DfcWtT
MIT LIBRARIES
111
3 9080 03130 1630
Massachusetts Institute of Technology
Department of Economics
Working Paper Series
The Monopoly of Violence:
Evidence from Colombia
Daron Acemoglu
James A. Robinson
Rafael J. Santos
Working Paper 09-30
November 30, 2009
Room
E52-251
50 Memorial Drive
Cambridge, MA02142
This paper can be downloaded without charge from the
Social Science Research Network Paper Collection at
http://ssrn.com/abstract = 1 520082
The Monopoly
of Violence: Evidence from Colombia*
Daron Acemoglu*
James A. Robinson*
November
Rafael
J.
Santos 5
2009.
Abstract
Many
states in Latin America, Africa
as the foundation of the state,
and Asia lack the monopoly
and thus the capacity to govern
of violence, identified
effectively. In this
by
Max Weber
paper we develop a new
perspective on the establishment of the monopoly of violence and the formation of the state.
We
build
a model to explain the incentive of central states to eliminate non-state armed actors (paramilitaries)
a democracy.
The model
is
premised on the idea that paramilitaries
elections. Since paramilitaries
they favor to eliminate them.
may
in
choose to and can influence
have preferences over policies, this reduces the incentives of the politicians
The model
also
shows that while
in
non-paramilitary areas policies are
targeted at citizens, in paramilitary controlled areas they are targeted at paramilitaries.
We
We
first
then investigate the predictions of our model using data from Colombia between 1991 and 2006.
present regression and case stud}' evidence supporting our postulate that paramilitary groups can
have significant
is
effects
on elections
for
the legislature and the executive. Next,
also broadly consistent with the implication of the
we show that
model that paramilitaries tend to
the evidence
persist to the extent
that they deliver votes to candidates for the executive whose preferences are close to theirs and that this
effect is larger in areas
where the Presidential candidate would have otherwise not done as
results illustrate that, consistent with our model, there appears to
some executives and
paramilitaries.
Finally,
between the executive and paramilitaries
in
we use
roll-call
well.
These
be a symbiotic relationship between
votes to illustrate a possible 'quid pro quo'
Colombia.
Keywords: State Capacity, Violence, Elections.
JEL Classifications: D7, Hll.
"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 benefitted from the suggestions of many seminar particiants
and scholars, particularly George Akerlof, Lee Alston, Ana Arjona, Isafas Chaves, Malcolm Deas, Jorge Dominguez,
Gustavo Duncan, June Erlick, Francisco Gallego, Francisco Gutierrez Sanin, Macartan Humphries, Stathis Kalyvas,
Sebastian Mazzuca, Jose Antonio Ocampo, Rafael Pardo, Gina Parody, Diana Rodriguez-Franco, Fabio Sanchez,
Abbey Steel and Hernan Vallejo. We are also grateful to Miriam Golden for her advice on the Italian literature.
Finally we thank Maria Angelica Bautista, Lorena Correa, Leopoldo Fergusson and Pablo Quel ubin for their superb
assistance with this research.
Massachusetts Institute of Technology, Department of Economics, E52-380, 50 Memorial Drive, Cambriudge
e-mail: daron@mit.edu.
* Harvard University, Department of Government and IQSS, 1737 Cambridge Street N309, Cambridge MA01238,
*
MA02142, USA.
USA;
§
e-mail: jrobinson@gov.harvard.edu.
Yale University,
Department of Economics, 28 Hillhouse Avenue, New Haven
rafael.santosvillagran@yale.edu
CT
06511,
USA;
email:
Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium Member Libraries
http://www.archive.org/details/monopolyofviolenOOacem
"Political
power grows out of the barrel of a gun."
—Mao Zedong.
Introduction
1
Many
scholars have argued that differences in state capacity are a key factor in comparative
economic and
political
development (see
Evans, 1989, 1995, Kohli, 2004).
for instance
Evans, Rueschemeyer and Skocpol, 1985,
Although state capacity
multi-faceted,
is
on Weber's famous notion of the state as "a human community that
monopoly of the legitimate use of physical force within a given
it
inevitably relies
(successfully) claims the
territory" (1946, p.
78). States
vary greatly in their capacities and whether or not they have such a monopoly of violence, and
there
is little
evidence that this variation has decreased over the recent past. For example, in the
1990s the state in Somalia, Sierra Leone, Liberia, the Congo and Rwanda, completely collapsed
and gave up any pretence of undertaking the tasks that we associate with
Colombia, Peru, Guatemala, El Salvador, and Nicaragua have
experiencing prolonged
Waziristan. Similarly, the Iraqi state in
do some states
fail
recently experienced or are
Baghdad
to establish this
little
now
control of the 'tribal areas' such as
exercises little authority in Kurdistan.
monopoly? The
social science literature
several key ideas, for instance, the inability of states to establish such
'difficult
America,
wars, with the writ of the state being absent from large parts of the
civil
country. In Pakistan the central state in Islamabad has
Why
all
states. In Latin
emphasizes
monopoly because
of
geography' (Herbst, 2000), 'rough terrain' (Fearon and Laitin, 2003), or simply poverty
(Fearon and Laitin, 2003).
It
has also suggested that inter-state competition and warfare (Hintze,
1975, Tilly, 1975, 1990, Brewer, 1988, Herbst, 2000, Bates, 2001, Centeno, 2002) and domestic
political
competition influence the incentives of politicians to build state capacity (Acemoglu,
Ticchi, Vindigni, 2006, Besley
and Persson, 2007).
Common
to
all
of these explanations
is
a type
of 'modernization' view, suggesting that as society modernizes and grows richer, state capacity
will simultaneously develop.
In particular 'state formation' involves eliminating
and establishing a monopoly of
violence, in the
same way that
after the
Wars
armed
actors
of the Roses the
victorious Tudors disarmed the English aristocracy (Storey, 1968).
Yet several of the examples above are quite puzzling from this point of view. In the case of
Pakistan, the tribal areas have existed since the formation of the country in 1947, and even though
they have been largely out of the control of the central state, they have also been represented
within
it.
Under the 1973 Constitution the
Assembly elected by the
had 8 representatives
tribal areas
tribal elders, or the Maliks.
in the National
Under General Musharraf's regime
this
was
increased to 12. In Iraq, while the peshmerga militia control the streets of Mosul, a coalition of
Kurdish
see, as
political parties keeps the
much
government
as one third of the legislature
in
power
in
may have been
Baghdad. In Colombia, as we
shall
elected in elections heavily influenced
by armed paramilitary groups. After many of these were arrested by the Supreme Court, the
Colombian President did
little to
These examples point to a
stop their alternates from voting in their absence.
different
path of state formation than the one taken by England
under the Tudors and subsequently enshrined in the social science
gest that state formation can take place without a
monopoly
literature. Instead,
they sug-
of violence being established. In this
paper we develop a new perspective on state formation, emphasizing the idea that aspects of state
weakness, particularly the lack of monopoly of violence in peripheral areas, can be an equilibrium
outcome. Moreover, in contrast to the implicit notion
common
ernization' need not automaticahy eradicate non-state
armed
in the previous literature,
actors.
'mod-
Although we believe that
the ideas proposed in this paper have relevance both in democratic or non-democratic contexts,
we develop a model formalizing
We
these notions in the context of a democratic country.
investigate several of implications of this
then
model using data from Colombia.
Our model begins from the observation
that in a democracy non-state
armed actors
(in
our
context, paramilitaries) can control citizens' voting behavior. Since paramilitaries naturally have
preferences over policies,
when they choose
to
tives of the politicians they favor to eliminate
become involved
in politics, this reduces the incen-
them. The model predicts that in non-paramilitary
areas policies are targeted at citizens while in paramilitary areas they cater to the preferences of
paramilitaries.
This implies that in paramilitary areas citizens obtain fewer public goods (and
other policies they value).
The model
further implies that paramilitaries will tend to persist to
the extent that they deliver votes to politicians they prefer
— and that
dent Alvaro Uribe
this effect is stronger in areas
otherwise not done as well.
Thus non-state armed
— in
the Colombian case, to Presi-
where these politicians would have
actors can persist because they can be in
a symbiotic relationship with specific politicians holding power: paramilitaries deliver votes to
politicians with preferences relatively close to theirs, while politicians they helped elect implicitly
or explicitly support laws and policies that they prefer.
We
empirically investigate the implications of our
ence, where two
cionarias de
main non-state armed
Colombia (FARC
model using the recent Colombian experi-
actors, the 'left-wing' guerrillas Fuerzas
Armadas Revolu-
—The Revolutionary Armed Forces of Colombia)
and
'right-wing'
paramilitary forces, which in 1997 coalesced into the Autodefensas Unidas de Colombia
United Self-Defense Organization of Colombia), have shaped the recent
first
provide evidence that paramilitaries, though interestingly not the
influenced electoral outcomes. In particular, after the
presence of paramilitaries in a municipality
parties' (that
is,
is
AUC
political landscape.
FARC, have
systematically
correlated with the rise of non-traditional 'third
and the
Socialists),
widely recognized to be often directly or indirectly associated with the paramilitaries
We
We
got involved in politics in 2001, the
parties other than the Liberals, the Conservatives,
2007, Valencia, 2007).
(AUC
also find that paramilitary presence
is
which are
(e.g.,
Lopez,
also associated with a greater
concentration of votes within a municipality in legislative elections and with greater support for
President Alvaro Uribe,
who has enacted
several key policies in line with the preferences of the
paramilitaries, in the presidential elections.
The
effect of paramilitaries
senator's
list
the senator
on the elections
receives a greater proportion of
is
more
is
its
with paramilitaries
for illegal connections
and to support the two clauses of the Justice and Peace Law that were highly
the paramilitaries.
senators whose
Column
1
list
1
Table
1
depicts
some
a
votes in areas with high paramilitary presence,
be subsequently arrested
likely to
when
further substantiated by the fact that
On
of the relevant information.
it
lenient towards
we placed the 20
got the greatest share of their votes in areas with high paramilitary presence. 2
shows that 45% of these senators belong to
'third political parties'.
Column
4 shows
that the two senators with the highest vote shares have been arrested and found guilty of links
May
with paramilitary groups. As of
3 are
under investigation,
those in
office at
all for
2009 another 4 senators are under
links with paramilitaries.
Column
3
arrest, while
a further
shows that the majority of
the time also supported the clauses of the Justice and Peace Law.
The evidence mentioned
so far
is
consistent with the assumptions of our model, that paramil-
itaries were actively involved in influencing elections.
paramilitaries should persist
particularly in areas
where
more where they
this politician
The main
prediction of our model
is
that
deliver votes to the executive that they prefer,
would otherwise not do
This
well.
nating paramilitaries would implicitly cost valuable votes in the election.
We
is
also
because elimi-
show that the
correlations in the data are broadly consistent with this prediction.
we examine the
Finally,
stitution to
roll-call
remove the one-term
votes in the Senate on the legislation for changing the Con-
limit
and allow presidential
re-election to illustrate a possible
channel for the 'quid pro quo' between legislators elected from high paramilitary areas and the
executive.
We
find evidence that the greater
in high paramilitary areas, the greater
ing the term limit.
Column
2 of Table
was the likelihood
1
20' senators voted in favor of re-election.
Our econometric
paramilitaries
and
was the proportion of votes a senator's
list
obtained
of the senator to vote in favor of
shows that of those who voted
all
remov-
but three of our 'top
3
analysis proceeds under the assumption that our measures of the presence of
guerillas are exogenous.
pretations to the conditional correlations
We
are therefore cautious about giving causal inter-
we uncover. For example,
it
may be
that paramilitaries
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.
Table 1 uses our main measures of paramilitary presence using data on attacks and conflict incidents. Appendix
Table Al reproduces Table 1 using a different measure of paramilitary presence, with very similar results.
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
who have
that President Uribe's relatively conservative policies are closer to those preferred by the paramilitaries,
strong conservative leanings and thus naturally have an interest
our theoretical and empirical results then exploit the fact that he
decisions.
may
in
maintaining him
in
power.
Some
of
take this into account in several of his key
select into areas
where people's preferences are 'conservative' and would naturally support the new
'third parties' or President Uribe, thus creating a positive association
paramilitary presence. Nevertheless, most of our findings
effects, so
that
between these variables and
come from panel data models with
fixed
there are time-invariant differences in political preferences across municipalities,
if
these will not influence our results.
different municipalities are,
In addition,
we use
direct controls for
and as already mentioned above,
of the paramilitaries over time,
we
how
'conservative'
on the persistence
in the regressions
see paramilitaries persist precisely in places which, in the ab-
sence of paramilitary coercion, were relatively unlikely to have voted for President Uribe. Finally,
the fact that paramilitary presence predicts the arrests of senators suggests that politicians are
not simply the perfect agents of underlying voter preferences, but are in fact implicated with the
non-state
armed
actors, as the case study literature also suggests.
Our empirical evidence comes from a
in
making claims about external
emphasized
in this
specific country,
validity.
4
Colombia; we must thus exercise caution
Nonetheless,
paper are useful in building a richer
we
believe the political
political explanation for
why many modern
(and in fact democratic) states do not establish a monopoly of violence in their
very
least,
mechanisms
territory.
5
At the
the theoretical ideas and the empirical evidence presented here show that the implicit
notion that 'modernization' in less-developed economies will naturally lead to the formation of
a Weberian state, mimicking the European experience, needs to be revised, refined or perhaps
even abandoned. Colombia has experienced over a century of sustained increases in
GDP
per-
capita, large increases in educational attainment, rapid urbanization, indeed all of the features of
modernization (Robinson and Urrutia
ed., 2007).
Yet the state has not established a monopoly
of violence.
In addition to the literature cited above, the arguments in this paper are related to the recent
political
economy
literature
on the determinants of state
capacity. For example,
Acemoglu (2005)
conceptualizes state capacity as the ability to tax citizens, and examines the consequences of
state capacity for economic growth
and
endogenous creation of capacity by an
have an incentive to choose
It
to a
welfare.
Acemoglu, Ticchi and Vindigni (2006) model the
elite facing
democratization.
inefficient state institutions to limit the
can also be argued that the political equilibrium of the
modern Weberian
state.
Whether
this
is
last
decade
in
They argue
amount
Colombia
that the elite
may
of redistribution they
will ultimately
pave the way
the case or not has no direct bearing on our analysis. Nevertheless,
we
suspect that the symbiotic relationship between the executive and non-state armed groups is not a transient
phenomenon. Chaves, Fergusson and Robinson (2009) show that the outcome of the 1922 Presidential election
was determined by fraud and violence with regional patterns strikingly similar to those seen in 2002 or 2006. In
this light, the mechanisms we isolate here may be the most recent incarnation of a process whose roots lie deep
in Colombian history. In this interpretation, it may have been the prior absence of state presence and authority
that led to the formation of the AUC in the first place, but in doing so it may have reproduced the same historical
pattern of behavior.
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.
will face
under democracy. In a related paper, Besley and Persson (2009) develop a model where
politicians have to decide
whether to build
None
fiscal capacity.
with the issue of establishing a monopoly of violence, which
Within
political science our
work
is
relates to the literature
which has emphasized how democratization at the national
of these papers are
concerned
the focus of our paper.
on 'subnational authoritarianism'
level
can coexist with highly author-
itarian local practices (O'Donnell, 1993, Gibson, 2005, Mickey, 2009).
This research has given
examples of some of the mechanisms contained in our model, but has not developed these ideas
formally, has not noticed the key predictions that our
model develops,
or provided
an econometric
investigation of these ideas.
The
on
literature
example, in
its stress
civil
war addresses some
of the issues
on the weakness of the state
most of the research on
civil
this
work the
implicitly, for
Fearon and Laitin, 2003). Nevertheless,
war focuses on the motivations which
people to take up arms against their governments
and Miguel, 2008). In
(e.g.,
we emphasize here
(see, e.g.,
fact that a state does not
lie
behind the decisions of
the excellent survey by Blattman,
have a monopoly of violence arises
because given the expected benefit of allocating resources to fighting rebels or insurgents,
worth paying the cost of eliminating them.
Our work owes
a great debt to the journalists, scholars and public officials
roles in bringing to light the involvement of paramilitaries
Particularly important has been the
(2007).
it is
not
who have played
key
6
and the
AUC
in politics in
work of the researchers whose essays appear
Sanchez and Palau (2006) also show that
political
competition
is
Colombia.
in
Romero
negatively correlated
with murders of politicians in municipal elections.
The paper proceeds
as follows. In the next section
we develop
a theoretical
model to examine
the incentives of politicians controlling the central state to eliminate or live with non-state armed
actors depending on whether they receive electoral support from these groups. Section 3 provides
a brief overview of the history and nature of non-state armed actors in Colombia.
describes the data
we use and provides some
basic descriptive statistics.
Section 4
Section 5 provides
regression evidence consistent with the effects of paramilitaries on electoral outcomes. Section 6
examines the key prediction of our theoretical model, that paramilitaries should persist more
in
areas where they deliver votes to the executive and where they would not have otherwise done
as well.
Section 7 provides some suggestive evidence on another implication of our model, the
symbiotic relationship between the executive and the paramilitaries. Section 8 concludes.
There
eds.
is
also a large literature about the origins of conflict in
Colombia
(see Bergquist,
Penaranda, Sanchez
1992, 2001, Deas, 1999, and Posada Carbo, 2003). Influenced by the wider academic literature on civil wars,
work has emphasized the importance of state weakness in the Colombian context as well (e.g., Waldmann,
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.
this
2007).
Model
2
we
In this section,
present a simple model to formalize the possible channels of interaction be-
tween central government and paramilitaries. 7 Motivated by the Colombian experience, our focus
be on democratic
will
where an incumbent
politics,
facing reelection
is
and decides whether
to
reconquer some of the areas under paramilitary control. The model will highlight how paramilitary preferences influence electoral outcomes because paramilitaries can coerce voters to support
one candidate over another.
then show how the
It will
comes influences the willingness of the democratic central government
the paramilitaries from different areas
monopoly
state with a
to reconquer
Finally,
we
also investigate
presence of paramilitaries affects the policy choices of the party in power.
evidence showing
how
and remove
—the conditions of the formation of the modern Weberian
of violence over the entire country.
communicate the main ideas
on electoral out-
effect of paramilitaries
Our purpose
Our empirical work
in the simplest possible way.
paramilitaries influence electoral outcomes
and how
how
will
the
is
to
then provide
this effect
on elections
interacts with the persistence of paramilitaries in certain areas.
We
consider a two-period model of political competition between two parties.
initially (at
=
t
0) in
power and at
consists of a large equal-sized
We
of individuals.
of the votes over
t
=
1, it
denote the collection of these regions by
all
regions wins the election at
Downsian models that
ideological stance
an election against
in
is
part}'
is
B. The country
number, N, of regions, with each region inhabited by a large number
ideological preferences and, in addition,
as in standard
competes
A
Party
fixed (and
may
t
—
l.
some regions
parties can
8
J\f.
Regions
The party
differ in
that wins the majority
terms of their policy and
are under paramilitary control.
make commitments
We
assume
to their policies, but their
capture dimensions of policies to which they cannot
make
commitments).
Electoral Competition without Paramilitaries
2.1
We
first
introduce the details of electoral competition at date
decisions at
t
=
0,
in particular,
to those concerning
expend the resources to reconquer some
t
=
1
and then return to the
whether the government
in
power
of the territories under paramilitary control.
To
will
start
with, let us ignore the regions that are under paramilitary control (these will be introduced below).
The
utility of individual
i
in region j
G
N
(i.e.
j
=
1,
...,N)
when party g G {A, B)
is
in
work we examine the impact of both the paramilitaries and the FARC on elections. However,
FARC and since case study evidence suggests that the FARC was much
electoral politics, in the theoretical model we focus on the relationship between politicians and the
In the empirical
since
less
we do not
involved in
paramilitaries.
find robust effects of the
We
return to this issue in subsection 2.5.
This implies that we are looking at a "presidential system," though the empirical evidence below comes mostly
from votes for senators and congressmen. Focusing on the presidential system simplifies the argument without any
major implications for our focus.
power
is
given by
Uij (q,
where q 6
Q C K
that
Y
yO-j
individual
—
(i.e.,
)
Y
is
is
Uj (q)
-Y
-
(dj
~6 9
+ 4,
)
a vector of policies, Uj denotes the utility of
is
over this policy vector,
99
#) =
8j
is
We
individuals in region j
the ideological bliss point of the individuals in region j
a penalty term for the ideological distance of the party in power
a function that's increasing in
—
9-j
9
also
an individual-specific
assume that each Uj
utility
term that
is
strictly
6
J\f
so
,
and
the
This ideological distance captures
).
policy choices not included in q (and to which the party cannot
election stage).
all
make
a
commitment
at the
concave and differentiate. Finally, £?
will play the role of
is
smoothing regional preferences over
We
the two parties as in standard probabilistic voting models (Lindbeck and Weibull, 1987).
assume that
i
where
<^
is
common
a
another and
£{j is
an
"valance" term determining the relative popularity of one party versus
iid
term. To simplify the discussion,
uniform distributions over
of individuals in region j
6
j_ j_
20'
q
is
A
party
B are the policy vectors of the two
= y
the ideological advantage of party
that £
for
will
£
and each
A
(§j
[e 3
relative to party
and
its
the fraction
!
- eA
B
in region j
also uniformly distributed, the probability that party
policies, the policies of the rival party,
have
and
parties,
-e B )- y
,
Eij
be
+ ^h(<r4 )-M<? B )+0J +£]
Oj
is
we assume that
Therefore, conditional on the realization of £
.
2</>
M who vote
^
where q A and
-i B - £ + £
A
A
£
J\f.
Now
using the fact
gets elected as a function of
ideological advantage
its
is
N
PA
where 6
is
(^
B \e)
=
+ ^j:[uj ( qA )-uj
l
j]
is
4=1, Party
is
maxP A
RA
^ +e
the vector of ideological biases in favor of party A. In the election at time
A's problem
where
(q
(q,q
party A's rent from holding
office.
max[l
-P A
qGQ
B \6)R A
(1)
,
Conversely, the problem of party
A
(q
,
q
\6)}R B
,
R B is party B's rent from holding office and we have used the fact
party B coming to power is the complement of that for party A. An
B
is
(2)
where
that the probability
of
electoral equilibrium
at time
t
=
1 is a tuple (q
A
,q
B
)
that solves problems (1) and (2) simultaneously (given the
Given the concavity and
ideological biases 6).
uniquely defined; moreover, as long as
it is
differentiability assumptions,
interior,
N
N
A
J2 Vuj (q ) =
J^ Vu
and
B
(q
)
=
0,
(3)
j=i
i=i
where Vuj denotes the gradient of function Uj with respect to the vector
be
satisfied
if
the solution
B
q
=
q
some
q"
q* for
election at time
t
=
1
even
may
Clearly, (3)
Strict concavity of each
if
the equilibrium
is
u
not
immediately
to see that strict concavity implies q
.
,
q.
not in the feasible set of policies, Q, and in this case, an obvious
is
complementary slackness generalization of (3) holds.
implies that A = B = q* It is also straightforward
q
is
the following equations
satisfies
it
an equilibrium
not interior.
Therefore, party
A
A
=
win the
will
with probability
N
ch
P A {q\q*\e)=- + ^Y. ei=2 + ^>
1
1
(
4)
j=l
where E#j denotes the expectation or the mean of
This then leads to the
8j across all regions.
following proposition, characterizing the equilibrium (proof in the text).
Proposition 1 Without paramilitaries,
where A = B — q* and q*
interior,
q
q
given by
Two
,
if
,
there exists a unique electoral equilibrium (at
Party
satisfies (3).
A
important points to note are as follows.
though as
is
is
well
special
known the
and
relies
fact that
on the
First,
without paramilitary presence, national
for the simpler specification.
under various
different
A
(which
is
currently in power).
We
all
regions determines
will
Next,
let
number
is
us suppose that a subset of the regions, denoted by
number
of total regions) by z.
that, as
we
will
document
of these regions by
The key
how
this
Z
and
Z C
N
are under paramilitary
their fraction (their ratio to the
feature of paramilitary-controlled areas for our purposes
in detail below, voting
explicit pressure of the paramilitaries.
next see
assumptions about paramilitary behavior.
Elections under Paramilitaries with Exogenous Preferences
total
prefer-
not important for the results here, and we therefore
2.2
Denote the
is fairly
two parties do not themselves have
Second, average ideological bias across
the probability of reelection for party
result changes
is
This feature
both parties choose the same policy vector (policy
fact that the
ences over policies. Whether they do or not
control.
1)
(4).
convergence)
opted
=
wins the election with probability
policies are chosen to cater to the preferences of all voters in all regions.
general,
t
is
Throughout the
not free but influenced by the implicit or
rest of this section
we impose
this feature.
We
with paramilitaries with "exogenous preferences," meaning that how the paramili-
start
taries influence the voting behavior of the citizens in the regions they control
with the case in which the support of the paramilitaries
will contrast
choices, studied in the next subsection. In particular,
exogenous. This
is
endogenous to the policy
is
we take the behavior
of the paramilitaries
(and the voting behavior of the citizens in paramilitary-controlled areas) as given.
suppose that
ular,
party
will vote for
each paramilitary-controlled region j £ Z, a fraction fhj of the voters
in
A
regardless of policies (so the voting behavior of these individuals in these
paramilitary-controlled regions
set
J
Z
=
J = N\Z
by
N — Z.
and the
insensitive to policies).
is
number
total
Let us also define
rtij
=
fhj
—
{
q
\ qB
6,
|
Let us denote the complement of the
of regions in this (non-paramilitary-controlled) set by
Then with an
1/2.
previous subsection, the probability that party
PA
m) = \ + tSLzA
A
£
will
Uj
[
identical reasoning to that in the
win the election
(/) _
(,B)
Uj
+
at
time
,.]
+£
m denotes the vector of m^s
We
=
t
1 is
£ mh
jez
jej
where
In partic-
(together with information on which j's are in the set Z).
again assume that both parties maximize the probability of coming to power and define
an electoral equilibrium
in the
same way. With an
identical
argument
to that before,
we obtain
the following proposition.
Proposition 2 Under paramilitaries with endogenous preferences, there exists a unique electoral
q* is interior, it satisfies ~^2j
equilibrium (at t = 1) where A = B = q*
Vuj (g*) = 0. Party
q
q
A
.
&j
If
wins the election with probability
p a (q*, q
Two
*
|
= \ + 4>Q- -
e,m)
~W
I
; e J]
features that are noteworthy relative to Proposition
longer cater to the preferences of
all
+
1
*EK
I
j
e z].
are as follows.
First, policies
no
regions. Since citizens in paramilitary-controlled areas cannot
reward or punish a government according to the policy proposals that
it
makes, both parties only
target their policies to the voters in the non-paramilitary-controlled areas.
This implies that,
endogenously, public goods and other amenities will be reduced in the paramilitary-controlled
areas beyond the direct effect of paramilitary presence. 9
Thus,
all else
equal,
we may expect
paramilitary presence to increase inequality across regions. Second, electoral outcomes will
be dependent on the influence of the paramilitaries on voting behavior, which
last
term
in
PA
{q*,q*
|
8,
m).
then the probability that party
equal.
The
The more
direct effect
more expensive,
If
A
paramilitaries prefer party A,
will
win the election (and stay
for
captured by the
meaning that E \mj
\
in
power)
is
areas are controlled by the paramilitaries, the stronger
may,
is
now
j € Z)
>
0,
greater, other things
is
this effect.
In the
example, stem from the fact that such investments and public good delivery become
damage infrastructure, law and order and the availability of public goods.
or paramilitaries directly
empirical work below,
we
will
provide indirect evidence consistent with Proposition 2 by showing
the influence of paramilitaries on electoral outcomes.
This
cratic politics. Nevertheless, this effect
exogenous preferences
discuss
how the
how
already highlights
last feature
was minimized by the model
We
for the paramilitaries.
will relax this
in this subsection
its
by assuming
assumption below. But
electoral role of paramilitaries affects the decision of the central
extend ("broadcast")
demo-
paramilitaries can have a major influence on
first
we
government to
power to peripheral areas controlled by the paramilitaries.
The State and the Paramilitaries
2.3
at time
Taking the electoral equilibrium
the government (party A) at time
—
t
4
=
access to guns
monopoly
its
of violence
and means of exercising
way and suppose
that at time
4
=
C Z
is
and thus remove the power of other groups with
model
this in the simplest possible
the objective of the governing party
0,
4
1Z
the ability and willingness of the central
is
(local) violence. Let us
Et. + pwhere
consider the decisions of
In particular, as discussed in the Introduction, a key
0.
dimension of the process of the formation of the state
government to establish
now
1 as given, let us
is
^,^!^,
(5)
a subset of the areas previously controlled by the paramilitary that are "recon-
quered" by the central government, and 7j
accrues to the government at time
4
=
the net benefit of reconquering area j £
is
and subtracts the potential
reconquest (spending on the military, potentially stability and loss of
of the governing party, party A, also includes the probability that
4
= 1.
In particular,
the subsequent electoral equilibrium at time
4
=
1,
=
votes from this region as opposed to receiving rhj
under paramilitary control.
equilibrium at date
utility taking the
4=1
date
A
A
party
rrij
will
+ 1/2
j
will
in
obtain a fraction 1/2
this
A
remain
game
at date
power, thus
2 is reconquered,
£
of the votes
subgame perfect equilibrium of
"real" cost of the
However, the objective
life).
it
some area
if
together with decisions by party
4=1
which
This net benefit includes the additional tax revenues or
0.
security gains that the central government will drive
enjoying rents from power at time
7Z,
had
is
4
then
+ <f)dj
this place
in
of the
remained
defined as an electoral
=
that maximizes
its
equilibrium as given.
This analysis in the preceding paragraph then establishes the following proposition.
Proposition 3 A subgame
Proposition 2 at time
perfect equilibrium involves the electoral equilibrium characterized in
4=1, and
all j
1
One
£
at time
Z
4
=
0,
Party
such that jj
+
A
(<fi8j
reconquers
-
could easily extend this so that these rents accrue both at.
+ P A (<?, q B 0) R A + Yl g-re li
functions will change to X)jctc 7j
\
\
10
t
>
rrij)
—
RA >
and
t
=
1,
and
in that case, the objective
slightly complicating the analysis.
and does not reconquer
any j 6
This proposition
is
Z
+
such that jj
(<p9j
—
rrij)
an important result of our analysis and
R <
will
be investigated
It
taries,
and thus establish the monopoly of violence envisaged as an
also
in
our empirical
implies that the willingness of the state to reconquer areas controlled by the paramili-
work.
modern
0.
state by
Max
Weber,
is
by the
affected not only
real costs
essential characteristic of the
and benefits of doing
if
many
of these paramilitary-controlled areas have
>
rrij
<f>9j,
then the state, currently
controlled by party A, will be reluctant to reconquer these areas, because doing so will
more
make
it
party to succeed in the upcoming elections (and moreover, this effect will
difficult for this
be stronger when rents from power
at
t
=
R
1,
are higher). Naturally, the areas that are most
,
valuable in the hands of the paramilitaries are those that have both low 9j and high
rrij;
that
areas that would have otherwise voted for party B, but paramilitaries are forcing citizens to
A
vote in favor of party A.
government that does not require electoral support
non-democratic" government) would have decided to reconquer
to the extent that
cfiiE
[8j
|
j £
government
may be
broadcast
less willing to
E
J\ <
ologically closer to the
in
its
[rrij
\
j
£ Z],
i.e.,
all
areas with jj
(e.g.,
>
0.
a "purely
Therefore,
to the extent that paramilitaries are ide-
power than the opposition party, a democratic government
power and reconquer areas under paramilitary control than
such a non-democratic government (or a government that
is
secure in
its position).
11
Note an important implication of the functional form assumptions we have imposed so
in particular the
probability"
.
in reality, this value,
its
is
constant and independent of
3 takes a simple form,
its
"expected winning
where the value of paramilitary
independent of this probability. With other functional forms, as
and thus the behavior of
this party
towards the paramilitary groups,
expected winning probability, for example, making
when
delivered by these paramilitary forces
it is
it
less
may
responsive to the votes
ex ante more likely to win the election.
Electoral Competition under Paramilitaries with Endogenous Preferences
2.4
The
is
As a consequence, Proposition
votes to the party in power
depend on
far,
uniform distributions of idiosyncratic preference and valance terms: the value
of additional votes to the party in power
tor
but
by the implications of this expansion of the authority of the state on electoral outcomes. In
particular,
is,
so,
discussion so far was for paramilitaries with endogenous preferences
m
as given.
and thus took the
vec-
Naturally, the willingness of the paramilitaries to coerce citizens to vote for
"Naturally, the net benefit of reconquering an area might be different for a non-democratic government. For
it might be jj > 7j instead of 7j
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
example,
,
governments to broadcast their power. On the other hand, the cost of doing so may also be higher for non-democratic
governments because 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 7j < ~1j12
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.
11
one candidate or another
ences.
We now
is
endogenous and depends on their policy and ideological prefer-
also
paramilitaries controlling region j
W
now
Y
9j is
is
Z
€
that, as with the citizens, the preferences of the
are given by
{q,e3)=w] {q)-Y(63
3
where
Suppose
investigate these issues.
-~63
+ i9
another function (possibly the same as Y) that
is
also increasing in
the policy preference of the group of paramilitaries controlling region
reasoning to that above,
let
=y
-
(9j
~9
B
)
-
Y
(dj
-
of 9j to highlight that this refers to the paramilitaries).
.J_ _L
a uniform distribution over
A
to party
B
is
Then
.
wj
'
And
in addition,
(q
(
1
.
all
all
may
suppose that if
— if
will
Z
still
win the election becomes
l3 \q
r-i
=
for all q
in the previous
two sections
€ Q, so that paramilitaries do not care
care about the ideological stance of the party in power).
similar reasoning to our analysis above, electoral competition will lead to the
policy choice for both parties,
has
ideological preferences, including those of the paramilitaries.
a special case of this model where Wj (q)
With a
A
model with paramilitaries with exogenous preferences
about policy (though they
9j instead
voters in their sphere of influence to vote for
J
where now 8 denotes the vector of
is
(we use
d>(l-z)
8
Z
Naturally, the
A
A )- Wj B
( q )+<
whichever party they prefer. Then the probability that party
B
similar
the probability that paramilitaries in region j €
Let us also assume that paramilitaries can force
q \<l
With a
and
given by
<t>
PA
99
fr
as the ideological leanings of the paramilitaries in region j in favor of party
party
-
us define
§.
will prefer
j.
6j
and when
it is
interior, this vector will
same
be given by the solution to
the following set of equations:
1
(1
-
z)
Vuj
(<f )
+ faVwj (q*) =
0.
Naturally, these equations hold in the complementary-slackness form
boundary of the
Therefore,
(6)
when
q*
may be
at the
feasible policy set Q.
we obtain the
following characterization of electoral equilibrium and efforts by the
state to reconquer paramilitary-controlled areas under the control of the paramilitaries (proof in
the text).
L2
Proposition 4 Under paramilitaries with endogenous preferences, there exists a unique electoral
A = B = q*
equilibrium at t = 1 where q
If q* is interior, it satisfies (6). Party A wins the
q
.
election with probability
PA
(q*,q*
\6,m) = ^+4>(l-
*)-E[fy
|
j
Moreover, the subgame perfect equilibrium involves Party
Z
€
all j
such that jj
+
(<j)6j
+ zj>E
£ J]
A
\o3
\jeZ
.
reconquering (at time
- 06U R A >
t
=
0)
0,
and not reconquering
any
Z
£
j
There are several new features
such that 7j
+
69j — 66j
(
in this proposition.
First,
)
R <
0.
when
paramilitaries adjust their
support depending on the policies and ideological stance of the two parties, the parties then
change their policies
That
is,
in order to
be more attractive to the paramilitaries' policy preferences.
rather than catering to the preferences of the citizens in the areas that are controlled
by the paramilitaries (which they would have done without the paramilitaries), parties appease
This result
the paramilitaries themselves.
is
the basis of the potential symbiotic relationship
between paramilitaries and the executive mentioned
increase the inequality
refrain
among
in the Introduction.
Moreover,
it
the regions, with the policies chosen specifically to support, or
from threatening, the paramilitaries and the areas where the paramilitaries are strongest.
Two
features determine
are:
the size of the paramilitary-controlled areas (the greater
how
slanted towards the paramilitaries equilibrium policies are.
paramilitaries in shaping equilibrium policy)
to policy concessions (the greater
is
6
and the
relative to
preferences relative to citizen preferences).
<f>,
is z,
These
the more influential are the
relative responsiveness of the paramilitaries
the more responsive are policies to paramilitary
In addition, because electoral competition makes
both parties cater to the wishes of the paramilitaries,
at the
end the paramilitaries ideological
preferences play a central role in whether they force the population to vote for party
B.
can further
A
or party
13
Finally,
we can
also allow
(in a credible fashion).
both parties or one of the parties to modify
The same
its
ideological stance
analysis as here will then imply that in order to attract votes
from paramilitary-controlled areas, one or both parties
may
decide to pander to the ideological
preferences of the paramilitaries.
13
The
result that
both parties modify their policies to partly cater to the wishes of the paramilitaries has an
logic, to Baron's (1994) result concerning the effects of a lobby
obvious similarity, and an identical mathematical
on the platforms of two competing parties.
13
2.5
Importance of Non-National Ambitions
An
important question in the context of Colombian politics
have become more involved
in influencing elections
is
why right-wing paramilitary groups
than left-wing
than the relatively well-organized FARC. One possible answer
is
made
this
so
that in contrast to the guerrillas,
the paramilitaries do not have national ambitions, making a coalition between them
executive controlling the central state more feasible.
more
guerrillas, in particular,
The model presented
and the
so far has implicitly
assumption, since we did not introduce the risk that the non-state armed actors
may
take over the central state.
A
state
simple
way
would be
of introducing this possibility
armed actors would become strong enough to challenge the central
state,
Such an overthrow of the central government by non-state armed actors
it.
weak African
of the
states,
FARC.
control
more
equation
(5)
we would expect $(2)
Naturally,
areas, they are
more
likely to
to be increasing in z, so that
when
in
these groups
pose such a national challenge. In that case,
A
we would
For example,
to incorporate this possibility.
it
A
Y,lj + [l-Hz)}P
(q
when &(z) >
0,
clear that
A
-q
and
sufficiently high
B \8)R A
-
there will be stronger incentives for party
to reconquer territories controlled by these non-state
is
uncommon
not
could be modified to
This specification makes
§{z)
is
perhaps overthrow
such as Somalia, Sierra Leone or Liberia, and has certainly been the objective
need to change the objective function of party
A
to have a probability $(2) that the non-
armed groups (thus reducing
sufficiently decreasing in z, this effect
z).
When
can more than compensate
for
the electoral advantage that local control by these groups creates for the party in power. Thus
factoring in the national ambitions of non-state
armed
actors reduces the
or a symbiotic relationship between these groups and the executive.
when non-state armed
reasoning suggests that
it
Expressed
actors have national ambitions,
geous for the central state to eliminate them (sooner or
promises that
room
later),
it
for a coalition
differently, this
will
be advanta-
thus any implicit or explicit policy
makes to such groups would be non-credible, making a
coalition
between them
impossible. This perspective suggests a natural reason for why, in Colombia, such a coalition
have been much more
with the
2.6
likely to arise
between the executive and the paramilitaries rather than
FARC.
Summary and
Empirical Predictions
In the rest of the paper,
electoral
may
outcomes
we
will investigate the effect of
in the early 2000s.
discussed above. In particular,
we
will
Our
investigation
Colombian paramilitary
is
forces
on the
motivated by the theoretical ideas
document the following broad patterns, which, though not
conclusive proof of the ideas developed here, are highly suggestive.
14
Consistent with Proposition
1.
2,
paramilitaries, once they
Colombia they
influencing electoral outcomes in the areas of
Consistent with Proposition
2.
we
3,
will
became
sufficiently powerful, started
controlled.
show that paramilitaries located
in areas that voted
for the current President in great numbers, but in past elections tended to vote for
liberal politicians, are
Consistent with Proposition
3.
line
A
we
4,
will
show that the President has proposed
have supported this
Armed
Brief Overview of Non-State
convulsed by a series of
and 1902
civil
actors.
During the 19th century the country was
wars which culminated in the
(see Pardo, 2004, for
and Robinson, 2009) and lasted
War
of
Thousand Days between 1899
Relative peace emerged subsequently (Mazzuca
an overview).
until the 1940s
Actors in Colombia
Armed Actors
Colombia has a long history of non-state armed
civil
legislation in
legislation.
Origins of Colombian Non-State
3.1
more
likely to persist.
with the preferences of the paramilitaries, and the Senators elected from high paramil-
itary areas
3
more
when the country again
war known as La Violencia, which slowly came to a halt
slipped into a murderous
after 1958
when
the Liberal and
Conservative parties signed a peace treaty and re-established elections, albeit under a highly
restrictive set of
Fals
power sharing institutions known
as the National Front (see
Guzman Campos,
Borda and Umaria Luna, 1962, Oquist, 1980, Pecaut, 2001, Roldan, 2002 on La Violencia
and Hartlyn, 1988, on the National Front). Though the National Front brought
year conflict between the Liberal and Conservative parties,
In 1964 out of the ashes of various Liberal
the Fuerzas
Armadas Revolucionarias de Colombia (FARC
Army) was
same year the
also created.
and 1970s, but began
revolutionary
These
to
expand rapidly
movements such
and Quintfn Lame. The 1980s
which
in
as the
also
— The
in the
guerillas,
Revolutionary
Ejercito de Liberacion Nacional
'left-wing' guerilla
an end the 100
issued in multifarious
and Communist
of conflict.
of Colombia). In the
it
to
(ELN
new forms
was formed
Armed
Forces
—National Liberation
groups were relatively small during the 1960s
1980s and they were joined by other left-wing
Movimiento 19 de Abril (M-19
saw the rapid expansion
—Movement
of April 19)
of 'right-wing' paramilitary forces
1997 coalesced into the Autodefensas Unidas de Colombia
(AUC — United
Self-Defense
Organization of Colombia).
These various non-state armed groups ranged over most of the
though estimates
vary,
of the 21st Century.
may have had around
They engaged
50,000
in kidnapping,
territory of
Colombia and
men and women under arms
massacres of
civilians,
at the start
drug production and
exportation, and regularly expropriated land and extorted income ('collected the taxes') from
Colombian
citizens.
They
also
engaged
in violent conflicts
15
with each other and with the armed
forces of the
and the
Colombian
AUC
state.
Map
1 in
the Appendix shows the distribution of the
across Colombia's 1,119 municipalities (using the 'attacks' data, described below,
The map
over the period 1997-2005).
uses different shades of red to indicate municipalities
The darkest red
in different quintiles of the distribution of guerrilla presence.
20%
FARC/ELN
of places with the
most intense
guerilla activity.
One can
indicates the
see that this covers
most
of
Colombia. Departments in the eastern planes ('Los Llanos'), such as Arauca and Casanare, have
high guerilla presence as do municipalities further south in Caqueta, where the negotiations took
place during the Presidency of Andres Pastrana.
However, high guerilla presence can also be
seen in the Northeast, in the department of Norte de Santander, further west in Antioquia and
in the
same
in
southwest
data.
many
One
of the
in
Map
Cauca.
Appendix examines paramilitary presence using the
2 in the
sees high paramilitary presence,
same
areas.
Paramilitary presence
in Cesar, inbetween
is
measured
in the
same way
as for the guerilla,
These include the eastern planes, Norte de Santander and Antioquia.
also high in municipalities in the coastal
Magdalena and Norte de Santander.
This
department of Magdalena and
is
natural, since paramilitaries
often formed to fight the left-wing guerrillas.
Paramilitaries and the
3.2
AUC
Colombia's paramilitaries are thought to originate from 1960s counterinsurgency measures and
Law
48 of 1968 which allowed the creation of self-defense militias by private citizens for the
purposes of protecting their properties and
Duncan, 2007,
the period of
for
lives (see
Romero, 2000, 2002, Rangel, 2005, and
overviews of the history and organization of the paramilitaries). Nevertheless,
La Violencia
is
littered
with various sorts of militias and spontaneous self-defense
groups (see Roldan, 2002, on Antioquia). Small groups of paramilitaries also emerged in places
where there were valuable resources whose exploitation was contested such as
in the
emerald
mines of Muzo, in the department of Boyaca and the drug plantations on the coast near Santa
Marta, Magdalena and
The
in the
nearby department of La Guajira.
escalation of paramilitaries in the early 1980s
cartels in Medellin
and Cali that faced threats
Martha Nieves Ochoa, the
In 1981
sister of
of
is
associated with the rise of the large drug
kidnapping and extortion from left-wing groups.
Jorge Luis Ochoa Vazquez, one of the founders of the
Medelh'n drug cartel (along with his brothers, Pablo Escobar, Carlos Lehder and Jose Gonzalo
Rodriguez Gacha), was kidnapped by the guerilla group M-19. In retaliation, the cartel formed an
armed paramilitary group
aim was
As
called
Muerte a Secuestradores
(MAS — Death
to Kidnappers)
whose
to eliminate kidnappers.
the wealth of the drug cartels grew,
many
of their
members began
to
buy up land and
ranches in rural areas. Here their interests began to fuse with those of traditional rural
also wished to protect themselves
from extortion and kidnappers
(see Gutierrez
elites
who
Sanin and Baron,
2005, for a study of this process in the Puerto Boyaca region). This led to collaboration in the
16
One
formation of paramilitary groups.
area of rapid expansion was the Magdalena Medio at the
eastern periphery of the department of Antioquia which saw the emergence of groups such as Los
Tangueros formed by the Castano brothers (Carlos, Fidel and Vicente). Five of Carlos Castaho's
11 siblings were killed by guerrillas, following the abduction
FARC
Jesus Castano, in 1981. In 1981,
and death
of their dairy-farmer father,
snatched the Castano's father and demanded a $7,500
ransom, which was raised by mortgaging the farm. Though the ransom was paid, the father's
corpse was found chained to a tree. Los Tangueros was founded to revenge this act. 14
In 1992 people
who had been
victimized by the drug lord Pablo Escobar founded a paramilitary
organization called 'Los Pepes' ('Perseguidos por Pablo Escobar'
—those
persecuted by Pablo
Escobar). This group, which played an important informal role in the hunt for Escobar and his
death in 1993 (Bowden, 2002) apparently included
had previously cooperated with the Medelhh
Campesinas de Cordoba y Uraba'
Castano brothers, even though they
of the
all
1994 the brothers formed the 'Autodefensas
cartel. In
(ACCU — Peasant
Self-Defense force of
Cordoba and Uraba).
This further expansion was facilitated in the same year by a law promoted by President Samper
to allow the creation of
important supported of
was
killed
CONVIVIR,
this
a national program of neighborhood watch groups.
program was Alvaro Uribe, then Governor
whose father
by the FARC.
In April 1997 the
AUC
existing paramilitary forces.
was formed by Carlos Castano and
The
of the country,
though as our data
will
FARC
show these
coming
to
power
in 2002, President
and
guerrilla
of Colombia (see Restrepo, Spagat and Vargas, 2004).
after
it
included possibly
90%
of the
creation of this national organization increased the effectiveness
of the paramilitaries considerably; as a result, the
Soon
of Antioquia,
An
ELN
were thrown out of large areas
groups are
still
active in
many
parts
15
Uribe began to negotiate the demobilization of
the paramilitaries, something he had promised during his election campaign. Decree 128 issued by
the President in January 2003 gave de facto amnesty for paramilitaries not under investigation for
human
92%).
rights violations
On
and
this
has been applied to the vast number of demobilizations (around
July 15, 2003 in Santa Fe de Ralito in Cordoba the government signed an agreement
with most of the groups of the
On November
AUC
to disarm
by the end of 2005. 16
25 2003 around 860 paramilitaries of Medelhn's Cacique Nutibara Bloc led by
'Don Berna' (Diego Fernando Murillo) demobilized. This process was further institutionalized by
the passing of the controversial Justice and Peace
14
There
is
also evidence suggesting involvement of the
Law
army
in
in
June 2005 which was signed into law
the training and organization of paramilitary
Law 48 was unconstitutional. One month later
Barco issued Decree 1194 which prohibited the creation, promotion or organization of paramilitary or
self-defense groups and declared such activities illegal.
15
The timing of the creation of the AUC was a consequence of the collapse of the Medellin and Cali drug cartels
which had previously exercised a large amount of control over the organizations.
16
The text of the agreement is available at the web page of the Office of the High Commissioner for Peace:
www.altocomisionadoparalapaz.gov.co/acuerdos/index.htm
groups, though in 1989 the Colombian Supreme Court declared that
President.
17
by President Uribe
guilty of
human
in the following
rights violations to
month. Article 29 of
between
5
and 8
this
law limits sentences to those found
years. Article 30 allows the
determine the place of detention, which need not be a prison. In
stitutional Court altered
in particular the
many
whether
is
it
2006 the Colombian Con-
Court stipulated that demobilizing combatants had to give a
Justice and Peace
Law have been
widely criticized by
full
confession of
Both the demobilization process and the
human
rights organization.
17
a great deal of controversy about whether the paramilitary demobilization
will lead to
the power of the
new armed groups
AUC; on
to
aspects of the law on the grounds that they were unconstitutional,
their activities in order for the law to apply to them.
There
May
government
this issue see
(or
is
real or
simply to the institutionalization/legitimation of
Pardo, 2007, International Crisis Group, 2007, Porch and
Rasmussen, 2008).
3.3
Soon
The Involvement
of Paramilitaries in Politics
after the foundation of the
AUC
in 1997 there appears to
influence electoral politics. This change
is
have been a strategic decision to
traced to a historic meeting in Santa Fe de Ralito in
2001 where members of the estado mayor (the governing body) of the
and members
document
of Congress signed a secret
Those who signed
this
AUC
along with politicians
calling for the 'refounding of the country'.
document included prominent paramilitary
leaders, such as Jorge
40 (Ro-
drigo Tovar Pupo) and Diego Vecino, and several politicians subsequently arrested for links with
paramilitaries, including Senators William
explicit
aim of the accord was to have the
Montes and Miguel de
AUC
la Espriella (see
Table
1).
An
play a more important role in electoral politics. 18
For instance the 2007 report of the International Federation for Human Rights notes: "The paramilitaries who
do fall under the Justice and Peace Act are tried at so-called "free version" hearings and may be sentenced to
no more than eight years of imprisonment, which they may serve in "work farms." They may even impose their
own conditions for "imprisonment," which flies in the face of the most basic principles of justice in view of the
seriousness of the crimes committed." The report also emphasizes "the lack of true willingness on the part of the
government to bring to trial and dismantle the paramilitary groups," and concludes that: "The paramilitaries
are not forced to confess to their crimes, disclose the truth about who supported their structures, or even show
repentance for their crimes. They have not been forced to turn in all of their weapons or hand over their assets
to compensate the victims, while the latter and their representatives have very limited access to hearings and are
hindered from participating in them. ...What is more, those victims who have attended the "free version" hearings
have not received adequate protection. Already, sixteen of them have been murdered with absolute impunity." See
also Human Rights Watch (2005) and Amnesty International (2005).
On the issue of handing in weapons the Inter-American Commission on Human Rights (2007) noted at some key
demobilizations in the department of Cesar that, "approximately 800 persons had passed through that circuit on
the previous day, but only 65 firearms were received. It also noted that from the 200 persons passing through the
circuit at La Mesa on the previous day, roughly 25 firearms were received. None of the weapons surrendered were
modern or in good condition" (p. 8).
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 offices 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
'
1
18
This change in the strategy of the
to investigate
how
electoral
AUC
be crucial to our empirical approach, allowing us
will
outcomes change
differentially in high paramilitary areas before
involvement in politics in 2001.
after their
The other
brand new
notable, and related, development during the 2002 election
political parties,
MIPOL. These
parties often
which we
had
the emergence of
is
Cambio Radical and
refer to as 'third parties,' such as
explicit or implicit links with the paramilitaries,
study evidence shows that paramilitary pressure was often to increase the vote
many
In
and the case
for these parties.
paramilitary-controlled areas they have replaced the traditional Liberal and Conservative
We
parties.
and
will use
the vote share of 'third parties' as a measure of paramilitary influence on
electoral outcomes.
Beginning
in
2005 there were increasing accusations of involvement of the
of 2002. Scandal
on March
officials
2006
10,
and
it
mounted
in
Cesar.
Jorge 40's computer
contained emails ordering his
during demobilization ceremonies.
links
and
his 2,000 strong block
into the
hands of government
further with the demobilization of Jorge 40
La Mesa,
19
men
fell
AUC in the elections
to recruit peasants to pretend to be paramilitaries
The computer
also listed over 500
murders and detailed many
between politicians and paramilitaries. These accusations led to intense scrutiny of the 2002
election results
changes
many
in voting patterns
municipalities.
some rather extraordinary
of which exhibit
and very high concentrations of votes
These include massive
features.
for
some candidates
in particular
20
Since then there have been
many
investigations of supposed links between politicians
and
paramilitaries and a large case study literature has emerged documenting such links (see in particular the research of Lopez, 2007, Valencia, 2007,
Zabala, 2007 and the
web
site
and the other essays
verdadabierta.com). As of
May
and the Senate were under investigation, 36 were arrested and
guilty of links with paramilitaries.
21
in
Romero. 2007, Serrano
29, 2009, 39
in detention,
members
and
11
of
Congress
had been found
In total this represents almost 1/3 of Colombian legislators.
Those previously arrested but released include Mario Uribe, President Uribe's cousin and main
Those arrested include Senator Carlos Garcia, the President
political adviser.
(U
for
Uribe
—though
investigation
and
this party is not officially recognized
arrest of these politicians has
of the
by the President as
"U party"
his party).
The
been undertaken primarily by the Supreme Court.
Magdalena (discussed in the Appendix), the pact with the paramilitaries involved 417 local
committed to support the candidates linked with the paramilitaries for the legislative elections
2002 through a movement called "Movimimiento la Provincia Unida" (Movement United Province) (Semana,
of the department of
politicians that
of
November
19
6,
2006).
See also Inter-American Commission on
Human
Rights (2007,
p.
5)
on the apparently fake demobilizations
in
Cesar.
*
instance the article in the Colombian weekly Semana "Votaciones
September 11, 2005.
This data is updated regularly on http://www.indepaz.org.co.
See
for
del 2002,"
1')
atipicas en las elecciones de congreso
Controlling the Vote
3.4
There
considerable case study evidence that following the meeting in Santa Fe de Ralito,
is
paramilitary groups actively tried and succeeded in influencing votes in national elections (that
is,
in
The testimony
the 2002 and 2006 elections).
groups replaced the authority of the state in
have been quite articulate about their
Mancuso
is
perhaps most
"What
I
said
many
areas
delivered justice,
all
the people
35%
of the Congress
representatives
translation)."
The
we had
to go into politics
there." (Salvatore
paramilitaries has revealed a large
politicians (detailed in
tions.
A
had
in areas
where there were
we were the ones
collecting taxes,
territorial control of the region
to
come and
and
deal with the political
Mancuso, second 53 of the Interview, authors
22
investigation into the 2002
different strategies
these the testimony of Salvatore
was elected
and we had the military and
who wanted
Of
'political project'.
states of the Self-Defense groups, in those states
we
and many of the paramilitary leaders
noting that
telling,
that
is
of major paramilitary leaders suggests that these
and 2006 election
number
results
and the testimony
of demobilized
of different 'pacts' between paramilitary leaders and
Lopez and Sevillano, 2008) and
also
demonstrates that a large number of
were used to guarantee that candidates preferred by paramilitaries
salient strategy
seems to have been to terrorize people into voting
In the municipality of San Onofre in the coastal department of Sucre,
won
elec-
for specific candidates.
23
for
example, this was
arranged by the paramilitary leader 'Cadena'.
"For the elections of 2002, the trucks sent by 'Cadena' went through neighborhoods
and
rural areas of
San Onofre picking people up. According
to
some people
in this
municipality in Sucre, thousands of peasants were taken to the corregimiento 'Plan
Parejo' so they could see the candidate for
elections: Jairo
Merlano
for
'Cadena' put in a bag
was going
to
kill
whom
they had to vote for in the legislative
Senate and Muriel Benito for the House of Representatives.
all
the
names
of the councilors, he took two
them and other people chosen randomly
if
a peasant from this town. The threat -seems to have been
and said that he
Muriel did not win"
effective:
,
says
each candidate
obtained 40,000 votes in Sucre." 24
The exact words in Spanish are "Lo que dije fue que el 35 % del congreso fue elegido en zonas
donde habi'an Estados de Autodefensas, en esos Estados nosotros fuimos los que cobramos impuestos,
impartiamos justicia, tenemos el control territorial y militar de la region y todas esta personas que
querian hacer poh'tica teni'an que venir y concertar con los representantes politicos que tenfamos ahi."
http://www.youtube.com/watch?v=OtsaMNqoa_k&feature=related
Interestingly the mayor and ex-mayor of San Onofre were both signatories of the Pact of Santa Fe de Ralito.
4
Quoted from "Redaction Nacional" El Tiempo November 11, 2006. In Spanish: "Los camiones enviados por
'Cadena' pasaron por barrios, corregimientos y veredas de San Onofre recogiendo gente. Fue para las elecciones
20
Sheer terror seems to have been used not just to induce people to vote
for particular
candidates
but also to keep them away from the polls so that ballot stuffing and other forms of manipulation
of vote totals could occur. Evidence of the use of coercion to keep people at
home and away from
the polls comes from La Jagua de Ibirico in the department of Cesar.
Another strategy, where coercion
also played
an important
role,
involved collecting people's
cedulas (national identity cards which a person must produce to vote) from their houses, using
them
to collect the ballots (the 'tarjeton')
Further evidence on
how
and
filling
them
in for people.
26
votes were delivered emerged during the testimony of Rafael Garcia
Torres, the former director of information services for the Presidential intelligence service, the
Administrative Security Department (DAS). Garcia, under investigation for links with paramilitaries, told
prosecutors that he had designed a computer program to use confidential information
"that told us the
list
of voters by any category, for example, by polling station, zone, municipality
and even by departments." With
this information in
hand
counterfeit ballots were created so
"that by the end of the elections they would include fake votes of the people
and
did not vote,
there were ballots favoring other candidates different from the ones from the paramilitary
if
group Bloque Norte they would be replaced by ballots
All in
make
who
all
for
our candidates". 2
'
the evidence indicates that paramilitary groups used a wide variety of strategies to
sure that their preferred candidates got elected. This ranged from terrorizing voters to vote
in particular ways, terrorizing
them
to stay
away from the
polls so they could stuff ballots, voting
instead of citizens by confiscating their identify cards, terrorizing politicians so that they would
y segun cuentan algunos habitantes de ese municipio de Sucre, centenaves de campesinos fueron llevados
les vieran la cara a los candidates por los que habi'a que votar en las
parlamentarias: Jairo Merlano para Senado y Muriel Benito Rebollo para Camara.
Cadena puso en una bolsa los nombres de todos los concejales, saco dos y dijo que los mataria a ellos y a otras
personas que iba a escoger al azar si Muriel no ganaba", dice un campesino de esa localidad. La amenaza parece
haber sido efectiva: cada candidato saco 40 mil votos en todo Sucre."
From 'Un Abrebocasa de estas Elecciones" by Cristina Velez in Votebien.com, February 2006.
' This phenomenon is discussed in the article "Aqui nadie es un santo" Semana, September 29, 2007.
According to the testimony given by Mrs. Judith Esther Salas Vallejo: "The [electoral] juries also marked
ballots, because they were told to do so. And this happened during the afternoon since the turnout was very low that
day and they had to deliver certain number of votes that did not been cast by then." This behavior was also witnessed
in Salamina and Remolino. One of the counsels sent by the paramilitaries told the juries that "the turnout is very low
and we have to deliver the votes they are expecting and therefore the juries had to collaborate. If they did not comply
these demands there would be consequences". See http://www.cambio.com.co/port.adacambio/712/ARTICULOdel 2002
hasta
el
corregimiento Plan Parejo para que
WEBNOTAJNTERIOR_CAMBIO-3444866.html.
Other evidence reveals both the extent to which the political project of the paramilitaries was planned and
how votes were guaranteed. Semana documented how paramilitary leaders and politicians divided up coastal
departments. For instance: "In Cesar, "Jorge 40" aimed to obtain two Senators and four Representatives. The
department was divided into 3 electoral districts, one in the north and one in the south. Each of the candidates
would campaign in only one of those districts and each of these would support two candidates for the House of
Representatives. The third district would be open to free competition. To organize the lists they did everything,
from threats and expropriation of land to kidnappings. They killed Luis Laborde, former mayor of Copey and
candidate to the House of Representatives, who did not want to "play" the rules of the game that were imposed
by the AUC. ""Jorge 40" announced this crime in a meeting in San Angel, in which there were 9 mayors of the
department" says a witness. Since that moment it was completely clear that anyone who did not comply with
"Jorge 40's" rules could not do politics" (Semana, November 25th 2006).
also
21
not run against their preferred candidates, and manipulating subsequent vote totals electronically.
The Involvement
3.5
of Other
Armed Actors
in Politics
AUC
more than any non-state armed actor has been
significantly influencing elections in Colombia.
Scattered pieces of evidence suggest that other
Our
empirical results will
show that the
non-state armed actors have been involved, but have not pursued such a consistent electoral
strategy. For instance the
Congress
drug
in
cartel
drug lord Pablo Escobar got himself elected as an alternate
1982 for the Liberal party and
had big impacts on
it is
members
very likely that he and
for
a
list
to
of the Medelh'n
though to our knowledge the data have
elections in Antioquia,
never been systematically analyzed. Nevertheless, electoral strategies can be dangerous. Escobar's
drew a
election
interests
1980s
lot of attention to
him, which was probably not to the advantage of his business
and no doubt strengthened the resolve of the state
when the FARC formed
to
move
against him. Similarly in the
a party to contest elections, the Union Patriotica (UP- Patriotic
Union) many of their supporters and candidates including their Presidential candidates in the
1986 and 1990 elections were murdered.
The FARC and ELN have
to
sway
AUC. We
opposed
However, their involvement
conjecture that this
is
As
also
in elections
mostly because the
to the institutions of the state in
national government.
made a
some
and have used their power
elections
or intimidate voters in favor of candidates they preferred.
killed politicians.
the
also certainly influenced
They have threatened and
has been more limited than that of
FARC
Colombia and
and
ELN
their project
emphasized by our theoretical discussion
have been ideologically
was
to overthrow the
in subsection 2.5, this
symbiotic relationship between them and the executive impossible. In contrast, the
are a continuation of local paramilitary groups allied with local elites and politicians.
light of our discussion in subsection 2.5,
it
was important that the
AUC
AUC
Again
in
did not want to overthrow
the state but to exercise control over certain peripheral areas and influence key legislation. This
created a platform for a symbiotic relationship between the central state and the
would have been impossible
3.6
The Colombian
for the
FARC
Political
Here we emphasize a few institutional
empirical strategy.
AUC, which
and ELN.
System
details of the
Colombian system that are important
Under the 1991 Constitution the President
of
Colombia was elected
four year term with no possibility of re-election. There has been a strong
our
for
one
against re-election
succeed himself was Rafael Nunez in 1886.
historically in
Colombia and the
Though under
the 1886 Constitution re-election was permitted
last President to
norm
for
if
not successive,
it
only happened
once with Alfonso Lopez Pumarejo being President between 1934 and 1938 and again between
1942 and 1945 (when he ended his second term early by resigning). The President
a national vote and
if
no candidate receives 50% of the vote
22
in the first
is
elected by
round a run-off election
is
held between the two candidates with the largest
It is also
important to note that
features strong Presidential
like
dominance
number
of votes in the
most Presidential systems
first
round.
America, Colombia's
in Latin
of the legislature with Presidents often ruling for long
periods by issuing decrees.
For the Senate there
is
a national constituency where 100 Senators are elected from
lists.
For
the Congress there are 32 multi- member districts with each district corresponding to a department.
The
representation of departments depends on their population and there are 162 Congressmen
in total. Historically in
situation
is
Colombia even traditional party
one where only one candidate
is
lists
are very personalized so the typical
elected from each
list.
This situation did not change
with the 2006 elections even though a reform in the electoral law stipulated that to win a seat
in the legislature a
this
list
had
to have at least
2%
At the same time
of the vote nationally.
as
law was introduced, the electoral system was changed to allow for open-list proportional
representation (with preference voting).
Thus even though the number
of lists
fell
dramatically,
personal politics continued unabated via preference voting.
The organization
Courts:
the
of the Higher Courts in
Supreme Court, the Conseil
Judiciary and the Constitutional Court.
d'
Colombia
rather intricate. There are four High
Etat (State Council), the Superior Council of the
While the
Colombian Constitution (although with some
is
first
two have been
in place since
the 1886
alterations), the latter ones were a creation of the
1991 Constitution. For the purposes of paramilitary involvement in politics
it is
relevant to briefly
understand the basics of the Supreme Court and the Constitutional Court. The Supreme Court's
main
role
is
to review
judgments of
inferior (provincial) courts
and decide whether there has been
a correct interpretation of the law and/or the facts of the cases. The Court
sections: criminal, civil
and
of trying and judging "the
(Article 235
#
labor.
The Criminal
members
Section of the
of the Congress: Senate
3 - Constitution). This gives the
is
Supreme Court
and Chamber
Some members
investigations for paramilitary links have started, have stepped
down from
is
also in charge
of Representatives"
Supreme Court the authority
of Congress accused of having links with paramilitary groups.
divided in three
to judge
members
of Congress, once
their public office to
avoid the jurisdiction of the Supreme Court and have subjected themselves to ordinary procedures
under "lower-scale" prosecutors and judges.
The
Constitutional Court has two types of roles: on the one
hand
in charge of reviewing
it is
the judgment of provincial and sectional judges regarding 'acciones de tutela,' which
is
a procedure
that any person can start -generally against public officials- in order to seek the protection of
his/her fundamental rights.
On
the other hand the Constitutional Court
is
in charge of judicial
review of legislation. In this role the Court determines whether a particular statute (law enacted by
Congress)
is
constitutional.
The
Constitution provides that some statutes should be automatically
reviewed by the Court, while in other cases the Court
the constitutionality of a particular statute.
will
only intervene
when a
In the case of the Justice
23
citizen challenges
and Peace Law, the
Constitutional Court reviewed
its
constitutionality thanks to a challenge posed
by a group of
citizens (under Article 241 of the 1991 Constitution).
There
is
a relatively high degree of autonomy of these High Courts vis-a-vis the Executive
Branch, particularly the Supreme Court. There
is
no
life
tenure for the Justices in Colombia:
instead they serve during one eight-year term and re-election to the same court
forbidden by the Constitution.
choosing a replacement from
When
lists
a Justice serves his time, the
Supreme Court
is
is
explicitly
in
charge of
prepared by the Administrative Section of the Superior Council
of the Judiciary (another High Court also in charge of administrative functions regarding the
judiciary). In turn, the justices of the Administrative Section are appointed
Court
Supreme Court
judge), by the
(1
and executive branches
impact of the
legislative
They can
some extent
to
judges elected for
life
(2 judges)
is
influence this process.
by the Constitutional
and by the Conseil d'Etat
therefore very indirect
(3 judges).
The
on the Supreme Court.
For instance the Constitutional Court has 9
by the Senate. However, the election takes place over
of candidates
lists
presented by the President (3 judges), the Supreme Court (3 judges) and the Conseil d'Etat
judges).
three of
(3
Thus, from the nine justices sitting on the Constitutional Court in a particular term,
them have come out from
lists
presented by the President.
The Data
4
4.1
Data Sources and Construction
The most important data
for the
paper are on the presence of non-state armed actors, specifically
paramilitaries and left-wing guerillas.
Our main measure
which we
of the presence of non-state
is
one based on
is
from Centro de Estudios sobre Desarrollo Economico (CEDE)
conflict incidents,
the Universidad de Los Andes in Bogota.
refer to as 'attacks' for short.
CEDE
collects
armed
The database
actors
of attacks
Facultad de Economi'a at
in the
data from the Observatory of
Human
Rights of the Vice-presidency and the National Department of Planning and aggregates variables
in several categories
by armed actor and type of action. The original data are a compilation of news
from newspapers and from reports of the national
by aggregating over many of these
variables.
police.
Our measure
of attacks
is
constructed
For each armed actor we simply add the following
variables: explosive terrorist acts, incendiary terrorist acts, other terrorist acts, assaults to private
property, attacks on civil organizations, political assassination attempts, road blockades,
armed
contact between state and non-state armed forces initiated by the latter, ambushes of civilians,
harassing (mainly threats to civilians), incursion into
armed
forces
wounded by the non-state armed group, murders
massacres, deaths of
forces,
'villages',
members
of the state
armed
forces,
kidnappings of politicians and kidnappings of
in the period 1997 to 2005.
24
overland piracy,
illegal
checkpoints,
of civilians, murders of politicians,
kidnappings of members of the armed
civilians.
We have
this variable for
each year
We
state
check the robustness of our results with an alternative measure of the presence of non-
armed
The database from which we constructed our
actors based on displaced people.
comes from Action Social an agency created by the presidency. This data
displaced measure
collected from people that report themselves
and are
classified as displaced in the Registro
de Poblacion Desplazada (unique register of displaced population)
in order to
is
Unico
obtain a set of
These data specify the municipality where the displacement originated, the year of the
subsidies.
displacement and the armed actor that originated
it.
We
have these data annually
for
the period
1997 and 2006.
We
use these data to construct various measures of the presence of non-state armed actors.
Because the time
quite noisy,
we
specifications.
is
series variation in
focus on 'averages' of these data, though
Our
first
main measure
total paramilitary attacks
the population measure
measure
is
a
dummy
is
identical
We
also exploit over-time variation in
in municipality
m per
We
some
1000 inhabitants where
the average population between the 1993 and 2005 censuses.
that takes the value of
measures
we
of paramilitary presence, referred to as paramilitary attacks,
between 1997 and 2005
above the 75th percentile.
and
both the attacks and the displaced measures appears to be
municipality
1 if
two similar measures with the displaced variable
also construct
for guerrilla attacks
m
Our second
has a value of paramilitary attacks
(FARC and ELN
combined).
take elections results from the Senate, Congress and Presidential elections 1991-2006 from
the Registraduria Nacional del Estado Civil.
politicians belonged,
parties into 'third',
Using the names of the political parties to which
we constructed party vote
'left'
and
shares in each municipality.
first
round election
outcomes by looking
29
For Presidential elections we took the
we
also investigate the effect of paramilitaries
at electoral concentration.
evidence from Magdalena, which
is
and calcu-
results.
In addition to the third party vote share,
electoral
then classified
'traditional' (Liberals or Conservatives) political parties,
lated the vote share of third parties in each municipality.
numbers from the
We
This
is
on
motivated by the case study
discussed in the Appendix, which shows
how
paramilitary
influence creates a highly concentrated vote share pattern in a few municipalities (where they
have presumably used coercion or manipulated the vote).
the vote share of the most popular
tables
it
is
from the Gacetas
del
is
in a municipality (for the Senate or Congress,
referred to as 'winning votes share').
roll-call votes
28
list
Electoral concentration
Senado.
30
Finally,
defined as
and
we obtained data from two
in the
crucial
31
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.
Given our focus on the impact of the AUC, when we examine votes for third parties, we do not consider parties
Democratico ('Democratic pole'), which are unconnected with the paramilitaries,
as 'third parties'. See Valencia (2007) for a similar distinction and calculation.
The results using a Herfindahl index to measure electoral concentration are very similar and omitted to save
in the left-wing coalition, the Polo
space.
1
Roll-calls are not taken for
most votes
in either
the Senate or Congress.
s,
As
controls,
we
collected data
measure of the extent of 'right-wing' support
election to construct a
was the son of the
on the vote share of Alvaro Gomez
we use the vote share
Patriotica in the
FARC, Pardo
same
we
is
earlier
Union Patriotica was the
1950s and
Conservative party.
for the
Jaime Pardo Leal the Presidential candidate
election. Since the
Leal's vote share
of our regressions,
of
Gomez
in a municipality.
wing conservative President Laureano Gomez from the
right
ran on a very conservative platform as the Presidential candidate
Similarly,
1986 Presidential
in the
for
unofficial political
the Union
wing of the
a good measure of 'left-wing' support in a municipality. In
also include interactions
between a
full set of
many
time dummies and various
municipality-level controls. These controls are the land Gini in 1985, the area of the municipality,
between 1993 and 2005, an
altitude, distance to state capital, average municipality population
index of
how
rural the population of the municipality
is
(in 1993),
an index of 'unfulfilled basic
needs' in 1993, proxying for the level of poverty in the municipality, and
cultivation in 1994
and opium cultivation
in 1994 (these controls are
dummies
from the
CEDE
for coca
database).
Descriptive Statistics
4.2
Table 2 shows the descriptive
statistics.
Columns
tion of the variables for the whole sample.
paramilitary areas where
we use
the attacks
1
and 2 report the mean and standard devia-
Columns
dummy
3-6 report the
same two
variables for high
defined in the last section to decide whether
or not a municipality has high paramilitary presence.
The
first
two
sets of
rows show the rapid increase
and 2000-2001 with some evidence that
show a
in
paramilitary presence between 1996-1997
this fell in 2004-2005. Interestingly, the
similar increase in guerilla presence (with
no tendency to
despite President Uribe's intensification of the war against the
and
5
it is
evident that there
presence, which
is
is
the most recent period,
FARC). Looking
we discussed above, paramilitary
large increase in the share of third parties after 2002,
high paramilitary areas.
18
is
columns 3
units were often
guerilla.
There are several noteworthy features of the data highlighted by Table
this
at
a positive correlation between paramilitary presence and guerilla
not surprising since, as
formed to combat the
fall in
next set of rows
Our measure
and
this increase
is
2.
Rows
9-12
show a
more pronounced
of electoral concentration also increases after 2002,
in
and
again largely concentrated in high paramilitary areas (rows 13-16). Finally, rows 17 and
show a noticeable
increase in the vote share of the winning presidential candidate in the high
paramilitary areas. These patterns give a preview of our regression evidence that will be presented
in the next section, as part of the evidence of paramilitary involvement in politics.
Rows
19-24 show that there are also
some notable
differences
between high and low paramili-
tary areas in terms of the covariates. Most importantly, low paramilitary areas appear to be more
'right-wing'. This
is
reassuring in connection with the concerns that our measure of paramilitary
presence will capture latent right-wing leanings. There are also some differences in terms of other
26
though these appear
covariates,
The Impact
5
We now
relatively small.
of Non-State Actors on Elections
armed actors on
investigate econometrically the impact of non-state
Our
starting with their impact on the vote share of third parties.
electoral outcomes,
be from
basic regressions will
a simple panel data model of the following form:
ym
where ym j
=
,t
d
t
+ 5m + a
X m|
between various geographic and
all
Gm
is
m
at time
|t
t,
e m% u
(7)
the dt's denote time effects, the
a vector of covariates, which contains the interactions
time dummies that are included in some specifications; e m ,t
as the corresponding
+
political controls at the municipality level (described above)
Pm
omitted factors. Most importantly,
and
t
7r
•
t
the outcome variable in municipality
is
Sm's are municipality fixed effects,
full set of
Pm + P G m + X^
t
measure of
is
is
an error term representing
our time-invariant measure of paramilitary presence
guerilla presence.
The term a t
Pm
therefore estimates
a potentially differential growth effect for every time period (relative to the base,
This specification
will
enable us to focus on whether there
example, the third party vote share) after the
(for
dependent variable
AUC
in elections,
will
be the vote share of third
and we have these data
and a
AUC
a change in an
is
become involved
initial date).
outcome variable
in politics.
Our
first
parties, proxying for the direct effect of the
for the 1991,
1994, 1998, 2002
and 2006
elections.
This enables us to include interactions with the 1994, 1998, 2002 and 2006 dummies and our
measures of paramilitary presence as a check against pre-existing trends (1991 election
omitted category). Our working hypothesis that the
AUC
to vote for certain lists (or used ballot stuffing) implies
We
also
Um.t
—
dt
+ 6m + at Pm ,t-i +
-
on the interaction between year
is
that, as
influenced elections and forced citizens
a >
after 2002.
t
C
•
Pm,t-\
effects
+
fit
Gm ,t-i +
G m ,t-i + X^
T]
effect of paramilitary
error.
above the 75th percentile.
prior to the election.
G m j-i
is
of year-to-year variations,
We
then
set
and
is
this
often
we estimate
(8)
dummy
Pm j-i
=
municipality
m is
1
if
Pmit—i
variables
constructed similarly. For the 1998 election we just use the
2002 we used data from 2000 and 2001, and so on. Equation
Pmi f_j
(8)
,t,
and the time-varying measures. The disadvantage of
To minimize the impact
and Gm.t-\ using the two years
direct effects of
+ em
and guerrilla presence, and focus
only using the election years for 1998, 2002 and 2006 and construct two
for
tt
•
noted above, year-to-year variation in paramilitary and guerrilla presence
due to measurement
1997 data,
the
experimented with empirical models of the following form
which include both a time-varying main
model
is
G m ,t-\-
27
(8) also includes
the
Paramilitary Effect on Elections
5.1
We
— Third Parties
investigate the impact of paramilitary presence on the vote share of third parties in the
first
Senate elections. More
specifically,
we estimate equations
(7)
and
(8),
corresponding to the vote share of third parties in municipality
y-mf,
at time
Our
t.
with the dependent variable
m in the elections for Senate
basic results, using the attacks measure, are reported in Table 3. In this
subsequent tables,
the municipality
all
standard errors are
level),
and
all
fully robust (allowing for arbitrary serial correlation at
and Tables 3-7 include a
of municipality
full set
and time dummies
in all
specifications.
Table 3 shows a robust positive and significant
effect in
both 2002 and 2006 of paramilitary
presence on the vote share of third parties. For example, column
a standard error of 3.14 and a similar estimate
mates are highly
statistically significant
for 2006, 6:2006
1
=
=
estimates 62002
Both
22.10 (s.e.=3.19).
and quantitatively large (the magnitudes
will
20.97 with
esti-
be discussed
below).
Column
2
adds our basic covariates (the land Gini
between 1993 and 2005, the index of rurality,
altitude, distance to state capital, average population
the index of unfulfilled basic needs,
right
and
left
dummies
leanings of the municipalities),
to allow for differential effects over time.
for coca
all
To save
in 1985, the area of the municipality,
and opium
cultivation,
interacted with a
space,
full set
highly significant (15.88 for 2002 and 10.79 for 2006). In column
still
dummies
of time
we do not report the
time interactions. The results in column 2 are similar to those in column
and our measures
coefficients
of
so as
on these
1,
slightly smaller,
3,
we include
though
interactions
with guerrilla presence as well as our main interactions of paramilitary presence and time. These
interactions are insignificant, consistent with the hypothesis that the left-wing guerrillas have
played a
much more
limited role in national politics.
The next three columns
using the attacks
dummy
very similar to the
(s.e.=1.98)
first
and 62006
=
re-estimate the
as the
same models
of the
measure of paramilitary and
three columns.
14.54
32
(s.e.
which are again
three columns but
guerilla presence.
For example, in column
= 1.99),
first
4,
The
we estimate
now
results are
62002
=
statistically highly significant.
13.71
The
estimates in columns 5 and 6 are very similar. These estimates also clearly show the quantitative
effects of paramilitary involvement.
They imply
that high paramilitary areas have, on average,
10 percentage points higher vote share for third parties after the
This
is
AUC's involvement
in politics.
a very sizable effect, particularly in view of the fact that the average vote share of third
parties before 2001
was about 15 percent (Table
2).
Prior to 2002, as the coefficients 61994 and 6199s illustrate there
ship between paramilitary presence and third party vote share.
are positive
and
significant in
column
1,
is
no robust positive relation-
Though both
their significance vanishes
of these coefficients
when we add the
covariates for
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 of parties, interactions with guerrilla presence are still insignificant.
28
columns
2
and
3.
the attacks data
added
are
in
In
column
4 6igg4
is
again positive and significant, though 61998
in levels these coefficients
columns
One concern with
5
and
become completely
is
not.
As with
insignificant once the covariates
6.
the results in columns 1-6
is
that the change in the coefficient of the time
interactions might reflect the changing importance of paramilitaries or guerrillas in certain areas.
Our data
and
are not ideal to investigate these issues, since the year-to-year variation in the military
guerrilla presence are
we estimate equation
specification
is
measured with considerable
some answers
(8) to provide
presented in column
The
7.
error.
Nevertheless, in columns 7
=
17.81 (s.e.=2.87)
and
6:2006
=
results here are consistent with those in the first 6
18.02 (s.e.=3.01),
Table 4 shows the results for the votes
about a 8 percentage points higher vote shares
3,
dummy
with the
statistically significant.
full set
statistically significant.
and
in fact
33
even
with high paramilitary areas showing
for third parties after the
of covariates
coefficients are
The general patterns and
Nevertheless, the results are weaker than in Table
the attacks
all
and are highly
for the Congress.
the point estimates are very similar to those in Table
politics.
8,
The most parsimonious
to these questions.
columns using our time-invariant measure of paramilitary presence. The estimated
&2002
and
3,
and
in
AUC's involvement
column
6
in
where we use
guerilla presence neither 62002
not 62006
Importantly, 01994 and 61998 are small and statistically insignificant in
specifications in Table
4.
Table 5 investigates the robustness of these results further (focusing on the votes for the
Senate; the results for the Congress are similar).
time interactions with the
full set
In this table,
all
specifications include the
of covariates introduced above, including the interactions with
guerrilla presence (which are reported in the table, while other covariates are again not reported
to save space).
The
as the presence
measure
first
column again estimates
for
(7),
both paramilitary and
but uses the levels of displaced people
guerilla presence.
supportive of our hypothesis, the estimates for the interactions between
and the displaced measure of paramilitary presence are
a large increase in the size of the coefficient in 2002, though
we use the dummy
similar to those in
similar.
all
statistically significant
estimates are different because the scale of the displaced variable
is
Now we
it
is
see that, less
the time
dummies
(though the point
different).
However, there
halves in 2006.
In
column
2
version of the displaced measure of paramilitary presence with results very
column
1
except that
Because we are now using a
now
dummy
the quantitative magnitudes of 62002 and 62006 are
variable for measuring paramilitary presence, the
quantitative effects can also be directly compared to those in Table
3.
The
estimates suggest that
high paramilitary areas experienced about 12 percentage points increase in the vote share of third
parties in 2002
33
and 2006, which
is
broadly consistent with the quantitative effects in Table
3.
is negative. We conjecture that this is because paramiliappear to have had a very strong effect on elections in departments on the Caribbean coast, such as Magdalena,
Sucre, Corboda 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.
Notice that the direct effect of paramilitary presence
taries
29
we again use the displaced measure, but now
In column 3
the data estimating model
column
In
positive.
4,
exploit the time series variation in
Here the estimates are imprecise and
(8).
we use a
though
insignificant,
still
and extract the principal component of the
different strategy
attacks and displaced measures, and use this principal component as our measure of paramilitary
presence.
The advantage
thus their
common component may
very
much
in line
of this strategy
is
contain
that both attacks and displaced numbers are noisy,
more information. The
with those of Tables 3 and
4.
We
find
dummy
that takes the value
this
approach are
no impact of paramilitary presence prior
In column
to 2002 but a positive and highly significant one afterwards.
strategy and construct a
from
results
1 if in
5,
we use
yet another
our data a municipality has high
paramilitary presence according to both our attacks data and our displaced data (according to
measure
this
areas for which
all
we have data
data are missing receive a value of zero).
measures.
The
though di998
now
is
results using this
now
as in
combined
column
5.
dummy
numbers but
construct a similar
dummy
dummy
for
which the attacks
for the guerilla
presence
are again very similar to our basic finding
statistically significant. Finally, in
using a time-varying
numbers
We
for displaced
column
6,
we estimate the model
(8),
but
constructed combining information from attacks and displaced
In contrast to the imprecise results in
column
3,
the estimates
now
are
consistent with the rest of our results and are highly significant.
A
major concern
specific areas
for
our empirical strategy
is
that paramilitary groups
may be
selecting into
based on their political or ideological characteristics; paramilitary presence
may
be
capturing a trend towards third parties unrelated to the paramilitaries' involvement in politics
(this has to
be a trend, since
all
specifications include a full set of municipality fixed effects).
For example, there might be a trend in 'conservative ideology' in certain areas, and the increased
vote share of third parties
While we cannot
do not find
it
is
a reflection of this trend in the areas where they have situated.
rule out this alternative explanation for the patterns
shown
in Tables 3-5,
we
very plausible. First, the specifications with covariates flexibly control for potential
differential trends in 'right-wing ideology'
by interacting the vote share of Alvaro Gomez in the
1986 Presidential election with a
of time
full set
dummies
(as well as for potential differential
trends in 'left-wing ideology'). Remarkably, this has almost no effect on the estimates. Second,
the case study evidence discussed in Section 3 supports our interpretation.
OveraU, Tables 3-5 provide robust correlations consistent with our basic hypothesis that
lowing the
AUC's
decision to
fluenced electoral outcomes.
become involved
in politics, paramilitaries
The next two subsections provide
have systematically
fol-
in-
further evidence consistent with
this pattern.
5.2
Results on Electoral Concentration: Senate and Congress
Motivated by the case study evidence from Magdalena, discussed
in the
Appendix, we hypothesize
that paramilitary involvement also translated into greater electoral concentration (greater vote
30
share for the winning
by estimating regressions similar to
of the most popular
(8),
measure to save space (the
Table 6 investigates this issue
elections).
with the dependent variable being the vote share
m
Senate or the Congress in municipality
we again have data from 1991
these regressions,
upon
and
(7)
in either the
list
and Congress
in the Senate
list
to 2006.
results with the displaced
at
time
t.
For
Throughout, we focus on the attacks
measured are very similar and are available
request).
The
first
six
columns
in
Table 6 use data from the Senate, while the
Congress. Each of these sets of columns
with a
broken down into three sets of two which use
dummy
attacks data in levels, then the attacks
finally the
is
for
first
the
high paramilitary and guerilla presence, and
time varying attacks dummy. All columns include the
full set
last six are for the
of covariates (interacted
full set
of time dummies).
This table shows no differential trend in electoral concentration between high and low paramil-
There
itary areas before 2002.
guerrilla areas.
which
is
There
visible in
is
(8), cfr.
also
no
differential
trend at any point between high and low
a large spike in electoral concentration in 2002 in high paramilitary areas,
columns
only imprecisely estimated
equation
is
columns
and
1-4
This
7-10.
effect
in specifications that
5, 6,
11
and
12).
The
disappears in 2006, and
is
not visible or
use the time-varying measures (the regression
quantitative effects in 2002 are large, correspond-
ing to about a 4-5 percentage points increase in the vote share of the winning candidate
This
is
again a very large
elections
An
effect, as
is
the vote share of the winning
list
(list).
and Congress
in Senate
was around 30 percent before 2001.
obvious question
is
paramilitary areas in 2006.
why
there
no
is
The answer
differential effect
to this question
is
on electoral concentration in high
suggested by our discussion of the
evidence from Magdalena in the Appendix and by the arguments advanced by Valencia (2007)
and Lopez (2007):
after the experience of 2002,
places such as Magdalena, paramilitaries
which was subsequently highly scrutinized
became much more
skilled at
manipulating the outcomes
in order to guarantee the election of their candidates without drawing so
As we show
themselves.
in the
Appendix, even
if
electoral concentration
much
fell
in
attention to
Magdalena
2006, Senators supported by the paramilitaries were re-elected and with a rather similar
of votes, but
of Table
either
1,
now spread more
in
in
number
evenly throughout the department. Indeed, recalling the evidence
the presence in the legislature after 2006 of so
under investigation or arrested
for connections
many
Senators and Congressmen
with paramilitaries
continuing impact of the paramilitaries on elections. This interpretation
is
is
direct evidence of a
also
supported by the
evidence presented in the previous subsection (Tables 3-5), which showed the robust association
between paramilitary presence and the
2006.
Had
have seen
rise of third parties in
the elections of both 2002 and
the effect of the paramilitaries on the electoral outcomes weakened in 2006,
this in a decline in the vote share of third parties.
The
results
from the votes
executive, presented in the next subsection, further corroborate this interpretation.
31
we would
for the
Results for the Executive
5.3
Table 7 presents estimates from regression models similar to
,t
time
Once again we focus on the attacks measure
again very similar).
The
basic robust finding
is
in 2006. For example, the estimate in
column
about 11 percentage points more votes
m at
measure are
that in both 2002 and in 2006 the vote share of the
in high paramilitary areas
vote share of the winning candidate in 1998 (Pastrana); this effect
is
the period 1998-2006,
(the results with the displaced
winning candidate (Alvaro Uribe) was systematically higher
effect
(8) for
defined as the share of votes of the winning Presidential candidate in municipality
with y m
t.
and
(7)
4,
11.45
(s.e.
=
is
in fact considerably stronger
1.67) suggests that
in high paramilitary areas
than the
than Pastrana
Uribe obtained
did.
The same
present in models that exploit the time- varying measures of paramilitary presence (models
as in (8)).
The
pattern with a stronger effect in 2006
is
President Uribe was favored by
plausible.
paramilitary groups already in 2002, but after his support for policies in line with these groups'
interests during his first term, the support of the paramilitary groups for his election
much
stronger. This pattern
is
became
thus consistent both with the notion that paramilitaries continued
to heavily influence elections after 2002,
and
documented further below,
also with the hypothesis,
that a symbiotic relationship between the executive and the paramilitaries developed after certain
key legislations proposed by Uribe. 34
The evidence
in this subsection
also particularly important since
is
it
suggests that, consistent
with the assumptions of our theoretical model, the executive, President Uribe, electorally benefited
from the presence of paramilitaries. 35
Predicting Arrests
5.4
As noted
in the Introduction, since the so called 'Parapolftica'
many Congressmen and
with
illegal
scandal
first
broke
in
Colombia,
Senators have been investigated, arrested for and even found guilty of links
paramilitary organizations.
A
useful 'reality check'
on whether the evidence reported
so far indeed represents the influence of paramilitaries on election outcomes
is
to see
whether
Senators elected in areas under paramilitary control have explicit links with the paramilitaries
Nevertheless, as always, there
may be
other interpretations of this finding.
For instance, to the extent that
who had previously suffered
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.
"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, Raul 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.
President Uribe had been successful in de-mobilizing the paramilitaries, then people
32
and have voted
for legislation favoring paramilitary interests. In this subsection,
we
investigate the
presence of explicit links, exploiting the fact that Colombian judiciary, particularly the Supreme
Court,
The
broadly independent and has prosecuted politicians with links with the paramilitaries.
is
voting behavior of these Senators
We
discussed in the next subsection.
is
adopt a simple empirical strategy to measure the extent to which Senators relied on the
support of paramilitaries for their election.
that Senate
list
First,
define u>ip to be the proportion of total vote
receives in municipalities with high paramilitary presence,
I
paramilitary presence by using our
dummy
of paramilitary presence. Similarly
we
To
as paramilitary presence.
look at Senators
who
where we measure high
variable constructed from our time-invariant
define uiig as the proportion of total vote that
in municipalities with high guerilla presence,
way
we
where high
guerilla presence
is
investigate the links between Senators
measured
list
I
in
receives
the same
and paramilitaries, we
We
are arrested for connections with paramilitary groups.
the proportion of the Senators of
list
measure
define A; as
that have been arrested for alleged connections with
I
paramilitary groups and then estimate
&l
=
P
p + A
U>l t
W; )G
+ Xj
•
7T
+
£
t
(9)
.
Based on our hypothesis that Senators and Congressmen receiving a high fraction of
in paramilitary areas have explicit connections with paramilitaries,
the covariate vector XJ we also use party identity.
'bad control' (Angrist and Pischke, 2009, pp.
Though,
in general, this
interest,
u>i
t
we
include
the explanatory power of
it
u>i
to
p
and party
is
>
In
0.
an example of
of a third party
and the outcome
is
variable,
all of
(thus acting as the channel through which paramilitary presence
is
arrests).
Table 8 shows the results from estimating
arrests
p,
to find p
examine whether being a member of a third party absorbs
impacting politician behavior and
between
member
64-68), since being a
an outcome variable inbetween the causal variable of
A;, in this case
we expect
their votes
identity.
(9).
In column
As we saw above, there
paramilitary presence and the rise of third parties.
Column
is
1
1,
we look
at the relationship
a close correspondence between
confirms
this,
showing that third
party Senators are significantly more likely to be arrested for links with the paramilitaries than
Liberals, Conservatives
and
Socialists (liberals are the
In the next three columns, columns 2-4,
omitted category).
we present estimates
of (9) for the Senate.
the most parsimonious specification where we regress the proportion of Senators on a
arrested on ui^p and u\ q.
We
see that p
suggesting that the higher the share of
greater
is
column
3
leaning'
=
its
the proportion of Senators on the
we add
and
1.38 (standard error=0.42)
votes that a
list
who
list
is
'left-leaning' areas.
We define the 'right-leaning'
and
list
2
is
who were
statistically significant
obtained in paramilitary areas, the
are arrested for links with paramilitaries. In
as a covariate the proportion of its vote share that the
36
Column
list
obtained in
'right-
Neither of these controls are significant in themselves and they
'left-leaning'
dummies again by looking
33
at the vote shares of
Alvaro
Gomez and
do not influence the coefficients of
In column 4
paramilitaries
The
is
third party
interest.
we then add the party dummies
is still
significant,
whether the entire
The estimates show
working through third parties.
dummy
to investigate
but so
that this
is
effect of the
not the case.
the share of votes from paramilitary areas.
is
This suggests that the only channel of influence of paramilitaries on politicians behavior
through third party
affiliation (which, in hindsight, is
is
not
not surprising, since in several municipali-
paramilitaries supported liberal or conservative candidates).
ties,
The next
four columns report estimates of (9) for the Congress. These are broadly consistent
with the results
areas
for the Senate; the
proportion of the votes that a
positively correlated with the proportion of
is
arrested,
though the
results are typically
somewhat
list
Congressmen on that
less precisely
is
in high paramilitary
list
estimated and
Overall, these results are important firstly because they provide
strategy for measuring the effects of the paramilitaries
won
some
who have been
less significant.
verification that our
indeed capturing what they are supposed
Secondly, they also show that our findings so far are unlikely to be driven by some omitted
to.
characteristic of municipalities (making paramilitary presence 'endogenous' to voters' preferences).
The
fact that paramilitary presence
is
positively correlated with arrests suggests that there
is
indeed a close relationship between paramilitaries and politicians such that paramilitary groups are
either forming a coalition with local politicians, or are themselves running as candidates. In either
case,
it
seems highly unlikely that politicians, particularly politicians elected
in high paramilitary
areas, are perfect agents of voters, supporting policies in line with their interests.
seems that these politicians are pursuing
5.5
As
Voting
in the
policies systematically preferred
Senate
lists
We
do this in the very specific but revealing context of a
vote to re-introduce two clauses of the Justice and Peace Law.
10%
at the time of the passing of the law
and Article
paramilitaries should be considered as 'sedition'.
71,
37
would imply that the paramilitaries had committed
Jaime Pardo Leal
These were Article
reduction in the sentences of demobilized paramilitaries
eligible to extradition.
These two
articles
in 1986; then, similar to the
we constructed dummy
we examine
that received large shares of their votes in paramilitary areas vote in
systematically different ways.
stipulated a
it
by paramilitaries.
a final demonstration of the potential influence of the paramilitaries on elections,
whether Senators from
Instead,
The main
which
who had been charged
significance of sedition
political crimes
were part of a
and
70,
which specified that the crimes of former
construction of the
variables for 'right-leaning'
roll-call
first
dummies
'left-leaning' areas
is
that
it
and would therefore not be
draft of the law presented
by the
for paramilitary and guerrilla presence,
depending on whether a municipality is
above the 75th percentile of votes for the corresponding candidate in the 1986 presidential elections.
At the time of the votes these clauses were actually Article 61 (sentence reduction) and Article 64 (sedition)
but this changed in the final law.
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
34
President but they were both rejected in the
The rejection of these two
articles
first
commissions of the Senate and the Congress.
was then appealed
a commission was formed to inform the Senate on
The members
of the
In response to this appeal
in the Senate.
how
to proceed with the appeal pleading.
commission were Mario Uribe (under arrest
2008 for connections with
in
paramilitaries and a cousin of President Uribe), Mauricio Pimiento (arrested
see Table 1),
Juan Gomez, Miguel de
Carrizosa, and
had
Hernando Escobar. The members
went
in favor of re-introducing the
To the extent that there
in paramilitary areas
two
articles, this
is
two
and found
it
was and we have the
articles into the Justice
and the proportion of Senators on the
list
who voted
I
it is
also evidence of the 'quid
Column
More
on these
specifically
roll-call votes,
we now
4.
40
we estimate versions
of
define A; to be the proportion of
that voted in favor of re-introducing these two clauses.
columns 9-12 of Table
in
of the paramilitaries
the previous sub-section.
list
received
list
would be further evidence that paramilitaries have indeed influenced election
To examine the impact
Senators on
and Peace law. 39
to re-introduce these
pro quo' between paramilitaries and the executive, consistent with Proposition
(9) in
roll-call for this
a positive correlation between the proportion of votes a
outcomes, consistent with Proposition 2 of our model. Implicitly,
model
guilty,
Table Al), Jesus
guilty, see
commission concluded that the appeal
of the
by the plenary of the Senate which
to be approved
vote. This vote
la Espriella (arrested
and found
The
results are reported
8.
9 shows that senators from third parties
and the conservatives are most
likely to
support the re-introduction of the two controversial articles of the Justice and Peace Law. The
were as
fact that conservatives
likely as third
party candidates to support these clauses
related to the fact that in the Senate they were also allied with President Uribe.
columns show a positive
significant correlation
got a large proportion of
the Peace and Justice
12
votes and the proportion of Senators
its
Law more
when we introduce the party
working through third party
Note
also that in
none
'pro-paramilitary'. (This effect
is
The remaining
in areas
who voted
probably
where a
in favor of
list
making
no longer significant in column
identity variables, suggesting that a large part of the effect
affiliation).
Third party identity
of the specifications
orientation on these votes.
between paramilitary presence
is
Overall,
we
is
is
is
again positive and significant.
there any effect of guerilla presence or right or
left
interpret this as evidence in support of our hypothesis
that politicians receiving support from paramilitaries have in turn supported legislation in line
with the interests of these groups.
extradition unconstitutional.
in hiding,
The day
after this
gave himself up to the authorities.
was written into the Constitution, Pablo Escobar, who had been
However, the Constitution was subsequently amended to allow
extradition.
However the Supreme Court
said 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
we discussed earlier.
As we discussed in Section 3, a large literature has heavily criticized the Justice and Peace Law as being very
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.
paramilitary leaders as
4
35
The Persistence
6
of the Paramilitaries
Having provided some evidence consistent with the assumption that paramilitaries have had
we now turn
systematic effects on elections in Colombia,
model.
to two predictions of our theoretical
model suggests that to the extent that paramilitaries
First, the
President, the President will have a greater incentive to allow
where they
to
are.
them
deliver votes to the
to stay in control of the areas
Second, this effect will be stronger in places which President Uribe did not expect
do well without intervention by the paramilitaries.
We
We
focus on the 2002 Presidential election and the subsequent persistence of paramilitaries.
municipality
is
A
only municipalities that had paramilitary presence in 2000-2001.
restrict attention to
classified as
having paramilitary presence
if it
experienced any paramilitary related
incidents (attacks or displacement depending on the measure used) during either 2000 or 2001.
We
use a time varying measure of paramilitary presence to capture presence before the election,
denoted
this
is
Pm] t<2002,
measured
and presence
sum
as the
after the election,
denoted
Pmi t>2002-
In our baseline specification
of either attacks or displacements during a two-year
window, the
periods 2000 and 2001 (t<2002) or between 2004 and 2005 (t>2002) in either case divided by
the population of the municipality.
we show.
variables as
We
Our
ways of measuring these
results are robust to different
can use this variable to explicitly examine how the persistence of
paramilitary presence depends on the extent of voting for President Uribe, thus testing the
prediction.
To
test the
second implication of the model we argue that even though
was Governor of Antioquia he was nominally a representative
a candidate against
both
cases).
him
when Uribe
of the Liberal Party, in fact his key
supporters, in the absence of coercion, were conservative voters
and order. Direct evidence comes from the
who
liked his
fact that the Conservative
emphasis on law
Party chose not to run
either in 2002 or 2006, while the Liberal Party did (Horacio
As a consequence, Uribe could
first
Serpa
in
anticipate doing well in places where Conservative
President Andres Pastrana had received a high vote share in 1998.
We
can therefore test the
second hypothesis by interacting Pastrana's vote share in 1998 with Uribe's vote share.
More
formally,
•Pm,«>2002
where v^
we estimate the
= a Pm .t<2002 + P
2002 is the
•
following model:
^,2002
+7
'
"£,2002
'
u m,1998
+
vote share of President Uribe in municipality
share of Pastrana in 1998.
Our model
predicts that
/3
>
lead to greater paramilitary presence after 2002), and 7
S
'
u m,1998
+ Xm
'
«"
m in 2002 and v^ 199g
(a greater share of votes for
<
+ ^m
is
(10)
the vote
Uribe would
—so that the higher was Pastrana's
vote share in 1998, the less Uribe would benefit from the support of the paramilitaries and thus
according to our theoretical model, the more likely are the paramilitaries to be eliminated.
An
alternative measure of the paramilitary-induced vote advantage of Uribe, which
closely related to the predictions of Proposition 3,
is
36
to define the variable
max{0, v^
2002 ~ v
is
more
m
i99s}>
which directly captures the vote advantage of Uribe
if
any
(in
— the greater
be positive
more we would expect
The
results
state
first
v^ 200 2 — v^
is
reasoning,
]998
,
we expect the
we examine
robustness,
The
all
(10) are
component
of attacks
In
3.
this variable to
and
the
column
1
in Table 9.
Throughout
effects are evaluated at the
this
sample
columns of Table 9 use the attacks measure non-
first six
The
displaced measure.
final
two use the
and displaced.
results are broadly consistent
Proposition
shown
main
of the
armed actor presence, while the next four use the
The
on
coefficient
the more Uribe gained relative to Pastrana
from estimating various versions of
to facilitate interpretation.
principal
2002 relative to Pastrana's vote in 1998.
paramilitaries to persist.
table and the next where
means
With the same
municipality m).
in
with our model, in particular, with the predictions of
of Table 9
we estimate the simplest version
Consistent
of (10).
with these predictions, the estimate of Uribe's vote share of value to the sample mean,
estimated at 0.14 (with a standard error=0.08). This estimate, which
/?,
is
marginally significant,
is
suggests that other things equal, the greater the vote share for President Uribe in the 2002
election in municipality
Quantitatively this
is
m, the greater the paramilitary presence
in the municipality after 2002.
10%
a sizable, though not implausible, effect implying that a
increase in
Uribe's vote share in an average municipality under paramilitary control will increase paramilitary
presence by 0.014. This
whole sample, which
municipalities which
a large effect
is
0.04,
is
had
compared
though smaller
for the
to the
mean
of paramilitary presence in the
mean when we
restrict attention to
positive paramilitary presence, this being 0.15.
Moreover, again consistent with the predictions of Proposition
teraction between Uribe's and Pastrana's vote shares
5%, 7
=
is
all else
equal, paramilitaries were
where President Uribe received a high share of
of the votes of President Pastrana in the 1998 election, he
lower vote share.
the coefficient on the in-
negative and statistically significant at
—0.63 (standard error=0.33). This implies that,
likely to persist in areas
3,
Column
2 adds covariates to the basic
only those
41
votes,
more
and on the basis
would have been expected to receive a
model of Column
1
while
Column
3 adds
controls for guerilla presence in 2000-2001. In these columns, the interaction term remains negative
and has a
similar quantitative
magnitudes to that in Column
significant (in several other specifications in this
and the next
1,
but
is
no longer
table, the interaction
statistically
term
is
again
significant).
Column
4 reports a regression
model more
directly inspired
by Proposition
3,
which makes the
persistence of paramilitaries a function of the vote gain of President Uribe relative to President
Pastrana, max{0,
in
column 3
In
(coefficient
Columns
Since this
^2002 ~ v m
is
in municipalities
5
and
6,
=
199s)-
0.25, s.e.
This variable
=
positive though again marginally insignificant
0.15), consistent with our theoretical predictions.
we use the logarithms
the effect at the sample
is
of
mean and 7 <
0,
with lower Pastrana share.
37
all
of the variables so
we
lose a lot of observa-
the impact of Uribe's vote share
is
significantly higher
tions (all the zeros).
is
Column
In
5 with this specification the direct effect of Uribe's vote share
not significant, but the interaction term
negative and highly significant (this specification
is
Column
also includes guerrilla presence before 2002 as in
max{0, v^
—
error
2002
and highly
all
The
significant.
levels,
but not with
Column
the coefficient on
6,
=
0.56,
interaction term
m,1998 }, the estimated effect
final
same models using the displaced measure
standard
and
positive
is
in fact
first
principal
more supportive
the results are
is
components of attacks and displaced.
is
positive
and
precisely estimated
and the
The
coefficient
max{0,i^
These
direct effect
statistically significant,
negative and significant. In this specification using
now much more
v^ 20 o2 —
significant.
of the predictions of our model.
mean
positive
always negative, and statistically significant
also
and highly
of Uribe's vote share evaluated at the sample
interaction term
is
is
Moreover, when we use the specification with max{0,
two columns use the
results are similar,
of paramilitary
of these specifications the direct effect of Uribe's vote share
logs.
v:
The
In
a g am positive and marginally significant (coefficient
four columns examine the
guerilla presence. In
with
*s
Kn 1998)
3).
0.30).
The next
and
—x
and the
~ v m i99s}>
now highly
2 002
on
this table
9.
In particular,
is
significant.
Table 10 investigates the robustness of the basic results in Table
two-year window instead of the three-year window
in
case divided by the population of the municipality).
9.
we
is,
define
uses a
Pm ,t<2002
an d
2000 and 2001 (t<2002) or between 2003, 2004, 2005 (t>2002) in either
-Pm,t>2002 using 1999,
Table
Table 9 (that
it
In particular, the interaction
term
is
The
now more
results are broadly similar to those in
precisely estimated
and
is
statistically
significant in almost all specifications.
OveraU, we interpret the results in this section as providing some support to our Proposition
3 that incumbent politicians in
where these groups
power
deliver votes
will
tend to refrain from eliminating paramilitaries in areas
and that
this effect is stronger
where they would not have
otherwise done as well.
The Symbiotic Relationship between the Paramilitaries and
7
the Executive
Econometric Evidence
7.1
The
last section
itaries
more
provided some evidence on a possible symbiotic relationship between the paramil-
and the executive. In places where the paramilitaries delivered votes
likely to survive.
extensively,
it is
they were
investigate this symbiotic relationship, or the 'quid pro quo,'
useful to study key legislation the executive brought to the legislature,
paramilitary presence
first issue,
To
in 2002,
may have
influenced voting on these legislations.
We
and how
can only examine the
what laws the executive proposed, using case study evidence. To look
38
more
at the second
.
how paramilitary presence may have
issue,
influenced voting patterns,
we again examine
roll-call
votes.
We focus on
a very salient and relevant roll-call for which
we have
data: the vote on whether or
not to change the Constitution to drop the single-period term limit on the President.
who were
more
elected with support from paramilitaries were
in the Constitution, then this
would be
inclined to support this change
though naturally not
direct,
Senators
If
definitive,
evidence that
the paramilitaries supported Uribe either as quid pro quo or because he would naturally choose
more
policies
As
I
our previous analysis of
in
equation
in line with their interests
(9) in
subsections 5.4 and
and preferences.
roll-call
5.5.
votes
We now
we use a simple empirical strategy based on
define A; to be the proportion of Senators
on
list
that voted in favor of changing the Constitution to allow President Uribe to run for re-election.
All of the remaining variables are defined as before.
Table 11 looks at the
of Table
first
8,
roll-call
The
vote for re-election.
except that the two parts of the table
column again estimates a simple regression
now
structure of this table
(all relative to
similar to that
use the attacks and displaced measures.
of A;
on party dummies.
It
Liberals).
Columns
members
of Left
2-4 then use the simple
model
with the attacks data, while 5-7 estimate the same model using the displaced data.
(9)
either strategy,
and even holding party
affiliation constant, there is
presence of paramilitaries in areas where Senate
of Senators on the
run again. This
list
lists
The
shows that members
of third parties tended to vote in favor of re-election, as did Conservatives, while
parties tended to vote against
is
Using
a robust positive effect of the
received a high vote share on the propensity
to vote in favor of changing the Constitution to allow President Uribe to
effect is statistically significant in all
columns, except in column 7 where
we
also
include party identity (this suggests that in this instance most of the effects of paramilitaries on
politicians behavior
show that there
on
is
might be working through third party
no robust impact of the presence of
affiliation).
guerillas
These columns also again
on the voting behavior
of Senators
this measure.
7.2
Case Study Evidence
There
is
a large case study literature which
executive and paramilitary politicians.
Peace
Law which took
is
consistent with the notion of a quid pro quo between
One obvious arena
is
the formulation of the Justice and
place after the secret negotiations at Santa
Fe de
Ralito.
This law was
widely criticized both inside and outside Colombia as being far too lenient on paramilitaries.
Here we focus on a very revealing more recent
the symbiotic relationship.
which
speak to
Even though many Congressmen and Senators have been
arrested,
also
they are replaced in the legislature by their alternates (in Spanish 'suplentes')
same
list
at the
To change
time of the election. In consequence, their political influence
this situation
seem
to
series of events
members
who appear on
is little
diminished.
of Congress proposed a political reform in 2008 to
39
the
remove
when many
these politicians and their alternates from the legislature. This initiative was killed
politicians failed to appear for a debate so that a
appear
failed to
for the vote
quorum was not
was widely blamed on President Uribe
(see the
Gustavo Petro and Rafael Pardo ("Entierro de quinta" Semana, June
notes
"If
Uribismo
projects, such as a
lost its majorities in
new reform
to that
(used sarcastically) in the Constitution
term of
This experience shows
office.
Congress,
would be
it
fact that
Senators
remarks of Senators
2008)
7,
difficult to get the
of the Constitution."
'little article'
is
The
reached.
43
42
,
and Semana
approval of key
The
'little article'
the change to allow President Uribe to run for a third
how
President Uribe relies on the legislative support of
politicians deeply implicated with the paramilitaries to pass key bills.
44
After this project failed the executive proposed an alternative political reform in August 2008
which involved changing the Constitution to take away the power of the Supreme Court to try
Congressmen. As Jose Miguel Vivanco, Americas director
is
at
Human
Rights Watch put
it
"Uribe
brazenly trying to take the power to investigate Congressmen away from the one institution
that has done the most to uncover and break paramilitary influence in the Congress
proposal serves no real purpose, other than to help
(Human
of Uribe's coalition get off the hook"
Rights Watch, 2008).
8
Conclusions
Why
are
many
broadcast
its
power
famous monopoly of
states in less-developed societies unable to establish Weber's
violence in their territories?
of this
members
This
...
The standard explanation
power throughout the
territories that
it
relies
on the
inability of the central state to
nominally controls and views an extension
to the periphery as a natural by-product of 'political modernization'. In this
we developed an
alternative perspective, suggesting that the central state can develop (even
'modernize') without establishing such a monopoly of violence because there
relationship between the parties controlling the central state
in the peripheries of the country.
The
origins of this symbiotic relationship
where non-state armed actors can control
political preferences, they
their ideological
and policy
may be
a symbiotic
and non-state actors exercising power
actors can provide support to those controlling the central state. This
in democracies
paper
elections.
is
is
that non-state
armed
particularly important
Since they naturally have
can (credibly) deliver votes for the national politicians in line with
biases. Politicians elected
with the implicit support of these non-state
actors will then have less incentive to eliminate them, leading to an equilibrium without a
monopoly
of violence of the central state.
We
full
developed this idea theoretically and provided
http://www.semana.com/noticias-nacion/entierroquinta/112489.aspx
In Spanish "Si el uribismo perch'a sus mayorias en el Congreso, se le hari'a dificil lograr el visto bueno para
proyectos clave, como una nueva reforma a ese 'articulito' de la Constitucion" (Semana, June 7, 2008).
44
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.
40
empirical support using recent political events from Colombia.
Our model makes a
series of predictions,
which we investigated using Colombian data. Our
empirical evidence began by using a variety of strategies to confirm that paramilitaries have
indeed had a significant impact on elections in Colombia. Following the foundation of the
there
is
a sharp increase in the vote share of third parties, which were explicitly or implicitly
The same
associated with paramilitaries, in areas that have high paramilitary presence.
also
AUC,
showed an increase
municipality).
We
in electoral concentration (the vote share of the
most 'popular'
also presented evidence that high paramilitary presence in areas
received large proportions of their votes predicts
Peace Law, and whether they get arrested
that paramilitary presence
is
how they voted on key
for illicit links
areas
list in
where senators
clauses of the Justice
with the paramilitaries.
a
We
and
also found
correlated with vote share of the winning presidential candidate,
Alvaro Uribe.
The two key
predictions of our theoretical
model are that paramilitaries should
extent that they deliver votes to politicians whose preferences are closer to their
where these
effect is larger in areas
politicians
would otherwise not do
well.
persist to the
own and
These
that this
results illustrate
the possible presence of a symbiotic relationship between the executive and the paramilitaries.
We further illustrated this
list
won
relationship by showing that the proportion of the votes which a senate
in paramilitary areas is positively correlated
with the proportion of Senators on the
list
that voted to change the Constitution to allow President Uribe to run for a second term.
Nevertheless, neither our results nor the case study literature suggests the presence of a formal
'coalition'
between paramilitary groups and President Uribe. They are consistent, however, with
the idea that Uribe's policies are closer to those preferred by paramilitaries than those of other
parties which naturally gives the paramilitaries an interest in generating support for him.
Jairo Angarita former leader of the
deputy declared
in
AUC's Simi and San Jorge
September 2005, he was proud to work
"reelection of the best President
we have
in
and Salvatore Mancuso's
for the
ever had" 45
.
Naturally, different interpretations of the data are possible.
people
blocs
As
For instance,
it
is
possible that
paramilitary areas are naturally pro 'law and order' or more 'conservative' so that pres-
ence of paramilitaries
is
endogenous to 'voter preferences'.
If this
were true our results connecting
the presence of paramilitaries to Presidential vote share could be due to omitted variable bias.
This might also undermine our interpretation of the
ators
who voted
roll-call
votes since
it
would imply that Sen-
to change the Constitution to allow President Uribe to run again did so because
their voters' preferred
it.
These alternative hypotheses do not appear plausible explanations
the patterns, however. First, our identification comes from the timing of the formation of the
'El Tiempo, September
7,
2005.
41
for
AUC
and
their decision to develop
an electoral strategy (relying on intimidation, murder and manipu-
lation to get their candidates elected to national office). For example, prior to these events, there
was no
significant differences
between electoral concentration
taries, indicating that political preferences did
The timing
of the formation of the
Cali drug cartels and
is
AUC
not systematically
with or without paramili-
differ
between these two areas.
stems mostly from the collapse of the Medellm and
unlikely to reflect a change in political preferences on the part of voters.
Second, to the extent that voter preferences
others, this
in areas
would be controlled
for
differ
by our use
with some areas being more conservative than
of fixed effects as long as these differences are
time- invariant. Moreover, the case study literature suggests something of a 'reality check'
argument since
ments such
as
it
on
this
provides extensive evidence of coercion and manipulation of elections in depart-
Magdalena, Cesar, Corboda and Sucre with high paramilitary presence. Third, the
fact that paramilitary presence predicts arrests of Senators
are simply representing voter preferences.
These
makes
results, along
it
improbable that politicians
with the case study literature,
suggests that politicians formed coalitions with paramilitaries, or paramilitaries generated their
own
politicians. Fourth,
we
tried to explicitly control for the time varying effects of conservative
leanings of certain municipalities.
An
armed
actors on public
the paramilitaries, for example, are effectively the local state,
how do they do
important area
good provision.
If
for future investigation
is
the impact of non-state
at providing public goods or services relative to the legitimate state?
is
mixed on
this.
Some, such as Gutierrez Sam'n and Baron (2005,
of the development of the paramilitaries in Puerto
The
p.
case study evidence
13) argue in their study
Boyaca that they undertook "investments
in
patronage (drug stores, climes, health campaigns, educational services)." Duncan's (2007) also
supports this idea (see also the
—probably not entirely credible—claims
other anecdotes mitigate against
A
final
and
it
in
in
Castaho, 2001). Yet
Casanare, for example, and in La Guajira.
crucial question concerns the external validity of these findings.
nisms we discuss only relevant
in
Are the mecha-
Colombia, or do they also influence the creation of the monopoly
of violence in other states? Although our arguments in this paper contrast with the emphasis of
much
of the existing literature
on state formation, there are several other instances where non-
state actors appear to play a similar role to that of paramilitaries in Colombia. For example, as
mentioned
in the Introduction, the existence of the tribal areas in
the state has not formed in Pakistan, but
to that in Colombia.
A
related
is
Waziristan does not imply that
indicative of a different equilibrium pattern similar
example comes from
Italy
and the
role of the
votes to the Christian Democratic Party (Walston, 1988). Finally, the long
South
after the Hayes- Tilden
the 1960s also presents
many
Mafia
in delivering
autonomy
of the
US
agreement of 1877 until the Civil Rights and Voting Rights acts of
parallels with the
Colombian
case.
Southern
political elites
were
able to establish a one-party undemocratic political system based on the disenfranchisement and
coercion of blacks.
The power thus consolidated gave them
42
great influence in national political
institutions
where they could both impede
plied votes, which
legislation they did not like,
was exactly what the Hayes-Tilden deal was about.
in favor of this interpretation
is
but where they also sup-
A
telling piece of
evidence
provided by Franklin D. Roosevelt's reaction to a proposal to
pass legislation to attempt to restrict lynching. In response he argued that Southern legislators
"are chairmen or occupy strategic places on
come out
for the anti-lynching bill
in Frederickson, 2001, p.
20).
now, they
There
is
most of the Senate and House committees
will
block every
therefore
identified here are of quite general applicability.
43
bill I
...
If
I
ask Congress to pass" (quoted
some evidence that the mechanisms we have
References
Acemoglu, Daron (2005)
Monetary Economics,
"Politics
and Economics
in
Weak and
Strong States," Journal of
52, 1199-1226.
Acemoglu, Daron, Davide Ticchi and Andrea Vindigni (2006) "Emergence and
sistence of Inefficient States"
NBER Working Paper
Per-
#12748, forthcoming Journal of the European
Economic Association.
Acemoglu, Daron, Davide Ticchi and Andrea Vindigni (2010)
Wars" forthcoming Journal of
Amnesty
the
"Persistence of Civil
European Economic Association.
International (2005) "Colombia: Flawed Demobilization Offers de Facto Amnesty,"
http://www.amnestyusa.org/all-countries/colombia/ flawed-demobilization-offers-de-facto- amnesty /page. do'
Angrist, Joshua D. and Jorn-Steffen Pischke (2009) Mostly Harmless Econometrics:
An
Empiricist's Companion, Princeton: Princeton University Press.
Baron, David
American
P.
(1994) "Electoral competition with informed and uninformed voters,"
Political Science Review, 88, 33-47.
Bates, Robert H. (2001) Prosperity and Violence: The
New
York:
WW.
Norton
&
Political
Economy
of Development,
Co.
Bergquist, Charles, Ricardo Penaranda, Gonzalo Sanchez eds. (1992) Violence
Colombia: the contemporary
crisis in historical perspective,
Wilmington,
Del.:
SR
Books.
Bergquist, Charles, Ricardo Penaranda, Gonzalo Sanchez eds. (2001) Violence
Colombia, 1990-2000: waging war and negotiating peace, Wilmington, Del.:
Besley,
Timothy and Torsten Persson (2009) "The
Rights, Taxation, and Politics,"
American Economic Review,
in
in
SR Books.
Origins of State Capacity: Property
99, 1281-1244.
Blattman, Christopher and Edward Miguel (2008)
"Civil
War," forthcoming
in the
Journal of Economic Literature.
Bowden, Mark (2002) Killing Pablo, New York: Penguin.
Brewer, John (1988) The Sinews of Power: war, money, and
New
the English state, 1688-1783,
York: Knopf.
Castaho, Carlos (2001) Mi
confesion: Carlos Castano revela sus secretos, Bogota: Editorial
Oveja Negra.
Centeno, Miguel (2002) Blood and Debt: War and
versity Park:
Penn State University
the Nation-State Latin America, Uni-
Press.
Who Counts
NBER Working
Chaves, Isaias, Leopoldo Fergusson and James A. Robinson (2009) "He
Elects:
Determinants of Fraud in the 1922 Colombian Presidential Election,"
Paper #15127.
Deas, Malcolm D. (1999) IntercamMos
violentos: reflexiones sobre la violencia politico en
Colombia, Bogota, Colombia: Taurus.
II
Duncan, Gustavo (2007) Los
Senores de
Guerra, Bogota: Editorial Planeta Colombiana.
la
A Comparative
Evans, Peter B. (1989) "Predatory, Developmental and Other Apparatuses:
Political
Economy
Perspective on the Third World State," Sociological Forum,
4,
561-587.
Evans, Peter B. (1995) Embedded Autonomy, Princeton: Princeton University Press.
Evans, Peter B., Dietrich Rueschemeyer and Theda Skocpol eds. (1985) Bringing
Back
the State
New
In:
Cambridge University
Perspectives on the State as Institution and Social Actor,
Ferro,
y
York:
Press.
Fearon, James D. and David D. Laitin (2003)
American
New
"Ethnicity, Insurgency,
and Civil War,"
Political Science Review, 97, 75-90.
Juan Guillermo (2002)
El orden de
la
FARC-EP
guerra: las
entre la organizacion
Bogota: Centro Editorial Javeriano: Pontificia Universidad Javeriana, Facultad de
la politica,
Estudios Ambientales y Rurales: Colciencias.
Frederickson, Kari
(2001) The Dixiecrat Revolt and
the
End
1932-
of the Solid South:
1968, Chapel Hill: University of North Carolina Press.
Gibson, Edward L. (2005) "Boundary Control: Subnational Authoritarianism
cratic Countries,"
World
Politics, 58, 101-132.
Gutierrez Sanin, Francisco (2008)
in the
Colombian War,"
Politics
&
"Telling the Difference: Guerrillas
and Paramilitaries
Society, 36, 3-34.
Gutierrez Sanin, Francisco and Mauricio Baron (2005) "Re-stating the
itary territorial control
Demo-
in
and
political order in
Colombia, 1978-2004,"
LSE
State: Paramil-
Crisis States
Program,
Working Paper #1.
Guzman Campos, German, Orlando
La
Fals Borda,
and Eduardo
Umana Luna
violencia en Colombia: estudio de un proceso social, Bogota: Ediciones Tercer
Hartlyn, Jonathan (1988) The
politics of coalition rule in
Colombia,
(1962)
Mundo.
New York: Cambridge
University Press.
Herbst, Jeffrey
I.
(2000) States and Power in Africa, Princeton: Princeton University
Press.
Hintze, Otto (1975) The
historical essays of Otto Hintze, edited with
Gilbert, with the assistance of Robert
Human
an introd. by Felix
M. Berdahl, New York: Oxford University
Rights Watch (2005) "Colombia: Letting Paramilitaries
off
Press.
the Hook,"
http://www.hrw.org/legacy/backgrounder/americas/colombia0105/
Human
tions,"
Rights
Watch
(2008) "Colombia:
Proposal Threatens 'Parapolitics' Investiga-
http://hrw.org/enghsh/docs/2008/08/04/colombl9533.htm
Inter- American
of the Justice
Commission on
Rights (2007) "Report on the Implementation
and Peace Law," http://www.cidh.oas.org/ pdf%20nles/III%20Informe%20proceso%20desmoviliza
International Crisis
N°20, 10
Human
May
Group
(2007) Colombia's
2007
45
New Armed
Groups, Latin America Report
Kline,
Harvey
F. (2007) Chronicle of a failure foretold: the peace process of Colombian
President Andres Pastrana, Tuscaloosa: University of
Kohli, Atul (2004) State- directed development:
global periphery,
New
Alabama
Press.
power and industrialization in
political
York: Cambridge University Press.
Lindbeck, Assar and Jorgen Weibull (1987) "Balanced-Budget Redistribution
Outcome
as the
of Political Competition," Public Choice, LII, 272-297.
Lopez, Claudia (2007) "La Ruta de
de Antioquia," in Mauricio Romero
Acuerdos
the
Politicos,
ed.
la
Para
Expansion Paramilitar y
Politica:
Bogota: Corporation Nuevo Arco
La Ruta de
Iris:
Expansion Paramilitar y
los
Intermedio.
Lopez, Claudia and Oscar Sevillano (2008) "Balance
poration Nuevo Arco
la
Transformation Polftica
la
Politico de la Parapolitical' Cor-
iris.
Mazzuca, Sebastian and James A. Robinson (2009)
sharing in the Origins of
Modern Colombia," Hispanic American
W.
Mickey, Robert
"Political Conflict
and Power-
Historical Review, 89, 285-321.
The Democratization of Authoritarian
(2009) Paths out of Dixie:
Enclaves in America's Deep South, forthcoming Princeton University Press.
O'Donnell, Guillermo (1993) "On the
Problems:
A
and Some Conceptual
State, Democratization,
Latin American View with Glances at
Some Post-Communist
World
Countries,"
Development, 21, 1355-70.
Oquist, Paul H. (1980) Violence,
conflict,
and
politics in
New
Colombia,
York: Academic
Press.
Pardo Rueda, Rafael (2004) La historia de las guerras, Barcelona: Vergara.
Pardo Rueda, Rafael (2007) Fin del paramilitarismo: les posible su desmonte?
D.C.: Ediciones
B
Colombia.
Pecaut, Daniel (2001) Orden
y 1953, Bogota:
Grupo
Editorial
y violencia: evolucion socio-politica de Colombia entre 1930
Norma.
Pizarro Leongomez, Eduardo (1991) Las
binacion de todas
Bogota,
la.s
formas de
Internationales: Tercer
Mundo
lucha, Bogota:
FARC
UN,
(1949-1966): de
la
autodefensa a
la
com-
Instituto de Estudios Politicos y Relaciones
Editores.
Pizarro Leongomez, Eduardo (2004) Una democracia asediada: balance y perspectivas
del conflicto
armado en Colombia, Bogota: Grupo
Posada Carbo, Eduardo (2003)
Editorial
'Ilegitimidad' del
Norma.
Estado en Colombia: sobre
los
abusos
de un concepto, Bogota, D.C.: Alfaomega Colombiana: Ideas para la paz.
Porch, Douglas and Maria Jose Rasmussen (2008) "Demobilization
in
Colombia: Transformation or Transition?" Studies in Conflict
Rangel Suarez, Alfredo (1998) Colombia,
Universidad de
los
&
guerra en elfin de
of Paramilitaries
Terrorism, 31, 520-540.
siglo,
Bogota:
TM
Editores:
Andes, Facultad de Ciencias Sociales.
Rangel Suarez, Alfredo (2005)
El poder paramilitar, Bogota:
46
Fundacion Seguridad
&
Democracia: Planeta.
Restrepo, Jorge, Michael Spagat and Juan Fernando Vargas (2004) 'The Dynamics
of the
Colombian
A New Data
Civil Conflict:
Robinson, James A. (2007) "Un
and Miguel Urrutia
Fondo de
la
Set,"
Oeconomicus,
21, 396-428.
Pais Tipico de America Latina," in
Economia Colombiana
eds.
Homo
del Siglo
XX: Un
James A. Robinson
Analysis Cuantitativo, Bogota:
Cultura Economica.
Robinson, James A. and Miguel Urrutia eds. (2007) Economia Colombiana
XX: Un
Andlisis Cuantitativo, Bogota and Mexico D.F.:
Mary
Roldan,
(2002) Blood and
Durham: Duke University
fire:
in
Fondo de Cultura Economica.
La Violencia
in Antioquia,
Colombia, 1946-1953,
Press.
Romero, Mauricio (2000) "Changing
and the Paramilitaries
del Siglo
Identities
and Contested
Settings:
Regional Elites
Colombia," International Journal of Politics, Culture, and Society,
14,
51-69.
Romero, Mauricio (2002)
Paramilitaries y autodefensas, 1982-2003, Bogota:
Editorial
Planeta Colombiana.
Safford,
New
Frank and Marco Palacios (2002) Colombia: fragmented
land, divided society,
York: Oxford University Press.
Mar Palau (2006) "Conflict, Decentralisation and Local
Governance in Colombia, 1974-2000," CEDE Universidad de Los Andes, Documento 2006-46.
Sanchez G., Gonzalo and Donny Meertens (2001) Bandits, peasants, and politics: the
Sanchez
,
Fabio and Maria Del
case of "La Violencia" in Colombia, Austin: University of Texas Press.
James C. (1972)
Scott,
Comparative Political Corruption, Prentice-Hall; Englewood
Cliffs
NJ.
Serrano Zabala, Alfredo (2007) La
politica,
batalla final de Carlos Castaiio:
Bogota: Editorial Oveja Negra: Quintero Editores.
Storey, R. L. (1968) The reign of Henry VII,
Taussig, Michael T. (2003) Law
York:
secretos de la para-
New
Tilly,
New
York: Walker.
in a lawless land: diary of a "limpieza" in Colombia,
New
Press.
Charles ed. (1975) The Formation of national
States in Western Europe, Princeton:
Princeton University Press.
Tilly,
Charles (1990) Coercion, Capital and European States
AD
900-1990, Cambridge
MA:
Basil Balckwell.
Valencia,
in
Leon (2007)
Mauricio Romero ed.
Politicos,
"Los Caminos de
Para
Politica:
nal of Conflict
(2007)
and Violence,
1,
"Is there
Allianza entre Los Paramilitaries y los Poh'ticos,"
La Ruta de
Bogota: Corporation Nuevo Arco
Waldmann, Peter
la
Iris:
la
Expansion Paramihtar y
los
Acuerdos
Intermedio.
a culture of Violence in Colombia?" International Jour-
61-75.
47
Walston, James (1988) The Mafia and
New
Clientelism:
Roads
to
Rome
in post-war Calabria,
York: Routledge.
Weber,
Max
(1946)
"Politics as a Vocation," in
From Max Weber: Essays
translated and edited by H.H. Gerth and C. Wright Mills,
48
New
in Sociology,
York: Oxford University Press.
Appendix: The Evidence from Magdalena
we examine voting patterns
In this Appendix,
Department
in the
of Magdalena,
where various
paramilitary groups have been very active over our period of investigation. In
Map 3 we reproduce
the data on electoral concentration, defined as the vote share of the
won
Magdalena
in
2002 and 2006 elections
for the 1998,
list
that
We
for the Senate.
the most votes,
also place
on the
map
the proportion of votes that Senator Luis Eduardo Vives (14th on Table Al) and Congressman
Alfonso Campo, one of the signatories at Sante Fe de Ralito,
shows the dramatic increase
specific politicians
that Vives and
won
in
each municipality.
in electoral concentration in 2002. In particular, in the
were winning essentially 100% of the votes cast
Campo won
The map
dark red areas
in a municipality.
One
sees
very large proportions of votes in the southernmost municipalities
such as San Sebastian and Guamal. In the dark red municipalities to the north Senators Salomon
Saade and Dieb Maloof won nearly
Vives, Maloof and Saade
Al).
tried
and found
Concentration
all
all
of the votes (2nd
and 5th
1,
3rd and 4th in Table
represented third parties and Vives and Maloof have been
guilty of links with paramilitaries, while
falls in
in Table
Saade
is
currently under investigation.
2006 however and Vives wins large proportions of his vote in municipalities
to the north, such as El
Pinon and San Antonio, where
in
2002 he had won no votes. Campo's
vote also becomes more evenly distributed over the Department, though he was not re-elected in
The pattern
2006.
both
similar for the other Senators
re-elected, with almost the
Magdalena
in
is
in
same number
2002 and 54,609 in 2006.
In 2006 Jose Joaquin Vives, a
and despite of
this
fall,
of votes as in 2006. For
Maloof and Vives were
example Vives got 53,759
46
Congressman
for
Magdalena
prior to 2002
whose vote share
collapsed in the 2002 elections, noted
"Certain Congressmembers have credentials that are built upon the blood of innocent citizens
be
free. It
in the
...
must
In
Magdalena circumstances
also
still
do not
exist for the next election to
be said clearly and directly that democracy has been kidnapped
department and remains kidnapped through a relationship created between
certain political groups
Overall, the data for
and the
self-defense forces."
Magdalena show a
4
'
significant increase in electoral concentration in 2002
with several Senators and Congressmen who subsequently were charged with links to paramilitaries,
wanning close to 100% of the vote in a few selected municipalities. In 2006 concentration
4
Prior to the 2006 congressional elections two pro-Uribe parties dropped five candidates from their lists who
were accused of having direct links with paramilitaries. The U party dropped Habib Merheg, Dieb Maloof and Luis
Eduardo Vives and Cambio Radical dropped Jorge Caballero and Jorge Castro. Expelled candidates re-appeared
on the lists of smaller pro-Uribe parties and Merheg and Maloof won seats in Congress with the Colombia Viva
Party.
*'
Quoted
in the article
'Mas
alia del control territorial' El Espectador,
49
January
22, 2006.
went down, but interestingly Senators Maloof and Vives were
votes spread out
much more
evenly across the department.
50
re-elected,
though now with
their
U
CQ
e
c
3
— b
> S
C5
N?
s
8
CO
CU
>s
CU
_c
—
ir.
o
on
m m
ob no
O
U~>
oo
r*>
•*
LO
3
cn
(N
<tf
— —
L/-)
m o
ON
r*->
3^
^'
NO
ON
ci
f—^
_aj
3
«3-
<sfr
(N
CO
£
£
O
cu
_3
"es
O
u
_
u
c
u
a.
"Co
5
o
u
^
a,
>:>
3
S
s-
T3
(B
en
u
£>
3
a
—
u
CO
CU
~o
"O
o
V)
Xfi
r3
.Sj
u
«
U
fc
i—
J
ao
CO
r3
SB
I
t
o
>
C
< <
cu
to
cu
-M
<
V)
-a
u
a
c
>
T3
l)
r3
E
—
CO
U
>
c
CU
c
CO
CU
cu
aj
o
c
CJ
23
E
2
CU
01
u
>1
U
>1
VI
cu
CU
LO
cu
a
>1
CO
CO
CU
CO
o
CO
d>
CO
>i
>%
>.
>1
>-l
U
Z.
—
s.
>
I—
I"?
"-
C
03
E
'5
at
co
U
is
co
CO
cu
0)
CO
i>
CO
a>
CO
o»
CO
cu
O
o
i—
5
3
—
S
^-^
tN
o
o
IN
E
—
r
C
C
J3
'—
O
CO
CL
CJ
re
a,
LO
E
_CQ
X!
- <
1c
2c
o
cu
w
CO
o
U
u
rt
cu
'J
"cS
a
r
<4-
m
_
CU
6
1
"o
o u
c
'
^
>.
as
co
5
c
CQ
-J
CU
1)
<U
CU
(LI
o
e
g
c
CU
o
c
g
J
CQ
U
B.
5
"H
c
-^
oi
<
5
5
33
cm
Si
5
<
2
5
Q
O
<
<
<
< u
si
< <
H a
- o
j
w "3
a
a!
tL,
<
N <
Q o ci
W
Z u H
u
<
z
2
O
-
r^
,
z
1
_
o
1
CO
<
2
>
<
3
Q
? 2 z
<
<
Q <
Bi
w C
~
z
is
Q
D z
C w 3
—1
,
i
U Z
Z i
D O
<
C/3
_
DD
Z o
<
—
a.
o
CQ
Q
5
<
Z
S
w
•
M
2
z
y
z
Cu
lo
U J
J J CJ
D
2 3 z
< O
j S2 u
d 3
cS
O
3
->
^
3
c
->
1)
E
CO
U
--'
Cm
5
E
U
— S
X
u
c
n
cu
CU
r5
CO
—
>-
E
o
D
u
^3
a
B
N
'^
cu
-5
'j
CB
c
u
B
c
a
CQ
>-
E
<
is
~
-5
u
^
E
cu
iu
CQ
-
CO
—
r
CO
1
t
CU
t~-
~3
u
CQ
U
E
-5
cr
"J
u
CU
CO
r3
C/3
z
C
_3
CO
CU
a
CO
o d
u a
B
'.j
>
O
C
S
N
—
O
o
'""
s
CQ
-1
CO
CO
Z
<
z
~
—
3
o
-J
-J
B
-£.
o,
a
/.
U
jz
CO
S3
u
?
U
U
-
CO
'u
cu
CI
11
CO
O
—
aj
*
IS
*-
>^ _C
r"
CO
CZl
3
CQ
—D.
—
cu
CO
cu
~
<
_
ca
>
N
Z N
H 'Z
< <
2
N
§
C
C
5
R
r^
"5
<
Q
<
_
>
CO
u
—
C/5
<
pi
5
3 ;_
^_ H
2 m eg
- y. K3
< U <
CQ
U
— N
D W
< < s
E
—
CO
E
c
y:
c
—
B
B
r
3
c
=
c
U
Z
es
r-
a>
CO
_3
I
o
E
U
CO
CO
.£
3
CO
_.
es
O
(N
CO
-;
fed
u
Ui
it.
a.
cu
E
B
u
u
es
e
B
Q.
a.
A
.£P
>^
K o
a
o
o
u
E
rl
«
^c
co
LO
I)
^z:
o
r—
M
-t
CO
O
"*^
CU
CO
"5
CQ
ra
u
CO
CQ
CU
CU
Z- 15
O
CO
C3
>
L-.
Xi
CU
CU
rz
ID
< <
B
>
o
_o
U3
,fiP
cu
1>
b
2
-
Q,
CO
C5
>*
"
C5
CD
u
CU
-^f-
s
O
—
o (N
'*
CD
I—
— oo 3 rj o — ro
o r- m oo (N
(N 3- <N o
-r IN M — O
n - — oo^g\o6o6odvo\bin
Cu
>^
E
u
—
E
3<
u
o
>
3
rr
3
g
ri
"3
—
'
-o
c
Ih
Q
a
.3
- Vj
CO
GO
>
o —
O —
O CN
O
© o
©
co
on cn
o cs
on
— © © o
ro
no
cn
<
Q
—
inNOONONONONNONO
NOOOCO-HCNCNOONO'xf
NOoorN]—
— — (S (S-ooonoiocn^IN - — — <N(N
ON
.C-)
NO
'
K!
o
CN
i
H
'
» = S S S —
CO
no
vo
—
O
—
t>
CN
On
1
^
>
o
-
-r-
-O
u
8P
b^
^ I
-3
I4-|
O
a
CO
0.
a
jj
o
'I
•:=
£
CO
n.
•a
o
to
u
2
5
o
o o
o —
o o
©
© © o ©
cs
— on
—
© tt
© o o d
m o
on
d
- m
On
in
P>
—h
oo
rj
On
m
n
-vf
tt
co
tj-
f^
in
co
r^
in
IT,
*0"
NO
M
M
rn
!£
u-,
CN
NO
NO
O
ITN
on
n n
•^~r-~in
,
^1-
NO_iNr~— —
1
a. _u
-'
•-;
to
wu
s
o
c/l
oo
.s a
-o -a
TD
c
t_
.2
r3
a
s
n
>
o m
o cn
CN
© © © ©
«n
c^
CN
— On oo — On
co — O On co — CN
o
rnoONrNjoooNr^NOrO'
— cn—'rN) — — r" — rNitsNO
CO
-r
no
a © a
r-»
(vl
'
go
e
co
co
u
Oh
o o o
s
N
oo
—
'
tcn
©'
n
^
*r co
©' d
CN
CN
On
m
tI-
OO
m —
o
TT
—
—
in rs
on
O
H m m m
(N
co
NO
r^
—
CNl
t~;
*
i
t~-
!0
co
r-~
~*~
— 3- CN o
— O CO o
00
n ^ in
NO CN tN
oo — d d a d »n in
d o
CN — —
— CO
r- __ N-^
cn
On
no
vt-
vn
in
in
nu
a
c
u
o.
to
u
U
1—
Ih
O
CJ
— CS ~
_ ^
a
"2
eg
7,
.3
>
ON
cn
—
o — n
rs
(N
odd
co
©
cn
~co
co
"*
r^
oo
d O ©
Q
oi/">tooor^r~-<N~-'f*")r^
Ojie^^tNffiqoot^M;
— on oo
— on
— —3
— r^
— uH
— tN
MMN N
iri
i
I
*-<
in CO
m no
* —
* oo
—
ON NO
K
d
—
—
—
CN
^ CNOCN
Tt
r-~
in
"3-
CN
•—
a o<
K
3
o
9
— CO o a—
'-T.
GO
fNl
IAi
co
CN
B
B
CO
u
2
— O
O © co
O
CN
*3-
©do©
ooofN)ONrN]ONt^ON»n
oo »
^ - prnfNrn
- —
N
O CO
N co
—
co
^
ddo
rNJNo'vnnSoNON'ncNcN
mcNmmmcNm'O-io
NO
u,
3
§1
o ^
to
—
-a
So
ca
tn
S
S
-a
s
n
S
to
CN
—
£ «
C)
GO
o
NO
CO
CO
CO
CN
CI
a
— »n
o
a K
in 2 d
K
d
o
CN
c
•-
JO
^—
o>
ON
ON
—
-a
CO
5
JC
o
d
J3
—
CO
o
o
CN
p
£
3
NO
r-
o
e
in
=
-
1
a.
CO
ex
~
o>n
u
9-
£
I
s
CO
„
co
«
CO
m
o
o
CN
t"~
o
2
1
S
S
O
o
_
O
o
IN §
in
a
a
CN
PQ
CO
t- —
»
o
5>
o
—
n
j:
O
co
9 a
~^
JO
.s
I-
ca
>
r\i
u
(U
E
c
u
u
«
-J
~ — —
.
LO
—-
r
o
—
l
(U
Cd
a
3
cn
EC
~
o
CN
u-i
cn
T
6
a
CN
u
r_
CC
C
C
c
c
<L>
CD
t/i
0)
cn
<U
<D
W5
U
CJ
0h
Cl.
CO
CO
crt
<L>
0*
&,
C-3
CO
to
CO
CO
CO
o.
o-
o»
<—
in
co
m
—
<
CD
u
u
u
u
oS
-s
-s
-s
u
.s
.s
u
u
C
o
.S
C/3
C/3
U
U
C
C/3
OJ
>
U
'£
I)
C/3
aU a
U
-.j
co
a o 6 6
m no t^ oo
.S
;.j
-j
CO
CO
GO
00
J3
u — w
<u
w H
U o o S
U U
-
60
•a
00
.£
oo
c
5
B
-a
-a
*-
b
0- Oh
oo 00 00
c = .S
E
.fa
ft
C
—
c-
CO
cd
,g
CD
2
.5
eg
CO
3
CO
03
-J
£ & £ £ $ £
o-Nm^in\o>»
— ~
in
.5
c3
U
.6
*i
—
.a
3
u
CO
o
(H
&
Ph
U
,B
u
.1
ex
.s
—o
s
c
3
E3
O
cd
CN
CO
u
u
c
r > > > >
ea
•a
.=
.S
"H
—
00 *3
CfH
tj
re"
"j8
.sa
-r
jo
.2
o
CO
3
5
'H_
2
s
>
CO
|H
u
..
a
c
>.
o-
U
U
a
.5
5
.e
—
z g OO
c £
a
1 S o
| U>
u E
3
5 < 5 Q < £ a 5
£
cNco-3-mNor~-ooON
ono
— OJ —
rj CN CN CN CN CN CN CN CN
.03
^ 2
co
ST"
CO
ON
EX
o
I
0.
o>
o -r
O
~
—
cs
C
o
1
•*
a
CL>
U U
.S
£
CO
a-
03
a
1
V3
1/3
a
a
CN
4
a
b b b b
C3
3
u
v
u
3
<
M
03
PQ
,
K
a
a
ON
ON
o
o a
4.
CO
o
t/1
1*
"H.
g
C/l
!0
LX'
CO
3
3
3
_3
1
Z T)
>
CD
JS
CO
G
O
u
CL>
u
3
o
3
u
CO
CJ
a,
co
au
cS
CO
3
E
—3
—
-
1
m—
O
©
ON
i
a
o
;—.
© o
cn
« o
on
£;
rn
£;
en
-7
O
NO
ON
CN
CN
CN
CN
CN
r-~
—
o
Z
—
e
= ^
'
-*-•
i
°
NO
•
'
-G
m e g a
NO
CO
-
^
J2
S a S
<
no
o
o
(N
i
eo
.S
cf
I
CO
On
ON
CO
r~
>
<u
u
a
«
!
£
v~)
m
—
l>
o
o
9
OO
'O
ri
Z
r^
NO
CO
(N
m
1
H
S
<w
O
u
I
a
-1
(N
.
-^
bo D.
ed
Ifi.
[
c
—o
C
<D
> C O c
-a
o
o
^o
-a
2
o>
J
to
hj
ra
g
c_
-j
**
o
o
.
m —
m on
CN
— — ©
r«i
no
co
" ~
m m
o —
LJ
•
'
CN
co
—
rN
ov
ON
;
:
u o
C
•-
£a
H—
C
3
U
£
-^
60
§
£•
«-
a-
c E
E^g
I
J3
•
en
Zi^ 2
>
O-
— r^^ON^-NO
— -^O — O^ao^
On -C
On —
00
CD I
a
on
S.a
On
•C
Q
es
aj
o
-
•a
vo
o
oo
NO
— O — O — O —
1
IS.
— £
on
00
e
c
j= := 73
NO
ri
-
=
u
in
CO
>
Q
On
ON
r-
i)
a.
o
o
O
CN
On
c
^h "O
i
to
r^ -J
cn
iS
'*"
"3 To
._•
.s
»
T, u "5
—~
<3
n.
n.
-
OJ
§5
a.
,,
-
—
I*a
m ca
a
iOJ
On
ON
_H
e > u
T3 U ° C
? O J3 o a a
.2 .2
NO
O
O
(N
r
s
c
a.
i
s
iDC-T'oco'r"co~r5'oN
On
0-
^ —
in
— 2
•-
:
o
S —
cf - 1
w ~S iS
3
S
a
-a
u
si
>
E-2
a
o
E
_• a
'o
oj
o-
E >
a o °- « 2 'EL
o 3 u E 's-s
B .^ o «
bO
NO
o
O
©
u
-=
>—
r•/">
5n
—
— O 2 m a V
no * cn °° cn n o
'
— o
cn
2
ci
ONO*SnOI^OU-i
NmOioqiN.^yo—
oo<9ooooo
o
2 m
u —
o
5
tf)
u
>
0.
o-
-r
C
° -^
o
— 'C
y
^:
-"
a. °- -5
^>
=
o ^
r.
I
a
§^
u —
S
>,
1/1
E
E
S
"a
U
ON
o
o
Os
<*
£
-=
a
S
w
=
sc
«"|, 1
—o
C
u.
cs
" a u
? 3
~ u = d>
o
S 5 E
°fi
J2
on
r»
m ON
©
t *
r-
O —
<
©'
CO
~
a
•^
CN
.E
§§
r»-
"3-
'
.
£?•&•?
Sa —s —^
D-.9-
"3
-a
^
~
a S a
E E
§ §•=.!£
E _ a »
o
a —
£ « a .a r^"
"3
s
iri*(N|^^t"Ov
*ONu-irjON";—
o
.
*
d
*'
d
°
d,
H
Eu
3>
—
d.
en
<j
ra
V)
J5
a a
^
^
t,
a.
^ = "a
^^
„a -a •— - - u
o h —
a
«
a
.a =
a
a »
g§5»
- a °'i
o o
-*
-o
cj
l~
i-
a>
-
TCJ
3-
—
'
^ S
.|
CO
cn
ON
a;
c;
o
—
cn
X
X
X
<u
0)
u
e
u
u
E
o
a
^
n
X
•J
NO
o
c
<u
U
-y.
ON
On
tu
o
X
E3
'-»
—
X
u
o
c
u
Cn)
£1
O
S
r
o
o
<N
X
X
u
O
c:
r^
u
u
>
4_i
O
i
a
a
o o o
CO
<S
O
ft)
^
£
"
re
E E
ra
u,
—
J
'J
o
c5
O
U
.S
3
>
u
.— .— •£
o
1
E
Oh
to
b
^
a a
3
cd
t-
O. Q.
„,
dj
<j
3
c
T3
-j
~
E
eg
'-'-""> 8-8
u u l>
OJ
c. c. a
ooa
ta
^3
a S c
h-T
I
P
U
-—
ri
«,
—
*
in
CO
ON
a
E -o a
—a a
no
lis
O .Em —
e b
3
3
co
to
aj
-
Q
O
* O
CN ^
•-^
r-
>*
r-
*
-r
OO
ci
—
t~
CN
"0"
UT*
l>
r-j
'CN
00
<N
o
Z
o
J2
a)
2:
CO
o
o
3
-C
.S
u
IS
CN
ic
on
|
rn
o
ON
r*-i
Tt
t-
1
Z
*
o
OO
co
>
E
a:
5 s
—
"JS
~
.M
<D
c
c
Q
tn
NOr^ct}o«o*NOcnr-«—
—
«(NO\ —
in
to
o.
„
3
E |
— fNrnfSr-r*1^-rM
s
U W
-a
I
=
CO
Q>
-a
•
5
»n
rn
M
CO
(N
°
u
<n
t^
\D
in
N
O^ m
t--
fN
r-~
—
o
Z
(N
Q.
u
I
JS
-2
'"S
3 o
3
c
— «J
D.
:=
c
r
O g
U°
CO
OO
IS. a
—
'
-a >n
ecu
>
i_
SO3
C
CJ
L.
CO
dj
O
HI
c r»
S
i 8
Im *
-a
—
-U tc fN
M
^u
i£
CJ
-
^
^ fs
Ov ^- —
cn
u
££:§>
:|1
il
s
—
3
CO
»>
C
^
5
~|
'5
"2
&i 5
in
'
>
3=
S *>
u E
rj
U
CO
*~
ca
J2
-j
ON
""
a
>.!t;
vO
S-gj
3
> c w
u 'b .Id
x: C3 ,tm
a ^_j
w >
u u D.
a
S
^
— *co
«
o
u
OIJ
'
c3n
^^
•
u
o
c
bfl
b
>
NO
a m oo
•—
..1
.-
<
\o
*""
c
C o '=
e
^n _2
S-
U
co
cu
£
CO
-5
*"»
cj
-
03
ri
««
H
* SPbJ
(U
r-
5£
-°
s
2
o O
S —B
la o- 5
— 2 e 2
c -S
13
5 u o 5.2
° a >
*c H o g « u .—
op
S.g
oj
N O \D ^ O
*i m oo © cN cs
pi k * \d ri
m
o
r- o
©'©'©'
oo
rj-
" on
O © no
© © — © O
ov
r-~
cn
NO
CJ
oo
"3
o
03
1
S
ON
"-
rn
fN
in
»n
Sn
-T
vd
-a
.2
3
S3
§•
a.
q o. 3
2 _-•&
> 3 CO O
S3
co
D- Q.
o S
S O
-^
CO
—
Sw
E
3
H
E cf a
.£
CO
—
|'S»
73 "O .3 —
O
o
2
^'
CJ
u
>.
3=
^3
CO
mis'n.myo.mn
ri N o N
2-*
o
Z
12
CO
^
S- D..3
.2 :§ cT
_
2
.a
t &
X
-Si
a.
3
U
—
=
u
'-'
-J
u
5\ ,0
b
(0
CO
Q
"5
—
X
-
o
CN
X
2
o u
B §
—
X
c
o
O
—
rs
X
x
X
J
.s
o
O
P
c
5 -b
en
2
S
t/1
-i^
u
CO
1c
CO
E
—
to
Cl.
=-
c0
o
=
a
—
U
U
o
3
o
co
CJ
3
a
t—
a
a
z
u
u
9
o
a
o
U
.2CJ
'3
CJ
CJ
-a
c
«
r-
-a
NJ
'5 C'
ON
O
3 -3 ON
" S 3
1)
D. "O
=
=3
S
t3
co
"O
u
2 3
-0
1—
O
z S
co
'7"
CJ
si
cj
CO
CO
b
CO
o
O
o
q. d. _e
'
CJ
=
1=9
CO
m
CJ
co
o3
-3
Ou
E
—
_
co
^ a S300'S_3-S
o
-°
ti S Cm 3 J2
O -3 O TO
3 u
S3 £ b b —s 1 1o =
IS
j=
cj
CO
>
r.
'
s
3
n
s
CJ
&
u
L_
—
b
B,
"«
c
u
[A
ON
ON
'-a
to
3 — B .. 8
£
3 3
" W tE E
.8 £
I •a
U 3 *"
E ,£ •=, a
2 >>
S
ON
ON
CO
> 3 CD —
E-g-E
gffl C 2. & °
.-b
15 IS
.s
3
O
CO
Sgb m^-a3;
« co "O
c a S
c
=
S
<3
'
n
ND
<
«5
>,
U g
crt
£P oo
.3 g
D.
0-
-Si
to
CO
CO
a u 3
CO
co
w
D.
a "
ti pL ,8 ""
ON os <*- "",
O- ON °
"«
•S -3
q
fN
.t:
2
a
I
3C
©
a
2
cj
CO
CO
CJ
CO
~
00
" 3
bflCn
e e a TJ'C
CO
O 3
^So
3:
O
O
"
1
o°
ON
0Xh£
"9
CJ
>*
-a
I
o -
£ ^
°°
2 « ^
2
i?S
fl?
B.S-1'
§ o
5 "t
c*
<~~
VD
ON
>*
S
3
.s
CJ
en
3
ca
cj
o
o
n
-S
i_
Q
5
.S3
U
en
C
< Q
O
OJ
TJ
_ „
*
c
^ _a
3 u c
—
g .a
r~-
»-J
3
O —
-1-
«/n
5
n
C!
CI
m
en
O
ON
O
r*1
in
—
*
53
— o —
o
,
>
-a i
.5
-
o On—
.—.
(N 00
O
_j
£S £
/">
-
5-
ca
u
O
O
o_
—
2 *
<
O
V)
(N
(N
"i
1
.
«n
H ^ n 2
f*l
r-i
i
o
5 o
IS
OOOOO
C-
CO
S
—
> u a
>>
cj
CO
CJ
U
-2 -a
.52
-
'&
ca
V3
—
~0
to
-^
-Si
X
X
o
c
O
O
eN
rs
X
X,
—
CJ
-s-
D>
0-
—
x
00
ON
CS
3>
—
c
™
CO
a
CN
x
J=
0)
E
ca
in
Z
o
o
ts
a
X
a.
a,
3
O
c
ca
C
-a <n •-
ca
^
2
-3
B o
<u
c
co. S
S
CO
U
Si
g
O
4
ov
u u gD. JS °
co
5= §
> ^ E
co
C
V.
,_
cS
>>
= E
>,
00
M
a. to
u
.
B o
00 "»
ca on
•
A
-=
>
> E
£
a
-3
.£•
u
c
a
'J
CO
ca
•
c
—
e
;e.n|
en
'o
" U
ca
cj
5
c
_
53
ca
n.
^
n
ca
5
<u
ca
T3
ca
o>
VI
.2
>
:•=
e
5
=
^»3
_2
E a
y
fc
10
O
s =
It£-
s O
r O
—
S3
illSfc
0> —
—
^
e u
»
ca
c2s||
n.
3
O
ca
ffl
ca
j=
^.
Bm
cS
c
S
e
"^
"S
.
•
3
ON
-0
c
'-
>
—
u
^3
ON
•CNO
t?
o —
ca
c c •=
c > p 13
ra
3 co co
CJ
"
— L3 TO
CO
-^ .; o ' o
O uQ u^ CJc
_>
"
ca
c -a
•i •ca
a ••- ^ —
co
3
u £ u —
CL £ ia
>
.„<+•
to
cj
CJ
1-
S
CJ
S
ca
U
(U
ca
on .2
ca
CO
a.
e
,
o
"2 «a '§
u
E
1*
=
.2
C
u
u C
S3
CJ
-3
a
00
i
ca
O
•
xi
to
0)
co
CJ
—
J S3 TO
3
w_
en
ca
—
ca
^-C
CO
r^
.9-
,
cj
A" -5
_
'-3
-J
-O
la
=5
-£
IS
TO
3 O
0-
a
O X cj
O u o
c
>>
O
o
=
u
=
—
fe
O
.2
ca
zj
<; -a
ca
1-
CN
^
-S
ON "
~
a
.s
n>
J
O
?&—
*—
CO
v.
-s
T3
D,
t_
ON
ON
J
p
5
*.£
ca
3o-
O
gj
XL
Cl Q.
c u g .ST3
O .2
TO
-r
"O
3 u
B..tS
"•*
_V5
OJ
ca
ca
>
•=
—
jj.se
Si
5
>
CJ
£
ca
-2
B H
S Q.:=
Q
u
—n. •=> >^cS
5,'i
•2 o S
"a =
d.
c "2 «S 3 -r
e —u s u c S
on
Cm E
o - U
o c
e
on
:
a c
S,
e u
o x
°.oPo°.oPo
9 o 9 o o .9 o
<
bo
.5
ca
1)
OrjO^oo^-tN^-
— no
m 'n
o no
o
=> o
-
—
£a
2
e-3
S
cn
•= on
..= c
no
TO
ca
o
o
~
3
c
S
ca
b!
2
.
'£
I
2 a
"°-§
c
-°
—
s£
'
UH
o
o
•=
-1
ON
3 E $
-
ri
-a-
£
=
1
OJ
,
inorrc^^oN — P
— — "7 H ~T
<V H
CD
E 3 E —
o E > 3 S
a
u ° O
to
w
C
JS
cj
'
E
3
t>
iri
<
—
«*
SO
—
OJ
a
1
o
n.
2
C 5-^
O C
u c
U CL £
s
je *(
&
u.
«
^
—
3"5i
« ~
cj
dj
& u
cir
>^i
<
U
0-
q
u <
00
.£ -3
>
e/-j
CJ
-a
e
c-
•
Sco
73
O
CL CS j3
—
[~-r~-^•->,>^c_J;cgl-ool/^^-lCN.NoHcN
a.
ca
Ml a
•*
•
J2
C
sS
3
CJ
_
<*-
U
J
u
S
t:
c°
en
cj
CJ
C
3 c «
«
c >
CJ —
3 E
U
ra
u
3
£ a ^
d_
3
n.
to
cn
^ aj c
CJ
" 3 c £
Is
'—
._
Q
K S
o. -a
D.
"
la
CL
o
o
_
CN
(N
co
a.
oo
os
J2
ca
Ov
co
H7 c
AJN
ca
<
o
°
oo
On
_J
^vl
CN
>.
?n
15
_:
ON
00
£
O
10
ON
On
sn
OO
on
CN
r~
000
£
a
CO
3
3
SO
—
_:
ON
00
H -
S<
O O0
t
O S M "1
cn — P -i — -i
OO
On
st
o
o
CN
=?
-
to
Cl>
^:
CJ
U
OssooocNmmCNist
— c>sO — tONtsorN
— © © —
— o —
<
os
as
to
on
4»
NNtflNntNoopToo
©mcNsnmoo^-:o©
>.
-C
—
i
ON
ON
o
a
o
[8
<
U
—
00
.sa
(N
ci
st
so
sr
sn
— © —
©
os
r~
st
r-
t-
g —
os
° —
ON
ON
O0
st
***
u
o
u J2
s
s
£
<
Tj
<
o
o
co
fj ©
«
S
9 in
cj
CO
so
-c
cj
£
u
<
oo
on
—
60
.5
o
1 8
3
0.
J.
OS
>>
E
£
3
st
°°.
'V
^m C
©
cn
so
r©'
°°
CO
—
r,
r
J3 CT
OS
H v
-'
r<->
CJ
73
-T 7D
^.2O
O
n ?
2
9 H
ea
£
en
SO
"">
"?
U
~
CJ
OX)
.S3
^2
o
ca
CO
cj
to
.52
S
JS
55
-Q o
t 3 "
3
« 1 3
3
CL)
.8
>N
cn
5 5
ca
3
© on m 00 on tn r^
^MrNir^t^inONm
P — o — sr — © —
£
rN)
rs
:
r*i
oc
-T
—
rNi
—
to
on
rn
ON
St
Q
J
.O
a
00
—
CO
_
<
:
CO
r^
On
tN
o —
'
CN
"Sf
in sn
st
-i
to
1
E
g
st
;
SO
t-~
r^
CN
CJ
^N
— sooo©©
~ On — st (N
in
C*1
© o © ©
—
-T
©
Os
.3
NO
Q.
o
00
—
-y
—
st
M^
O
-f
O
0N
—
X
OO
On
On
—
X
CN
©
©
CN
X
©
©
(N
X
St
OS
OS
OO
Os
ON
y
—
CU
a.
ca
C3
0.
b
£
CO
Lh
ca
c
CB
0-
to
b
M
On
ca
E
=
ca
E
ca
u
-
E
k_
ca
Ui
ca
ca
13
n.
PU
B.
CL
'Z
ca
ca
im
to
O
©
(N
/.
SO
©
©
x
C-)
on
MO
&'",'>
a.3: ~^ 5
CO
s
2
ca
T!
SJ
3
od
S
'o
u
3
3
o
>
—
o —
855-5
c c
3 3
'
cj
cj
g,
"
^ M-^«
ii 73
3
—
tH
3
-—
*-
2
«
-3
CJ
CJ
-
to
"
"3
co
c£ 73
-
5
.3
iZ
ca
>> 73
a
-71
C
§ c°-
^>,
CO
C
2
L-
D.
-^
3
«
3
- SO
x 2?
On
CJ
73
.8
—
3
is 13
3
£?
^
j-
'P."
CO
ON
CJ
CJ
CJ
3 2
Cu
S
"
•
CL
to
CO
CJ
CJ
CO
=
3 <
u.
cj
cj
S|&-(
E 73
a, c~ u «
o "
E
5
ca
CO
CJ
<*-
to
o
a s
2 S —CO3
O CJ
>
to
CD
o
—
.2.0
o
w
~1 83
a
—
V)
CO
,—
S
1
to
CJ
ca
a
w
en
73
CJ
'^
E
2 1.2
5
o
.3 I
'E 3
U
—
fc
'J
r)
-3
llll
E ,| "
CJ
D rE 3
SO
g
On
m
3
c -t
to
3
<
—
3
E
5b
"
© b ^
cj
CN
3> g
•- H
'C -3
CJ
to.
© —
to
.2
"
to
u 3
m
© —
—
t
—
ca
O
o
•
D. 15 ,° jn, 3
CO «M
O
-3 O to o
•S U S 60
© —
m
'
"^ 33
3 £ 3
Nt
— —
'
u
Op -o
22
£
—
t-
M
_
1-
^2
*
Si
o
o
(N
E u
S
s
ft
o.
l
3
'
to
OS
OS
c2
aO
S
-
b
> Q
>>
bd
>
1
•— ON
m
m so m
— © —
in
to
I
Q.
to
J."
03
u S
cj
so
fc
'
13
cu
CO
SO
-o
CO
In O CO
u o
°° fN« LO
£8°
c
tN
o
1
I
Os
OS
P © © ©
ON
ON
OO
st
3
S
(N
CN
J2
=
CO
so
0000
sf
6S
S B
3
M to
3 a. 00
1c >o .S C CJ„
co = § & CJto 73-3
o
> CJ c
Q. CJ
a = OS e
^
os _E CO
so
o
o
CN
3
.2
O M E
cj 'S
E
•- 3
>
J=
iJ2
(4-
1-
o
3 t*S
^
CJ
CO
3
I
05
a.
.j
ON
ON
00
CNl
CNl
o.
Ss
™
o
S
1CL
r
Q
at
a.
n
•
o
D.
to
ca
(L>
u
I-
5
— —» J=
*
to
Q.
OS
60
*
I
-a
t
u 3
so
0-
V
0.
g
l_
CO
u
rt
c >
M
fl E
5 3| -3
,0
3 n
au3
?
On
H *
CJ
o
o
^_
<N O
_
ON
Cj
« S.1
E
3
b
a Q
>
vo
BLi
—
— S
<^ m oo
o _; o
H PK N
— »
On "~ ON "l —
^"
~"
o
60
.S
1
- b &c m
1
so
5
-%
CO
n
CJ
'
2_
u
-o
c
W ra
S
tu
_
r^
jj
_
c -r
= CJ
st
"—
.
ON cK
On CJ
(_
Cu
CJ
CO
CJ
5
On
O-
b
2 3O
3
o
CJ
5.
.ti
CL_3
.©
O
O 3
r~
a. D.
O
o
o
CN
o>
^>
so
rn
-*
2
in
to
fi
^)
<"1
—
«
<*i
•
-—
T —
^f
•o
1
'
rs
"^
"?
S
° ~
r- f- <N
— — (N
V)
B
o
>. ^'
>
o
Z
u
ca
en
a
CO
U
3
O
O
<N
2
E
H>M
1 a
rs
'j
(S
o
s
^3
OS
in
en
t>
»
oo
>
00
in
m
£j
I
m
m
Cfl
•""
a a
D.
a
o
CJ
—5
lS
>,
eg
E E
'w
-3
-rs
rs
eg
'> m'
CO
q
eg
IS
- IS
CN
rs
«s>
I
oo
Os
CO
eg
'-I
to
Cfl
rs
d
a
•*«•
=
- 3 E
'-B
E E TD
c = s
rs
'
-rs
Q.
eg
O. n.
O 'o
I—
'
a
IS
tt-1
o
e
a.
CJ
rs
rs
eg
eg
r>
3>
rs.
CO
q
a
'S
rs
rs
eg
CO
u
E
o
I—
= rs
a
u E _rs
3
2 O
Q.O o
"J
E egu
rs
p
•n,
D
so
u,
eg
t
— n
— «
CN
O o
CN — MS —
i
OO
Os
3*
—
^J"
f*-
1/1
—
**
so
*r
;
eg
D.
CO
CO
-£
_<:
rs
T3
z_
1
a
^5
b
£>
2
ca
'
eg
-rs
CO
£—
r-J-.
Q. Q.
I
so
CN
-'
a
—
—
— — (N
F^ M -
3 | >i
E CO
a
O E
E
a
c
MS
(0
O
o
O
"O
—
<3
CO
O
U
l«
a.
b
2
CO
1
U
u a
bo a
CO
3
©
<n
J a
a
—
3
O
b
E
Os
JS
<
:=
CN
lO
-a
rs
—
-r.
r—
rs
"rs
rs
5
CO
O
o
a
"E
CJ
U
>
'i
V
o 3
trt
'S
U
t_
o
.
OS
*°.
CN
E c
o
-rf
Z
a
D.
eg
i—
ea
eg
CJ
eg
i—
rs
m
—
rs
jj
£
-
em
r-
-*
-Si
a
^3
c S
2
.?
^.
eu
X
eg
U
cj
o
u
eg
e
en
u
"3
-
o
X
u
o
o
CN
X
o
rs
EA
eg
u.
u
eg
co
s>
[A
b
b
£3
CO
b
a
i
-J
3
S
u
rs
eg
co
eg
u
>
-rs
rs
cfl
eg
o
E
a
ba
E
B
Ui
E
a
Im
01
C3
C3
u
Oh
Oh
=
,9
I—
'-J
_rs
t/i
eg
a
U
W
3
J
c
'J
ao
S
eg
t/i
n.
i—
gj
i
U D
=.
-
z
-C
a
z
u
3
i
1~-
C^
eg
CO
-
J3
CO
O
eg
Z
61
C
i—
o
OH
CJ
>S
eg
—
rs
rs
eg
rs
r>i
t,
rs
^
i3
D.
3-
O
—
2
rs
C
is
>
u
s.
o
.o
-^.
E
E
rs
-3
o
3
rs
i—
E.
eg
SI,
rs
c
j
1
"E
_2
^s
-r
r>
j -~
m
CO
— —3
rs
•
rs
eg
CO
£
CO
eg
CJ
"E
-J
o
a
cS
s.
cfl
u_
-rs
E-
£ a
Ss
eg
'is
2
rs
rr-
eg
>
-
eg
rs
'=
CO
-C
t
00
2
-3
5
eg
rs_
E
rs
tN CN
c
O
--
rs
'j
1—
D.
MO
r>
CO
-J
<L>
w
u T3
c
_^
CO
c
—
u
-E
CJ
i—
s=
eg
a3c 3 O en
O
a —
CJ
ij
rs_
o
eg
eg
—
o-
rs
*—
o
eg
i/-i
S
n
C3>
rs
sz
—
D 5EL
™ tn
r-
u a u
"2
2
—
rs
eg
—
rs
l—
eg
l_
0-
—
J—
CN
rl
(N
73
-5
c
c
CJ
CO
ss
o
s
rs
— a
rs
is
CJ
D.
a
in"
1
b
CC
r_>
n
O
3
tH
o
CO
SJ
3
-J
rs
u E
> rs.
rs
O
n -3
a
rs.
3
3 -rs
a
U
eg
5
ST.
u
.eg
2
£ a
E
CO
bo
~ —
>.
rs.
c_
-
rs
s»
rs-
E
3
CD
CJ
rs
eg
i—
'"
rrs
U
eg
eg
CO
5 <
n
eg
-3
9J
"rs
3
i-
"is
CO
^
n
-S3
14—
~
X
-J
i-
eg
rs
c
o>
o
c2
E
in
a.
rs
m
_rs
E
C c
u a
i
-5
rs
co
eg
-J
IT
o
ea
u
r
si
eg
B
JS
m
rs
£
a E
—s
E
<L>
rs
-3 .s
CI
'J
ir.
CO
w
3
eg
>
J3
CO
•5
J3
ea
0.
>.
rs
srj
3 -
c CO
a U
l—
_;
rs
OO
u — -
rs
CJ
k-i
-<
eg
_rs
eg
D.
(U
~
-r
rr
CO
eg
u c
o
e u_
CO
C/g
- — as
—
Z
r-
—
H5
eg
en
<a-
rs
«n"
eg
-rs
0>
-rs
Cl
rs
3 u
'
r~~
;
l/;
rr*
LSS
•s
eg
a bH
M i
a -*T
i—
c
M - *~
a —
oo
Os
Os
eg
a
> s
2 a
a
60
eg
rs
E
o
o
o
CN
CJ
DO
rS
s
5
o
E u u G
-3
—
J
c
CO
o
o
CN
1—
u
rs
E
3
O
u
rr.
2
-3
rs
—u
o
rs
>
3S
rs
CS
o
a
c
a
a
^o
u
X.
CO
<3
a
00
•*~
2
c
ft)
.o
-o
o
.c
1)
GO
tS
^
ft)
CO
^
«9
ft)
<U
o
£
£
.o
1
2
ts
-^,
^>i
^
.U
u
^
Q
CN
r»
-t
odd
U1
^ d
d
^
M
,<~
^
*j
S
S
CJ
U
in
m no
o — o
On
oo
no
r~
[2
r»
^ d
a
CA
CA
oo
s.
Tt
r-
°°.
10
d d 9 d
d
c
CA
o
C
S;
2
a
g
a
— tN f£ <N rf m
•* i— "J — r^ —
o o ? o o o
ft>
"«
3
a
ft)
ft)
ft>
co
OS
a;
ft.
« 'ca
= CA
U
-o
^ d
u
Oi)
c
03
o
° o
3
— m
— f~
<n " o
^ O
o o o o 9 d 9 d
(N
o oo
o m
o
o o -t
o On
On
o=
oo
odd
:
m m r» ~ r) 2; no
o
cN — o o H — °. o
d d d d 9 d 9 d
'
u
1/1
ft)
5d
c
— d
I
B
o
u
n
— o
°. o
o
g
01
fti
b
o
5 S o o no
o m
o
u
odd
u
m
1/3
CA
ra
C
p
£
B E
i7
3
— o
ft>
O
o
u-
cy-j
U
>
s
Q
u
u
t-~
'5
u
C
~
ro
-:d9d9d9d
u
* d
B.
CO
B
c^
.a
,g
o
m n
r^
m
— d 9 d
oo
I
u
CO
r)
rf-
2
* d
NO
l/
!
c
~
_
CA
CA
CA
u
n
—
o <n
"—J
re
i—
re
3cd
B
ay
>
NO
* d
r5
c
c
5
c
>
l*-
cS
u
C
eg
CA
t3
ha
<
LTj
-J
BJ
ft)
is.
a
CO
B
u
03
=3
So
OS
-J
3
krt
ai
Qtj
re
JB
|
c
3 3
OJ
^1
re
a
Cl
re ^5
J3 J3 u
3 u
bfi
-3
CA
O
O
O
CJ
-3
is
CA,
re
5
i—
u
1—
CJ
re
£p a.
—
CA
ex
re
CJ
re
C
'-j
CA
M
'S
VI
CA
o
3
'«
c E
a
CJ
OJ
J3
jo
cm
CJ
ro
.5
B
Cm
re
0)
£
u o
ex
fi
re
w
1)
ay
2
eg
>
C
OJ
>N
CJ
O
U L>
3
u
43
re
o u o
3 B
a
5
£
-O
(J
B
en
OJ
re
_
o
re
i—
re
>
—
o.
re
re
u
re
CA
U
-3
c-
|
a>
3
a.
E E
re
— d Tl
a u £
-Li
-J
a
2
u B
o
S
"
CA
o
u
re
re
u
a,
0)
CJ
?
^
CA
3 u
u
3 U E
re
E
b
re
5
OJ
n o
B
o
D.
c
Tn u
CA
1)
jB
1
>
re
re
P.
B
liH
"3
o
0)
.3
U
c
O Q
c
c Ed
O
re
re
u
j=
s
o o
c
ir,
JB
CA
3
u
o >
a aID aIDa
a3
o
re
CA
3
BB
3 a
B 3
re
X
'
CA
CA
M
o3
3
1—
o
>
O
U u
3 M
CJ
o
CJ
*M Cm
u-
£
O i~
o
_re
re
a)
B.
<L)
CA
o.
fc
3 c
u
JO a.
c u
C/3
u
3
CJ
CA
C8
*CA
rt
D. Cm
CA
CJ
CA
C
(L)
o
0)
_J
CJ
B E
3 3
a
o
E
CA
1m
o 3
u
B
3
3 g
CA
CA
-J
o
CA
av
U
> d
3
«^
Xre a*« ot- reCJ re3
CA
c E O
'Si
aB
U
c
u
Q
OJ
1-
CJ
OJ
-T)
s
CJ
CA
> "5
5
a
^re
OJ
jB
I)
u
t/i
o
O
re
O T3
C
O o
CA
-D
-a
-£ ri
Cm
re
re
CA
<
3
u B
re
J3
!_.
^C
i-'
5
D.
>
u
aj
OJ
—
c
— On
O
O On
9 d d d
~":
3
o
cm
O
o
r)
>
>
:>
u
in
-~
3
re
^ u o
cr,
U o
5
o
CJ
-a
Cm
i)
o
c
c
ft)
"a
—
IB
u £
ja
ca
C
u
u C g o
OJ
u.
o E
$ 1—
03
B
a
u
U
* d
CJ
J3
O
•->
c/n
'33
m M >n n
5> * o
o o o o d H P
re
aj
Cm
o
CA
aj
0)
'5j
o
>
u
1—
C
o
~
f
— S
°. C:
O (N a-o
O O o
T3
D.
>.
re
U 3 <*u 3 to
c o
E o
0)
a,
E U
O u CA Bo u
CA
|M
V 3 3
CA
B
]
r-'i
nU
jS
o
u
r
ac
CA
OO T3
cn.
.g
>
u
u
c
o
C
ait
-3
5
b
u
Ul
3
—
a.
o
a sOJ V M
re
B =
B
—R
o
CA
—
CA
XI
u U
u
c
>o
£
o
>
oij
u
CN|
u
o
— d
^o
03
B
a
u
'qj
CO
ra
re
en
CA
P
u
o
•J
^g ^S
'5
SP
o
u
"o
5«
^
OJ
1*4
C
™"
o
c a
B c:
> c c
o S
B
r>
-a
•3
0)
a
p
OJ
'S
o
u
c
CA
"J
<5J
.g
CJ
5
u-
m
CO
03
r-
CD
>
o
c
o
=
ft>
-^;
ft)
CO
M
n-i
•*».
3
5!
m
o-
i<i
60
"a
o
O
oo
a
CJ
ft)
<t S * n
m —
^j — m —
o d "odd
s
Si
•^
CA
O
ft)
•2
e
au
"re
'a
(1)
"5 J
re
CA
3 ?7
3
3
CN
b
'-
1
en
w
Q
U
CO
O
oo
cj
n
—
V)
I
00
— * n
•J
P
U
U
oo
JJ
>
NO
o
=
—
!o
M5
c
^ n — q O
N
5
NO o
o
U
^
E
9
2
—
d
CJ
U £
O
CN
E
3
CN
ca
on
m
NO
o (N o
o o o o
ts
© a
en
CD
>-
CD ZZ
*
o
to O
a.
.a
o
a
en
2
CN
3
O
o
Ml
Cfl
CO
a
u
U
m — ~ ©
©
©
©o©©q©©o©'o
m
JQ
cj
•—
cn
cn
i
CD
o
•*
CJ
S3
E
TJ
eg
b9
CD
CD
C
cd
i_
"3
S
_o
p.
CD
u
JZi
o
—
D.
m N
n
o N
o o
o
o o © o
~
CO
o
£
CN
'-I
to
CO
en
OJ
Sh
ctt
CD
C
-J
-J
ri
o—
J
c
3
E
U
JO
•£
r*1
S>
l«
r-
t°
en
CD
-z
X
c
to
—CD
zr.
c
5
T
e
s
CN
^=
a
8-
c
E/j
ba
u-
•J
—
U
—
—
CJ
5r>
w
a E 2 I
2s
CD
en
OO
On
O
o —
si
C —
g
I
o.
a
^>
g
4
n^
la
ca
I-
—
^
2
a
cu -a
o.
ao
o
D
o o
?
'£
-'
t/i
~
~
a
1
U
en
CD
en
CD
D.
CD
1 3
J
C
3
CJ
CD
q
—
O
~
tt
CJ
^
c
Crt
q o
9 P.
o
<
ON
~
'
:
3
.-2
OS
CD
.
!
£
On
ON
_
a
>>
JO ST
os
« 7 ^
— - <
¥
jS
a
c
s
,§sT
— noO^CNiz-jCN
OOnO«00_;NO©©©©
d ° ci 7 c. © © © o
On00)£>
'-.
en
en
o
U
CJ
CO
<n
°^
— —
NO o
NO
CD
ca
CM
o
o
vO
en
.2
o
CI
n o o
© d
O 3 o a
en
CD
>
OO
o
o
2
U
— n
r^r^- — ;2cNcnrN©o
o * t in n — © —
—
oooo<9ooooo
.j Tj
i>
cj
(-
i-o
Cu
-T
OS
<n
cd
u
ON
w
u
p
g
o
M
on __ cn
m
cn on © cn
— oo
— q ©
mrr
o q — * no
©'©q©©©©©'q©
Oo
on
CN
,o
rj
»*-
*-
cd
ui
js
**
a
— "*mTf —
b
n
en
CD
>
—
ON
U
o^oqm^
©©©'©©'©©©
^j-oofNoo^mcNr^
—
—
M.
en
3
CN
c
cu
"-i
1)
S3
<N
CD
ON
CN
CN
o
re
u
re
00
a
7.
CD
35
u
p
-
CD
>
CD
cB
<C
s
rS
D
>
CO
M
—
cd
a,
CN
p
—
eu
o
§
CN
i
^"
CJ
CD
en
CD
CD
CJ
—
b
CD
en
2
a
U
c
g
i—
-
a
O
—
co
--j
eg
3
Oh
o
^
—
o
c
e a
—
c
•—
O
-a
e2
^
_5
-
i;
SD. «
ca
'S.
O
e^
L_
tj_J
a
"^
°
-=
O
"2
a
ca
5
o
a
1
>.
~^
2
e|
cu
—
i-
1
-a
u
-a
5
r^
i2
c
ca
^
—
a
-.
5
^
-o
CD
O
> C
_T"
y
u
j
CD
a
j=
"
3
ca
E
T3
—
a
CJ
CD
CD
js
ca
a
H
o
ca
>
CD
CB
JS
cj
en
'ej
CJ
en
cB
p
a
en
.3
5
o
cj
o
U
3
—
—
CD
>, -a
di
5
~5
o
JS
-s CN
00
E
c
Q-
;>
a
T3
3>
or
ca
5
3
CD
3
—
°
— O
™
en
CD
-j;
ca
c
o
°
•=
—
cd
en
CD
Cj
2
=
S P.
C3
u
a
2
JO
CD
CN
-i
OS
E
=
cd
rs
^^
CD
•^
c
CD
<>
CD
ca
c
a.
ca
P
-
'
CD
a
-a
c_j
—
2 — o
eg
CD
D
.s
O
>
Im
^
CQ
CD
CB
rS
8
O
o
cn
CQ
a
a^
ca
.a
•-J
t"
—
s
^ ?,
o
q
= o
CN CN
C
3
£
C
U
en
ca
en
*CD
i—
-J
~3
en
CN
a
~
CD
u
cj
X
u
I—
«N
5)
Q
1
0.
i
"3
5
£a
en
a
c >
n
_.
2
CD
5
3
00
S^
°^
0)
CN
i
>
ca
ca
.2
O
o
r
-a
'«>
O.
o
c
en
"O
o
CJ
>
O
JS
3
00
oj
*^
&
ca
CJ
C
C
J=
«J
~i
ci
C
n
CN
t<N
oo
"
^-C
-r^
-a
J3
.t3
O
CO
2
!?
3a —_
>
CB
-t-
oq
©qq'©'©©'oo
G
CO
cj
en
CD
•
i
—
r-
—
:z
r
i
0/j
U
u
CJ
o
u
r<-ltN0O
Zt:
-
CD
Jl
<n
1m
i/ia-NC^O;—
~r.
ca
—
£i.
CD
b
o
n
CD
*—
,
as
3
n.
CN
cu
_,
ca
a.
Q.
-t
-N
_>> •£
Ih
O
00 O (N
<* — o q
© o © ©
—
cz.
c
ca
2
o
°
Ui
-i
en
.a
—
C/3
1/1
=
D.
O
3
>
I
So
•«•
oo
»
c;
fi
tn
IN
o
re
<
M
O
_
a
-S
—o H
E
>.
00
'
tN
CD
Jo
o.
c_
—
- 2
CD
B
^< o
Tl
%
'0
E
;
>
O. JS
CD
CJ
ON NO
— 00
— ^
< .3 —
en
3
c
1
CA
CA
2
O
cj
U
so
ON
vo
B
,
CA
.
so^t-^^o^c^im — c^o
6096^-609°
.2
E
U
m
o ON
— S
o
m CN
6 6 "9 6
-
m
00
o
03
u
GO
e
1
CA
CA
o
Ui
.Q
T3
Tj
o
u
r \
(J
wn
en
m
CN O cN O
6 6 6 6
**o
u
JS
50
n.
.S3
a
e
1
CA
CA
mON0N'd-9i^
- O N - N q
M
.2
u~|Vj:>u"
CN
O
O
6666966666
o
U
r<">
(N
>
50
E
1
CA
CA
O
.2
"5
-3
CD
u
au
in
ON
— T
o O O
CN
^D
CN
OOOO
.
.S
a.
c
CA
CA
o
I—
.sa
00
en
.2
u
50
CA
CA
,o
u
CA
CA
.2
i
on
2
U
—
vo
000
cn
vo
n-
on
cn
—
o
— 3;
9 —
6696
'
)9,
g\
O
CN
6666906699,
-->
50
B
O
o
CN
'rt
ON
ro
OO
— 9 O
6696
—
GO
— O ON
° CN
» t
t - 2
q
<*~i
669060669°
o
U
CA
CA
S
U
i
1)
GO
cio'i'cNfi'oNor^
O
"
66906066
'J3
co
E
O
CA
CA
-rj
U
GO
o
rn
—
.
r^CNcNON^ON— —
B
Tj
wd
r~
00
5 m
6 6
o
U
U
^g
«i
,
u
o
^o
**
a
[NifNio^Jr^dddo
— o
—
O
66609666
3
<^.
<^
r<1
"*
CD
s
J=
GO
>
O
rl
O
O
CN
CO
CD
co
In
«O
CD
CO
3
E
CO
E
CO
-CA
CD
(m
CO
CO
£fi
D
6
CD
GO
CD
CD
>
a
>
a;
CN
CD
*->
CD
CA
CO
CO
cS
eu
CD
CD
m
CD
CD
CA
CD
CD
X
CD
l_
CO
go
E
0.
CO
CO
>
CO
CD
—
CO
Q
GO
CN
-9
kH
u
c
CD
CA
CD
a.
_3
£
CO
CD
-
u
c
o
N » n
n
— S
"1 — (N —
o 9 o o ©
u
m
ooPooopo
*0
<:
^D
oo
<*">
.
cn
c
Q
o
o
So
oo
**
—J
o
in
O
* <1 fS
m
in
o — o 9 o
oo
r-
jo
S
C/5
o
t_
£ 9
o
o- *
o
<^ 00
m 9 <n
p — o 9 o
M
~ — 2
— O
en
m D
o
o
o
o O 9
M3
>-<
'
<
.2
>>
o
O
=!
cd
Q.
fll
-j
O
—
II
o
-a
<S)
g g
>1
E =
S
cd
*
(L)
in
U
3S
8
5 g
"
-a
o c^
U O
"
«
c
9
> CH 5
*;
o
>
00
5 S°! ^
*
r»
O 9 O
2
C-
T3
>>=
O
u
u
S
is
—
X>
g
C C
il>
;
Tt
en
^
-©
—:©"7<=>— d9<=>
VD
(N
-3
,S
U o
**
hin~oooor<i.N
o
.2
U
:
P
—
co
o o
c
qj
M
-C
E
K
2
5
IS
ill
'J
a.
f
u
o =
B O
u
" S u
U
Cd.
u
-a
!L>
-S?
3
'-^
'j
S
=
r:
u
« c T5
o
2 «
.2
-j
te"
^ ©
DQ
o
f
C
|||
§
S
||s
- 5 °
'
r-i
o
H
S c s
nun
—
o
u
2
_•
§-§
~
—«
—:
^ w
£ 9
o
a.
.S
» u
u
jc
a
t^
£2
-a
-5
tS
cd
c o e
2
,
3 £ E
3 3
"« O
a u
a u
5
=o
n.
'""
_ed
cs
Cj
_3
j-
cd
o s .^ s
c ° **
u
>- —
—
w u
o
„
c
g o tS-9-Q c
o S vs
aj
<«
JS
—
-
p
111
- ^ - —
n fn
©
o
o o
o ©
oo
u
o
I-T
<N
^ S
d
!-<
SO
y:
§§
u
*-
CJ
O
c
=2'S
ii o g
•£«*W
o
'S3 i.
t-
CJ
C
CjJ
-8
5
•5
~
^
bo
(D
u
v)
,j2
.Si
t-l
r5
>
£
rz
Oh
-o
1)
CQ
c
cd
cd
IN
>%
o
CN
£
c
** -£
c3
eft
CJ
CJ
CJ
J
^J
—
D.
"3
E to U
CJ
o
—
u
'->
-o
tn
60
-s-
-
_
^_,
«
O-S
+L in
3|
^
3
£
65
CD
U
Q
i
3
Q
i—i
<
CO
2q
Cu
a
s
g
•a
'
—g
Cd
cd
I—
1—
cd
—
cd
==
5
3
W
<
O'
Q
u
—
r3
C
1
bd
aU
3
C/3
J3
O
l
si
g
C
u
>
gi
§?
K
—
,53
.2
CS
eft
u
3
cj &
CJ
3 § - 3
?
7 -3 «
-r-
3 c "i- .«
O cd CO %
c o >
60
*-
•a
>>
J=
o
*—
i
S l"
° -SO
u «
^
ES
II
o §
§
•£ T3
'
CO
u £
(L»
r-
eft
e°
cd
t/D
r-
I
__
CJ
ra
s
crt
01
t?>
1
CJ
Q.££
K
1
o
.
>>
n
- 'J
c 5
c u o
£.£ >
^
^>
So
(2
en
CJ
a. *-
a
3
b'^iS
s Cd
u
**
eft
-^
.O. CN
t-«
.
«
Ov
o
£
2
CJ
cj
•3
u
.
C
M
o
[^
x~~
>o
r-~
— vo
in as \D oo m
r4^rtN-*(N
ooNCNt-^t^inoovoooot^m
vp
cd
B
r->
-q-
0^vo^Oinincn'3-c'">r<">r<-)C-<->rNfNCNCN<N[NCN(N
> E
S?
—
cn — r~
m m
p-
L^l
<u
U
C
1)
'J
-
o
<u
o
ts
SP
S
^ '3& ^
'5
'3
a
w oW a—
•a
-a
u
'
a 1
& a
o
CD
DB
u
o
T3
s
-Odd
'
VI
•a
T3
.5
•a
S
0)
<D
u
R
<
.Si
o
u
u
43
*3
u
>
O
s
< < <
cn
cn
t5
fa
< <
<
C
g
on
•-
."3
to
g
B
S
R
S
J±
^o
rs
o
>
£
sR
9D.
S
C3
S2£
0)
i-
oo
S
T3
-j
o
<u
u
R
Q.
cn
cn
3
ed
S
a
>
Q
R
U
u
3
J=
«
in
t—
£>
a
-a
4)
-o
02
U
O
u
3
c
3
1
R
R
'J
o.
Pm
3
r3
J3
B
R
O
U
u
R
|J
D.
^"v
^ f)
U
cn
CD
cn
cn
<d
cn
1)
>%
>>
>1
>,
u
CD
CD
CD
1>
CD
u
B
CD
u
II
E
o
n
R
-a
B
R
=
R
R
-
Crt
CD
Cn
CD
cn
CD
cn
CD
C/1
U
CD
<U
CD
1)
>>
>-%
Wj
<D
>-.
a
>
>
R
00
R
O
Xi
u
o
R
JS
D.
cn
cy
Cw
R
-
u
w
o
e
u
c
c
,2
<u
a)
cd
i
S
rt
-
cq
B
c
O
b
Oh
aj
X
—B
^
£f
—
£3
®S
e
u
BQ
aj
O
t;
'5
u
3
j3
^5
.5
o
U
ja
s
5
3
W
y
E
u
R
o
c
cn
-a
o
—
e
H
a
Q.
o
«g
R
1)
T3
I
3
E
aE
R
E
1)
cn
-a
E
E
OJ
t/3
a,
&
E
R
r-
-
U
O
R
u
-—
ID
E
-*"
.a
-a
E
R
E
<
-
it:
oo
_
^ W <
a 2 Q o
oo
OJ
c
ry
OJ
2 y
O
_
—
O
_
<
0Q
OJ
"J
3
o
00
2 _ oo
o 2
o
_ - d
at uj
_
J
< < <
U
2
z
—
<
00
c
<
S
> <
_ u
<
OJ
£$
?»^3
Z
Q U
< <
_.
o
u
o
OJ
OJ
y
>
bj
oo
5
S
o-
o
0-.
-J
_E
o O
«;
OJ
oa
<
a
< J -7
<
E^ cn B
pa
S
§
-a
R
at
o
y:
1)
Cl
3
S.
D
p
_=
cm
B
00
O
E
o
R
B
u
>-
>
u
_o
—B
00
2
B
U
a,
u
3
T3
o
o
R
ON
o
°
<**
o
i
_
-
r-
o
C
-a
•-=,
2b
R
g
U
cn
5
5
2^—
Map
1:
Guerrilla Attacks,
1997-2005
SaniAndtes
l
Barrafiqtifllai,.
Ayanj&o^J
wl aadafdna
!
r
moo1ft
e—
v
I
I
r^Wichada;
v'Vaie.
Tolima"
-<[-
y
1
1
j"
IVIeia-
7^
"*0
^gg& MW^gu^pgo
f&pjwi
Caqtu?ta
i^mayo
Cr^PiuB
>-.
Attacks by the guerrilla
per 1.000 inhabitants
_ij (1 61,26.45]
[_
__
2
IJ
(.67,1.61]
(22,67]
(0..22]
[o.o]
r
s
Guatriia
z^
:
Map
2:
Paramilitary Attacks, 1997-2005
o
anjAndt
L^GuajIra
r
.^__
BarraHquilla
/"
.,-
.,.
AJafiticcx ...:'•]
">
,
(
4^Wlagdalena
„^&~y^ Cesar
/\ ^creP^ Lf'k
#1
* <^S \\ Bblivar¥ 'n
:
%:,
.
Cordova'^
)
.'
;,
Jpfe?
feJJ :/2N?- /\ "'Nte-DeStder
/-'{
\
RalitoV
£±{
,-iUJ
.-.'.
s
V
v'
i
fc
'-
Biicaratriafiga,_
H=rj
^Santander
>*
'
',
•
Medellin
Risaralda
,
*
'
'
/"'"Caldas
'
'
:
Antioquia
'.....
I
i
."'!
•A'^5
Qu d0
5
;
:
••.
.•
.
,
v:v
-
.-••
-o
H
.
.?
u " pl,namarca
Bogota
call
k~--~**H
Casanare
'•
*yy-[ .4
r
j.ri-s.
Arauca
,v>r
:.;-:-
:
—
Vichada
Meta
-:
^ ^S
W
^^C^
^"^7)
/; Narino'
^;^Vte
Caquet>
v.-l.._gutqmdyo
L
IrJ)
KI
(.19,3.62]
(.04, .19]
U
(0.04]
[0,0]
itary
~
XXlXpei^
\T^^T
Amazonas~'
Attacks by the parami
per 1.000 inhabitants
(
/
%
U
Map
Elections
3: Legislative
Winning Vote Share of
J
y
Clenaga
6.4K
(12.2,^8.6)
P.Viejo(Y6>5,5)
;
\
^.
(4.1,2.1)\
/
^-r<J
(12.3, 16.2)
(8.6, 11.4)
\-
—
7
(Congress)
2006
j.Sta,Marta (8.2, 24(2)
(7.3,
Campo
2002
&
Nuevo
Magdalena
Vives (Senate) and A.
L.
1998
Sitio
in
(2.5,
(1.2, .3)
-1 (6.2,17.3)\
S^-v\
8.1k
K.
(20.1,2.6)
(1.5.7, 8.6)
——"-4
Remolino
PS^v^^taca
(6.5,
1
y~^
Salamina'(3T9";'f5,2)
(0^(4.7,171.
jri)
_/
/(;;
Antonio (9.5,7.6);,
(-7,
Pedraza"(3,Z,.
i2.9V<^~j
-—.Chivolo
\
—->S
,
\^~J
'
J
'
'
';
"
'.
(1.2,43.1)^
:
(«w)
--v
V\
Ariguani (22,^1)
(.9,
,
—
/
'
,
(.9, .1)
.,
P>",
38-2JL-S"
,"(.5,'.)
"
vi.1
I
Banco
ancc
V... t --r'"'' 7
(20.9, 19.6)]
(18.2,
(10.8,2.8)
9_2)
(16.9, 5.4)
i
-s47 8.9, 74-6),
v
^^
"
to the
.9)\
V
~"\.{78.9, 49.7^
Numbers
/f
(20.2, 18.3):
.
(3.1, .3)
^Guarnal(10.8,
W&J-*^
\
.(1.8, .2)
Plato (7.4, 7.2)
VS
A
1
/
14.1)
1-8,51.4)
','•.
X 1,-4)
/
V
1?f"
"(5-'8..7-4V-
,
Y
'
(27.9, 1)1.5)
C
Tenerife (20.8>2.7)
\
:
'
.2, .1)
/
!
[(.9,31.2P^fk.,../(( 7 ,307)
r(
8 155)
0)
(.5,-5)
V
S.
-£
/{
are vote shares of (L. Vives, ACampo) & Colors Correpond
Vote Share of Winning
Winnina Candidate in Municipality
Munidnalitv m
t
~-''S„
Sflare
d Wmnmg candidate)
p
j";™"
(40.60]
4
[0.20]
B(20.40j
^-^p
'
*f>