
The text of the agreement is available at the web page of the Oce of the High Commissioner for Peace:
www.altocomisionadoparalapaz.gov.co/acuerdos/index.htm
17
Ly Iiosioonl UiiLo in lho following monlh. Ailiclo 20 of lhis law limils sonloncos lo lhoso founo
guilly of human iighls violalions lo Lolwoon ano 8 yoais. Ailiclo 80 allows lho govoinmonl lo
ooloimino lho placo of oolonlion, which nooo nol Lo a piison. In May 2006 lho ColomLian Con-
slilulional Couil alloioo many aspocls of lho law on lho giounos lhal lhoy woio unconslilulional,
in pailiculai lho Couil slipulaloo lhal oomoLilizing comLalanls hao lo givo a full confossion of
lhoii aclivilios in oiooi foi lho law lo apply lo lhom. Lolh lho oomoLilizalion piocoss ano lho
Juslico ano Ioaco Law havo Loon wiooly ciilicizoo Ly human iighls oiganizalion.
18
1hoio is a gioal ooal of conliovoisy aLoul wholhoi lho paiamililaiy oomoLilizalion is ioal oi
wholhoi il will loao lo now aimoo gioups (oi simply lo lho inslilulionalizalionlogilimalion of
lho powoi of lho AUC; on lhis issuo soo Iaioo, 2007, Inloinalional Ciisis Gioup, 2007, Ioich ano
Iasmusson, 2008, Zuckoiman, 2000).
3.3 The Involvement of Paramilitaries in Politics
Soon afloi lho founoalion of lho AUC in 1007 lhoio appoais lo havo Loon a slialogic oocision lo
inluonco olocloial polilics. 1his chango is liacoo lo a hisloiic mooling in Sanla Fo oo Ialilo in
2001 whoio momLois of lho estado mayor (lho govoining Looy) of lho AUC along wilh polilicians
ano momLois of Congioss signoo a sociol oocumonl calling foi lho iofounoing of lho counliy`.
1hoso who signoo lhis oocumonl incluooo piominonl paiamililaiy loaoois, such as Joigo 40 (Io-
oiigo 1ovai Iupo), Don Loina (using a psouoonym Aoolfo Iaz`) ano Diogo Vocino, ano sovoial
polilicians suLsoquonlly aiiosloo foi links wilh paiamililaiios, incluoing Sonalois William Monlos
(1aLlos 1 ano A1) ano Miguol oo la Lspiiolla (1aLlo A1).
19
1his chango in lho slialogy of lho
AUC will Lo ciucial lo oui ompiiical appioach, allowing us lo invosligalo how olocloial oulcomos
chango oiloionlially in high paiamililaiy aioas Lofoio ano afloi lhoii involvomonl in polilics in
2001.
1ho olhoi nolaLlo, ano iolaloo, oovolopmonl ouiing lho 2002 oloclion is lho omoigonco of
Liano now polilical pailios, which wo iofoi lo as third parties,` such as CamLio Iaoical ano
MIIOL. 1hoso pailios oflon hao oxplicil oi implicil links wilh lho paiamililaiios, ano lho caso
18
See Human Rights Watch (2005), Amnesty International (2005) and International Federation for Human Rights
(2007).
19
Although the meeting in Ralito was probably the most important one for the subsequent strategy of the
paramilitaries, it was not the only such pact between them and politicians during this period. In the southern
plains, paramilitary leader Hector Buitrago (Martin Llanos) organized a meeting in 2000 with all the candidates
running for the governor's and the mayor's oces and explicitly traded political support against key positions in
the local executive, allocations of public contracts, and a share in the resources of the municipality (\La Sombra
de Martin Llanos" Semana, October 8, 2007). In Puerto Berio, Antioquia, four congressmen from Santander met
with paramilitary leader Salvatore Mancuso in 2001 for a similar pact. In the Municipalities of Chivolo and Pivijay
of the department of Magdalena, the pact with the paramilitaries involved 417 local politicians that committed
to support the candidates linked with the paramilitaries for the legislative elections of 2002 through a movement
called \Movimimiento la Provincia Unida" (Movement United Province) (Semana, November 6, 2006).
18
sluoy ovioonco shows lhal paiamililaiy piossuio was oflon lo incioaso lho volo foi lhoso pailios.
In many paiamililaiy-conliolloo aioas lhoy havo ioplacoo lho liaoilional LiLoial ano Consoivalivo
pailios. Wo will uso lho volo shaio of lhiio pailios` as a moasuio of paiamililaiy inluonco on
olocloial oulcomos.
Loginning in 200 lhoio woio incioasing accusalions of involvomonl of lho AUC in lho oloclions
of 2002. Scanoal mounloo fuilhoi wilh lho oomoLilizalion of Joigo 40 ano his 2,000 sliong Llock
on Maich 10, 2006 in La Mosa, Cosai. Joigo 40`s compuloi foll inlo lho hanos of govoinmonl
omcials ano il conlainoo omails oiooiing his mon lo iociuil poasanls lo piolono lo Lo paiamililaiios
ouiing oomoLilizalion coiomonios.
20
1ho compuloi also lisloo ovoi 00 muioois ano oolailoo many
links Lolwoon polilicians ano paiamililaiios. 1hoso iovolalions loo lo inlonso sciuliny of lho 2002
oloclion iosulls many of which oxhiLil somo ialhoi oxliaoioinaiy foaluios. 1hoso incluoo massivo
changos in voling palloins ano voiy high conconlialions of volos foi somo canoioalos in pailiculai
municipalilios.
21
Sinco lhon lhoio havo Loon many invosligalions of links Lolwoon polilicians ano paiamililaiios
ano a laigo caso sluoy liloialuio has omoigoo oocumonling such links (soo in pailiculai lho
iosoaich of Lopoz, 2007, Valoncia, 2007, ano lho olhoi ossays in Iomoio, 2007, Soiiano ZaLala,
2007 ano lho woL silo voioaoaLioila.com). As of May 20, 2000, 80 momLois of Congioss ano lho
Sonalo woio unooi invosligalion, 86 woio aiiosloo ano in oolonlion, ano 11 hao Loon founo guilly
of links wilh paiamililaiios.
22
In lolal lhis iopiosonls almosl 18 of ColomLian logislalois. 1hoso
pioviously aiiosloo, ioloasoo ano lhon io-aiiosloo incluoo Maiio UiiLo, Iiosioonl UiiLo`s cousin
ano main polilical aovisoi. 1hoso aiiosloo incluoo Sonaloi Cailos Gaicia, lho piosioonl of lho U
paily" (U foi UiiLolhough lhis paily is nol omcially iocognizoo Ly lho piosioonl as his paily).
1ho invosligalion ano aiiosl of lhoso polilicians has Loon unooilakon piimaiily Ly lho Supiomo
Couil.
3.4 Controlling the Vote
1hoio is consiooiaLlo caso sluoy ovioonco lhal following lho mooling in Sanla Fo oo Ialilo,
paiamililaiy gioups aclivoly liioo ano succooooo in inluoncing volos in nalional oloclions (lhal
is, in lho 2002 ano 2006 oloclions). 1ho loslimony of majoi paiamililaiy loaoois suggosls lhal
lhoso gioups ioplacoo lho aulhoiily of lho slalo in many aioas. Many of lho paiamililaiy loaoois
havo Loon quilo ailiculalo aLoul lhoii polilical piojocl`. Of lhoso lho loslimony of Salvaloio
20
See also Inter-American Commission on Human Rights (2007, p. 5) on the apparently fake demobilizations in
Cesar.
21
See for instance the article in the Colombian weekly Semana \Votaciones atpicas en las elecciones de congreso
del 2002," September 11, 2005.
22
This data is updated regularly on http://www.indepaz.org.co.
10
Mancuso is poihaps mosl lolling, noling lhal
8/ of lho Congioss was olocloo in aioas whoio lhoio woio slalos of lho Solf-
Dofonso gioups, in lhoso slalos wo woio lho onos collocling laxos, wo oolivoioo juslico,
ano wo hao lho mililaiy ano loiiiloiial conliol of lho iogion ano all lho pooplo who
wanloo lo go inlo polilics hao lo como ano ooal wilh lho polilical iopiosonlalivos wo
hao lhoio."
23
1ho invosligalion inlo lho 2002 ano 2006 oloclion iosulls ano lho loslimony of oomoLilizoo
paiamililaiios has iovoaloo a laigo numLoi of oiloionl pacls` Lolwoon paiamililaiy loaoois ano
polilicians (oolailoo in Lopoz ano Sovillano, 2008) ano also oomonslialos lhal a plolhoia of oiloi-
onl slialogios woio usoo lo guaianloo lhal canoioalos piofoiioo Ly paiamililaiios won oloclions. A
salionl slialogy sooms lo havo Loon lo loiioiizo pooplo inlo voling foi spocilc canoioalos. In lho
municipalily of San Onofio in lho coaslal oopailmonl of Sucio,
24
foi oxamplo, lhis was aiiangoo
Ly lho paiamililaiy loaooi Caoona`: Foi lho oloclions of 2002, lho liucks sonl Ly Caoona` wonl
lhiough noighLoihooos ano iuial aioas of San Onofio picking pooplo up. Accoioing lo somo
pooplo in lhis municipalily in Sucio, lhousanos of poasanls woio lakon lo lho coiiogimionlo Ilan
Iaiojo` so lhoy coulo soo lho canoioalo foi whom lhoy hao lo volo foi in lho logislalivo oloclions:
Jaiio Moilano foi Sonalo ano Muiiol Lonilo foi lho Houso of Iopiosonlalivos.
Caoona` pul in a Lag all lho namos of lho councilois, ho look lwo ano saio lhal ho
was going lo kill lhom ano olhoi pooplo choson ianoomly if Muiiol oio nol win", says
a poasanl fiom lhis lown. 1ho lhioal sooms lo havo Loon oloclivo: oach canoioalo
oLlainoo 40,000 volos in Sucio."
25
Shooi loiioi sooms lo havo Loon usoo nol jusl lo inouco pooplo lo volo foi pailiculai canoioalos,
Lul also lo koop lhom away fiom lho polls so lhal Lallol slumng ano olhoi foims of manipulalion
of volo lolals coulo occui. Lvioonco of lho uso of cooicion lo koop pooplo al homo ano away fiom
lho polls comos fiom La Jagua oo ILiiico in lho oopailmonl of Cosai.
26
Anolhoi slialogy, whoio cooicion also playoo an impoilanl iolo, involvoo collocling pooplo`s
cooulas (nalional ioonlily caios which a poison musl pioouco lo volo) fiom lhoii housos, using
lhom lo collocl lho Lallols (lho laijolon`) ano llling lhom in foi pooplo.
27
23
Translation of the authors from http://www.youtube.com/watch?v=0tsaMNqoa k&feature=related
24
Interestingly the mayor and ex-mayor of San Onofre were both signatories of the Pact of Santa Fe de Ralito.
25
Quoted from \Redacci on Nacional" El Tiempo November 11, 2006. Translation by the authors.
26
From `Un Abrebocas de estas Elecciones" by Cristina Velez in Votebien.com, February 2006.
27
This phenomenon is discussed in the article \Aqu nadie es un santo" Semana, September 29, 2007.
20
Fuilhoi ovioonco on how volos woio oolivoioo omoigoo ouiing lho loslimony of Iafaol Gaicia
1oiios, lho foimoi oiiocloi of infoimalion soivicos foi lho Iiosioonlial inlolligonco soivico, lho
Aominislialivo Socuiily Dopailmonl (DAS). Gaicia, unooi invosligalion foi links wilh paiamili-
laiios, lolo piosoculois lhal ho hao oosignoo a compuloi piogiam lo uso conloonlial infoimalion
lhal lolo us lho lisl of volois Ly any calogoiy, foi oxamplo, Ly polling slalion, zono, municipalily
ano ovon Ly oopailmonls." Wilh lhis infoimalion in hano counloifoil Lallols woio cioaloo so
lhal Ly lho ono of lho oloclions lhoy woulo incluoo fako volos of lho pooplo who oio nol volo,
ano if lhoio woio Lallols favoiing olhoi canoioalos oiloionl fiom lho onos fiom lho paiamililaiy
gioup Lloquo Noilo lhoy woulo Lo ioplacoo Ly Lallols foi oui canoioalos".
28
All in all lho ovioonco inoicalos lhal paiamililaiy gioups usoo a wioo vaiioly of slialogios lo
mako suio lhal lhoii piofoiioo canoioalos gol olocloo. 1his iangoo fiom loiioiizing volois lo volo
in pailiculai ways, loiioiizing lhom lo slay away fiom lho polls so lhoy coulo slul Lallols, voling
insloao of cilizons Ly conlscaling lhoii ioonlify caios, loiioiizing polilicians so lhal lhoy woulo nol
iun againsl lhoii piofoiioo canoioalos, ano manipulaling suLsoquonl volo lolals oloclionically.
29
3.5 The Colombian Political System
Hoio wo omphasizo a fow inslilulional oolails of lho ColomLian syslom lhal aio impoilanl foi oui
ompiiical slialogy. Unooi lho 1001 Conslilulion lho piosioonl of ColomLia was olocloo foi ono
foui yoai loim wilh no possiLilily of io-oloclion. 1hoio has Loon a sliong noim againsl io-oloclion
hisloiically in ColomLia ano lho lasl piosioonl lo succooo himsolf was Iafaol N u noz in 1886.
1hough unooi lho 1886 Conslilulion io-oloclion was poimilloo if nol succossivo, il only happonoo
onco wilh Alfonso Lopoz Iumaiojo Loing piosioonl Lolwoon 1084 ano 1088 ano again Lolwoon
1042 ano 104 (whon ho onooo his socono loim oaily Ly iosigning). 1ho piosioonl is olocloo Ly a
nalional volo ano if no canoioalo iocoivos 0/ of lho volo in lho lisl iouno, a iun-ol oloclion is
holo Lolwoon lho lwo canoioalos wilh lho laigosl numLoi of volos in lho lisl iouno.
28
See www.cambio.com.co/portadacambio/712/ARTICULO-WEBNOTA INTERIOR CAMBIO-3444866.html.
Other evidence reveals both the extent to which the political project of the paramilitaries was planned and
also how votes were guaranteed. Semana documented how paramilitary leaders and politicians divided up coastal
departments, see Semana, November 25th 2006.
29
Our empirical results will show that the AUC more than any non-state armed actor has been signicantly
inuencing elections in Colombia. The FARC and ELN have also certainly inuenced some elections and have used
their power to sway or intimidate voters in favor of candidates they preferred. They have threatened and killed
politicians. However, their involvement in elections has been more limited than that of the AUC. We conjecture
that this is mostly because the FARC and ELN have been ideologically opposed to the institutions of the state
in Colombia and their project was to overthrow the national government. As also emphasized by our theoretical
discussion in subsection 2.5, this made a symbiotic relationship between them and the executive impossible. In
contrast, the AUC are a continuation of local paramilitary groups allied with local elites and politicians. Again in
light of our discussion in subsection 2.5, it was important that the AUC did not want to overthrow the state but to
exercise control over certain peripheral areas and inuence key legislation. This created a platform for a symbiotic
relationship between the central state and the AUC, which would have been impossible for the FARC and ELN.
21
Il is also impoilanl lo nolo lhal liko mosl piosioonlial sysloms in Lalin Amoiica, Colom-
Lia`s foaluios sliong piosioonlial oominanco of lho logislaluio wilh piosioonls making impoilanl
oocisions Ly issuing oocioos.
Foi lho Sonalo lhoio is a nalional consliluoncy whoio 100 sonalois aio olocloo fiom lisls. Foi
lho Congioss lhoio aio 82 mulli-momLoi oisliicls wilh oach oisliicl coiiosponoing lo a oopailmonl.
1ho iopiosonlalion of oopailmonls ooponos on lhoii populalion ano lhoio aio 162 congiossmon
in lolal. Hisloiically in ColomLia ovon liaoilional paily lisls aio voiy poisonalizoo so lho lypical
silualion is ono whoio only ono canoioalo is olocloo fiom oach lisl. 1his silualion oio nol chango
wilh lho 2006 oloclions ovon lhough a iofoim in lho olocloial law slipulaloo lhal lo win a soal
in lho logislaluio a lisl hao lo havo al loasl 2/ of lho volo nalionally. Al lho samo limo as
lhis law was inliooucoo, lho olocloial syslom was changoo lo allow foi opon-lisl piopoilional
iopiosonlalion (wilh piofoionco voling). 1hus ovon lhough lho numLoi of lisls foll oiamalically,
poisonal polilics conlinuoo unaLaloo via piofoionco voling.
1ho oiganizalion of lho Highoi Couils in ColomLia is ialhoi inliicalo ano wo oiscuss il in
lho Apponoix. 1ho ciucial poinl foi lho papoi is lhal unooi lho 1001 Conslilulion lhoy woio
sol up in a way which makos il voiy oimcull foi polilicians ano lho piosioonl lo ooloimino lhoii
composilion.
4 The Data
4.1 Data Sources and Construction
1ho mosl impoilanl oala foi lho papoi aio on lho piosonco of non-slalo aimoo aclois, spocilcally
paiamililaiios ano lofl-wing guoiillas. Oui main moasuio of lho piosonco of non-slalo aimoo aclois
is ono Lasoo on conlicl incioonls, which wo iofoi lo as allacks` foi shoil. 1ho oalaLaso of allacks
is fiom Conlio oo Lsluoios soLio Dosaiiollo Lconomico (CLDL) in lho Facullao oo Lconomia al
lho Univoisioao oo Los Anoos in Logola. CLDL collocls oala fiom lho OLsoivaloiy of Human
Iighls of lho Vico-piosiooncy ano lho Nalional Dopailmonl of Ilanning ano aggiogalos vaiiaLlos
in sovoial calogoiios Ly aimoo acloi ano lypo of aclion. 1ho oiiginal oala aio a compilalion of nows
fiom nowspapois ano fiom iopoils of lho nalional polico. Oui moasuio of attacks is consliucloo
Ly aggiogaling ovoi many of lhoso vaiiaLlos. Foi oach aimoo acloi wo simply aoo lho following
vaiiaLlos: oxplosivo loiioiisl acls, inconoiaiy loiioiisl acls, olhoi loiioiisl acls, assaulls lo piivalo
piopoily, allacks on civil oiganizalions, polilical assassinalion allompls, ioao Llockaoos, aimoo
conlacl Lolwoon slalo ano non-slalo aimoo foicos inilialoo Ly lho lalloi, amLushos of civilians,
haiassing (mainly lhioals lo civilians), incuision inlo villagos`, ovoilano piiacy, illogal chockpoinls,
aimoo foicos wounooo Ly lho non-slalo aimoo gioup, muioois of civilians, muioois of polilicians,
22
massacios, ooalhs of momLois of lho slalo aimoo foicos, kionappings of momLois of lho aimoo
foicos, kionappings of polilicians ano kionappings of civilians. Wo havo lhis vaiiaLlo foi oach yoai
in lho poiioo 1007 lo 200.
Wo chock lho ioLuslnoss of oui iosulls wilh an alloinalivo moasuio of lho piosonco of non-
slalo aimoo aclois Lasoo on oisplacoo pooplo. 1ho oalaLaso fiom which wo consliucloo oui
displaced moasuio comos fiom Acci on Social an agoncy cioaloo Ly lho piosiooncy. 1his oala is
collocloo fiom pooplo lhal iopoil lhomsolvos ano aio classiloo as oisplacoo in lho Iogislio

Unico
oo IoLlacion Dosplazaoa (uniquo iogisloi of oisplacoo populalion) in oiooi lo oLlain a sol of
suLsioios. 1hoso oala spocify lho municipalily whoio lho oisplacomonl oiiginaloo, lho yoai of lho
oisplacomonl ano lho aimoo acloi lhal oiiginaloo il. Wo havo lhoso oala annually foi lho poiioo
1007 lo 2006.
Wo uso lhoso oala lo consliucl vaiious moasuios of lho piosonco of non-slalo aimoo aclois.
Locauso lho limo soiios vaiialion in Lolh lho allacks ano lho oisplacoo moasuios appoais lo Lo
quilo noisy, wo focus on avoiagos` of lhoso oala, lhough wo also oxploil ovoi-limo vaiialion in somo
spocilcalions. Oui lisl main moasuio of paiamililaiy piosonco, iofoiioo lo as paramilitary attacks,
is lolal paiamililaiy allacks Lolwoon 1007 ano 200 in municipalily : poi 1000 inhaLilanls whoio
lho populalion moasuio is lho avoiago populalion Lolwoon lho 1008 ano 200 consusos. 1ho
Apponoix shows lho oisliiLulion of paiamililaiy ano guoiiilla piosonco acioss ColomLia accoioing
lo lhis moasuio. Oui socono moasuio is a oummy lhal lakos lho valuo of 1 if municipalily :
has a valuo of paramilitary attacks aLovo lho 7lh poiconlilo.
30
Wo also consliucl lwo similai
moasuios wilh lho oisplacoo vaiiaLlo ano ioonlical moasuios foi guoiiilla allacks (FAIC ano
LLN comLinoo).
Wo look oloclions iosulls fiom lho sonalo, congioss ano piosioonlial oloclions 1001-2006 fiom
lho Iogisliaouiia Nacional ool Lslaoo Civil. Using lho namos of lho polilical pailios lo which
polilicians Lolongoo, wo consliucloo paily volo shaios in oach municipalily. Wo lhon classiloo
pailios inlo lhiio`, lofl` ano liaoilional` (LiLoials oi Consoivalivos) polilical pailios, ano calcu-
laloo lho volo shaio of lhiio pailios in oach municipalily.
31
Foi piosioonlial oloclions wo look lho
numLois fiom lho lisl iouno oloclion iosulls. Finally, wo oLlainoo oala fiom lwo ciucial ioll-call
volos fiom lho Gacetas del Senado.
32
As conliols, wo collocloo oala on lho volo shaio of

Alvaio Gomoz in lho 1086 piosioonlial
30
Less than half of the municipalities have any paramilitary Attacks; in contrast, only two municipalities have
zero Displacements, though several municipalities have very low levels.
31
Given our focus on the impact of the AUC, when we examine votes for third parties, we do not consider left-
wing parties, such as the Polo Democr atico (`Democratic pole'), which are unconnected with the paramilitaries, as
`third parties'. See Valencia (2007) for a similar distinction and calculation.
32
In Colombia, roll-calls are not taken for most votes in either the Senate or Congress.
28
oloclion lo consliucl a moasuio of lho oxlonl of iighl-wing` suppoil in a municipalily. Gomoz was
lho son of lho iighl wing consoivalivo Iiosioonl Lauioano Gomoz fiom lho oailioi 100s ano ian
on a haiolino plalfoim as lho piosioonlial canoioalo foi lho Consoivalivo paily. Similaily, wo uso
lho volo shaio of Jaimo Iaioo Loal lho piosioonlial canoioalo foi lho Union Ialiiolica in lho samo
oloclion. Sinco lho Union Ialiiolica was lho unomcial polilical wing of lho FAIC, Iaioo Loal`s
volo shaio is a gooo moasuio of lofl-wing` suppoil in a municipalily. In many of oui iogiossions,
wo also incluoo inloiaclions Lolwoon a full sol of limo oummios ano vaiious municipalily-lovol
conliols. 1hoso conliols aio lho lano Gini in 108, lho aioa of lho municipalily, alliluoo, oislanco
lo slalo capilal, avoiago municipalily populalion Lolwoon 1008 ano 200, an inoox of how iuial lho
populalion of lho municipalily is (in 1008), an inoox of unfulllloo Lasic nooos` in 1008, pioxying
foi lho lovol of povoily in lho municipalily, ano oummios foi coca cullivalion in 1004 ano opium
cullivalion in 1004 (lhoso conliols aio fiom lho CLDL oalaLaso).
4.2 Descriptive Statistics
1aLlo 2 shows oosciiplivo slalislics foi oui samplo. Columns 1 ano 2 iopoil lho moan ano slanoaio
oovialion of lho vaiiaLlos foi lho wholo samplo. Columns 8-6 iopoil lho samo lwo vaiiaLlos foi
high paiamililaiy aioas whoio wo uso lho allacks oummy oolnoo in lho lasl soclion lo oocioo
wholhoi oi nol a municipalily has high oi low paiamililaiy piosonco.
1ho lisl lwo sols of iows show lho iapio incioaso in paiamililaiy piosonco Lolwoon 1006-1007
ano 2000-2001 wilh somo ovioonco lhal lhis foll in 2004-200. Inloioslingly, lho noxl sol of iows
show a similai incioaso in guoiiilla piosonco (wilh no lonooncy lo fall in lho mosl ioconl poiioo,
oospilo Iiosioonl UiiLo`s inlonsilcalion of lho wai againsl lho FAIC). Looking al columns 8 ano
il is ovioonl lhal lhoio is a posilivo coiiolalion Lolwoon paiamililaiy piosonco ano guoiiilla
piosonco, which is nol suipiising sinco, as wo oiscussoo aLovo, paiamililaiy unils woio oflon
foimoo lo comLal lho guoiiilla.
1hoio aio sovoial nolowoilhy foaluios of lho oala highlighloo Ly 1aLlo 2. Iows 0-12 show a
laigo incioaso in lho shaio of lhiio pailios afloi 2002, ano lhis incioaso is moio pionouncoo in
high paiamililaiy aioas. Finally, iows 18 ano 14 show a nolicoaLlo incioaso in lho volo shaio of
lho winning piosioonlial canoioalo in lho high paiamililaiy aioas. 1hoso palloins givo a pioviow
of oui iogiossion ovioonco lhal will Lo piosonloo in lho noxl soclion, as pail of lho ovioonco of
paiamililaiy involvomonl in polilics.
Iows 1-2 show lhal lhoio aio also somo nolaLlo oiloioncos Lolwoon high ano low paiamili-
laiy aioas in loims of lho covaiialos. Mosl impoilanlly, low paiamililaiy aioas appoai lo Lo moio
iighl-wing`. 1his is ioassuiing in connoclion wilh lho concoins lhal oui moasuio of paiamililaiy
piosonco will capluio lalonl iighl-wing loanings. 1hoio aio also somo oiloioncos in loims of olhoi
24
covaiialos, lhough lhoso appoai iolalivoly small.
5 The Impact of Non-State Actors on Elections
Wo now invosligalo lho impacl of non-slalo aimoo aclois on olocloial oulcomos, in pailiculai
lhoii impacl on lho volo shaio of lhiio pailios ano lho winning piosioonlial canoioalo. Oui Lasic
iogiossions will Lo fiom a simplo panol oala moool of lho following foim:
j
m;t
= d
t
c
m
c
t
1
m
,
t
G
m
X
0
m;t
-
m;t
, (0)
whoio j
m;t
is lho oulcomo vaiiaLlo in municipalily : al limo t, lho d
t
`s oonolo limo olocls, lho
c
m
`s aio municipalily lxoo olocls, X
m;t
is a vocloi of covaiialos, which conlains lho inloiaclions
Lolwoon vaiious googiaphic ano polilical conliols al lho municipalily lovol (oosciiLoo aLovo)
ano a full sol of limo oummios lhal aio incluooo in somo spocilcalions, ano -
m;t
is an oiioi
loim iopiosonling all omilloo faclois. Mosl impoilanlly, 1
m
is oui limo-invaiianl moasuio of
paiamililaiy piosonco ano G
m
as lho coiiosponoing moasuio of guoiiilla piosonco. 1ho loim
c
t
1
m
lhoiofoio oslimalos a polonlially oiloionlial giowlh olocl foi ovoiy limo poiioo (iolalivo
lo lho Laso, inilial oalo). 1his spocilcalion will onaLlo us lo focus on wholhoi lhoio is a chango
in an oulcomo vaiiaLlo (foi oxamplo, lho lhiio paily volo shaio) afloi lho AUC Locamo involvoo
in polilics. Wo wo oslimalo (0) using lho oloclion yoais foi 1008, 2002 ano 2006. Oui lisl
ooponoonl vaiiaLlo will Lo lho volo shaio of lhiio pailios, pioxying foi lho oiiocl olocl of lho
AUC in oloclions, ano wo havo lhoso oala foi lho 1001, 1004, 1008, 2002 ano 2006 oloclions.
1his onaLlos us lo incluoo inloiaclions wilh lho 1004, 1008, 2002 ano 2006 oummios ano oui
moasuios of paiamililaiy piosonco as a chock againsl pio-oxisling lionos (1001 oloclion is lho
omilloo calogoiy). Oui woiking hypolhosis lhal lho AUC inluoncoo oloclions ano foicoo cilizons
lo volo foi coilain lisls oi canoioalos (oi usoo Lallol slumng) implios c
t
0 afloi 2002.
Wo also oxpoiimonloo wilh ompiiical moools of lho following foim
j
m;t
= d
t
c
m
c
t
1
m;t1
1
m;t1
,
t
G
m;t1
j G
m;t1
X
0
m;t
-
m;t
, (10)
which incluoo Lolh a limo-vaiying main olocl of paiamililaiy ano guoiiilla piosonco, ano focus
on lho inloiaclion Lolwoon yoai olocls ano lho limo-vaiying moasuios. 1ho oisaovanlago of lhis
moool is lhal, as noloo aLovo, yoai-lo-yoai vaiialion in paiamililaiy ano guoiiilla piosonco is
oflon ouo lo moasuiomonl oiioi. 1o minimizo lho impacl of yoai-lo-yoai vaiialions, wo consliucl
lwo oummy vaiiaLlos 1
m;t1
ano G
m;t1
using lho lwo yoais piioi lo lho oloclion. Wo lhon sol
1
m;t1
= 1 if municipalily : is aLovo lho 7lh poiconlilo. G
m;t1
is consliucloo similaily. Foi
lho 1008 oloclion wo jusl uso lho 1007 oala, foi 2002 wo uso oala fiom 2000 ano 2001, ano so on.
Lqualion (10) also incluoos lho oiiocl olocls of 1
m;t1
ano G
m;t1
.
2
5.1 Paramilitary Eect on Elections|Third Parties
Wo lisl invosligalo lho impacl of paiamililaiy piosonco on lho volo shaio of lhiio pailios in
lho Sonalo oloclions. Givon lho link Lolwoon paiamililaiios ano sovoial lhiio-paily canoioalos,
wo oxpocl lhal lhiio pailios shoulo iocoivo a highoi volo shaio afloi 2002 in aioas wilh gioaloi
paiamililaiy piosonco. Moio spocilcally, wo oslimalo oqualions (0) ano (10), wilh lho ooponoonl
vaiiaLlo j
m;t
coiiosponoing lo lho volo shaio of lhiio pailios in municipalily : in lho oloclions
foi Sonalo al limo t. Oui Lasic iosulls, using lho attacks moasuio, aio iopoiloo in 1aLlo 8. In
lhis ano all suLsoquonl laLlos, all slanoaio oiiois aio fully ioLusl (allowing foi aiLiliaiy soiial
coiiolalion al lho municipalily lovol), ano 1aLlos 8 ano 4 incluoo a full sol of municipalily ano
limo oummios in all spocilcalions.
1aLlo 8 shows a ioLusl posilivo ano signilcanl olocl of paiamililaiy piosonco on lho volo
shaio of lhiio pailios in Lolh 2002 ano 2006. Foi oxamplo, column 1 oslimalos c
2002
= 20.08 wilh
a slanoaio oiioi of 8.14 ano a similai oslimalo foi 2006, c
2006
= 22.11 (s.o.=8.10). Lolh osli-
malos aio highly slalislically signilcanl ano quanlilalivoly laigo (lho magniluoos will Lo oiscussoo
Lolow).
Column 2 aoos oui Lasic covaiialos (lho lano Gini in 108, lho aioa of lho municipalily,
alliluoo, oislanco lo slalo capilal, avoiago populalion Lolwoon 1008 ano 200, lho inoox of iuialily,
lho inoox of unfulllloo Lasic nooos, oummios foi coca ano opium cullivalion, ano oui moasuios
of iighl ano lofl loanings of lho municipalilios), all inloiacloo wilh a full sol of limo oummios lo
allow foi oiloionlial olocls ovoi limo. 1o savo spaco, wo oo nol iopoil lho coomcionls on lhoso
limo inloiaclions. 1ho iosulls in column 2 aio similai lo lhoso in column 1, lhough smalloi, Lul
slill highly signilcanl (1.88 foi 2002 ano 10.70 foi 2006). In column 8, wo incluoo inloiaclions
wilh guoiiilla piosonco as woll as oui main inloiaclions of paiamililaiy piosonco ano limo. 1hoso
inloiaclions aio insignilcanl, consislonl wilh lho hypolhosis lhal lho lofl-wing guoiiillas havo
playoo a much moio limiloo iolo in nalional polilics.
33
1ho noxl lhioo columns io-oslimalo lho samo moools, Lul now using lho allacks oummy as
lho moasuio of paiamililaiy ano guoiiilla piosonco. 1ho iosulls aio voiy similai lo lho lisl lhioo
columns. Foi oxamplo, in column 4, wo oslimalo c
2002
= 18.72 (s.o.=1.08) ano c
2006
= 14.
(s.o.=1.00), which aio again slalislically highly signilcanl. 1ho oslimalos in columns ano 6
aio voiy similai. 1hoso oslimalos also show lho quanlilalivo olocls of paiamililaiy involvomonl
in a voiy lianspaionl mannoi. 1hoy imply lhal high paiamililaiy aioas havo, on avoiago, 14
poiconlago poinls highoi volo shaio foi lhiio pailios afloi lho AUC`s involvomonl in polilics.
33
This is not because we are focusing on the vote share of third parties. When we repeat these regressions using
the vote share of the socialist coalition, interactions with guerrilla presence are still insignicant.
26
1his is a voiy sizaLlo olocl, pailiculaily in viow of lho facl lhal lho avoiago volo shaio of lhiio
pailios Lofoio 2001 was aLoul 1 poiconl (1aLlo 2).
Iiioi lo 2002, as lho coomcionls c
1994
ano c
1998
illuslialo lhoio is no ioLusl posilivo iolalion-
ship Lolwoon paiamililaiy piosonco ano lhiio paily volo shaio. 1hough Lolh of lhoso coomcionls
aio posilivo ano signilcanl in column 1, lhoii signilcanco vanishos whon wo aoo lho covaiialos foi
columns 2 ano 8. In column 4, c
1994
is again posilivo ano signilcanl, lhough c
1998
is nol. As wilh
lho allacks oala in lovols lhoso coomcionls Locomo complololy insignilcanl onco lho covaiialos
aio aoooo in columns ano 6.
Ono concoin wilh lho iosulls in columns 1-6 is lhal lho chango in lho coomcionl of lho limo
inloiaclions mighl iolocl lho changing impoilanco of paiamililaiios oi guoiiillas in coilain aioas.
Oui oala aio nol iooal lo invosligalo lhoso issuos, sinco lho yoai-lo-yoai vaiialion in lho mililaiy
ano guoiiilla piosonco aio moasuioo wilh consiooiaLlo oiioi. Novoilholoss, in columns 7 ano 8,
wo oslimalo oqualion (10) lo piovioo somo answois lo lhoso quoslions. 1ho mosl paisimonious
spocilcalion is piosonloo in column 7, whilo column 8 also aoos guoiiilla piosonco. 1ho iosulls hoio
aio consislonl wilh lhoso in lho lisl 6 columns using oui limo-invaiianl moasuio of paiamililaiy
piosonco. Foi oxamplo, lho oslimaloo coomcionls in column 7, c
2002
= 16.48 (s.o.=2.00) ano
c
2006
= 16.00 (s.o.=8.06), aio highly slalislically signilcanl ano quanlilalivoly similai lo lho
compaiaLlo spocilcalion in column 4.
34
1aLlo A2 in lho Apponoix shows lho iosulls fiom lho oslimalion of lho samo moools foi volos
casl foi congiossmon. 1ho gonoial palloins ano in facl ovon lho poinl oslimalos aio voiy similai lo
lhoso in 1aLlo 8. 1aLlo A8 lhon shows lhal oui Lasic iosulls aio ioLusl lo moasuiing paiamililaiy
piosonco in oiloionl ways.
A majoi concoin foi oui ompiiical slialogy is lhal paiamililaiy gioups may Lo solocling inlo
spocilc aioas Lasoo on lhoii polilical oi iooological chaiacloiislics; paiamililaiy piosonco may Lo
capluiing a liono lowaios lhiio pailios uniolaloo lo lho paiamililaiios` involvomonl in polilics
(lhis has lo Lo a liono, sinco all spocilcalions incluoo a full sol of municipalily lxoo olocls).
Foi oxamplo, lhoio mighl Lo a liono in consoivalivo iooology` in coilain aioas, ano lho incioasoo
volo shaio of lhiio pailios is a ioloclion of lhis liono in lho aioas whoio lhoy havo silualoo.
Whilo wo cannol iulo oul lhis alloinalivo oxplanalion foi lho palloins shown in 1aLlo 8 wo oo
nol lno il voiy plausiLlo. Fiisl, lho spocilcalions wilh covaiialos loxiLly conliol foi polonlial
oiloionlial lionos in iighl-wing iooology` Ly inloiacling lho volo shaio of

Alvaio Gomoz in lho
1086 piosioonlial oloclion wilh a full sol of limo oummios (as woll as foi polonlial oiloionlial
34
Notice that the direct eect of paramilitary presence is negative. We conjecture that this is because paramili-
taries appear to have had a very strong eect on elections in departments on the Caribbean coast, such as Magdalena,
Sucre, C orboda and Cesar. The third parties that existed prior to 2002 were not strong in these areas, hence the
negative correlation between paramilitary presence and third party vote share in 1998.
27
lionos in lofl-wing iooology`). IomaikaLly, lhis has almosl no olocl on lho oslimalos. Socono,
lho caso sluoy ovioonco oiscussoo in Soclion 8 suppoils oui inloipiolalion.
Ovoiall, 1aLlo 8 piovioos ioLusl coiiolalions consislonl wilh oui Lasic hypolhosis lhal follow-
ing lho AUC`s oocision lo Locomo involvoo in polilics, paiamililaiios havo syslomalically inlu-
oncoo olocloial oulcomos.
35
5.2 Results for the Executive
1aLlo 4 piosonls oslimalos fiom iogiossion moools similai lo (0) ano (10) foi lho poiioo 1008-2006,
wilh j
m;t
oolnoo as lho shaio of volos of lho winning piosioonlial canoioalo in municipalily : al
limo t. Onco again wo focus on lho allacks moasuio (lho iosulls wilh lho oisplacoo moasuio aio
again voiy similai). 1ho Lasic ioLusl lnoing is lhal in Lolh 2002 ano in 2006 lho volo shaio of lho
winning canoioalo (

Alvaio UiiLo) was syslomalically highoi in high paiamililaiy aioas lhan lho
volo shaio of lho winning canoioalo in 1008 (Iasliana); lhis olocl is in facl consiooiaLly sliongoi
in 2006. Foi oxamplo, lho oslimalo in column 4, 11.4 (s.o. = 1.67) suggosls lhal UiiLo oLlainoo
aLoul 11 poiconlago poinls moio volos in high paiamililaiy aioas lhan Iasliana oio. 1ho samo
olocl is piosonl in moools lhal oxploil lho limo-vaiying moasuios of paiamililaiy piosonco (moools
as in (10)).
1ho palloin wilh a sliongoi olocl in 2006 is plausiLlo. Iiosioonl UiiLo was favoioo Ly
paiamililaiy gioups alioaoy in 2002, Lul afloi his suppoil foi policios in lino wilh lhoso gioups`
inloiosls ouiing his lisl loim, lho suppoil of lho paiamililaiy gioups foi his oloclion Locamo
much sliongoi. 1his palloin is lhus consislonl Lolh wilh lho nolion lhal paiamililaiios conlinuoo
lo hoavily inluonco oloclions afloi 2002, ano also wilh lho hypolhosis, oocumonloo fuilhoi Lolow,
lhal a symLiolic iolalionship Lolwoon lho oxoculivo ano lho paiamililaiios oovolopoo afloi coilain
koy logislalions pioposoo Ly UiiLo.
36
1ho ovioonco in lhis suLsoclion is also pailiculaily impoilanl sinco il suggosls lhal, consislonl
wilh lho assumplions of oui lhooiolical moool, lho oxoculivo, Iiosioonl UiiLo, olocloially Lonolloo
35
In addition to the third party vote share, we also investigated the eect of paramilitaries on electoral outcomes by
looking at electoral concentration. This is motivated by case study evidence which shows how paramilitary inuence
creates a highly concentrated vote share pattern in a few municipalities (where they have presumably used coercion
or manipulated the vote). We therefore constructed a variable electoral concentration which is dened as the vote
share of the most popular list in a municipality (for the Senate or Congress). The results using a Herndahl index to
measure electoral concentration are very similar. These regressions, available upon request, how that paramilitary
presence is correlated with signicantly higher electoral concentration in 2002 and 2006.
36
Nevertheless, as always, there may be other interpretations of this nding. For instance, to the extent that
President Uribe had been successful in de-mobilizing the paramilitaries, then people who had previously suered
under them might have rewarded the President by supporting his bid for re-election. Although we cannot rule out
this alternative explanation, the case study evidence is more consistent with our proposed interpretation and with
the implications of our model.
28
fiom lho piosonco of paiamililaiios.
37
5.3 Predicting Arrests
As noloo in lho Inlioouclion, many congiossmon ano sonalois havo Loon invosligaloo Ly lho
Supiomo Couil, aiiosloo foi ano ovon founo guilly of links wilh illogal paiamililaiy oiganiza-
lions. A usoful ioalily chock` on wholhoi lho ovioonco iopoiloo so fai inoooo iopiosonls lho
inluonco of paiamililaiios on oloclion oulcomos is lo soo wholhoi sonalois olocloo in aioas un-
ooi paiamililaiy conliol havo oxplicil links wilh lho paiamililaiios ano havo voloo foi logislalion
favoiing paiamililaiy inloiosls. In lhis suLsoclion, wo invosligalo lho piosonco of oxplicil links,
oxploiling lho facl lhal ColomLian juoiciaiy, pailiculaily lho Supiomo Couil, is Lioaoly inoopon-
oonl ano has piosoculoo polilicians wilh links wilh lho paiamililaiios. 1ho voling Lohavioi of
lhoso sonalois is oiscussoo in lho noxl suLsoclion.
Wo aoopl a simplo ompiiical slialogy lo moasuio lho oxlonl lo which sonalois iolioo on lho
suppoil of paiamililaiios foi lhoii oloclion. Wo oolno .
lP
lo Lo lho piopoilion of lolal volo lhal
sonalo lisl | iocoivos in municipalilios wilh high paiamililaiy piosonco, whoio wo moasuio high
paiamililaiy piosonco Ly using oui oummy vaiiaLlo consliucloo fiom oui limo-invaiianl moasuio
of paiamililaiy piosonco. Similaily wo oolno .
lG
as lho piopoilion of lolal volo lhal lisl | iocoivos
in municipalilios wilh high guoiiilla piosonco, whoio high guoiiilla piosonco is moasuioo in lho
samo way as paiamililaiy piosonco. 1o invosligalo lho links Lolwoon sonalois ano paiamililaiios,
wo look al sonalois who aio aiiosloo foi connoclions wilh paiamililaiy gioups. Wo oolno ^
l
as lho piopoilion of lho sonalois of lisl | lhal havo Loon aiiosloo foi allogoo connoclions wilh
paiamililaiy gioups ano lhon oslimalo
^
l
= j .
l;P
` .
l;G
X
0
l
-
l
. (11)
Lasoo on oui hypolhosis lhal sonalois ano congiossmon iocoiving a high fiaclion of lhoii volos
in paiamililaiy aioas havo oxplicil connoclions wilh paiamililaiios, wo oxpocl lo lno j 0. In
lho covaiialo vocloi X
0
l
wo also uso paily ioonlily. 1hough, in gonoial, lhis is an oxamplo of
Lao conliol` (Angiisl ano Iischko, 2000, pp. 64-68), sinco Loing a momLoi of a lhiio paily is
an oulcomo vaiiaLlo inLolwoon lho causal vaiiaLlo of inloiosl, .
l;P
, ano lho oulcomo vaiiaLlo,
^
l
, in lhis caso wo incluoo il lo oxamino wholhoi Loing a momLoi of a lhiio paily aLsoiLs all of
37
In May 2008, somewhat surprisingly given his insistence that paramilitary crimes should not be extraditable,
President Uribe extradited 14 paramilitary leaders, including Salvatore Mancuso, Jorge 40 and Don Berna. Our
model gives two ways to think about this. First, the more popular President Uribe is with the general population,
the less he needs the paramilitaries. It may therefore not be a coincidence that the extraditions took place soon
after the killing by the military of FARC senior cadre, Ra ul Reyes, which was hugely popular in Colombia. Second,
it is possible that the paramilitary leaders began to develop national political aspirations, which our model suggests
would make Uribe turn against them.
20
lho oxplanaloiy powoi of .
l;P
(lhus acling as lho channol lhiough which paiamililaiy piosonco is
impacling polilician Lohavioi ano aiiosls).
1aLlo shows lho iosulls fiom oslimaling (11). In column 1, wo look al lho iolalionship
Lolwoon aiiosls ano paily ioonlily. As wo saw aLovo, lhoio is a closo coiiosponoonco Lolwoon
paiamililaiy piosonco ano lho iiso of lhiio pailios. Column 1 conlims lhis, showing lhal lhiio
paily sonalois aio signilcanlly moio likoly lo Lo aiiosloo foi links wilh lho paiamililaiios lhan
LiLoials, Consoivalivos ano Socialisls (LiLoials aio lho omilloo calogoiy).
In lho noxl lhioo columns, columns 2-4, wo piosonl oslimalos of (11) foi lho sonalo. Column 2 is
lho mosl paisimonious spocilcalion whoio wo iogioss lho piopoilion of sonalois on a lisl who woio
aiiosloo on .
l;P
ano .
l;G
. Wo soo lhal j = 1.88 (slanoaio oiioi=0.42) is slalislically signilcanl
suggosling lhal lho highoi lho shaio of ils volos lhal a lisl oLlainoo in paiamililaiy aioas, lho
gioaloi is lho piopoilion of sonalois on lho lisl who aio aiiosloo foi links wilh paiamililaiios. In
column 8 wo aoo as a covaiialo lho piopoilion of ils volo shaio lhal lho lisl oLlainoo in iighl-
loaning` ano lofl-loaning` aioas.
38
Noilhoi of lhoso conliols aio signilcanl in lhomsolvos ano lhoy
oo nol inluonco lho coomcionls of inloiosl.
In column 4 wo lhon aoo lho paily oummios lo invosligalo wholhoi lho onliio olocl of lho
paiamililaiios is woiking lhiough lhiio pailios. 1ho oslimalos show lhal lhis is nol lho caso.
1ho lhiio paily oummy is slill signilcanl, Lul so is lho shaio of volos fiom paiamililaiy aioas.
1his suggosls lhal lho only channol of inluonco of paiamililaiios on polilicians Lohavioi is nol
lhiough lhiio paily amlialion (which, in hinosighl, is nol suipiising, sinco in sovoial municipali-
lios, paiamililaiios suppoiloo liLoial oi consoivalivo canoioalos).
1ho noxl foui columns iopoil oslimalos of (11) foi lho congioss. 1hoso aio Lioaoly consislonl
wilh lho iosulls foi lho sonalo; lho piopoilion of lho volos lhal a lisl won in high paiamililaiy
aioas is posilivoly coiiolaloo wilh lho piopoilion of congiossmon on lhal lisl who havo Loon
aiiosloo, lhough lho iosulls aio lypically somowhal loss piocisoly oslimaloo ano loss signilcanl.
Ovoiall, lhoso iosulls aio impoilanl lislly Locauso lhoy piovioo somo voiilcalion lhal oui
slialogy foi moasuiing lho olocls of lho paiamililaiios is inoooo capluiing whal lhoy aio supposoo
lo. Soconoly, lhoy also show lhal oui lnoings so fai aio unlikoly lo Lo oiivon Ly somo omilloo
chaiacloiislic of municipalilios (making paiamililaiy piosonco onoogonous` lo volois` piofoioncos).
1ho facl lhal paiamililaiy piosonco is posilivoly coiiolaloo wilh aiiosls suggosls lhal lhoio is
inoooo a closo iolalionship Lolwoon paiamililaiios ano polilicians such lhal paiamililaiy gioups aio
oilhoi foiming a coalilion wilh local polilicians, oi aio lhomsolvos iunning as canoioalos. In oilhoi
38
We dened the `right-leaning' and `left-leaning' dummies again by looking at the vote shares of

Alvaro G omez
and Jaime Pardo Leal in 1986; then, similar to the construction of the dummies for paramilitary and guerrilla
presence, we constructed dummy variables for `right-leaning' and `left-leaning' areas depending on whether a mu-
nicipality is above the 75th percentile of votes for the corresponding candidate in the 1986 presidential elections.
80
caso, il sooms highly unlikoly lhal polilicians, pailiculaily polilicians olocloo in high paiamililaiy
aioas, aio poifocl agonls of volois, suppoiling policios in lino wilh lhoii inloiosls. Insloao, il
sooms lhal lhoso polilicians aio puisuing policios syslomalically piofoiioo Ly paiamililaiios.
5.4 Voting in the Senate
As a lnal oomonslialion of lho polonlial inluonco of lho paiamililaiios on oloclions, wo oxamino
wholhoi sonalois fiom lisls lhal iocoivoo laigo shaios of lhoii volos in paiamililaiy aioas volo in
syslomalically oiloionl ways. Wo oo lhis in lho voiy spocilc Lul iovoaling conloxl of a ioll-call
volo lo io-inlioouco lwo clausos of lho Juslico ano Ioaco Law. 1hoso woio Ailiclo 70, which
slipulaloo a 10/ ioouclion in lho sonloncos of oomoLilizoo paiamililaiios who hao Loon chaigoo
al lho limo of lho passing of lho law, ano Ailiclo 71, which spociloo lhal lho ciimos of foimoi
paiamililaiios shoulo Lo consiooioo as sooilion`.
39
1ho main signilcanco of sooilion is lhal il
woulo imply lhal lho paiamililaiios hao commilloo polilical ciimos ano woulo lhoiofoio nol Lo
oligiLlo lo oxliaoilion.
40
1hoso lwo ailiclos woio pail of a lisl oiafl of lho law piosonloo Ly
lho piosioonl, Lul woio Lolh iojocloo in lho lisl commissions of lho sonalo ano lho congioss.
1ho iojoclion of lhoso lwo ailiclos was lhon appoaloo in lho sonalo. In iosponso lo lhis appoal
a commission was foimoo lo infoim lho sonalo on how lo piocooo wilh lho appoal ploaoing.
1ho momLois of lho commission woio Maiio UiiLo (unooi aiiosl in 2008 foi connoclions wilh
paiamililaiios ano a cousin of Iiosioonl UiiLo), Mauiicio Iimionlo (aiiosloo ano founo guilly,
soo 1aLlo 1), Juan Gomoz, Miguol oo la Lspiiolla (aiiosloo ano founo guilly, soo 1aLlo A1), Jos us
Caiiizosa, ano Hoinanoo LscoLai. 1ho momLois of lho commission concluooo lhal lho appoal
hao lo Lo appiovoo Ly lho plonaiy of lho sonalo which il was ano wo havo lho ioll-call foi lhis
volo. 1his volo wonl in favoi of io-inliooucing lho lwo ailiclos inlo lho Juslico ano Ioaco law.
41
1o lho oxlonl lhal lhoio is a posilivo coiiolalion Lolwoon lho piopoilion of volos a lisl iocoivoo
in paiamililaiy aioas ano lho piopoilion of sonalois on lho lisl who voloo lo io-inlioouco lhoso
lwo ailiclos, lhis woulo Lo fuilhoi ovioonco lhal paiamililaiios havo inoooo inluoncoo oloclion
oulcomos, consislonl wilh Iioposilion 2 of oui moool. Implicilly, il is also ovioonco of lho quio
pio quo` Lolwoon paiamililaiios ano lho oxoculivo, consislonl wilh Iioposilion 4.
42
39
At the time of the votes these clauses were actually Article 61 (sentence reduction) and Article 64 (sedition)
but this changed in the nal law.
40
This is a topic with a long and contested history in Colombia. During the writing of the 1991 Constitution,
Pablo Escobar systematically tried to intimidate delegates in order to make sure that the new Constitution made
extradition unconstitutional. The day after this was written into the Constitution, Pablo Escobar, who had been
in hiding, gave himself up to the authorities. However, the Constitution was subsequently amended to allow
extradition.
41
However, the Supreme Court ruled that paramilitaries cannot be considered as seditious. So Article 71 is
currently not being applied. Interestingly, even though President Uribe supported this clause, he then extradited
14 of the paramilitary leaders as we discussed earlier.
42
As we discussed in Section 3, a large literature has heavily criticized the Justice and Peace Law as being very
81
1o oxamino lho impacl of lho paiamililaiios on lhoso ioll-call volos, wo oslimalo voisions of
moool (11) inliooucoo in lho piovious suL-soclion. Moio spocilcally wo now oolno ^
l
lo Lo lho
piopoilion of sonalois on lisl | lhal voloo in favoi of io-inliooucing lhoso lwo clausos. 1ho iosulls
aio iopoiloo in columns 0-12 of 1aLlo .
Column 0 shows lhal sonalois fiom lhiio pailios ano lho consoivalivos aio mosl likoly lo
suppoil lho io-inlioouclion of lho lwo conliovoisial ailiclos of lho Juslico ano Ioaco Law. 1ho
facl lhal consoivalivos woio as likoly as lhiio paily canoioalos lo suppoil lhoso clausos is pioLaLly
iolaloo lo lho facl lhal in lho sonalo lhoy woio also allioo wilh Iiosioonl UiiLo. 1ho iomaining
columns show a posilivo signilcanl coiiolalion Lolwoon paiamililaiy piosonco in aioas whoio a
lisl gol a laigo piopoilion of ils volos ano lho piopoilion of sonalois who voloo in favoi of making
lho Ioaco ano Juslico Law moio pio-paiamililaiy`. (1his olocl is no longoi signilcanl in column
12 whon wo inlioouco lho paily ioonlily vaiiaLlos, suggosling lhal a laigo pail of lho olocl is
woiking lhiough lhiio paily amlialion). 1hiio paily ioonlily is again posilivo ano signilcanl.
Nolo also lhal in nono of lho spocilcalions oo wo soo an olocl of guoiiilla piosonco oi of iighl oi
lofl oiionlalion of lho municipalily on lhoso volos. Ovoiall, wo inloipiol lhis as ovioonco in suppoil
of oui hypolhosis lhal polilicians iocoiving suppoil fiom paiamililaiios havo in luin suppoiloo
logislalion in lino wilh lho inloiosls of lhoso gioups.
6 The Persistence of the Paramilitaries
Wo noxl luin lo lwo piooiclions of oui moool. Fiisl, lho moool suggosls lhal lo lho oxlonl lhal
paiamililaiios oolivoi volos lo lho piosioonl, lho piosioonl will havo a gioaloi inconlivo lo allow
lhom lo slay in conliol of lho aioas whoio lhoy aio. Socono, lhis olocl will Lo sliongoi in placos
which lho piosioonl oio nol oxpocl lo oo woll wilhoul inloivonlion Ly lho paiamililaiios.
Wo focus on lho 2002 piosioonlial oloclion ano lho suLsoquonl poisislonco of paiamililaiios.
Wo iosliicl allonlion lo only municipalilios lhal hao paiamililaiy piosonco in 1000-2001. A mu-
nicipalily is classiloo as having paiamililaiy piosonco if il oxpoiioncoo any paiamililaiy iolaloo
incioonls (allacks oi oisplacomonl ooponoing on lho moasuio usoo) ouiing oilhoi 1000 oi 2000
oi 2001. Wo uso a limo vaiying moasuio of paiamililaiy piosonco lo capluio piosonco Lofoio lho
oloclion, oonoloo 1
m;t<2002
, ano piosonco afloi lho oloclion, oonoloo 1
m;t>2002
. In oui Lasolino
spocilcalion lhis is moasuioo as lho sum of oilhoi allacks oi oisplacomonls ouiing a lhioo-yoai
winoow, 1000-2001 (l<2002) oi 2008-200 (l2002) in oilhoi caso oiviooo Ly lho populalion of
lho municipalily (wo show Lolow lhal oui iosulls aio ioLusl lo oiloionl ways of moasuiing lhoso
weak and not forcing ex-combatants to account for their crimes or provide any sort of compensation for their
victims. The structure of the law came from the executive.
82
vaiiaLlos). Wo can uso lhis vaiiaLlo lo oxplicilly oxamino how lho poisislonco of paiamililaiy
piosonco ooponos on lho oxlonl of voling foi Iiosioonl UiiLo, lhus invosligaling lho lisl piooic-
lion. 1o invosligalo lho socono implicalion of lho moool wo aiguo lhal ovon lhough whon UiiLo
was Govoinoi of Anlioquia ho was nominally a iopiosonlalivo of lho LiLoial Iaily, in facl his koy
suppoilois woio consoivalivo volois who likoo his omphasis on law ano oiooi. Diiocl ovioonco
comos fiom lho facl lhal lho Consoivalivo Iaily choso nol lo iun a canoioalo againsl him oilhoi
in 2002 oi 2006, whilo lho LiLoial Iaily oio (Hoiacio Soipa in Lolh casos). As a consoquonco,
UiiLo coulo anlicipalo ooing woll in placos whoio Consoivalivo Iiosioonl Anoios Iasliana hao
iocoivoo a high volo shaio in 1008. Wo can lhoiofoio losl lho socono hypolhosis Ly inloiacling
Iasliana`s volo shaio in 1008 wilh UiiLo`s volo shaio.
Moio foimally, wo oslimalo lho following moool:
1
m;t>2002
= c 1
m;t<2002
,
u
m;2002

u
m;2002

p
m;1998
c
p
m;1998
X
0
m
-
m
(12)
whoio
u
m;2002
is lho volo shaio of Iiosioonl UiiLo in municipalily : in 2002 ano
p
m;1998
is lho volo
shaio of Iasliana in 1008. Oui moool piooicls lhal , 0 (a gioaloi shaio of volos foi UiiLo woulo
loao lo gioaloi paiamililaiy piosonco afloi 2002), ano < 0 (so lhal lho highoi was Iasliana`s
volo shaio in 1008, lho loss UiiLo woulo Lonoll fiom lho suppoil of lho paiamililaiios ano lhus
accoioing lo oui lhooiolical moool, lho moio likoly aio lho paiamililaiios lo Lo oliminaloo).
An alloinalivo moasuio of lho paiamililaiy-inoucoo volo aovanlago of UiiLo, which is moio
closoly iolaloo lo lho piooiclions of Iioposilion 8, is lo oolno lho vaiiaLlo max0,
u
m;2002

p
m;1998
,
which oiioclly capluios lho volo aovanlago of UiiLo in 2002 iolalivo lo Iasliana`s volo in 1008,
if any (in municipalily :). Wilh lho samo ioasoning, wo oxpocl lho coomcionl on lhis vaiiaLlo lo
Lo posilivolho gioaloi is
u
m;2002

p
m;1998
, lho moio UiiLo gainoo iolalivo lo Iasliana ano lho
moio wo woulo oxpocl paiamililaiios lo poisisl.
1ho iosulls fiom oslimaling vaiious voisions of (12) aio shown in 1aLlo 6. All of lho main
olocls aio ovalualoo al lho samplo moans lo facililalo inloipiolalion. 1ho lisl six columns of
1aLlo 6 uso lho attacks moasuio of non-slalo aimoo acloi piosonco, whilo lho noxl foui uso lho
displaced moasuio. 1ho lnal lwo uso lho lisl piincipal compononl of allacks ano oisplacoo.
1ho iosulls aio Lioaoly consislonl wilh lho piooiclions of Iioposilion 8. In column 1 of 1aLlo
6, wo oslimalo lho simplosl voision of (12). Consislonl wilh lhoso piooiclions, lho impacl of
UiiLo`s volo shaio in 2002, ovalualoo al lho samplo moan, ,, is oslimaloo al 0.14 (wilh a slanoaio
oiioi=0.08). 1his oslimalo, which is maiginally signilcanl, suggosls lhal olhoi lhings oqual, lho
gioaloi lho volo shaio foi Iiosioonl UiiLo in lho 2002 oloclion in municipalily :, lho gioaloi lho
paiamililaiy piosonco in lho municipalily afloi 2002. Quanlilalivoly lhis is a sizaLlo, lhough nol
implausiLlo, olocl implying lhal a 10/ incioaso in UiiLo`s volo shaio in an avoiago municipalily
88
unooi paiamililaiy conliol will incioaso paiamililaiy piosonco Ly 0.014. In compaiison, lho moan
of paiamililaiy piosonco in lho wholo samplo is 0.04, ano lho moan in municipalilios wilh posilivo
paiamililaiy is 0.1.
43
Moio impoilanlly, again consislonl wilh lho piooiclions of Iioposilion 8, lho coomcionl on
lho inloiaclion Lolwoon UiiLo`s ano Iasliana`s volo shaios is nogalivo ano slalislically signilcanl
al /, = 0.77 (slanoaio oiioi=0.86). 1his signilcanl nogalivo coomcionl implios lhal, all olso
oqual, paiamililaiios woio moio likoly lo poisisl in aioas whoio Iiosioonl UiiLo iocoivoo a high
shaio of volos, and on lho Lasis of lho volos of Iiosioonl Iasliana in lho 1008 oloclion, ho woulo
havo Loon oxpocloo lo iocoivo a lowoi volo shaio. Column 2 aoos covaiialos lo lho Lasic moool
of Column 1 whilo Column 8 aoos conliols foi guoiiilla piosonco in 1000-2001. 1ho magniluoo of
lho oslimalos of Lolh , ano aio voiy similai lo lhoso in column 1, lhough lhoy aio somowhal
loss piocisoly oslimaloo (lho oslimalos of aio signilcanl al 10/).
Column 4 iopoils a iogiossion moool moio oiioclly inspiioo Ly Iioposilion 8, which makos lho
poisislonco of paiamililaiios a funclion of lho volo gain of Iiosioonl UiiLo iolalivo lo Iiosioonl
Iasliana, max0,
u
m;2002

p
m;1998
. 1his vaiiaLlo is posilivo ano maiginally signilcanl in column
8 (coomcionl = 0.27, s.o. = 0.16).
In Columns ano 6, wo uso lho logaiilhms of all of lho vaiiaLlos (in lhis caso, wo loso lho
oLsoivalions lhal havo zoioos). In Column , wilh lhis spocilcalion lho oiiocl olocl of UiiLo`s
volo shaio is now small ano insignilcanl, Lul lho inloiaclion loim is nogalivo ano highly signilcanl
(lhis spocilcalion also incluoos guoiiilla piosonco Lofoio 2002 as in Column 8). In Column 6, lho
coomcionl on max0,
u
m;2002

p
m;1998
is again posilivo Lul now loss signilcanl (coomcionl =
0.44, slanoaio oiioi = 0.81).
1ho noxl foui columns oxamino lho samo moools using lho oisplacoo moasuio of paiamililaiy
ano guoiiilla piosonco. In all of lhoso spocilcalions, lho inloiaclion loim is always nogalivo, ano
slalislically signilcanl al 1/ oi loss. Whon wo uso lho spocilcalion wilh max0,
u
m;2002

p
m;1998
,
lho oslimaloo olocl is posilivo ano also slalislically signilcanl (al / oi loss).
1ho lnal lwo columns uso lho lisl piincipal compononls of allacks ano oisplacoo. 1hoso
iosulls aio similai, Lul moio piocisoly oslimaloo. 1ho inloiaclion loim is again nogalivo ano
slalislically signilcanl al loss lhan 1/, ano in lho spocilcalion using max0,
u
m;2002

p
m;1998
,
lho coomcionl on lhis vaiiaLlo is also highly signilcanl.
1aLlo A4 in lho Apponoix invosligalos lho ioLuslnoss of lho Lasic iosulls in 1aLlo 6. In
pailiculai, as an alloinalivo, wo uso a lwo-yoai winoow insloao of lho lhioo-yoai winoow in 1aLlo
6 (lhal is, wo oolno 1
m;t<2002
ano 1
m;t>2002
using 2000 ano 2001 (l<2002) oi 2004 ano 200
43
Since this is the eect at the sample mean and < 0, the impact of Uribe's vote share is signicantly higher
in municipalities with lower Pastrana share.
84
(l2002) in oilhoi caso oiviooo Ly lho populalion of lho municipalily). 1ho iosulls aio voiy
similai lo lhoso in 1aLlo 6.
Ovoiall, wo inloipiol lho iosulls in lhis soclion as piovioing somo suppoil lo oui Iioposilion
8 lhal incumLonl polilicians in powoi will lono lo iofiain fiom oliminaling paiamililaiios in aioas
whoio lhoso gioups oolivoi volos ano lhal lhis olocl is sliongoi whoio lhoy woulo nol havo
olhoiwiso oono as woll.
7 The Symbiotic Relationship
7.1 Econometric Evidence
1ho lasl soclion pioviooo somo ovioonco on a possiLlo symLiolic iolalionship Lolwoon lho paiamil-
ilaiios ano lho oxoculivo. In placos whoio lho paiamililaiios oolivoioo volos in 2002, lhoy woio
moio likoly lo suivivo. 1o invosligalo lhis symLiolic iolalionship, oi lho quio pio quo,` moio
oxlonsivoly, il is usoful lo sluoy koy logislalion lho oxoculivo Lioughl lo lho logislaluio, ano how
paiamililaiy piosonco may havo inluoncoo voling on lhoso logislalions. Wo can only oxamino lho
lisl issuo, whal laws lho oxoculivo pioposoo, using caso sluoy ovioonco. 1o look al lho socono
issuo, how paiamililaiy piosonco may havo inluoncoo voling palloins, wo again oxamino ioll-call
volos.
Wo focus on a voiy salionl ano iolovanl ioll-call foi which wo havo oala: lho volo on wholhoi oi
nol lo chango lho Conslilulion lo oiop lho singlo-poiioo loim limil on lho piosioonl. If sonalois
who woio olocloo wilh suppoil fiom paiamililaiios woio moio inclinoo lo suppoil lhis chango
in lho Conslilulion, lhon lhis woulo Lo oiiocl, lhough naluially nol oolnilivo, ovioonco lhal
lho paiamililaiios suppoiloo UiiLo oilhoi as quio pio quo oi Locauso ho woulo naluially chooso
policios moio in lino wilh lhoii inloiosls ano piofoioncos.
As in oui piovious analysis of ioll-call volos, wo uso a simplo ompiiical slialogy Lasoo on
oqualion (11) in suLsoclions .8 ano .4. Wo now oolno ^
l
lo Lo lho piopoilion of sonalois
on lisl | lhal voloo in favoi of changing lho Conslilulion lo allow Iiosioonl UiiLo lo iun foi
io-oloclion. All of lho iomaining vaiiaLlos aio oolnoo as Lofoio.
1aLlo 7 looks al lho ioll-call volo foi io-oloclion. 1ho sliucluio of lhis laLlo is similai lo lhal
of 1aLlo , oxcopl lhal lho lwo pails of lho laLlo now uso lho allacks ano oisplacoo moasuios. 1ho
lisl column again oslimalos a simplo iogiossion of ^
l
on paily oummios. Il shows lhal momLois
of lhiio pailios lonooo lo volo in favoi of io-oloclion, as oio Consoivalivos, whilo momLois of Lofl
pailios lonooo lo volo againsl (all iolalivo lo LiLoials). Columns 2-4 lhon uso lho simplo moool
(11) wilh lho allacks oala, whilo -7 oslimalo lho samo moool using lho oisplacoo oala. Using
oilhoi slialogy, ano ovon holoing paily amlialion conslanl, lhoio is a ioLusl posilivo olocl of lho
8
piosonco of paiamililaiios in aioas whoio sonalo lisls iocoivoo a high volo shaio on lho pioponsily
of sonalois on lho lisl lo volo in favoi of changing lho Conslilulion lo allow Iiosioonl UiiLo lo
iun again. 1his olocl is slalislically signilcanl in all columns, oxcopl in column 7 whoio wo also
incluoo paily ioonlily (lhis suggosls lhal in lhis inslanco mosl of lho olocls of paiamililaiios on
polilicians Lohavioi mighl Lo woiking lhiough lhiio paily amlialion). 1hoso columns also again
show lhal lhoio is no ioLusl impacl of lho piosonco of guoiillas on lho voling Lohavioi of sonalois
on lhis moasuio.
7.2 Case Study Evidence
1hoio is a laigo caso sluoy liloialuio which is consislonl wilh lho nolion of a quio pio quo Lolwoon
oxoculivo ano paiamililaiy polilicians. Ono oLvious aiona is lho foimulalion of lho Juslico ano
Ioaco Law, which look placo afloi lho sociol nogolialions al Sanla Fo oo Ialilo. 1his law was
wiooly ciilicizoo Lolh insioo ano oulsioo ColomLia as Loing fai loo lonionl on paiamililaiios.
Hoio wo focus on moio ioconl ovonls, which appoai lo conlim lhis symLiolic iolalionship. Lvon
lhough many congiossmon ano sonalois havo Loon aiiosloo, lhoy aio ioplacoo in lho logislaluio Ly
lhoii alloinalos (in Spanish suplonlos`) who appoai on lho samo lisl al lho limo of lho oloclion.
In consoquonco, lhoii polilical inluonco is lilllo oiminishoo. 1o chango lhis silualion momLois
of congioss pioposoo a polilical iofoim in 2008 lo iomovo lhoso polilicians ano lhoii alloinalos
fiom lho logislaluio. 1his inilialivo was killoo whon many polilicians failoo lo appoai foi a ooLalo
so lhal a quoium was nol ioachoo. 1ho facl lhal sonalois failoo lo appoai foi lho volo was
wiooly Llamoo on Iiosioonl UiiLo (soo lho iomaiks of Sonalois Guslavo Iolio ano Iafaol Iaioo
(Lnlioiio oo quinla" Semana, Juno 7, 2008),
44
ano Semana nolos If UiiLismo losl ils majoiilios
in congioss, il woulo Lo oimcull lo gol lho appioval of koy piojocls, such as a now iofoim lo lhal
lilllo ailiclo` of lho Conslilulion."
45
1ho lilllo ailiclo` (usoo saicaslically) in lho Conslilulion
is lho chango lo allow Iiosioonl UiiLo lo iun foi a lhiio loim of omco. 1his oxpoiionco shows
how Iiosioonl UiiLo iolios on lho logislalivo suppoil of polilicians oooply implicaloo wilh lho
paiamililaiios lo pass koy Lills.
46
Afloi lhis piojocl failoo lho oxoculivo pioposoo an alloinalivo polilical iofoim in Augusl 2008
which involvoo changing lho Conslilulion lo lako away lho powoi of lho Supiomo Couil lo liy
congiossmon. As Joso Miguol Vivanco, Amoiicas oiiocloi al Human Iighls Walch pul il UiiLo
is Liazonly liying lo lako lho powoi lo invosligalo congiossmon away fiom lho ono inslilulion lhal
44
http://www.semana.com/noticias-nacion/entierro-quinta/112489.aspx
45
Semana, June 7, 2008.
46
In September 2009 both houses of the legislature voted to hold a national referendum on whether to allow Uribe
to run for a third term. Socialists and members of the Liberal party abstained in these votes so we do not have
further roll-calls on this issue.
86
has oono lho mosl lo uncovoi ano Lioak paiamililaiy inluonco in lho congioss ... 1his pioposal
soivos no ioal puiposo, olhoi lhan lo holp momLois of UiiLo`s coalilion gol ol lho hook" (Human
Iighls Walch, 2008).
8 Conclusions
Why aio many slalos in loss-oovolopoo sociolios unaLlo lo oslaLlish WoLoi`s famous monopoly of
violonco in lhoii loiiiloiios 1ho slanoaio oxplanalion iolios on lho inaLilily of lho conlial slalo lo
Lioaocasl ils powoi lhioughoul lho loiiiloiios lhal il nominally conliols ano viows an oxlonsion
of lhis powoi lo lho poiiphoiy as a naluial Ly-piooucl of polilical moooinizalion`. In lhis papoi
wo oovolopoo an alloinalivo poispoclivo, suggosling lhal lho conlial slalo can oovolop (ovon
moooinizo`) wilhoul oslaLlishing such a monopoly of violonco Locauso lhoio may Lo a symLiolic
iolalionship Lolwoon lho pailios conliolling lho conlial slalo ano non-slalo aclois oxoicising powoi
in lho poiiphoiios of lho counliy. 1ho oiigins of lhis symLiolic iolalionship is lhal non-slalo aimoo
aclois can piovioo suppoil lo lhoso conliolling lho conlial slalo. 1his is pailiculaily impoilanl
in oomociacios whoio non-slalo aimoo aclois can conliol oloclions. Sinco lhoy naluially havo
polilical piofoioncos, lhoy can (ciooiLly) oolivoi volos foi lho nalional polilicians in lino wilh
lhoii iooological ano policy Liasos. Iolilicians olocloo wilh lho implicil suppoil of lhoso non-slalo
aclois will lhon havo loss inconlivo lo oliminalo lhom, loaoing lo an oquiliLiium wilhoul a full
monopoly of violonco of lho conlial slalo. Wo oovolopoo lhis iooa lhooiolically ano pioviooo
ompiiical suppoil using ioconl polilical ovonls fiom ColomLia.
Oui moool makos a soiios of piooiclions, which wo invosligaloo using ColomLian oala. Oui
ompiiical ovioonco Logan Ly conliming lhal paiamililaiios havo inoooo hao a signilcanl impacl
on oloclions in ColomLia. Following lho founoalion of lho AUC, lhoio is a shaip incioaso in lho
volo shaio of lhiio pailios, which woio oxplicilly oi implicilly associaloo wilh paiamililaiios, in
aioas lhal havo high paiamililaiy piosonco. Wo also piosonloo ovioonco lhal high paiamililaiy
piosonco in aioas whoio sonalois iocoivoo laigo piopoilions of lhoii volos piooicls how lhoy voloo
on koy clausos of lho Juslico ano Ioaco Law, ano wholhoi lhoy gol aiiosloo foi illicil links wilh
lho paiamililaiios. Wo also founo lhal paiamililaiy piosonco is coiiolaloo wilh volo shaio of lho
winning piosioonlial canoioalo,

Alvaio UiiLo.
1ho lwo koy piooiclions of oui lhooiolical moool aio lhal paiamililaiios shoulo poisisl lo lho
oxlonl lhal lhoy oolivoi volos lo polilicians whoso piofoioncos aio closoi lo lhoii own ano lhal lhis
olocl is laigoi in aioas whoio lhoso polilicians woulo olhoiwiso nol oo woll. 1hoso iosulls illuslialo
lho possiLlo piosonco of a symLiolic iolalionship Lolwoon lho oxoculivo ano lho paiamililaiios.
Wo fuilhoi illuslialoo lhis iolalionship Ly showing lhal lho piopoilion of lho volos which a sonalo
87
lisl won in paiamililaiy aioas is posilivoly coiiolaloo wilh lho piopoilion of sonalois on lho lisl
lhal voloo lo chango lho Conslilulion lo allow Iiosioonl UiiLo lo iun foi a socono loim.
Novoilholoss, noilhoi oui iosulls noi lho caso sluoy liloialuio suggosls lho piosonco of a foimal
coalilion` Lolwoon paiamililaiy gioups ano Iiosioonl UiiLo. 1hoy aio consislonl, howovoi, wilh
lho iooa lhal UiiLo`s policios aio closoi lo lhoso piofoiioo Ly paiamililaiios lhan lhoso of olhoi
pailios which naluially givos lho paiamililaiios an inloiosl in gonoialing suppoil foi him. As
Jaiio Angaiila foimoi loaooi of lho AUC`s Sin u ano San Joigo Llocs ano Salvaloio Mancuso`s
oopuly ooclaioo in SoplomLoi 200, ho was piouo lo woik foi lho
iooloclion of lho Losl Iiosioonl wo havo ovoi hao".
47
Naluially, oiloionl inloipiolalions of lho oala aio possiLlo. Foi inslanco, il is possiLlo lhal
pooplo in paiamililaiy aioas aio naluially pio law ano oiooi` oi moio consoivalivo` so lhal pios-
onco of paiamililaiios is onoogonous lo voloi piofoioncos`. If lhis woio liuo oui iosulls connocling
lho piosonco of paiamililaiios lo piosioonlial volo shaio coulo Lo ouo lo omilloo vaiiaLlo Lias. 1his
mighl also unooimino oui inloipiolalion of lho ioll-call volos sinco il woulo imply lhal sonalois
who voloo lo chango lho Conslilulion lo allow Iiosioonl UiiLo lo iun again oio so Locauso lhoii
volois` piofoiioo il. 1hoso alloinalivo hypolhosos oo nol appoai plausiLlo oxplanalions foi lho pal-
loins, howovoi. Fiisl, oui ioonlilcalion comos fiom lho liming of lho foimalion of lho AUC ano
lhoii oocision lo oovolop an olocloial slialogy (iolying on inlimioalion, muiooi ano manipulalion
lo gol lhoii canoioalos olocloo lo nalional omco). Foi oxamplo, piioi lo lhoso ovonls, lhoio was
no signilcanl oiloioncos Lolwoon olocloial conconlialion in aioas wilh oi wilhoul paiamililaiios,
inoicaling lhal polilical piofoioncos oio nol syslomalically oiloi Lolwoon lhoso lwo aioas. 1ho
liming of lho foimalion of lho AUC sloms moslly fiom lho collapso of lho Mooollin ano Cali oiug
cailols ano is unlikoly lo iolocl a chango in polilical piofoioncos on lho pail of volois. Socono, lo
lho oxlonl lhal voloi piofoioncos oiloi wilh somo aioas Loing moio consoivalivo lhan olhois, lhis
woulo Lo conliolloo foi Ly oui uso of lxoo olocls as long as lhoso oiloioncos aio limo-invaiianl.
Moioovoi, lho caso sluoy liloialuio suggosls somolhing of a ioalily chock` on lhis aigumonl sinco
il piovioos oxlonsivo ovioonco of cooicion ano manipulalion of oloclions in oopailmonls such as
Magoalona, Cosai, CoiLooa ano Sucio wilh high paiamililaiy piosonco. 1hiio, lho facl lhal
paiamililaiy piosonco piooicls aiiosls of sonalois makos il impioLaLlo lhal polilicians aio simply
iopiosonling voloi piofoioncos. 1hoso iosulls, along wilh lho caso sluoy liloialuio, suggosls lhal
polilicians foimoo coalilions wilh paiamililaiios, oi paiamililaiios gonoialoo lhoii own polilicians.
Fouilh, wo liioo lo oxplicilly conliol foi lho limo vaiying olocls of consoivalivo loanings of coilain
municipalilios.
47
El Tiempo, September 7, 2005.
88
A lnal ano ciucial quoslion concoins lho oxloinal valioily of lhoso lnoings. Allhough oui
aigumonls in lhis papoi conliasl wilh lho omphasis of much of lho oxisling liloialuio on slalo
foimalion, lhoio aio sovoial olhoi inslancos whoio non-slalo aclois appoai lo play a similai iolo lo
lhal of paiamililaiios in ColomLia. Foi oxamplo, as monlionoo in lho Inlioouclion, lho oxislonco
of lho liiLal aioas in Waziiislan ooos nol imply lhal lho slalo has nol foimoo in Iakislan, Lul is
inoicalivo of a oiloionl oquiliLiium palloin similai lo lhal in ColomLia. A iolaloo oxamplo comos
fiom Ilaly ano lho iolo of lho Mala in oolivoiing volos lo lho Chiislian Domocialic Iaily (Walslon,
1088). Finally, lho long aulonomy of lho US Soulh fiom lho Hayos-1iloon agioomonl of 1877
unlil lho Civil Iighls ano Voling Iighls acls of lho 1060s also piosonls many paiallols wilh lho
ColomLian caso. Soulhoin polilical olilos woio aLlo lo oslaLlish a ono-paily unoomocialic polilical
syslom Lasoo on lho oisonfianchisomonl ano cooicion of Llacks. 1ho powoi lhus consolioaloo gavo
lhom gioal inluonco in nalional polilical inslilulions whoio lhoy coulo Lolh impooo logislalion
lhoy oio nol liko, Lul whoio lhoy also supplioo volos, which was oxaclly whal lho Hayos-1iloon
ooal was aLoul. A lolling pioco of ovioonco in favoi of lhis inloipiolalion is pioviooo Ly Fianklin D.
Ioosovoll`s ioaclion lo a pioposal lo pass logislalion lo allompl lo iosliicl lynching. In iosponso
ho aiguoo lhal Soulhoin logislalois
aio chaiimon oi occupy slialogic placos on mosl of lho Sonalo ano Houso commilloos
... If I como oul foi lho anli-lynching Lill now, lhoy will Llock ovoiy Lill I ask Congioss
lo pass" (quoloo in Fioooiickson, 2001, p. 20).
1ho oxamplos suggosl lhal lho mochanisms wo ioonliloo hoio may Lo impoilanl in a vaiioly
of oiloionl conloxls ano invosligaling lhom in oiloionl sollings woulo Lo a fiuilful aioa foi fuluio
iosoaich.
References
Acemoglu, Daron (2005) Iolilics ano Lconomics in Woak ano Sliong Slalos," Journal of
Monetary Economics, 2, 1100-1226.
Acemoglu, Daron, Davide Ticchi and Andrea Vindigni (2006) Lmoigonco ano Ioi-
sislonco of Inomcionl Slalos" NLLI Woiking Iapoi =12748, foilhcoming Journal of the European
Economic Association.
Acemoglu, Daron, Davide Ticchi and Andrea Vindigni (2010) Ioisislonco of Civil
Wais" foilhcoming Journal of the European Economic Association.
Amnesty International (2005) ColomLia: Flawoo DomoLilizalion Olois oo Faclo Amnosly,"
hllp:www.amnoslyusa.oigall-counliioscolomLia
80
Angrist, Joshua D. and Jorn-Steen Pischke (2009) Mostly Harmless Econometrics:
An Empiricist's Companion, Iiincolon: Iiincolon Univoisily Iioss.
Baron, David P. (1994) Llocloial compolilion wilh infoimoo ano uninfoimoo volois,"
American Political Science Review, 88, 88-47.
Bergquist, Charles, Ricardo Pe~ naranda, Gonzalo Sanchez eds. (1992) Violence in
Colombia: the contemporary crisis in historical perspective, Wilminglon, Dol.: SI Looks.
Bergquist, Charles, Ricardo Pe~ naranda, Gonzalo Sanchez eds. (2001) Violence in
Colombia, 1990-2000: waging war and negotiating peace, Wilminglon, Dol.: SI Looks.
Besley, Timothy and Torsten Persson (2009) 1ho Oiigins of Slalo Capacily: Iiopoily
Iighls, 1axalion, ano Iolilics," American Economic Review, 00, 1281-1244.
Blattman, Christopher and Edward Miguel (2010) Civil Wai," Journal of Economic
Literature, 48, 8-7.
Brewer, John (1988) The Sinews of Power: war, money, and the English state, 1688-1783,
Now Yoik: Knopf.
Deas, Malcolm D. (1999) Intercambios violentos: reexiones sobre la violencia poltica en
Colombia, Logola, ColomLia: 1auius.
Dube, Oeindrila and Suresh Naidu (2009) Lasos, Lullols, ano Lallols: 1ho Llocl of
U.S. Mililaiy Aio on Iolilical Conlicl in ColomLia," mimoo Sanla Fo Inslilulo.
Duncan, Gustavo (2007) Los Se~ nores de la Guerra, Logola: Loiloiial Ilanola ColomLiana.
Evans, Peter B. (1989) Iiooaloiy, Dovolopmonlal ano Olhoi Appaialusos: A Compaialivo
Iolilical Lconomy Ioispoclivo on lho 1hiio Woilo Slalo," Sociological Forum, 4, 61-87.
Evans, Peter B. (1995) Embedded Autonomy, Iiincolon: Iiincolon Univoisily Iioss.
Evans, Peter B., Dietrich Rueschemeyer and Theda Skocpol eds. (1985) Bringing
the State Back In: New Perspectives on the State as Institution and Social Actor, Now Yoik:
CamLiiogo Univoisily Iioss.
Fearon, James D. and David D. Laitin (2003) Llhnicily, Insuigoncy, ano Civil Wai,"
American Political Science Review, 07, 7-00.
Frederickson, Kari (2001) The Dixiecrat Revolt and the End of the Solid South: 1932{
1968, Chapol Hill: Univoisily of Noilh Caiolina Iioss.
Gibson, Edward L. (2005) Lounoaiy Conliol: SuLnalional Aulhoiilaiianism in Domo-
cialic Counliios," World Politics, 8, 101-182.
Gutierrez Sann, Francisco and Mauricio Baron (2005) Io-slaling lho Slalo: Iaiamil-
ilaiy loiiiloiial conliol ano polilical oiooi in ColomLia, 1078-2004," LSL Woiking Iapoi.
40
Herbst, Jerey I. (2000) States and Power in Africa, Iiincolon: Iiincolon Univoisily
Iioss.
Hintze, Otto (1975) The historical essays of Otto Hintze, ooiloo wilh an inlioo. Ly Folix
GilLoil, wilh lho assislanco of IoLoil M. Loioahl, Now Yoik: Oxfoio Univoisily Iioss.
Human Rights Watch (2005) ColomLia: Lolling Iaiamililaiios ol lho Hook,"
hllp:www.hiw.oiglogacyLackgiounooiamoiicascolomLia010
Human Rights Watch (2008) ColomLia: Iioposal 1hioalons Iaiapolilics` Invosliga-
lions," hllp:hiw.oigonglishoocs20080804colomL1088.hlm
Inter-American Commission on Human Rights (2007) Iopoil on lho Implomonlalion
of lho Juslico ano Ioaco Law," hllp:www.cioh.oas.oig
International Crisis Group (2007) Colombia's New Armed Groups, Lalin Amoiica Iopoil
N

20, 10 May 2007

Kohli, Atul (2004) State-directed development: political power and industrialization in the
global periphery, Now Yoik: CamLiiogo Univoisily Iioss.
Lindbeck, Assar and Jorgen Weibull (1987) Lalancoo-Luogol IooisliiLulion as lho
Oulcomo of Iolilical Compolilion," Public Choice, LII, 272-207.
Lopez, Claudia (2007) La Iula oo la Lxpansion Iaiamililai y la 1iansfoimacion Iolilica
oo Anlioquia," in Mauiicio Iomoio oo. Para Poltica: La Ruta de la Expansion Paramilitar y los
Acuerdos Polticos, Logola: Coipoiacion Nuovo Aico Iiis: Inloimooio.
Lopez, Claudia and Oscar Sevillano (2008) Lalanco Iolilico oo la Iaiapolilica," Coi-
poiacion Nuovo Aico iiis.
Mickey, Robert W. (2009) Paths out of Dixie: The Democratization of Authoritarian
Enclaves in America's Deep South, foilhcoming Iiincolon Univoisily Iioss.
O'Donnell, Guillermo (1993) On lho Slalo, Domocializalion, ano Somo Concoplual
IioLloms: A Lalin Amoiican Viow wilh Glancos al Somo Iosl-Communisl Counliios," World
Development, 21, 18-70.
Pardo Rueda, Rafael (2007) Fin del paramilitarismo: >es posible su desmonte? Logola,
D.C.: Loicionos L ColomLia.
Posada Carbo, Eduardo (2003) `Ilegitimidad' del Estado en Colombia: sobre los abusos
de un concepto, Logola, D.C.: Alfaomoga ColomLiana: Iooas paia la paz.
Porch, Douglas and Mara Jose Rasmussen (2008) DomoLilizalion of Iaiamililaiios
in ColomLia: 1iansfoimalion oi 1iansilion" Studies in Conict & Terrorism, 81, 20-40.
Rangel Suarez, Alfredo (1998) Colombia, guerra en el n de siglo, Logola: 1M Loiloios:
Univoisioao oo los Anoos, Facullao oo Cioncias Socialos.
41
Rangel Suarez, Alfredo (2005) El poder paramilitar, Logola: Funoacion Soguiioao \
Domociacia: Ilanola.
Restrepo, Jorge, Michael Spagat and Juan Fernando Vargas (2004) 1ho Dynamics
of lho ColomLian Civil Conlicl: A Now Dala Sol," Homo Oeconomicus, 21, 806-428.
Robinson, James A. and Miguel Urrutia eds. (2007) Economa Colombiana del Siglo
XX: Un Analisis Cuantitativo, Logola ano Moxico D.F.: Fonoo oo Culluia Lconomica.
Roldan, Mary (2002) Blood and Fire: La Violencia in Antioquia, Colombia, 1946-1953,
Duiham: Duko Univoisily Iioss.
Romero, Mauricio (2000) Changing Ioonlilios ano Conlosloo Sollings: Iogional Llilos
ano lho Iaiamililaiios in ColomLia," International Journal of Politics, Culture, and Society, 14,
1-60.
Romero, Mauricio (2002) Paramilitaries y autodefensas, 1982-2003, Logola: Loiloiial
Ilanola ColomLiana.
Sanchez , Fabio and Mara Del Mar Palau (2006) Conlicl, Doconlialisalion ano Local
Govoinanco in ColomLia, 1074-2000," CLDL Univoisioao oo Los Anoos, Documonlo 2006-46.
Serrano Zabala, Alfredo (2007) La batalla nal de Carlos Casta~ no: secretos de la para-
poltica, Logola: Loiloiial Ovoja Nogia: Quinloio Loiloios.
Storey, R. L. (1968) The Reign of Henry VII, Now Yoik: Walkoi.
Tilly, Charles ed. (1975) The Formation of national States in Western Europe, Iiincolon:
Iiincolon Univoisily Iioss.
Valencia, Leon (2007) Los Caminos oo la Allianza onlio Los Iaiamililaiios y los Iolilicos,"
in Mauiicio Iomoio oo. Para Poltica: La Ruta de la Expansion Paramilitar y los Acuerdos
Polticos, Logola: Coipoiacion Nuovo Aico Iiis: Inloimooio.
Waldmann, Peter (2007) Is lhoio a culluio of Violonco in ColomLia" International Jour-
nal of Conict and Violence, 1, 61-7.
Walston, James (1988) The Maa and Clientelism: Roads to Rome in Post-war Calabria,
Now Yoik: Ioulloogo.
Weber, Max (1946) Iolilics as a Vocalion," in From Max Weber: Essays in Sociology,
lianslaloo Ly H.H. Goilh ano C. Wiighl Mills, Now Yoik: Oxfoio Univoisily Iioss.
Zuckerman, Sarah (2009) Bankruptcy, Guns or Campaigns?: Explaining Armed Organi-
zations' Post-War Trajectories, IhD Dissoilalion, Dopailmonl of Iolilical Scionco, MI1.
42



Third Parties
(1)
Reelection
(2)
Justice and
Peace Law
(3)
Status
(4)
% Votes In
Paramilitary Zones
(5)
MAURICIO PIMIENTO BARRERA yes yes yes Arrested (Guilty) 68.30
DIEB NICOLAS MALOOF CUSE yes yes yes Arrested (Guilty) 56.93
ALVARO ARAUJO CASTRO yes yes Arrested 54.78
JUAN CARLOS MARTINEZ SINISTERRA yes yes Arrested 51.22
SALOMON DE JESUS SAADE ABDALA no yes Investigated 41.40
CARLOS ARTURO CLAVIJO VARGAS yes Arrested 39.33
JUAN GOMEZ MARTINEZ yes yes 34.96
ISABEL CELIS YAEZ no 33.96
PIEDAD CORDOBA no no no 33.20
GERMAN HERNANDEZ AGUILERA no yes yes 31.46
FLOR MODESTA GNECCO ARREGOCES yes yes yes 31.27
RUBEN DARIO QUINTERO VILLADA yes Arrested 30.03
BERNARDO ALEJANDRO GUERRA HOYOS no no 29.48
HUGO SERRANO GOMEZ no no 29.21
WILLIAM ALFONSO MONTES MEDINA yes yes yes Arrested (Not Guilty) 28.48
LUIS GUILLERMO VELEZ TRUJILLO no yes yes 28.44
CONSUELO DE MUSTAFA no yes 28.22
JOSE RENAN TRUJILLO GARCIA no yes yes 26.80
VICTOR RENAN BARCO LOPEZ no yes yes Investigated 26.11
GUILLERMO GAVIRIA ZAPATA no no yes Investigated 25.07
Senator
Table 1: Top 20 Senators By Vote Share in Paramilitary Areas

Notes: Senators that obtained the twenty highest shares of votes in municipalities with high paramilitary presence. High paramilitary presence is measured by a
dummy that takes the value of one if the municipality had a total number of attacks by the paramilitaries per 1.000 inhabitants above the 75
th
percentile in the
1997-2001 period. A Yes indicates that the senator belongs to a third party in the election of 2002 (column (1)), voted yes to approve reelection (column (2)) or yes
to reintroduce Sedition and Reduction of Sentences articles in the Justice and Peace Law (column (3)). The status of the senator (column (4)) is that on May 21 of
2009 and is taken from Indepaz http://www.indepaz.org.co (for reelected senators) and from the news. A blank space in columns (2) or (3) means that the senator
did not vote on the measure.
Mean
Standard
Deviation Mean
Standard
Deviation Mean
Standard
Deviation
(1) (2) (3) (4) (5) (6)
(1) Paramilitary Presence 0.12 (0.29) 0.42 (0.45) 0.02 (0.03)
(2) Paramilitary Presence 1996-1997 0.00 (0.02) 0.01 (0.04) 0.00 (0.01)
(3) Paramilitary Presence 2000-2001 0.04 (0.12) 0.15 (0.20) 0.01 (0.02)
(4) Paramilitary Presence 2004-2005 0.03 (0.13) 0.11 (0.25) 0.00 (0.01)
(5) Guerrilla Presence 1.03 (2.03) 1.82 (2.76) 0.77 (1.63)
(6) Guerrilla Presence 1996-1997 0.13 (0.31) 0.27 (0.48) 0.08 (0.20)
(7) Guerrilla Presence 2000-2001 0.21 (0.43) 0.44 (0.65) 0.14 (0.29)
(8) Guerrilla Presence 2004-2005 0.23 (0.87) 0.33 (0.67) 0.19 (0.92)
(9) Third Parties Vote Share in the Senate Before 2001 14.72 (15.90) 11.22 (13.06) 15.95 (16.61)
(10) Third Parties Vote Share in the Senate After 2002 32.88 (19.65) 38.59 (20.78) 30.90 (18.85)
(11) Third Parties Vote Share in the Congress Before 2001 26.80 (21.75) 22.84 (19.37) 28.17 (22.36)
(12) Third Parties Vote Share in the Congress After 2002 35.42 (21.48) 38.45 (22.12) 34.39 (21.19)
(13) Winning Presidential Candidate Vote Share Before 2001 42.29 (22.73) 39.21 (23.29) 43.38 (22.46)
(14) Winning Presidential Candidate Vote share after 2001 52.35 (21.47) 54.58 (21.66) 51.58 (21.37)
(15) Preferences for the Right 42.26 (27.47) 36.73 (28.32) 44.25 (26.91)
(16) Preferences for the Left 5.05 (11.41) 6.12 (11.47) 4.67 (11.37)
(17) Land Gini in 1985 70.11 (10.58) 69.63 (10.51) 70.28 (10.61)
(18) Unfullfiled Basic Needs in 1993 53.87 (19.56) 55.33 (16.43) 53.36 (20.55)
(19) Average Population in 1993 and 2005 34828.42 (202618.10) 17851.39 (23238.26) 40534.65 (233578.10)
(20) Rural Index in 1993 62.63 (24.50) 60.65 (20.29) 63.33 (25.81)
(21) Distance to the States' Capital 129.66 (105.54) 146.12 (115.71) 123.85 (101.16)
(22) Altitude 1157.38 (897.89) 983.56 (785.11) 1218.72 (927.16)
(23) Precipitation 1912.28 (1061.04) 2235.12 (1146.02) 1798.36 (1005.68)
(24) Dummy of Opium Cultivation in 1994 0.10 (0.29) 0.05 (0.23) 0.11 (0.31)
(25) Dummy of Coca Cultivation in 1994 0.05 (0.23) 0.04 (0.20) 0.06 (0.24)
Whole Sample High Paramilitary Presence Low Paramilitary Presence
Table 2: Descriptive Statistics
Variable

Notes: Each variable in rows (1) to (8) is the sum of paramilitary or guerrilla attacks per 1.000 inhabitants over the corresponding time period averaged over different
samples of municipalities. Whole sample columns report means and standard deviations of variables in the entire sample, High Paramilitary Presence columns restrict the
sample to municipalities where the 1997-2005 attacks by the paramilitaries dummy takes the value of one (see footnote of table 3 for definition of this variable), Low
Paramilitary Presence columns restrict the sample to municipalities where the 1997-2005 attacks by the paramilitaries dummy takes the value of 0.
Dependent Variable is Vote
Share obtained by Third
Parties in the Elections for
the Senate
Panel 1991-2006
(1)
Panel 1991-2006
(2)
Panel 1991-2006
(3)
Panel 1991-2006
(4)
Panel 1991-2006
(5)
Panel 1991-2006
(6)
Panel 1998-2006
(7)
Panel 1998-2006
(8)
Paramilitary Presence -10.21 -9.58
(2.72) (2.81)
Paramilitary Presence X 1994 4.94 0.79 0.57 4.16 1.92 1.34
(1.54) (1.48) (1.61) (1.25) (1.24) (1.31)
Paramilitary Presence X 1998 4.52 0.56 0.55 2.64 -0.19 -0.11
(1.99) (2.11) (2.21) (1.66) (1.69) (1.80)
Paramilitary Presence X 2002 20.98 15.88 15.79 13.72 10.62 10.46 16.48 15.56
(3.14) (3.18) (3.23) (1.98) (1.94) (2.00) (2.90) (3.04)
Paramilitary Presence X 2006 22.11 10.79 10.29 14.55 8.48 8.31 16.90 16.01
(3.19) (3.03) (3.04) (1.99) (1.66) (1.73) (3.06) (3.20)
Guerrilla Presence -1.14
(1.76)
Guerrilla presence X 1994 0.19 2.49
(0.56) (1.54)
Guerrilla Presence X 1998 -0.00 -0.33
(0.66) (1.85)
Guerrilla Presence X 2002 0.08 0.67 2.38
(0.69) (1.98) (2.13)
Guerrilla Presence X 2006 0.45 0.70 2.99
(0.61) (1.80) (2.33)
Controls Interacted with Year No Yes Yes No Yes Yes No No
Dummies
Observations 5379 4915 4915 5379 4915 4915 3286 3286
Table 3: Paramilitary Presence and Third Parties Share of Votes in the Elections for the Senate
Armed Actors Presence is Measured by:
Attacks Attacks Dummy Time Varying Attacks Dummy


Notes: Robust Standard errors clustered at the municipality level in parentheses. Panel regressions with full set of municipality and year dummies. Dependent variable is share of votes of
third parties lists (not Conservative, nor Liberal, nor from the left) in the elections for the Senate. We report results with three different measures of paramilitary presence: i. The sum of
paramilitary attacks per 1,000 inhabitants in municipality m during the 1997-2005 period in columns (1), (2) and (3); ii. A time invariant dummy that takes the value of one if the sum of
paramilitary attacks per 1,000 inhabitants in municipality m during the 1997-2005 period is above the 75
th
percentile in columns (4), (5) and (6); iii. A time varying attacks dummy that
takes the value of one in municipality m and time t if time varying measure of attacks over population is above the 75
th
percentile (calculated over all municipalities and years) in
columns (7) and (8). When guerrilla presence is included, in columns (3), (6) and (8), it is measured as the corresponding paramilitary presence measure. Columns (2), (3), (5) and (6)
include the following controls interacted with time dummies: altitude, distance to the state capital, precipitation, average population between 1993 and 2005, rurality index in 1993, land
gini in 1985, unfulfilled basic needs in 1993, dummy for coca cultivation in 1994, dummy for opium cultivation in 1994, preferences for the Right in 1986 and preferences for the Left in
1986.



Dependent Variable is Winning
Presidential Candidate Vote
Share
Panel 1998-2006
(1)
Panel 1998-2006
(2)
Panel 1998-2006
(3)
Panel 1998-2006
(4)
Panel 1998-2006
(5)
Panel 1998-2006
(6)
Panel 1998-2006
(7)
Panel 1998-2006
(8)
Paramilitary Presence -6.92 -6.91
(3.59) (3.65)
Paramilitary Presence X 2002 10.16 5.31 7.43 3.11 1.26 2.14 8.87 10.49
(1.99) (1.53) (1.59) (1.45) (1.11) (1.13) (3.58) (3.65)
Paramilitary Presence X 2006 21.60 13.67 12.32 11.45 8.17 6.66 12.53 12.23
(2.41) (1.71) (1.64) (1.67) (1.21) (1.20) (3.77) (3.86)
Guerrilla Presence -3.54
(1.61)
Guerrilla Presence X 2002 -1.73 -3.71 -5.53
(0.34) (1.14) (1.73)
Guerrilla Presence X 2006 1.22 6.47 1.70
(0.41) (1.45) (2.21)
Controls Interacted with Year No Yes Yes No Yes Yes No No
Dummies
Observations 3297 2951 2951 3297 2951 2951 3297 3297
Tables 4: Paramilitary Presence and Winning Presidential Candidate Share of Votes
Armed Actors Presence is Measured by:
Attacks Attacks Dummy Time Varying Attacks Dummy

Notes: Robust Standard errors clustered at the municipality level in parentheses. Panel regressions with full set of municipality and year dummies. Dependent variable is share of votes
of the winning presidential candidate. We report results with three different measures of paramilitary presence: i. The sum of paramilitary attacks per 1,000 inhabitants in municipality m
during the 1997-2005 period in columns (1), (2) and (3); ii. A time invariant dummy that takes the value of one if the sum of paramilitary attacks per 1,000 inhabitants in municipality m
during the 1997-2005 period is above the 75
th
percentile in columns (4), (5) and (6); iii. A time varying attacks dummy that takes the value of one in municipality m and time t if time
varying measure of attacks over population is above the 75
th
percentile (calculated over all municipalities and years) in columns (7) and (8). When guerrilla presence is included, in
columns (3), (6) and (8), it is measured as the corresponding paramilitary presence measure. Columns (2), (3), (5) and (6) include the following controls interacted with time dummies:
altitude, distance to the state capital, precipitation, average population between 1993 and 2005, rurality index in 1993, land gini in 1985, unfulfilled basic needs in 1993, dummy for coca
cultivation in 1994, dummy for opium cultivation in 1994 , preferences for the Right in 1986 and preferences for the Left in 1986.



Cross
Section
(1)
Cross
Section
(2)
Cross
Section
(3)
Cross
Section
(4)
Cross
Section
(5)
Cross
Section
(6)
Cross
Section
(7)
Cross
Section
(8)
Cross
Section
(9)
Cross
Section
(10)
Cross
Section
(11)
Cross
Section
(12)
Dummy Conservative 0.05 0.12 -0.01 0.09 0.41 0.38
(0.12) (0.14) (0.06) (0.09) (0.12) (0.14)
Dummy Left -0.10 -0.03 -0.06 -0.02 -0.59 -0.54
(0.06) (0.07) (0.03) (0.04) (0.12) (0.14)
Dummy Third Parties 0.19 0.21 0.06 0.08 0.34 0.33
(0.09) (0.09) (0.05) (0.05) (0.13) (0.13)
Share of Votes From:
Paramilitary Areas 1.38 1.03 0.92 0.26 0.20 0.18 0.83 1.33 0.93
(0.42) (0.51) (0.45) (0.12) (0.13) (0.13) (0.41) (0.53) (0.54)
Guerrilla Areas -0.59 -0.13 0.02 -0.01 0.03 0.07 -0.86 -1.63 -1.26
(0.72) (0.81) (0.75) (0.07) (0.07) (0.08) (0.73) (0.89) (0.83)
Right Oriented Areas -0.41 -0.55 -0.31 -0.44 1.37 0.52
(0.42) (0.39) (0.12) (0.21) (0.46) (0.47)
Left Oriented Areas -0.27 -0.29 -0.09 -0.10 -0.13 -0.05
(0.18) (0.18) (0.06) (0.06) (0.27) (0.23)
Observations 96 96 96 96 162 162 162 162 57 57 57 57
R-squared 0.06 0.13 0.15 0.21 0.01 0.04 0.06 0.08 0.41 0.04 0.13 0.45
Senate Congress
Table 5: Senators and Congressmen Arrested, Justice and Peace Law and Votes from High Paramilitary Presence Areas Table 5: Senators and Congressmen Arrested, Justice and Peace Law and Votes from High Paramilitary Presence Areas
Dependent Variable is the Fraction of Arrested Senators/Congressmen in list l.
Dependent Variable is the Fraction of
Senators in List l that Voted Yes for
Reintroducing the Articles of Sedition and
Reduction of Sentences in the Justice and
Peace Law .
Senate

Notes: Robust standard errors in parentheses. Left hand panel: OLS regressions relating arrests of senators/congressmen to votes obtained in areas with presence of non-state armed
actors. Dependent variable is the proportion of senators on list l arrested for being involved with the paramilitary. Right hand panel: OLS regressions linking votes in the senate to votes
obtained in areas with presence of non-state armed actors. Dependent variable is the proportion of senators on list l that voted yes. The vote is for reintroducing the articles of Sedition
and Reduction of Sentences in the Justice and Peace Law (since only three lists have more than one candidate in the senate in the legislature of 2002-2006 and since candidates in the
same list voted in the same manner, the dependent variable is a dummy). Both Panels: To measure the share of votes of list l from a given area we first create dummies for places with
high presence of paramilitary and guerrilla, Right oriented preferences or Left oriented preferences (municipality m is a high presence area if the value of the corresponding variable in
municipality m is above the 75
th
percentile; paramilitary and guerrilla presence measures are the sum of attacks per 1,000 inhabitant in the 1997-2001 period, just before the elections of
2002). Then, with each of these dummies, we compute the share of votes in national elections obtained by list l in areas where the dummy takes the value of one.
Dependent Variable is Paramilitary
Presence in 2003-2005
Cross-
Section
(1)
Cross-
Section
(2)
Cross-
Section
(3)
Cross-
Section
(4)
Cross-
Section
(5)
Cross-
Section
(6)
Cross-
Section
(7)
Cross-
Section
(8)
Cross-
Section
(9)
Cross-
Section
(10)
Cross-
Section
(11)
Cross-
Section
(12)
Max{0, Uribe-Pastrana vote share} 0.27 0.44 9.95 0.25 1.85
(0.16) (0.31) (3.26) (0.13) (0.69)
Uribe Vote Share 0.14 0.13 0.13 0.03 2.58 0.15 0.66
(0.08) (0.10) (0.10) (0.22) (2.46) (0.09) (0.43)
Patrana Vote Share -0.23 -0.07 -0.08 0.12 -0.36 0.29 -0.93
(0.09) (0.12) (0.13) (0.39) (3.87) (0.14) (0.60)
Uribe Vote Share X Pastrana Vote Share -0.77 -0.62 -0.63 -0.46 -19.11 -0.20 -3.84
(0.36) (0.39) (0.41) (0.19) (7.36) (0.07) (1.31)
Paramilitary Presence in 1999-2001 0.41 0.40 0.40 0.39 0.31 0.29 0.12 0.12 0.23 0.23 0.36 0.35
(0.16) (0.17) (0.19) (0.18) (0.11) (0.11) (0.04) (0.04) (0.05) (0.05) (0.12) (0.12)
Guerrilla Presence in 1999-2001 -0.00 -0.00 -0.06 -0.04 0.06 0.06 0.26 0.26 -0.01 -0.00
(0.02) (0.02) (0.09) (0.10) (0.03) (0.03) (0.05) (0.05) (0.09) (0.09)
Controls No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Observations 319 309 309 309 134 134 622 622 524 524 654 654
R-squared 0.21 0.24 0.24 0.24 0.55 0.52 0.25 0.26 0.49 0.46 0.25 0.25
Table 6: Persistence of Paramilitaries and Vote Share for Alvaro Uribe
Sample is Restricted to Municiaplities with Paramilitary Presence in 1999-2001
Armed Actors Presence is Measured by:
Attacks Log Attacks Displaced Log Displaced
Principal
Component Attacks
and Displaced

Notes: Robust standard errors in parentheses. Cross Section regressions restricting the sample to municipalities with paramilitary presence in 1999-2001. Dependent variable is
paramilitary presence in 2003-2005. We report results with three measures of paramilitary presence: i. Attacks by the paramilitaries in columns (1) to (6) is the sum of paramilitary
attacks per 1,000 inhabitants in municipality m during the 2003-2005 period (dependent variable) and during the 1999-2001 period (paramilitary presence before 2002 variable), ii.
Displaced by the paramilitaries in columns (7) to (10) is the sum of people displaced by the paramilitary per 1,000 inhabitants in municipality m during the 2003-2005 period (dependent
variable) and during the 1999-2001 period (paramilitary presence before 2002 variable), iii. The principal component of attacks by the paramilitary and displaced by the paramilitary in
column (11) and (12). Guerrilla presence before 2002 is measured as paramilitary presence before 2002. In columns (5), (6), (9) and (10) all variables are in logs. Uribe and Pastrana vote
shares are the vote shares of lvaro Uribe in 2002 and Andrs Pastrana in 1998 (first round), respectively. These two variables are measured in a scale from zero to one for ease of
exposition (to report fewer decimals) and they are also demeaned to interpret the derivatives at the mean of the interactions in all columns except in columns (4), (8) and (12). In these
columns, the variable of interest is the maximum between zero and the difference between lvaro Uribes vote share in 2002 and Andrs Pastranas vote share in 1998 in municipality m.
All specifications include the same controls as in Table 3: altitude, distance to the state capital, precipitation, average population between 1993 and 2005, rurality index in 1993, land gini
in 1985, unfulfilled basic needs in 1993, dummy for coca cultivation in 1994, dummy for opium cultivation in 1994, preferences for the Right in 1986 and preferences for the Left in
1986.
Dependent Variable is the Fraction
of Senators in List l that Voted Yes
for Changing the Constitution to
Allow the Reelection of the President
Cross
Section
(1)
Cross
Section
(2)
Cross
Section
(3)
Cross
Section
(4)
Cross
Section
(5)
Cross
Section
(6)
Cross
Section
(7)
Dummy Conservative 0.48 0.36 0.33
(0.11) (0.12) (0.12)
Dummy Left -0.52 -0.48 -0.50
(0.11) (0.11) (0.12)
Dummy Third Parties 0.31 0.30 0.28
(0.13) (0.12) (0.13)
Share of Votes From:
Paramilitary Areas 1.26 1.79 1.61 1.02 1.28 0.63
(0.41) (0.55) (0.60) (0.41) (0.44) (0.36)
Guerrilla Areas -0.92 -1.87 -1.39 -0.88 -1.05 -0.21
(0.73) (0.82) (0.80) (0.79) 0.78 (0.65)
Right Oriented Areas 1.81 1.11 1.55 0.88
(0.36) (0.34) (0.34) (0.32)
Left Oriented Areas -0.17 -0.02 -0.27 -0.16
(0.24) (0.21) (0.23) (0.21)
Observations 76 76 76 76 76 76 76
R-squared 0.38 0.07 0.21 0.45 0.04 0.17 0.39
Attacks
Table 7 : Reelection and Senators Elected from High Paramilitary Presence Areas
Armed Presence Measured By:
Displaced

Notes: Robust standard errors in parentheses. OLS regressions linking votes in the Senate to votes obtained in areas with presence of
non-state armed actors. Dependent variable is the proportion of senators in list l that voted yes (since only three lists have more than one
candidate in the senate in the legislature of 2002-2006 and since candidates in the same list voted in the same manner, the dependent
variable is a dummy). The vote is for changing the constitution to allow the president to be elected for a second consecutive term. To
measure the share of votes of list l from a given area we first create dummies for places with high presence of paramilitary, guerrilla,
right-oriented preferences or left-oriented preferences (municipality m is a high presence area if the value of the corresponding variable
in municipality m is above the 75
th
percentile; paramilitary and guerrilla presence measures are the sum of attacks per 1,000 inhabitant in
the 1997-2001 period, just before the elections of 2002). Then, with each of these dummies, we compute the share of votes in national
elections obtained by list l in areas where the dummy takes the value of one. Columns (2) to (4) use attacks to define the presence
dummies, columns (5) to (7) use displaced.
Appendix for \The Monopoly of Violence: Evidence From Columbia"
(Not-For-Publication)
Alternative Modeling Assumption
An alloinalivo moooling assumplion, as noloo in foolnolo 0 in lho loxl, is lo assumo lhal paiamil-
ilaiios in iogion , ? aio aLlo lo foico a fiaclion :
j
of lho populalion lo volo foi lhoii favoiilo
canoioalo (say paily A), whilo lho iomaining 1 :
j
slill volo fiooly, wilh oxaclly lho samo
piofoioncos as spociloo in lho loxl. In lhis caso, oqualion (6) will Lo ioplacoo wilh
1
A
_

A
,
B
[ , m
_
= Ii
_
\ _

2
_
= Ii
_
_
N

j=1
:
j

N

j=1
(1 :
j
)
_
1
2
n
j
_

A
_
n
j
_

B
_
0
j

_
_

2
_
_
= Ii
_
_
1
2
N

j=1
:
j

N

j=1
(1 :
j
)
_
n
j
_

A
_
n
j
_

B
_
0
j

_
_ 0
_
_
=
1
2
c
_

N
j=1
:
j
2

N
j=1
(1 :
j
)

N
j=1
(1 :
j
)
_
n
j
_

A
_
n
j
_

B
_
0
j
_

N
j=1
(1 :
j
)
_
,
whoio lho lisl lino usos lho facl lhal ano lho facl lhal

j2Z
:
j
=

N
j=1
:
j
, ano wo uso lho
convonlion lhal :
j
= 0 foi , ano lho lasl oqualily again usos lho facl lhal is unifoim ovoi
_

1
2
,
1
2
_
. In lhis caso, lho oquiliLiium policios of Lolh pailios will Lo givon Ly

lhal maximizos

N
j=1
(1 :
j
) n
j
(), ano if inloiioi, lhis will salisfy
N

j=1
(1 :
j
) \n
j
(

) = 0,
so lhal now volois in lho paiamililaiy-conliolloo aioas also gol somo woighl. Novoilholoss, in lino
wilh oui iosulls in lho loxl, lho highoi is :
j
(i.o., lho fiaclion of lho populalion unooi lho conliol
of lho paiamililaiios), lho lowoi is lhis woighl. In oquiliLiium, wo naluially havo
1
A
(

[ , m) =
1
2
c
_

N
j=1
:
j
2

N
j=1
(1 :
j
)

N
j=1
(1 :
j
) 0
j

N
j=1
(1 :
j
)
_
.
Ono aovanlago of lhis alloinalivo foimulalion is lhal wo oLlain oxaclly lho samo policy vocloi
ano lho samo pioLaLilily of paily A winning lho oloclion as

N
j=1
:
j
0, whoioas wilh oui
Lasolino moool in lho loxl, lhis iosull is only oLlainoo as wo lako lho limil whoio [?[ 0.
1ho iosulls on poisislonco also conlinuo lo apply in lhis mooiloo solup. In pailiculai, lho cioss
pailial of 1
A
(

[ , m) wilh iospocl lo :
j
ano 0
j
is nogalivo (i.o., 0
2
1
A
(

[ , m) ,0:
k
00
k
=
c

j6=k
(1 :
j
) ,
_

N
j=1
(1 :
j
)
_
2
. 1his implios lhal lho conliiLulion of paiamililaiy volos lo
A-1



Third Parties
(1)
Reelection
(2)
Justice and
Peace Law
(3)
Status
(4)
% Votes In
Paramilitary Zones
(5)
MAURICIO PIMIENTO BARRERA yes yes yes Arrested (Guilty) 68.30
DIEB NICOLAS MALOOF CUSE yes yes yes Arrested (Guilty) 56.93
ALVARO ARAUJO CASTRO yes yes Arrested 54.78
JUAN CARLOS MARTINEZ SINISTERRA yes yes Arrested 51.22
SALOMON DE JESUS SAADE ABDALA no yes Investigated 41.40
CARLOS ARTURO CLAVIJO VARGAS yes Arrested 39.33
JUAN GOMEZ MARTINEZ yes yes 34.96
ISABEL CELIS YAEZ no 33.96
PIEDAD CORDOBA no no no 33.20
GERMAN HERNANDEZ AGUILERA no yes yes 31.46
FLOR MODESTA GNECCO ARREGOCES yes yes yes 31.27
RUBEN DARIO QUINTERO VILLADA yes Arrested 30.03
BERNARDO ALEJANDRO GUERRA HOYOS no no 29.48
HUGO SERRANO GOMEZ no no 29.21
WILLIAM ALFONSO MONTES MEDINA yes yes yes Arrested (Not Guilty) 28.48
LUIS GUILLERMO VELEZ TRUJILLO no yes yes 28.44
CONSUELO DE MUSTAFA no yes 28.22
JOSE RENAN TRUJILLO GARCIA no yes yes 26.80
VICTOR RENAN BARCO LOPEZ no yes yes Investigated 26.11
GUILLERMO GAVIRIA ZAPATA no no yes Investigated 25.07
Senator
Table 1: Top 20 Senators By Vote Share in Paramilitary Areas

Notes: Senators that obtained the twenty highest shares of votes in municipalities with high paramilitary presence. High paramilitary presence is measured by a
dummy that takes the value of one if the municipality had a total number of attacks by the paramilitaries per 1.000 inhabitants above the 75
th
percentile in the
1997-2001 period. A Yes indicates that the senator belongs to a third party in the election of 2002 (column (1)), voted yes to approve reelection (column (2)) or yes
to reintroduce Sedition and Reduction of Sentences articles in the Justice and Peace Law (column (3)). The status of the senator (column (4)) is that on May 21 of
2009 and is taken from Indepaz http://www.indepaz.org.co (for reelected senators) and from the news. A blank space in columns (2) or (3) means that the senator
did not vote on the measure.
Mean
Standard
Deviation Mean
Standard
Deviation Mean
Standard
Deviation
(1) (2) (3) (4) (5) (6)
(1) Paramilitary Presence 0.12 (0.29) 0.42 (0.45) 0.02 (0.03)
(2) Paramilitary Presence 1996-1997 0.00 (0.02) 0.01 (0.04) 0.00 (0.01)
(3) Paramilitary Presence 2000-2001 0.04 (0.12) 0.15 (0.20) 0.01 (0.02)
(4) Paramilitary Presence 2004-2005 0.03 (0.13) 0.11 (0.25) 0.00 (0.01)
(5) Guerrilla Presence 1.03 (2.03) 1.82 (2.76) 0.77 (1.63)
(6) Guerrilla Presence 1996-1997 0.13 (0.31) 0.27 (0.48) 0.08 (0.20)
(7) Guerrilla Presence 2000-2001 0.21 (0.43) 0.44 (0.65) 0.14 (0.29)
(8) Guerrilla Presence 2004-2005 0.23 (0.87) 0.33 (0.67) 0.19 (0.92)
(9) Third Parties Vote Share in the Senate Before 2001 14.72 (15.90) 11.22 (13.06) 15.95 (16.61)
(10) Third Parties Vote Share in the Senate After 2002 32.88 (19.65) 38.59 (20.78) 30.90 (18.85)
(11) Third Parties Vote Share in the Congress Before 2001 26.80 (21.75) 22.84 (19.37) 28.17 (22.36)
(12) Third Parties Vote Share in the Congress After 2002 35.42 (21.48) 38.45 (22.12) 34.39 (21.19)
(13) Winning Presidential Candidate Vote Share Before 2001 42.29 (22.73) 39.21 (23.29) 43.38 (22.46)
(14) Winning Presidential Candidate Vote share after 2001 52.35 (21.47) 54.58 (21.66) 51.58 (21.37)
(15) Preferences for the Right 42.26 (27.47) 36.73 (28.32) 44.25 (26.91)
(16) Preferences for the Left 5.05 (11.41) 6.12 (11.47) 4.67 (11.37)
(17) Land Gini in 1985 70.11 (10.58) 69.63 (10.51) 70.28 (10.61)
(18) Unfullfiled Basic Needs in 1993 53.87 (19.56) 55.33 (16.43) 53.36 (20.55)
(19) Average Population in 1993 and 2005 34828.42 (202618.10) 17851.39 (23238.26) 40534.65 (233578.10)
(20) Rural Index in 1993 62.63 (24.50) 60.65 (20.29) 63.33 (25.81)
(21) Distance to the States' Capital 129.66 (105.54) 146.12 (115.71) 123.85 (101.16)
(22) Altitude 1157.38 (897.89) 983.56 (785.11) 1218.72 (927.16)
(23) Precipitation 1912.28 (1061.04) 2235.12 (1146.02) 1798.36 (1005.68)
(24) Dummy of Opium Cultivation in 1994 0.10 (0.29) 0.05 (0.23) 0.11 (0.31)
(25) Dummy of Coca Cultivation in 1994 0.05 (0.23) 0.04 (0.20) 0.06 (0.24)
Whole Sample High Paramilitary Presence Low Paramilitary Presence
Table 2: Descriptive Statistics
Variable

Notes: Each variable in rows (1) to (8) is the sum of paramilitary or guerrilla attacks per 1.000 inhabitants over the corresponding time period averaged over different
samples of municipalities. Whole sample columns report means and standard deviations of variables in the entire sample, High Paramilitary Presence columns restrict the
sample to municipalities where the 1997-2005 attacks by the paramilitaries dummy takes the value of one (see footnote of table 3 for definition of this variable), Low
Paramilitary Presence columns restrict the sample to municipalities where the 1997-2005 attacks by the paramilitaries dummy takes the value of 0.
Dependent Variable is Vote
Share obtained by Third
Parties in the Elections for
the Senate
Panel 1991-2006
(1)
Panel 1991-2006
(2)
Panel 1991-2006
(3)
Panel 1991-2006
(4)
Panel 1991-2006
(5)
Panel 1991-2006
(6)
Panel 1998-2006
(7)
Panel 1998-2006
(8)
Paramilitary Presence -10.21 -9.58
(2.72) (2.81)
Paramilitary Presence X 1994 4.94 0.79 0.57 4.16 1.92 1.34
(1.54) (1.48) (1.61) (1.25) (1.24) (1.31)
Paramilitary Presence X 1998 4.52 0.56 0.55 2.64 -0.19 -0.11
(1.99) (2.11) (2.21) (1.66) (1.69) (1.80)
Paramilitary Presence X 2002 20.98 15.88 15.79 13.72 10.62 10.46 16.48 15.56
(3.14) (3.18) (3.23) (1.98) (1.94) (2.00) (2.90) (3.04)
Paramilitary Presence X 2006 22.11 10.79 10.29 14.55 8.48 8.31 16.90 16.01
(3.19) (3.03) (3.04) (1.99) (1.66) (1.73) (3.06) (3.20)
Guerrilla Presence -1.14
(1.76)
Guerrilla presence X 1994 0.19 2.49
(0.56) (1.54)
Guerrilla Presence X 1998 -0.00 -0.33
(0.66) (1.85)
Guerrilla Presence X 2002 0.08 0.67 2.38
(0.69) (1.98) (2.13)
Guerrilla Presence X 2006 0.45 0.70 2.99
(0.61) (1.80) (2.33)
Controls Interacted with Year No Yes Yes No Yes Yes No No
Dummies
Observations 5379 4915 4915 5379 4915 4915 3286 3286
Table 3: Paramilitary Presence and Third Parties Share of Votes in the Elections for the Senate
Armed Actors Presence is Measured by:
Attacks Attacks Dummy Time Varying Attacks Dummy


Notes: Robust Standard errors clustered at the municipality level in parentheses. Panel regressions with full set of municipality and year dummies. Dependent variable is share of votes of
third parties lists (not Conservative, nor Liberal, nor from the left) in the elections for the Senate. We report results with three different measures of paramilitary presence: i. The sum of
paramilitary attacks per 1,000 inhabitants in municipality m during the 1997-2005 period in columns (1), (2) and (3); ii. A time invariant dummy that takes the value of one if the sum of
paramilitary attacks per 1,000 inhabitants in municipality m during the 1997-2005 period is above the 75
th
percentile in columns (4), (5) and (6); iii. A time varying attacks dummy that
takes the value of one in municipality m and time t if time varying measure of attacks over population is above the 75
th
percentile (calculated over all municipalities and years) in
columns (7) and (8). When guerrilla presence is included, in columns (3), (6) and (8), it is measured as the corresponding paramilitary presence measure. Columns (2), (3), (5) and (6)
include the following controls interacted with time dummies: altitude, distance to the state capital, precipitation, average population between 1993 and 2005, rurality index in 1993, land
gini in 1985, unfulfilled basic needs in 1993, dummy for coca cultivation in 1994, dummy for opium cultivation in 1994, preferences for the Right in 1986 and preferences for the Left in
1986.



Dependent Variable is Winning
Presidential Candidate Vote
Share
Panel 1998-2006
(1)
Panel 1998-2006
(2)
Panel 1998-2006
(3)
Panel 1998-2006
(4)
Panel 1998-2006
(5)
Panel 1998-2006
(6)
Panel 1998-2006
(7)
Panel 1998-2006
(8)
Paramilitary Presence -6.92 -6.91
(3.59) (3.65)
Paramilitary Presence X 2002 10.16 5.31 7.43 3.11 1.26 2.14 8.87 10.49
(1.99) (1.53) (1.59) (1.45) (1.11) (1.13) (3.58) (3.65)
Paramilitary Presence X 2006 21.60 13.67 12.32 11.45 8.17 6.66 12.53 12.23
(2.41) (1.71) (1.64) (1.67) (1.21) (1.20) (3.77) (3.86)
Guerrilla Presence -3.54
(1.61)
Guerrilla Presence X 2002 -1.73 -3.71 -5.53
(0.34) (1.14) (1.73)
Guerrilla Presence X 2006 1.22 6.47 1.70
(0.41) (1.45) (2.21)
Controls Interacted with Year No Yes Yes No Yes Yes No No
Dummies
Observations 3297 2951 2951 3297 2951 2951 3297 3297
Tables 4: Paramilitary Presence and Winning Presidential Candidate Share of Votes
Armed Actors Presence is Measured by:
Attacks Attacks Dummy Time Varying Attacks Dummy

Notes: Robust Standard errors clustered at the municipality level in parentheses. Panel regressions with full set of municipality and year dummies. Dependent variable is share of votes
of the winning presidential candidate. We report results with three different measures of paramilitary presence: i. The sum of paramilitary attacks per 1,000 inhabitants in municipality m
during the 1997-2005 period in columns (1), (2) and (3); ii. A time invariant dummy that takes the value of one if the sum of paramilitary attacks per 1,000 inhabitants in municipality m
during the 1997-2005 period is above the 75
th
percentile in columns (4), (5) and (6); iii. A time varying attacks dummy that takes the value of one in municipality m and time t if time
varying measure of attacks over population is above the 75
th
percentile (calculated over all municipalities and years) in columns (7) and (8). When guerrilla presence is included, in
columns (3), (6) and (8), it is measured as the corresponding paramilitary presence measure. Columns (2), (3), (5) and (6) include the following controls interacted with time dummies:
altitude, distance to the state capital, precipitation, average population between 1993 and 2005, rurality index in 1993, land gini in 1985, unfulfilled basic needs in 1993, dummy for coca
cultivation in 1994, dummy for opium cultivation in 1994 , preferences for the Right in 1986 and preferences for the Left in 1986.



Cross
Section
(1)
Cross
Section
(2)
Cross
Section
(3)
Cross
Section
(4)
Cross
Section
(5)
Cross
Section
(6)
Cross
Section
(7)
Cross
Section
(8)
Cross
Section
(9)
Cross
Section
(10)
Cross
Section
(11)
Cross
Section
(12)
Dummy Conservative 0.05 0.12 -0.01 0.09 0.41 0.38
(0.12) (0.14) (0.06) (0.09) (0.12) (0.14)
Dummy Left -0.10 -0.03 -0.06 -0.02 -0.59 -0.54
(0.06) (0.07) (0.03) (0.04) (0.12) (0.14)
Dummy Third Parties 0.19 0.21 0.06 0.08 0.34 0.33
(0.09) (0.09) (0.05) (0.05) (0.13) (0.13)
Share of Votes From:
Paramilitary Areas 1.38 1.03 0.92 0.26 0.20 0.18 0.83 1.33 0.93
(0.42) (0.51) (0.45) (0.12) (0.13) (0.13) (0.41) (0.53) (0.54)
Guerrilla Areas -0.59 -0.13 0.02 -0.01 0.03 0.07 -0.86 -1.63 -1.26
(0.72) (0.81) (0.75) (0.07) (0.07) (0.08) (0.73) (0.89) (0.83)
Right Oriented Areas -0.41 -0.55 -0.31 -0.44 1.37 0.52
(0.42) (0.39) (0.12) (0.21) (0.46) (0.47)
Left Oriented Areas -0.27 -0.29 -0.09 -0.10 -0.13 -0.05
(0.18) (0.18) (0.06) (0.06) (0.27) (0.23)
Observations 96 96 96 96 162 162 162 162 57 57 57 57
R-squared 0.06 0.13 0.15 0.21 0.01 0.04 0.06 0.08 0.41 0.04 0.13 0.45
Senate Congress
Table 5: Senators and Congressmen Arrested, Justice and Peace Law and Votes from High Paramilitary Presence Areas Table 5: Senators and Congressmen Arrested, Justice and Peace Law and Votes from High Paramilitary Presence Areas
Dependent Variable is the Fraction of Arrested Senators/Congressmen in list l.
Dependent Variable is the Fraction of
Senators in List l that Voted Yes for
Reintroducing the Articles of Sedition and
Reduction of Sentences in the Justice and
Peace Law .
Senate

Notes: Robust standard errors in parentheses. Left hand panel: OLS regressions relating arrests of senators/congressmen to votes obtained in areas with presence of non-state armed
actors. Dependent variable is the proportion of senators on list l arrested for being involved with the paramilitary. Right hand panel: OLS regressions linking votes in the senate to votes
obtained in areas with presence of non-state armed actors. Dependent variable is the proportion of senators on list l that voted yes. The vote is for reintroducing the articles of Sedition
and Reduction of Sentences in the Justice and Peace Law (since only three lists have more than one candidate in the senate in the legislature of 2002-2006 and since candidates in the
same list voted in the same manner, the dependent variable is a dummy). Both Panels: To measure the share of votes of list l from a given area we first create dummies for places with
high presence of paramilitary and guerrilla, Right oriented preferences or Left oriented preferences (municipality m is a high presence area if the value of the corresponding variable in
municipality m is above the 75
th
percentile; paramilitary and guerrilla presence measures are the sum of attacks per 1,000 inhabitant in the 1997-2001 period, just before the elections of
2002). Then, with each of these dummies, we compute the share of votes in national elections obtained by list l in areas where the dummy takes the value of one.
Dependent Variable is Paramilitary
Presence in 2003-2005
Cross-
Section
(1)
Cross-
Section
(2)
Cross-
Section
(3)
Cross-
Section
(4)
Cross-
Section
(5)
Cross-
Section
(6)
Cross-
Section
(7)
Cross-
Section
(8)
Cross-
Section
(9)
Cross-
Section
(10)
Cross-
Section
(11)
Cross-
Section
(12)
Max{0, Uribe-Pastrana vote share} 0.27 0.44 9.95 0.25 1.85
(0.16) (0.31) (3.26) (0.13) (0.69)
Uribe Vote Share 0.14 0.13 0.13 0.03 2.58 0.15 0.66
(0.08) (0.10) (0.10) (0.22) (2.46) (0.09) (0.43)
Patrana Vote Share -0.23 -0.07 -0.08 0.12 -0.36 0.29 -0.93
(0.09) (0.12) (0.13) (0.39) (3.87) (0.14) (0.60)
Uribe Vote Share X Pastrana Vote Share -0.77 -0.62 -0.63 -0.46 -19.11 -0.20 -3.84
(0.36) (0.39) (0.41) (0.19) (7.36) (0.07) (1.31)
Paramilitary Presence in 1999-2001 0.41 0.40 0.40 0.39 0.31 0.29 0.12 0.12 0.23 0.23 0.36 0.35
(0.16) (0.17) (0.19) (0.18) (0.11) (0.11) (0.04) (0.04) (0.05) (0.05) (0.12) (0.12)
Guerrilla Presence in 1999-2001 -0.00 -0.00 -0.06 -0.04 0.06 0.06 0.26 0.26 -0.01 -0.00
(0.02) (0.02) (0.09) (0.10) (0.03) (0.03) (0.05) (0.05) (0.09) (0.09)
Controls No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Observations 319 309 309 309 134 134 622 622 524 524 654 654
R-squared 0.21 0.24 0.24 0.24 0.55 0.52 0.25 0.26 0.49 0.46 0.25 0.25
Table 6: Persistence of Paramilitaries and Vote Share for Alvaro Uribe
Sample is Restricted to Municiaplities with Paramilitary Presence in 1999-2001
Armed Actors Presence is Measured by:
Attacks Log Attacks Displaced Log Displaced
Principal
Component Attacks
and Displaced

Notes: Robust standard errors in parentheses. Cross Section regressions restricting the sample to municipalities with paramilitary presence in 1999-2001. Dependent variable is
paramilitary presence in 2003-2005. We report results with three measures of paramilitary presence: i. Attacks by the paramilitaries in columns (1) to (6) is the sum of paramilitary
attacks per 1,000 inhabitants in municipality m during the 2003-2005 period (dependent variable) and during the 1999-2001 period (paramilitary presence before 2002 variable), ii.
Displaced by the paramilitaries in columns (7) to (10) is the sum of people displaced by the paramilitary per 1,000 inhabitants in municipality m during the 2003-2005 period (dependent
variable) and during the 1999-2001 period (paramilitary presence before 2002 variable), iii. The principal component of attacks by the paramilitary and displaced by the paramilitary in
column (11) and (12). Guerrilla presence before 2002 is measured as paramilitary presence before 2002. In columns (5), (6), (9) and (10) all variables are in logs. Uribe and Pastrana vote
shares are the vote shares of lvaro Uribe in 2002 and Andrs Pastrana in 1998 (first round), respectively. These two variables are measured in a scale from zero to one for ease of
exposition (to report fewer decimals) and they are also demeaned to interpret the derivatives at the mean of the interactions in all columns except in columns (4), (8) and (12). In these
columns, the variable of interest is the maximum between zero and the difference between lvaro Uribes vote share in 2002 and Andrs Pastranas vote share in 1998 in municipality m.
All specifications include the same controls as in Table 3: altitude, distance to the state capital, precipitation, average population between 1993 and 2005, rurality index in 1993, land gini
in 1985, unfulfilled basic needs in 1993, dummy for coca cultivation in 1994, dummy for opium cultivation in 1994, preferences for the Right in 1986 and preferences for the Left in
1986.
Dependent Variable is the Fraction
of Senators in List l that Voted Yes
for Changing the Constitution to
Allow the Reelection of the President
Cross
Section
(1)
Cross
Section
(2)
Cross
Section
(3)
Cross
Section
(4)
Cross
Section
(5)
Cross
Section
(6)
Cross
Section
(7)
Dummy Conservative 0.48 0.36 0.33
(0.11) (0.12) (0.12)
Dummy Left -0.52 -0.48 -0.50
(0.11) (0.11) (0.12)
Dummy Third Parties 0.31 0.30 0.28
(0.13) (0.12) (0.13)
Share of Votes From:
Paramilitary Areas 1.26 1.79 1.61 1.02 1.28 0.63
(0.41) (0.55) (0.60) (0.41) (0.44) (0.36)
Guerrilla Areas -0.92 -1.87 -1.39 -0.88 -1.05 -0.21
(0.73) (0.82) (0.80) (0.79) 0.78 (0.65)
Right Oriented Areas 1.81 1.11 1.55 0.88
(0.36) (0.34) (0.34) (0.32)
Left Oriented Areas -0.17 -0.02 -0.27 -0.16
(0.24) (0.21) (0.23) (0.21)
Observations 76 76 76 76 76 76 76
R-squared 0.38 0.07 0.21 0.45 0.04 0.17 0.39
Attacks
Table 7 : Reelection and Senators Elected from High Paramilitary Presence Areas
Armed Presence Measured By:
Displaced

Notes: Robust standard errors in parentheses. OLS regressions linking votes in the Senate to votes obtained in areas with presence of
non-state armed actors. Dependent variable is the proportion of senators in list l that voted yes (since only three lists have more than one
candidate in the senate in the legislature of 2002-2006 and since candidates in the same list voted in the same manner, the dependent
variable is a dummy). The vote is for changing the constitution to allow the president to be elected for a second consecutive term. To
measure the share of votes of list l from a given area we first create dummies for places with high presence of paramilitary, guerrilla,
right-oriented preferences or left-oriented preferences (municipality m is a high presence area if the value of the corresponding variable
in municipality m is above the 75
th
percentile; paramilitary and guerrilla presence measures are the sum of attacks per 1,000 inhabitant in
the 1997-2001 period, just before the elections of 2002). Then, with each of these dummies, we compute the share of votes in national
elections obtained by list l in areas where the dummy takes the value of one. Columns (2) to (4) use attacks to define the presence
dummies, columns (5) to (7) use displaced.
lho pioLaLilily of winning is gioaloi whon 0
j
is lowoi, ano lhus lho main iosull of Iioposilion 8
conlinuos lo apply (lhough lho oxacl chaiacloiizalion of lho sol of municipalilios lhal lho slalo
ioconquois is moio complicaloo in lhis caso).
The Structure of Colombian High Courts
Wo now oiscuss in moio oolail lho sliucluio of lho ColomLian Hig Couils lo illuslialo why lhoy
aio aLlo lo acl inooponoonlly of lho piosioonl. 1hoio aio foui High Couils: lho Supiomo Couil,
lho Consoil o` Llal (Slalo Council), lho Supoiioi Council of lho Juoiciaiy ano lho Conslilulional
Couil. Whilo lho lisl lwo havo Loon in placo sinco lho 1886 ColomLian Conslilulion (allhough
wilh somo alloialions), lho lalloi onos woio a cioalion of lho 1001 Conslilulion. Foi lho puiposos
of paiamililaiy involvomonl in polilics il is iolovanl lo Liioly unooislano lho Lasics of lho Supiomo
Couil ano lho Conslilulional Couil. 1ho Supiomo Couil`s main iolo is lo ioviow juogmonls of
infoiioi (piovincial) couils ano oocioo wholhoi lhoio has Loon a coiiocl inloipiolalion of lho law
anooi lho facls of lho casos. 1ho Couil is oiviooo in lhioo soclions: ciiminal, civil ano laLoi.
1ho Ciiminal Soclion of lho Supiomo Couil is also in chaigo of liying ano juoging lho momLois
of lho Congioss: Sonalo ano ChamLoi of Iopiosonlalivos" (Ailiclo 28 = 8Conslilulion). 1his
givos lho Supiomo Couil lho aulhoiily lo juogo momLois of Congioss accusoo of having links
wilh paiamililaiy gioups. Somo momLois of Congioss, onco invosligalions foi paiamililaiy links
havo slailoo, havo sloppoo oown fiom lhoii puLlic omco lo avoio lho juiisoiclion of lho Supiomo
Couil ano havo suLjocloo lhomsolvos lo oioinaiy piocoouios unooi lowoi-scalo" piosoculois ano
juogos.
1ho Conslilulional Couil has lwo lypos of iolos: on lho ono hano il is in chaigo of ioviowing
lho juogmonl of piovincial ano soclional juogos iogaioing accionos oo lulola,` which is a piocoouio
lhal any poison can slailgonoially againsl puLlic omcialsin oiooi lo sook lho pioloclion of
hishoi funoamonlal iighls. On lho olhoi hano lho Conslilulional Couil is in chaigo of juoicial
ioviow of logislalion. In lhis iolo lho Couil ooloiminos wholhoi a pailiculai slalulo (law onacloo Ly
Congioss) is conslilulional. 1ho Conslilulion piovioos lhal somo slalulos shoulo Lo aulomalically
ioviowoo Ly lho Couil, whilo in olhoi casos lho Couil will only inloivono whon a cilizon challongos
lho conslilulionalily of a pailiculai slalulo. In lho caso of lho Juslico ano Ioaco Law, lho
Conslilulional Couil ioviowoo ils conslilulionalily lhanks lo a challongo posoo Ly a gioup of
cilizons (unooi Ailiclo 241 of lho 1001 Conslilulion).
1hoio is a iolalivoly high oogioo of aulonomy of lhoso High Couils vis-`a-vis lho Lxoculivo
Lianch, pailiculaily lho Supiomo Couil. Supiomo Couil Juslicos soivo ono oighl-yoai loim
wilh no io-oloclion. 1ho Supiomo Couil is in chaigo of choosing ioplacomonl juslicos fiom lisls
piopaioo Ly lho Aominislialivo Soclion of lho Supoiioi Council of lho Juoiciaiy (anolhoi High
A-2
Couil also in chaigo of aominislialivo funclions iogaioing lho juoiciaiy). In luin, lho juslicos of
lho Aominislialivo Soclion aio appoinloo Ly lho Conslilulional Couil (1 juogo), Ly lho Supiomo
Couil (2 juogos) ano Ly lho Consoil o`Llal (8 juogos). 1ho impacl of lho logislalivo ano oxoculivo
Lianchos is lhoiofoio voiy inoiiocl on lho Supiomo Couil. 1hoy can lo somo oxlonl inluonco lhis
piocoss. Foi inslanco lho Conslilulional Couil has 0 juogos olocloo foi lifo Ly lho Sonalo. Howovoi,
lho oloclion lakos placo ovoi lisls of canoioalos piosonloo Ly lho piosioonl (8 juogos), lho Supiomo
Couil (8 juogos) ano lho Consoil o`Llal (8 juogos). 1hus, fiom lho nino juslicos silling on lho
Conslilulional Couil in a pailiculai loim, lhioo of lhom havo como oul fiom lisls piosonloo Ly
lho piosioonl.
Robustness Checks on Table 3
1aLlo A2 shows lho iosulls foi lho volos foi lhiio pailios in lho Congioss. 1ho gonoial palloins
ano in facl ovon lho poinl oslimalos aio voiy similai lo lhoso in 1aLlo 8, wilh high paiamililaiy
aioas showing aLoul a 8 poiconlago poinls highoi volo shaios foi lhiio pailios afloi lho AUC`s
involvomonl in polilics. Novoilholoss, lho iosulls aio woakoi lhan in 1aLlo 8, ano in column 6
whoio wo uso lho allacks oummy wilh lho full sol of covaiialos ano guoiiilla piosonco noilhoi
c
2002
nol c
2006
slalislically signilcanl. Impoilanlly, c
1994
ano c
1998
aio small ano slalislically
insignilcanl in all spocilcalions in 1aLlo A2.
1aLlo A8 invosligalos lho ioLuslnoss of lho iosulls piosonloo in 1aLlo 8 fuilhoi (focusing on
lho volos foi lho Sonalo; lho iosulls foi lho Congioss aio similai). In lhis laLlo, all spocilcalions
incluoo lho limo inloiaclions wilh lho full sol of covaiialos inliooucoo aLovo, incluoing lho inloi-
aclions wilh guoiiilla piosonco (which aio iopoiloo in lho laLlo, whilo olhoi covaiialos aio again
nol iopoiloo lo savo spaco). 1ho lisl column again oslimalos (0) fiom lho loxl, Lul usos lho
lovols of oisplacoo pooplo as lho piosonco moasuio foi Lolh paiamililaiy ano guoiiilla piosonco.
Now wo soo lhal, loss suppoilivo of oui hypolhosis, lho oslimalos foi lho inloiaclions Lolwoon all
lho limo oummios ano lho oisplacoo moasuio of paiamililaiy piosonco aio slalislically signilcanl
(lhough lho poinl oslimalos aio oiloionl Locauso lho scalo of lho oisplacoo vaiiaLlo is oiloionl).
Howovoi, lhoio is a laigo incioaso in lho sizo of lho coomcionl in 2002, lhough il halvos in 2006.
In column 2 wo uso lho oummy voision of lho oisplacoo moasuio of paiamililaiy piosonco wilh
iosulls similai lo lhoso in column 1 oxcopl lhal now lho quanlilalivo magniluoos of c
2002
ano
c
2006
aio voiy similai. Locauso wo aio now using a oummy vaiiaLlo foi moasuiing paiamililaiy
piosonco, lho quanlilalivo olocls can also Lo oiioclly compaioo lo lhoso in 1aLlo 8. 1ho oslimalos
suggosl lhal high paiamililaiy aioas oxpoiioncoo aLoul 12 poiconlago poinls incioaso in lho volo
shaio of lhiio pailios in 2002 ano 2006, which is Lioaoly consislonl wilh lho quanlilalivo olocls
in 1aLlo 8.
A-8
In column 8 wo again uso lho oisplacoo moasuio, Lul now oxploil lho limo soiios vaiialion in
lho oala oslimaling moool (10). Hoio lho oslimalos aio impiociso ano insignilcanl, lhough slill
posilivo. In column 4, wo uso a oiloionl slialogy ano oxliacl lho piincipal compononl of lho
allacks ano oisplacoo moasuios, ano uso lhis piincipal compononl as oui moasuio of paiamililaiy
piosonco. 1ho aovanlago of lhis slialogy is lhal Lolh allacks ano oisplacoo numLois aio noisy,
lhus lhoii common compononl may conlain moio infoimalion. 1ho iosulls fiom lhis appioach aio
voiy much in lino wilh lhoso of 1aLlos 8 ano 4. Wo lno no impacl of paiamililaiy piosonco piioi
lo 2002 Lul a posilivo ano highly signilcanl ono afloiwaios. In column , wo uso yol anolhoi
slialogy ano consliucl a oummy lhal lakos lho valuo 1 if in oui oala a municipalily has high
paiamililaiy piosonco accoioing lo Lolh oui allacks oala ano oui oisplacoo oala (accoioing lo
lhis moasuio all aioas foi which wo havo oala foi oisplacoo numLois Lul foi which lho allacks
oala aio missing iocoivo a valuo of zoio). Wo consliucl a similai oummy foi lho guoiiilla piosonco
moasuios. 1ho iosulls using lhis comLinoo oummy aio again voiy similai lo oui Lasic lnoing
lhough c
1998
is now slalislically signilcanl. Finally, in column 6, wo oslimalo lho moool (10), Lul
now using a limo-vaiying oummy consliucloo comLining infoimalion fiom allacks ano oisplacoo
numLois as in column . In conliasl lo lho impiociso iosulls in column 8, lho oslimalos now aio
consislonl wilh lho iosl of oui iosulls ano aio highly signilcanl.
Robustness Checks on Table 6
1aLlo A4 oslimalos oqualion (12), Lul insloao of oolning 1
m;t<2002
ano 1
m;t>2002
using 2000 ano
2001 (l<2002) oi Lolwoon 2004 sono 200 (l2002) as in 1aLlo 6, il usos a lhioo-yoai winoow
(1000, 2000 ano 2001 foi 1
m;t<2002
ano 2008, 2004 ano 200 foi 1
m;t>2002
). 1ho iosulls in 1aLlo
A4 aio Lioaoly similai lo, in facl sliongoi lhan, lhoso in 1aLlo 6.
A-4





Notes: Senators that obtained the twenty highest shares of votes in municipalities with high paramilitary presence. High paramilitary presence is measured by a dummy that takes
the value of one if the municipality had a total number of displaced by the paramilitaries per 1,000 inhabitants is above the 75th percentile in the 1997-2001 period. A Yes
indicates that the senator belongs to a third party in the election of 2002 (column (1)), voted yes to approve reelection (column (2)) or yes to reintroduce Sedition and Reduction
of Sentences articles in the Justice and Peace Law (column (3)). The status of the senator (column (4)) is that on May 21 of 2009 and is taken from Indepaz
http://www.indepaz.org.co (for reelected senators) and from the news. A blank space in columns (2) or (3) means that the senator did not vote on the measure.








Third Parties
(1)
Reelection
(2)
Justice and
Peace Law
(3)
Status
(4)
% Votes In
Paramilitary Zones
(5)
FLOR MODESTA GNECCO ARREGOCES yes yes yes 65.67
JAIRO ENRIQUE MERLANO FERNANDEZ yes yes yes Arrested (Not Guilty) 65.37
SALOMON DE JESUS SAADE ABDALA no yes Investigated 65.23
DIEB NICOLAS MALOOF CUSE yes yes yes Arrested (Guilty) 61.17
MAURICIO PIMIENTO BARRERA yes yes yes Arrested (Guilty) 59.71
ALVARO ALFONSO GARCIA ROMERO yes Arrested (Guilty) 52.46
CARLOS ARTURO CLAVIJO VARGAS yes Arrested 50.94
JUAN CARLOS MARTINEZ SINISTERRA yes yes Arrested 42.02
JESUS LEON PUELLO CHAMIE no yes 38.94
WILLIAM ALFONSO MONTES MEDINA yes yes yes Arrested (Not Guilty) 34.22
ALVARO ARAUJO CASTRO yes yes Arrested 32.74
PIEDAD DEL SOCORRO ZUCCARDI DE GARCIA no yes yes 31.72
VICENTE BLEL SAAD yes yes yes Arrested 29.51
LUIS EDUARDO VIVES LACOUTURE yes yes yes Arrested (Guilty) 27.86
PIEDAD CORDOBA no no no 27.63
GABRIEL ACOSTA BENDEK yes 26.05
HUGO SERRANO GOMEZ no no 25.09
MIGUEL ALFONSO DE LA ESPRIELLA BURG yes yes yes Arrested (Guilty) 23.96
LUIS ALBERTO GIL CASTILLO yes yes Arrested 23.78
RUBEN DARIO QUINTERO VILLADA yes Arrested 23.33
Table A1: Top 20 Senators By Vote Share in Paramilitary Areas Using Displaced
Senator


Notes: Robust Standard errors clustered at the municipality level in parentheses. Panel regressions with full set of municipality and year dummies. Dependent variable is share of votes of
third parties lists (not Conservative, nor Liberal, nor from the left) in the elections for the Congress. We report results with three different measures of paramilitary presence: i. The sum
of paramilitary attacks per 1,000 inhabitants in municipality m during the 1997-2005 period in columns (1), (2) and (3); ii. A time invariant dummy that takes the value of one if the sum
of paramilitary attacks per 1,000 inhabitants in municipality m during the 1997-2005 period is above the 75
th
percentile in columns (4), (5) and (6); iii. A time varying attacks dummy that
takes the value of one in municipality m and time t if time varying measure of attacks over population is above the 75
th
percentile (calculated over all municipalities and years) in columns
(7) and (8). When guerrilla presence is included, in columns (3), (6) and (8), it is measured as the corresponding paramilitary presence measure. Columns (2), (3), (5) and (6) include the
following controls interacted with time dummies: altitude, distance to the state capital, precipitation, average population between 1993 and 2005, rurality index in 1993, land gini in
1985, unfulfilled basic needs in 1993, dummy for coca cultivation in 1994, dummy for opium cultivation in 1994, preferences for the Right in 1986 and preferences for the Left in 1986.


Dependent Variable is Vote Share
obtained by Third Parties in The
Elections for the Congress
Panel 1991-2006
(1)
Panel 1991-2006
(2)
Panel 1991-2006
(3)
Panel 1991-2006
(4)
Panel 1991-2006
(5)
Panel 1991-2006
(6)
Panel 1998-2006
(7)
Panel 1998-2006
(8)
Paramilitary Presence -3.00 -2.23
(3.98) (4.06)
Paramilitary Presence X 1994 2.53 0.16 0.06 2.24 1.27 0.86
(2.29) (2.33) (2.35) (1.99) (2.10) (2.17)
Paramilitary Presence X 1998 -0.61 -1.51 -1.53 -1.78 -2.86 -3.66
(2.34) (2.31) (2.39) (2.11) (2.10) (2.20)
Paramilitary Presence X 2002 13.63 9.55 8.81 7.67 5.78 3.94 9.07 7.69
(4.05) (4.07) (4.06) (2.92) (3.02) (3.16) (4.26) (4.39)
Paramilitary Presence X 2006 15.17 6.22 6.23 7.53 2.98 2.55 7.88 6.92
(2.31) (2.08) (2.20) (2.24) (2.09) (2.17) (4.53) (4.63)
Guerrilla Presence -1.39
(1.92)
Guerrilla Presence X 1994 0.08 1.68
(0.74) (2.27)
Guerrilla Presence X 1998 0.02 3.41
(0.78) (2.40)
Guerrilla Presence X 2002 0.63 7.86 4.01
(1.08) (3.29) (2.26)
Guerrilla Presence X 2006 -0.01 1.85 2.93
(0.69) (2.38) (2.84)
Controls Interacted with Year No Yes Yes No Yes Yes No No
Dummies
Observations 5363 4899 4899 5363 4899 4899 3289 3289
Table A2: Paramilitary Presence and Third Parties Share of Votes in The Elections for the Congress
Armed Actors Presence is Measured by:
Attacks Attacks Dummy Time Varying Attacks Dummy

Notes: Robust Standard errors clustered at the municipality level in parentheses. Panel regressions with full set of municipality and year dummies.
Dependent variable is share of votes of third parties lists (not Conservative, nor Liberal nor from the left) in the elections for the Senate. We report
results with six different measures of paramilitary presence: i. The sum of people displaced by the paramilitary per 1,000 inhabitants in municipality
m during the 1997-2007 period in columns (1); ii. A time invariant dummy that takes the value of one if the sum of people displaced by the
paramilitary per 1,000 inhabitants in municipality m during the 1997-2005 period is above the 75% percentile in column (2); iii. A time varying
displaced dummy that takes the value of one in municipality m and time t if a time varying measure of attacks over population is above the 75
th

percentile (calculated over all municipalities and years) in column (3); iv. The principal component of attacks and displaced in column (4); v. A
time invariant attacks and displaced dummy that takes the values of one if both attacks and displaced time invariant dummies take the value of 1 in
column (5); vi. A time varying attacks and displaced dummy that takes the values of one if both attacks and displaced time varying dummies take
the value of 1 in column (6). Guerrilla presence is measured accordingly. All specifications include the following controls interacted with time
dummies: altitude, distance to the state capital, precipitation, average population between 1993 and 2005, rurality index in 1993, land gini in 1985,
unfulfilled basic needs in 1993, dummy for coca cultivation in 1994, dummy for opium cultivation in 1994, preferences for the Right in 1986 and
preferences for the Left in 1986.
Dependent Variable is Vote Share
obtained by Third Parties in The
Elections for the Senate
Panel 1991-2006
(1)
Panel 1991-2006
(2)
Panel 1998-2006
(3)
Panel 1991-2006
(4)
Panel 1991-2006
(5)
Panel 1998-2006
(6)
Displaced
Displaced
Dummy
Time Varying
Displaced
Dummy
Principal
Component of
Attacks and
Displaced
Attacks and
Displaced
Dummy
Attacks and
Displaced Time
Varying Dummy
Paramilitary Presence 3.73 -4.00
(3.72) (6.40)
Paramilitary Presence X 1994 0.13 4.65 1.17 1.82
(0.03) (1.50) (0.61) (1.57)
Paramilitary Presence X 1998 0.12 5.32 1.18 4.03
(0.04) (2.02) (0.73) (2.22)
Paramilitary Presence X 2002 0.32 11.74 3.02 5.67 14.59 13.81
(0.05) (2.52) (4.31) (0.95) (2.60) (6.72)
Paramilitary Presence X 2006 0.16 10.25 2.97 3.20 10.20 12.60
(0.06) (2.09) (3.99) (0.89) (2.27) (6.70)
Guerrilla Presence -14.16 -10.32
(3.56) (4.37)
Guerrilla Presence X 1994 -0.00 1.44 0.53 2.87
(0.02) (1.70) (0.92) (1.88)
Guerrilla Presence X 1998 -0.01 -1.84 0.06 -0.60
(0.02) (2.28) (1.13) (2.39)
Guerrilla Presence X 2002 -0.08 -1.12 13.11 -1.78 -1.69 9.82
(0.03) (2.59) (4.09) (1.21) (2.56) (4.79)
Guerrilla Presence X 2006 -0.02 -2.41 15.56 -0.02 -0.49 13.03
(0.03) (2.27) (3.76) (1.05) (2.31) (4.62)
Observations 4915 4915 3286 4915 4915 3286
Armed Actors Presence is Measured by:
Table A3: Paramilitary Presence and Third Parties Share of Votes in the Elections for the Senate. Robustness to Alternative Definitions of
Paramilitary Presence.

Notes: Robust standard errors in parentheses. Cross Section regressions restricting the sample to municipalities with paramilitary presence in 2000-2001. Dependent variable is
paramilitary presence in 2004-2005. We report results with three measures of paramilitary presence: i. Attacks by the paramilitaries in columns (1) to (6) is the sum of paramilitary
attacks per 1,000 inhabitants in municipality m during the 2004-2005 period (dependent variable) and during the 2000-2001 period (paramilitary presence before 2002 variable); ii.
Displaced by the paramilitaries in columns (7) to (10) is the sum of people displaced by the paramilitary per 1,000 inhabitants in municipality m during the 2004-2005 period (dependent
variable) and during the 2000-2001 period (paramilitary presence before 2002 variable); iii. The principal component of attacks by the paramilitary and displaced by the paramilitary in
columns (11) and (12). Guerrilla presence before 2002 is measured as paramilitary presence before 2002. In columns (5), (6), (9) and (10) all variables are in logs. Uribe and Pastrana
vote shares are the vote shares of lvaro Uribe in 2002 and Andrs Pastrana in 1998 (first round), respectively. These two variables are measured in a scale from zero to one for ease of
exposition (to report fewer decimals) and they are also demeaned to interpret the derivatives at the mean of the interactions in all columns except in columns (4), (8) and (12). In these
columns, the variable of interest is the maximum between zero and the difference between lvaro Uribes vote share in 2002 and Andrs Pastranas vote share in 1998 in municipality m.
All specifications include the same controls as in Table 3: altitude, distance to the state capital, precipitation, average population between 1993 and 2005, rurality index in 1993, land gini
in 1985, unfulfilled basic needs in 1993, dummy for coca cultivation in 1994, dummy for opium cultivation in 1994 , preferences for the Right in 1986 and preferences for the Left in
1986.
Dependent Variable is Paramilitary
Presence in 2004-2005
Cross-
Section
(1)
Cross-
Section
(2)
Cross-
Section
(3)
Cross-
Section
(4)
Cross-
Section
(5)
Cross-
Section
(6)
Cross-
Section
(7)
Cross-
Section
(8)
Cross-
Section
(9)
Cross-
Section
(10)
Cross-
Section
(11)
Cross-
Section
(12)
Max{0, Uribe-Pastrana vote share} 0.25 0.56 10.16 0.39 2.57
(0.15) (0.30) (2.95) (0.13) (0.83)
Uribe Vote Share 0.14 0.15 0.15 0.11 4.09 0.32 1.17
(0.08) (0.09) (0.08) (0.27) (1.98) (0.10) (0.49)
Patrana Vote Share -0.22 -0.09 -0.09 0.07 -0.85 0.31 -1.30
(0.08) (0.10) (0.11) (0.41) (2.81) (0.17) (0.66)
Uribe Vote Share X Pastrana Vote Share -0.63 -0.41 -0.42 -0.46 -12.68 -0.10 -3.65
(0.33) (0.33) (0.36) (0.22) (5.60) (0.09) (1.46)
Paramilitary Presence in 2000-2001 0.42 0.42 0.42 0.40 0.35 0.34 0.04 0.03 0.22 0.21 0.37 0.35
(0.17) (0.18) (0.19) (0.18) (0.12) (0.12) (0.02) (0.02) (0.06) (0.06) (0.15) (0.14)
Guerrilla Presence in 2000-2001 -0.00 0.00 0.00 0.01 0.05 0.05 0.21 0.21 -0.08 -0.08
(0.02) (0.02) (0.10) (0.10) (0.02) (0.02) (0.06) (0.06) (0.09) (0.09)
Controls No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Observations 299 291 291 291 88 88 616 616 503 503 643 643
R-squared 0.25 0.27 0.27 0.27 0.64 0.61 0.19 0.20 0.43 0.41 0.21 0.22
Table A4: Persistence of Paramilitaries and Vote Share for Alvaro Uribe. Robustness Checks.
Armed Actors Presence is Measured by:
Attacks
Principal
Component Attacks
and Displaced
Sample is Restricted to Municiaplities with Paramilitary Presence in 2000-2001
Displaced Log Attacks Log Displaced







Amazonas
Antioquia
Arauca
Atlantico
Bolivar
Boyaca
Caldas
Caqueta
Casanare
Cauca
Cesar
Choco
Cordova
Cundinamarca
Guainia
Guaviare
Huila
La Guajira
Magdalena
Meta
Narino
Nte De Stder
Putumayo
Quindio
Risaralda
San Andres
Santander
Sucre
Tolima
Valle
Vaupes
Vichada
Barranquilla
Ralito
Bucaramanga
Medellin
Bogota
Cali
S Vte Caguan
Attacks by the guerrilla
per 10 000 inhabitants
(1.61,26.45]
(.67,1.61]
(.22,.67]
(0,.22]
[0,0]
Figure A1: Guerrilla Attacks, 1997-2005






Amazonas
Antioquia
Arauca
Atlantico
Bolivar
Boyaca
Caldas
Caqueta
Casanare
Cauca
Cesar
Choco
Cordova
Cundinamarca
Guainia
Guaviare
Huila
La Guajira
Magdalena
Meta
Narino
Nte De Stder
Putumayo
Quindio
Risaralda
San Andres
Santander
Sucre
Tolima
Valle
Vaupes
Vichada
Barranquilla
Ralito
Bucaramanga
Medellin
Bogota
Cali
S Vte Caguan
Attacks by the paramilitary
per 10 000 inhabitants
(.19,3.62]
(.04,.19]
(0,.04]
[0,0]
Figure A2: Paramilitary Attacks, 1997-2005