OBSERVARE
Universidade Autónoma de Lisboa
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
!
QUANTITATIVE DETERMINANTS OF THE FARC-EP GUERRILLA VIOLENCE
IN COLOMBIA
This work was supported by the Observatory of Human Rights and International Humanitarian
Law (ODHDIH by its abbreviation in Spanish) of the Vice-Presidency of the Republic of Colombia
Jerónimo Ríos Sierra
jriossie@universidadean.edu.co
Associate Professor at Universidad EAN (Colombia) and advisor at Ibero American
States Organization. Corresponding Author
Camilo Vargas
cvargasw@universidadean.edu.co
Associate Professor at Universidad EAN (Colombia)
Paula Bula
pbulagal@universidadean.edu.co
Associate Professor at Universidad EAN (Colombia)
Amalia Novoa Hoyos
anovoah@universidadean.edu.co
Associate Professor at Universidad EAN (Colombia)
Abstract
The purpose of the following study is to explain the activism of the Revolutionary Armed
Forces of Colombia People's Army (FARC-EP) within the framework of the Colombian
internal armed conflict between 2002 and 2012. In addition to being the time of the greater
armed intensity, the investigation tries to explain the impact of different social, economic or
institutional variables that, from a statistical exercise with regressions, show how it is
possible to find foundation in understanding why the activism of this guerrilla responds in a
greater manner to some scenarios than to others. Drawing on a multivariate exercise with
institutional sources, a comprehensive exercise on guerrilla violence in Colombia is
conducted, which escapes from unidirectional explanations and juxtaposes different
variables in order to seek for an answer, with greater complexity, to what and how the logic
of the guerrilla activism from the FARC-EP has been understood during the last decade and
a half.
Keywords
Colombian Armed, Conflict, Determinats of Violence, FARC-EP, Political Violence.
How to cite this article
Ríos-Sierra, Jerónimo; Vargas, Camilo; Bula, Paula; Novoa Hoyos, Amalia (2018).
"Quantitative determinants of the FARC-EP guerrilla violence in Colombia". JANUS.NET e-
journal of International Relations, Vol. 9, Nº. 1, May-October 2018. Consulted [online] on
the date of last consultation, DOI: https://doi.org/10.26619/1647-7251.9.1.9
Article received on November 15, 2017 and accepted for publication on February 1,
2018
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
138
!
QUANTITATIVE DETERMINANTS OF THE FARC-EP GUERRILLA VIOLENCE IN
COLOMBIA
This work was supported by the Observatory of Human Rights and International Humanitarian
Law (ODHDIH by its abbreviation in Spanish) of the Vice-Presidency of the Republic of Colombia
Jerónimo Ríos Sierra
Camilo Vargas
Paula Bula
Amalia Novoa Hoyos
1. Introduction
This work presents, from a strictly quantitative approach, the analysis of some
variables that have determined guerrilla violence in Colombia during the last decade.
This, with the aim of identifying not only the patterns of direct violence exercised by
the FARC-EP, but also to understand why violence happens to a greater or lesser
degree, and according to which contexts.
The idea, therefore, is to analyze some of the variables that, considering the prolific
literature on the internal armed conflict in Colombia, have traditionally been identified
as those with the greatest influence to generate guerrilla activism and presence. So,
socioeconomic, institutional, geographical, and other variables will be aggregated, same
that in the end will allow us to understand why the internal armed conflict ended up
becoming territorial in certain regions of the country, to such an extent that the military
solution in favor of the State became impossible, and ended up making necessary to
overcome the conflict through peaceful means of conflict resolution, culminating in the
signing of the Peace Agreement between the guerrilla and the Colombian Government
on November 24, 2016.
Based on the above, the work is organized around four parts. The first one, identifies
the most relevant literature that has studied the internal armed conflict from different
angles and perspectives, paying special attention to some of the factors that are
identified as explanatory of the violence in Colombia, and that also construct a
theoretical approach that understands that the violence generated from the armed
conflict in Colombia responds to objective circumstances, especially of a socio-economic
nature (Sánchez, 2009). Then, the details of the methodological analysis with which the
econometric model, description of variables, their operationalization, as well as the
origin and sources of information are explained. In the third part, the analysis and
results that would explain how and why the guerrilla violence has occurred in the last
years are presented; to finally, as a corollary, present possible lines of research based
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
139
!
on this work which, in particular, allow to identify not only the most vulnerable
scenarios after the armed violence in Colombia, but also an eventual framework of the
post-conflict such as the present one.
2. Status of the issue and theoretical framework
The Colombian armed conflict has been one of the most investigated in Latin America,
given its longevity of more than half a century, so there are a wide variety of
approaches that have tried to explain it and address it. Thus, among many others,
historical studies, military studies, geographic-political approaches to violence, as well
as strictly economic approaches could be highlighted as major fields of study.
From the historical contributions of armed violence, the works of Pécaut (2008),
Aguilera (2010), and Pizarro (2011) about the FARC-EP become mandatory references;
just as in the case of the ELN, the investigations by Medina (1996), or Hernández
(2006) stand out; and in relation to paramilitarism, the publications of Medina (1990),
Romero (2003), Duncan (2006), or Ronderos (2014), among others. All of them deal
with the origins and evolution of armed groups that have been protagonists in the
internal armed conflict, taking into account not only the root causes of their
appearance, but also the organizational, economic and violence factors that determined
the particular evolution.
From the military studies, there is another neat research line on violence in Colombia,
focused on determining what the dynamics of violence were in terms of military
strategies or types of military security and cooperation policies. For example, works
such as those by Blair (1993) or Leal (1994), focus on analyzing the role of the Public
Force, the influence of national security doctrines, and the concurrence of internal
enemies, resulting from the influence of the American thought. Additionally, Ramírez
(2000) and Rangel (2003) focus on the importance of US military cooperation policies,
while others such as Echandía (1999, 2006), Salas (2010, 2015) or Ríos (2016a, 2016b
) prioritize the study of the military strategy of the guerrillas, according to the change
factors and the different stimuli concurrent in the internal conflict.
Political geography has also been an interesting starting point in the understanding of
the Colombian conflict, mainly because of its attempt to understand how areas used to
plant coca and other resources have operated as explanatory factors of violence in
Colombia (Betancourt, 1991; Observatorio Geopolítico de las Drogas, 1996; Echandía,
1996). These visions, on the other hand, have been complemented by local approaches
such as those developed by the Center for Research and Popular Education (CINEP by
its abbreviation in Spanish) in the works developed by García (2003) on peace actors
and violence in Bajo Cauca region of Antioquia; by Guzmán (2003), on Valle del Cauca
and Cauca; or by Gutiérrez (1998; 2002) on the relationship of violence and political
system in Cundinamarca. Also, we must highlight the contributions from Vásquez,
Vargas, and Restrepo (2011) concentrated on the South area of the country; or Torres
(2011) and Rodríguez Cuadros (2015), in particular case studies focused on Putumayo
and Nariño, respectively. On the other hand, García and Aramburo (2011), with a
marked geographical imprint, address the complexity of armed violence in the East
area and the Urabá region in Antioquia; since González et al. (2012) focus on Eastern
Colombia, and particularly in Antioquia, Boyacá, Santander, Norte de Santander and
Arauca; and in a second document, does the same work about the Caribbean region of
the country (González et al., 2014).
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
140
!
From the approaches that are closer to the economy, it may be noted, first of all, that
there have been two currents that have dominated the explanation of the Colombian
armed conflict. On one side, those focused on analyzing the cost of violence in
Colombia and, in front of them, those that have analyzed, either with a more
qualitative or more quantitative approach, the economic factors that have stimulated
violence.
In reference to the first works, some outstanding contributions emerge from the second
half of the nineties; there are works that are focused on showing the negative
repercussion that the conflict has on the economy of the country and, by extension, on
the continuation of the social problems. It is thus possible to highlight the contributions
of Castro et al. (2000); Trujillo and Badel (1998); Granada and Rojas (1995), or Deas
and Gaitán (1995), who agree, in brief, in indicating that the costs for Colombia of this
armed conflict, in a macroeconomic study focused in the 1990s, would amount between
2 and 4 % of the Gross Domestic Product (GDP); a figure well below the 15.1% that
Sánchez and Díaz (2005) quantified with respect to the percentage of GDP represented
by illicit activities in Colombia. In more current literature, we can not overlook the work
of Otero (2007. p. 10), who, being focused on the Democratic Security Policy, refers to
an impairment of 4.5% of GDP, according to a conflict quantification exercise which
states that between 1958 and 2012, it has left behind more than 220,000 deaths,
25,000 missing, 27,000 people kidnapped, 5.7 million displaced persons, almost 2,000
massacres and 5,000 attacks against public infrastructure (Centro Nacional de Memoria
Histórica, 2013). In sum, an economic impact, only between 2000 and 2003, of $35
billion Dollars in security costs, plus another $2.3 billion in direct costs of war. Nor can
we ignore other works, equally aimed at the economic deceleration that has involved
armed violence in Colombia, as proposed by Álvarez and Rettberg (2008), Sánchez et
al. (2009) or, more recently, from a sub-national perspective, the work of Querubín
(2013).
From the group of works that are oriented on explaining violence based on economic
factors, the work of Sánchez (1987) is one of the mandatory starting points, as it is the
first one to alert the close relationship between structural violence and the emergence
of armed conflict. This hypothesis, followed by Molano (1987), Reyes (1988) or Ramírez
(1990) will open a line of research in the 1990s that is predominant in the Colombian
academy, known as the violentology, which will be developed within the National
University of Colombia.
Theoretically, this line has a linking point with the theoretical developments that try to
understand the violence of an armed conflict from an unfailing correspondence, not
only to legitimize the existence of an armed struggle but also, to identify a rational
calculation and a use of economic resources as an input to sustain the aforementioned
violence for the sake of a military victory (Montenegro & Posada, 2005). And, in fact,
between one and the other, as will be seen below, is where the approach of this work
comes into force.
That is to say, the explanatory contribution of this work evidences that there are
structural factors from origin that allow explaining and understanding certain enclaves
of greater guerrilla entrennchment since, due to the evolution of the internal armed
conflict, there are other factors that, besides being derived from violence and not so
much from origin, they also take on an argumentative force when it comes to
understanding why guerrilla violence is concentrated in certain scenarios rather than in
others. Thus, for the case of Colombia, all the works coincide in highlighting the
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
141
!
economic inequality (Candelo et al., 2000), land concentration (Ibáñez and Querubín,
2004), or forced displacement and accumulation of land (Reyes, 2009).
Finally, the internal armed conflict can not be understood without the revenues of legal
sources, and mostly illegal, that fuel every armed confrontation against the State but
also construct particular dynamics of legitimacy (Collier, 2000; Collier & Hoefler 2004;
Bates, 2008). Something that, for that matter, Yaffe (2011, pp. 193) for the case of
Colombia and taking up contributions such as Ballentine and Nitzchke (2003) or
Ballentine and Sherman (2003), implies that:
…although the struggle for access to economic resources can be a
perpetuating element of armed conflicts, it is not the main cause of
their emergence (since) they agree in the fact that the origin of
violent conflicts is in the resentments generated by the bad
management of resources, the unequal distribution of wealth
derived from these resources, and the government policies that
prevent many sectors from benefiting from these fortunes. Yaffe
(2011, pp. 193).
That is to say, both conditions originating from structural violence, translated into
institutional abandonment, inequality, or socio-economic backwardness, such as
conditions that are linked to the curse of resources for the survival of violence would
come together as explanatory factors of the guerrilla activism in Colombia. An activism
that, according to what is stated in the following section, seeks to be explained in the
light of a list of variables that, in the beginning, should serve to understand how the
dynamics of violence in Colombia are produced.
3. Methodological design
The methodological design is inspired by two investigations of mandatory reference.
First, the work by Sánchez and Díaz (2005) that focuses on analyzing the economic
effects of the armed conflict in Colombia and investigates the evolution of the armed
activity of the FARC-EP, the ELN, and the United Self-Defense Forces of Colombia,
between 1995 and 2002, throughout a vast sample of municipalities. In this work, the
consequences of the activities of these groups are estimated on the forced internal
displacement, human capital (in terms of education and infant mortality),
socioeconomic variables (based on Unsatisfied Basic Needs and the Gini coefficient), in
addition to geographic variables (distance and infrastructure), and fiscal activity
(transfers and public investment). The technique that was used, and that also serves
as a reference, is matching estimators, whereas what is compared is the reality of a
municipality with the armed activity against a simulated municipality without armed
activity, but that in other aspects maintains conditions that are very similar to the first
one.
On the other hand, the contribution of Botello (2014) seeks to explain the
determinants of the homicide rate at national level, transcending itself from the armed
conflict, between 1993 and 2005; this time, based on a Tobit model, which is used
when the information of the dependent variable can be divided into two groups
(municipalities without homicides and with homicides), we use explanatory variables
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
142
!
that include the average income of the municipalities, the number of inhabitants, and
the size of the urban sector.
Based on the above, the assumption on this occasion is to estimate and quantify the
determinants of the armed conflict in Colombia although, unlike the previous ones, the
dependent variable is the number of armed actions of the FARC-EP, whilst it is about
explaining what factors explain their activism in recent years. In this way, and contrary
to both, the period that is analyzed is more recent, covering the decade between 2002
and 2012, since it is the one of the greatest guerrilla violence in the recent history of
the country, and so, emphasize in this way the explanatory imprint that pose the
cultivation of coca, the mining activity, the operational capacity of the Armed Forces,
or the legal changes focused towards the end of the conflict, among other factors.
3.1. Description of variables
All the indicators that are presented in this paper are analyzed by department
according to an annual periodicity that, as noted before, covers from 2002 until 2012.
In the case of most of the quantitative variables, the natural logarithm (symbolized by
an "L" that precedes a variable) is applied. The number of clashes with the FARC-EP
guerrillas and the armed contacts initiated by the Public Force were obtained from the
Observatory of Human Rights and International Humanitarian Law (ODHDIH by its
abbreviation in Spanish) of the Vice-Presidency of the Republic, whose information, in
turn, was processed by the Department of Security Administration until 2011, and from
then on by the General Command of the Military Forces. Thus, based on this indicator,
armed actions are understood as the number of attacks against the Armed Forces,
ambushes, harassment, attacks against the civilian population, and acts of terrorism
carried out by the FARC-EP. An activism that, according to the data, would be mainly
concentrated in the departments of Cauca, Nariño, Antioquia, Caquetá, Arauca, and
Putumayo, where the annual average of fifty actions was exceeded by all of them.
The number of coca crops is measured per cultivated hectare, based on the
calculations made by the United Nations Office on Drugs and Crime (UNODC), which
coordinates all the information that nourishes the Illicit Crop Monitoring System in
Colombia (SIMCI by its abbreviation in Spanish), in which all data on illicit drugs is
concentrated. Similarly, departments such as Cauca, Nariño, Antioquia, Caquetá,
Putumayo, Guaviare, and Meta would be the coca-growing departments of Colombia
par excellence, surpassing the 4,000 hectares of cultivated coca plantations annual
average in the period under study.
On the other hand, Cundinamarca and the departments of Santander, Valle del Cauca,
and Antioquia would be the ones with the highest real GDP per capita, exceeding the
$8,000,0000 Colombian Pesos (COP) per year, approximately $3,000 Dollars (US),
which would contrast with departments such as Vaupés, Chocó, Guainía, Guaviare,
Nariño, and Sucre, where the same indicator is less than COP$3,700,000 in annual
average (approximately US$1,200). On this occasion, it would be the National
Administrative Department of Statistics (DANE
1
by its abbreviation in Spanish) who will
work as the source from which information on this type of indicator can be extracted,
making also possible to calculate the population density by department.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
1
Information can be found on www.dane.gov.co in the section Estadísticas por tema and then in Cuentas
Nacionales.
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
143
!
Data about inequity, specifically in terms of ownership and distribution of land tenure,
and covering all data and departments, with exception of the data for year 2012, are
obtained through the Ministry of Agriculture of Colombia. In this way, the departments
with the greatest inequality would be Antioquia, Meta, Arauca, Cauca, Valle del Cauca,
and Boyacá, with a coefficient that exceeds the value of 0.82 in annual average. This,
compared to the departments of Orinoquia and Amazonia, which like Vaupés, Vichada,
Guainía, or Guaviare, share much more egalitarian trends (coefficient of 0.47 annual
average).
The socioeconomic conditions of violence are integrated, fundamentally, based on
three variables. On the one hand, educational performance is measured on the basis of
the results of the mathematics tests from the exams that are given to enroll Higher
Education (SABER 11)
2
. These tests measure the level of quality of education, which is
key for stimulating, or not, actions of guerrilla recruitment. In this way, this indicator
was analyzed for all years, and the results showed a high performance from
Cundinamarca, Santander, Boyacá, Nariño, and Valle del Cauca (above 44.5 annual
average) compared to other departments such as Chocó, Vaupés, Amazonas, and
Magdalena, where the lowest scores were recorded (below 42 annual average).
Also, the departmental tax collection per inhabitant would measure institutional
capacities to combat violence, although it is available only for the years between 2009
and 2012, thanks to the figures that are housed in the National Planning Department
(DNP
3
by its abbreviation in Spanish). Tax revenues are high in San Andrés,
Cundinamarca, Antioquia, and Boyacá (above COP$149,000 annual average) compared
to Putumayo, Cauca, Nariño, and La Guajira (below COP$70,000 annual average) that
is where the tax collection is much lower.
Besides, the Ministry of Health and Social Protection
4
provided the infant mortality
rate, which measures the number of deaths in children less than 1 year of age per
1,000 births, from 2000 to 2005. The best indicators of infant mortality are in
Santander, San Andrés, Arauca, Casanare, and Valle del Cauca (below 15 deaths in the
annual average), as opposed to Chocó, Vichada, Caquetá, and Guainía (above 29
deaths in the annual average).
Dichotomous variables were created to measure changes throughout time. This was
the case of "Santos", who took the value of 1 for 2011 and 2012 (periods in which
Juan Manuel Santos served as president) and the variable "justice" to which value 1
was assigned from 2006 to 2012, seeking to find the possible effects of Law 975 of
2005 on Justice and Peace in guerrilla violence in Colombia, while leaving with it the
demobilization of more than 31,000 paramilitaries.
Finally, variables were constructed to measure specific characteristics of the
departments, especially in attention to the center/periphery binomial. Distancia
measures the length in kilometers between the capital of a department and the capital
of Bogotá. Frontera, Mar, Andino, Minorías, and Minero are qualitative variables that
serve to compare the group of departments that share a border with some country,
have access to the sea, where minorities are representative within their population or
with respect to which the mining sector becomes especially important within the
departmental GDP.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2
Information can be found on www.icfesinteractivo.gov.co/historicos/
3
Information can be found on www.dnp.gov.co/programas/desarrollo-territorial/evaluacion-y-seguimiento-
de-la-descentralizacion/Paginas/desempeno-fiscal.aspx
4
Information can be found on www.minsalud.gov.co/estadisticas
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
144
!
4. ANALYSIS OF THE INVESTIGATION
4.1. Determinants of conflict (specification 1)
The model that has been proposed for analyzing the determinants of the conflict is the
following (equation 1):
a) Illegal crops CI, government GOB, real GDP per capita PIBRPC, and the presence of
mining MIN determine the evolution of the conflict CO. The error u captures all
independent variables not explicitly included in equation (1).
b) Illegal Crops CI: when there is an increase in the coca crops, it requires the
augmentation of troops from illegal armed groups, leading to greater conflict (positive
beta 1). Coca crops were calculated using the variable LCOCA.
c) Government GOB: higher tax revenue per capita reflects a greater presence of the
Government and a lower possibility of conflict (negative beta 2). In this scenario, it is
also important to consider that a larger army can generate a greater conflict (positive
beta 2). In this case, LRECTRIPC is the variable used to measure tax collection.
d) Economic Performance DESECON: since there is much part of the conflict with
guerrillas occurring outside of urban centers, to a greater economic development
reduced the incentives to generate conflict (negative beta 3). The variable used is
LPIBRPC.
e) Miner MINERO: the departments that are most dependent on the mining sector,
attract the attention of illegal armed groups given the ability to generate additional
income. This in turn translates into greater conflict in protecting these new resources
(beta 4 positive). The mining activity was estimated using the variable MINERO.
4.2. Determinants of conflict (specification 2)
Another model proposed to analyze the determinants of the conflict is the following
(equation 2):
( )
titititititi
vINLEYDILDESLCILCO
,,543,2,10,
)()()()( ++++++=
αααααα
Illegal crops CI, inequality DES, distance DI, justice JUS, and initiative IN determine
the evolution of the conflict CO. The error v captures all the independent variables not
explicitly included in equation (2).
a) Illegal Crops CI: when there is an increase in the coca crops, it requires the
augmentation of troops from illegal armed groups, leading to greater conflict (positive
alfa 1). Coca crops were calculated using the variable LCOCA.
b) Inequality DES: according to the literature, inequality and especially rural inequality
are one source of conflict, but also of legitimacy to justify guerrilla action (alpha 2
positive). The inequality was estimated using the variable LGINIT.
c) Distance DIS: the economic performance and the presence of the Government are
also associated with the peripheralization of the conflict. A greater distance indicates
the peripheralization of the conflict due to a lower presence of the Government and a
reduced internal trade (alpha 3 positive). The variable used is LDISTANCE.
d) Laws LAW: the legal framework also poses solutions to the conflict. Laws allowing
the inclusion of illegal armed groups into civil society reduce future sources of violence
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
145
!
(negative alpha 4). In this case, JUSTICE is the variable that is used to measure
change in legal processes.
e) Initiative INI: as mentioned above, it is important to consider that a larger majority
of Public Force can generate intensification of the conflict (positive beta 5). The
initiative of the Military Forces is evaluated with the variable LFFMM.
5. Results of the investigation
5.1. Ordinary Least Squares (double logarithmic model)
Using the statistical program Eviews version 9.5, the following regressions in which the
dependent variable is the natural logarithm of the number of confrontations from the
FARC-EP guerrillas were performed (Tables 1 and 2). The Ordinary Least Squares
method allows us to know the marginal effect of an independent variable (coca crops)
on a dependent variable (conflict), keeping the other variables constant. In all cases a
Log-Log model is used since the coefficients have an intuitive interpretation
(elasticities), it reduces the possible problems of heteroscedasticity and allows to
compare variables in different units of measurement. The statistical program Eviews is
used given that it is one of the softwares that allows to model information from panel
data, and allows to analyze the information in an intuitive way.
In the first regressions (Table 1) the emphasis was on coca production, State presence
(tax collection per capita), and economic performance (real GDP per capita). In all
cases, the response of the conflict to coca cultivation is inelastic and positive. This
result can be explained in the following way: when the production increases, the
guerrilla group makes an additional effort to protect these lands from different threats
(FARC, ELN, paramilitary, and National Army).
The presence of the State attenuates the magnitude of the conflict in a considerable
way. When the departmental tax collection increases by 1%, the number of FARC-EP
clashes decreases by 1.93% (Model 4). When the Government carries out a greater
collection, it is probable that it will have a greater presence of Public Force in that
department and in this way restrain confrontations with the guerrilla group.
The economic growth also has a significant impact on fighting. When GDP per capita
increases by 1%, the number of FARC-EP clashes increases by 1.36% (Model 4). This
logic-challenging outcome can be explained by the productive structure of the region.
Departments in which the mining sector generates a large part of its GDP attract the
attention of illegal armed groups. Among the multiple sources of income of the FARC-
EP, the extraction of minerals such as gold in the Pacific or oil in the department of
Arauca can be included. When analyzing the variable MINERO * LPIBRPC, the results
show a positive sign pointing out how the economic growth of departments with high
dependence on mining generates more clashes with the FARC-EP.
As in the regressions in Table 1, the conflict response to coca crops is inelastic and
positive (Table 2). In addition to the above, measures of inequality (GINI of lands),
peripheralization of the conflict (distance to Bogotá), Law of Justice and Peace
(Justice), and Military Forces (initiative) were incorporated into the models 6, 7 and 8.
Among all variables, inequality has the largest impact on the number of confrontations
(models 6, 7 and 8). When the Gini coefficient of land per department increases by
1%, the number of FARC-EP clashes increases by 1.63% (Model 8). Thus, high levels
of inequality generate discontent and discomfort in such a way that when injustices are
perceived in the processes, the result is a greater conflict in those regions, according
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
146
!
to what is posed in some previous works like the one by Whitworth (2012), which is
centered in showing how there is a significant link between inequality and violence at
the local level.
The presence of the State can also be modeled by the distance between the capital of a
department and Bogotá. When the distance increases by 1%, the confrontations
increase by 0.10% (Model 8). The distance indirectly reflects the transport costs and
the influence of the capital city on a department, given by the easiness in the internal
commerce (even from the times of the colony, as proposed by Safford & Palacios,
2012.
On the other hand, the Justice and Peace Law reduced the intensity of the conflict by
43% when comparing the period between 2001 to 2005 and 2006 to 2012. Given the
strong acceptance of this Law by the United Self-Defense Groups of Colombia (AUC by
its abbreviation in Spanish), it would be expected for the guerrilla of the FARC-EP to
have less competition for coca crops (between 2002 and 2010, there were 31,810
group and 21,849 individual demobilizations carried out according to the Office of the
High Commissioner for Peace).
When the military forces increase their initiative in the conflict by 1%, the number of
conflicts with the FARC-EP increases by 0.66%. According to the above, the Army has
also had an offensive position in the conflict which generates a response from the
guerrilla and sharpens the conflict in the country. This poses a positive scenario for
peace as both actors lose the incentive to start a fight.
5.2. Binary models
Another way to understand the consequences of the Colombian internal armed conflict
is to estimate the probabilities of engagement with the FARC-EP. Under this scenario,
the departments that have had at least one contact with this group differ from those
without conflict. The technique that is used is the binary models that allow the
calculation of regressions when the dependent variable acquires only two values (1 if it
had conflict, 0 otherwise).
Tables 3 and 4 present the results of the Logit model including all the departments of
the sample. A qualitative variable that allows comparing the departments with coca
crops versus those that do not have them was constructed. The difference is
statistically significant in most cases, and in model 16, the coca-producing
departments are 39% more likely to have conflicts with the FARC.
When analyzing the incidence of inequality and the initiative of the military forces, one
can observe their importance as conflict sources. For each additional point in the Gini
coefficient of land, the probability of having a combat with the FARC increases by 63%.
For every military initiative of the Colombian Public Force, the probability of generating
a conflict with the FARC-EP increases by 7%.
5.3. Models with censored or truncated information
The Tobit model employs the best of the double logarithmic model (5.2.1) and the
binary models (5.2.2). Just like the binary models, the Tobit model divides the sample
into those departments that had conflict with those that did not have this condition,
but in this case the regression allows to differentiate the severity of the conflict
because the dependent variable assumes its original values (not binaries).
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
147
!
Similar to what is discussed in point 5.2.1, the conflict response to coca crops is
inelastic and positive. In addition, the inequality has the greatest impact on the
number of clashes (Table 6) and its coefficient is very similar to the one indicated in
5.2.1 (1.75% versus 1.63%). In spite of the above, the impact of the Justice and
Peace Law, and the initiative of the Military Forces is larger in the Tobit models (71%
compared to 43% in the first case and 0.66% compared to 0.71 %).
5.4. Assumptions review
Among the independent variables, the correlation coefficients register values that are
lower than 0.63, indicating that multicollinearity is not a serious problem. This
diagnosis was validated by the auxiliary regressions and the Klein rule.
When performing regressions with robust errors that correct problems of
heteroscedasticity, the signs and magnitudes of the estimators did not change
significantly.
The approximation to the normal standard of the Durbin Watson statistic shows that, in
general terms, most of the regressions do not present autocorrelation order 1 (positive
or negative).
With a significance of 1%, the Jarque Bera test of normality does not reject the null
hypothesis that the regression errors of models 2, 3, 4, 7 and 8 follow the normal
distribution. In the other cases, although the null hypothesis is rejected, the sample
size allows statistical inference (minimum sample size of 238).
6. Conclusions
This research carried out a quantitative approximation by analyzing some variables
that, segmented according to their socioeconomic, institutional, and geographic nature,
and pondering by their importance, explain much of the FARC-EP guerrilla activism that
has occurred in recent years.
On the socioeconomic side, the emphasis was placed on the production of coca and the
presence of the State (tax collection per capita), revealing a positive relationship
between the armed conflict and coca cultivation. The response of the conflict to coca
crops is less than 1%; a situation that can be explained by the additional efforts that
the FARC-EP guerrilla must carry out in order to protect these lands. This result,
although not so evident, is magnified when compared with groups of regions, showing
that coca-producing departments are 39% more likely to have clashes with the FARC.
The guerrilla group gained ground in the economic environment thanks to illicit crops
and criminal activities, before which the Colombian State has had serious difficulties to
stop these phenomena. In this study, it is determined that the tax collection has
influence in the conflict as, when it increases by 1%, the number of combats with the
FARC-EP decrease by 1.93%. That is to say, it could be affirmed that when the
Government carries out a greater collection, it counts on greater Public Force support
from which to dissuade the violence produced by the internal conflict.
Within the findings of the study, it is evident that the mining activity is also a
determinant of the armed conflict. When conducting the interaction of the percentages
of the mining GDP with the per capita GDP of a department, it shows a positive
relationship. The greater the mining wealth, the more conflicts are generated in that
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
148
!
region, perhaps also related by the fact that illegal mining has stood out as one of the
most important sources of income for the FARC-EP.
Amidst the socioeconomic variables, inequality plays a very important role in the
confrontations. When the Gini coefficient of land increases by 1%, the number of FARC-
EP combats increase by 1.63%. This, insofar as high levels of inequality generate
malaise and nonconformity, and when injustices are perceived in the processes, the
result is a greater conflict in those regions.
Likewise, the periferialization of the conflict (distance to Bogotá), and the Justice and
Peace Law together with the initiative of the Military Forces were, on the other hand,
included as institutional variables. The effect of distance and army initiative on the
conflict is positive, but below 0.2% in the first case and below 0.7% in the second
variable. Besides, it is evident that the guerrillas have been strengthened, thanks to the
absence of the State, in regions that are far from the great capitals and the
demographic centers. The further the department is from the capital the more reduced
the presence of the State, which is reinforced by the greater costs of transport. The
Justice and Peace Law, on the other hand, has a significant effect on the reduction of
the conflict over the conflict, comparing the period between 2001 to 2005 and 2006 to
2012, its severity was reduced by 43%.
Finally, it can be concluded that guerrilla violence in Colombia is mainly determined by
inequality, the presence of the State, cocaine crops, and armed demobilization
processes (Justice and Peace Law).
References
Aguilera, Mario (2010). Las FARC: la guerrilla campesina 1949-2010. Bogotá: Arfo.
Álvarez, Stephanie & Rettberg, Angelika (2008). “Cuantificando los efectos económicos
del conflicto: una exploración de los costos y los estudios sobre los costos del conflicto
armado colombiano”. Colombia Internacional, 67: 14-37.
Ballentine, Karen & Nitzschke, Heiko (2003). Beyond greed and grievance: Policy
lessons from studies in the political economy of armed conflicto. New York:
International Peace Academy.
Ballentine, Karen & Sherman, Jake (2003). The Political Economy of Armed Conflict:
Beyond Greed and Grievance. London, Lynne Rienner Publishers.
Bates, Robert (2008). “State failure”. Annual Review of Political Science, 11(1): 1-12.
Betancourt, Darío (1991). “Los cinco focos de la mafia colombiana (1968-1988).
Elementos para una historia”. Revista Folios, 2: 13-30.
Blair, Elsa (1993). Las fuerzas armadas. Una mirada civil. Bogotá: CINEP.
Botello, Héctor (2014). “Inequidad, violencia y pobreza en los municipios de Colombia”.
Perfil de Coyuntura Económica, 24: 33-48.
Candelo, Ricardo et al. (2000). “Propiedad rural y reforma agraria en Colombia: un
debate sobre el periodo 1985-1996”. Economía Colombiana y Coyuntura Económica,
278: 2332.
Castro, Manuel et al. (2000). “El conflicto armado: la estrategia económica de los
principales actores y su incidencia en los costos de la violencia 1990-1998”. Revista de
Planeación y Desarrollo, 30(3): 81-105.
Centro Nacional de Memoria Histórica (2013). ¡Basta Ya! Colombia: Memorias de
guerra y dignididad. Bogotá, CNMH Departamento de Prosperidad Social.
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
149
!
Collier, Paul (2000). Economic causes of civil conflict and their implications for policy.
Washington, World Bank.
Collier, Paul & Hoeffler, Anke (2004). “Greed and grievance in civil war”. Oxford
Economic Papers, 56(4): 563-595.
Deas, Malcom & Gaitán, Fernando (1995). Dos ensayos especulativos sobre la violencia
en Colombia. Bogotá: Tercer Mundo Editores.
Duncan, Gustavo (2006). Los señores de la guerra. Bogotá: Planeta.
Echandía, Camilo (1996). “La amapola en el marco de las economías de ciclo corto en
Colombia”. Análisis Político, 27: 3-19.
Echandía, Camilo (2006). Dos décadas de escalamiento del conflicto armado en
Colombia 1986-2006. Bogotá: Universidad Externado.
García, Clara (2003). El caso del Bajo Cauca antioqueño. Cómo ver las regiones.
Bogotá: CINEP.
García, Clara & Aramburo, Clara (2011). Geografías de la guerra, el poder y la
resistencia. Oriente y Urabá antioqueños, 1990-2008. Bogotá: CINEP.
González, Fernán et al. (2012). Conflicto y territorio en el Oriente colombiano. Bogotá:
CINEP.
González, Fernán et al. (2014). Territorio y conflicto en la Costa Caribe. Bogotá: CINEP.
Granada, Camilo & Rojas, Leonardo (1995). “Los costos económicos del conflicto
armado en Colombia: 1990-1994”. Revista de Planeación y Desarrollo, 26(4): 119-151.
Gutiérrez, Francisco (1998). La ciudad representada. Política y conflicto en Bogotá.
Bogotá: Tercer Mundo Editores.
Gutiérrez, Francisco (2002). Degradación o cambio. Evolución del sistema político
colombiano. Bogotá: Norma.
Guzmán, Álvaro (2003). Cuatro ensayos de coyuntura: Valle y Cauca a finales del siglo
XX. Cali: CEREC.
Hernández, Milton (2006). Rojo y Negro. Historia del ELN. Tafalla: Txalaparta.
Ibáñez, Ana & Querubin, Pablo (2004). “Acceso a tierras y desplazamiento forzado en
Colombia”. Documentos CEDE, 23: 1-114.
Leal, Francisco (1994). El oficio de la Guerra: la seguridad nacional en Colombia.
Bogotá: Tercer Mundo Editores.
Medina, Carlos (1990). Autodefensas, paramilitares y narcotráfico en Colombia.
Origen, desarrollo y consolidación. El caso Puerto Boyacá. Bogotá: Editorial de
Documentos Jurídicos.
Medina, Carlos (1996). ELN: una historia contada a dos voces, Bogotá, Rodríguez Quito
Editores.
Molano, Alfredo (1987). Selva adentro: una historia oral de la colonización del
Guaviare. Bogotá: Áncora.
Montenegro, Armando & Posada, Carlos (1995). La violencia en Colombia. Bogotá:
Alfaomega.
Observatorio de Derechos Humanos y Derecho Internacional Humanitario (ODHDIH)
(n.d.). Síntesis de la violencia y la confrontación armada en Colombia, 1998-2012 y
2015. Bogotá: Colombia.
Observatorio Geopolítico de las Drogas (1996). Informe Anual. Geopolítica Mundial de
las Drogas. París: OGD.
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
150
!
Otero, Diego (2007). Las cifras del conflicto armado en Colombia. Bogotá: INDEPAZ.
Pécaut, Daniel (2008). Las FARC, ¿una guerrilla sin fin o sin fines? Bogotá: Norma.
Pizarro, Eduardo (2011). Las FARC (1949-2011). De guerrilla campesina a máquina de
guerra. Bogotá: Norma.
Querubin, Pablo (2003). “Crecimiento departamental y violencia criminal en Colombia”.
Documentos CEDE. 12: 1-49.
Ramírez, William (1990). Estado, violencia y democracia, Bogotá: Universidad Nacional.
Ramírez, William (2000). Colombia, cambio de siglo: balances y perspectivas. Bogotá:
Planeta.
Rangel, Alfredo (2003). Fuerzas Militares para la guerra. La agenda pendiente de la
reforma militar. Bogotá: Fundación Seguridad y Democracia.
Reyes, Alejandro (1988). “Conflictos agrarios y luchas armadas en la Colombia
contemporánea: una visión geográfica”. Análisis Político, 5: 7-29.
Reyes, Alejandro (2009). Guerreros y campesinos. El despojo de la tierra en Colombia.
Bogotá: Norma.
Ríos, Jerónimo (2016a). “Dinámicas de la violencia guerrillera en Colombia”. Revista de
Ciencias Sociales, 22(3): 84-103.
Ríos, Jerónimo (2016b). La narcotización del activismo guerrillero de las FARC y el
ELN, 1998-2012”. Revista UNISCI, 41: 205-234.
Rodríguez Cuadros, José (2015). Génesis, actores y dinámicas de la violencia política
en el Pacífico nariñense. Bogotá: CINEP.
Romero, Mauricio (2003). Paramilitares y autodefensas 1982-2003. Bogotá: Planeta.
Ronderos, Teresa (2014). Guerras recicladas. Una historia periodística del
paramilitarismo en Colombia. Bogotá: Aguilar.
Safford, Frank & Palacios, Marco (2012). Colombia: País fragmentado, sociedad
dividida. Bogotá: Norma.
Salas, Luis (2010). “Corredores y territorios del conflicto armado colombiano: una
prioridad en la geopolítica de los actores armados”. Perspectiva Geográfica, 15: 9-36.
Salas, Luis (2015). “Lógicas territoriales y relaciones de poder en el espacio de los
actores armados: un aporte desde la geografía política al estudio de la violencia y el
conflicto armado en Colombia, 1990-2012”. Cuadernos de Geografía, 24(1): 157-172.
Sánchez, Gonzalo ([1987]2009). Colombia: violencia y democracia. Comisión de
Estudios para la Violencia. Bogotá: La Carreta.
Sánchez, Fabio, & Díaz, Ana (2005). Los efectos del conflicto armado en el desarrollo
social colombiano, 1990-2002. Bogotá: Universidad de Los Andes - CEDE.
Torres, María (2011). Estado y coca en la frontera colombiana. El caso de Putumayo.
Bogotá: CINEP.
Trujillo, Edgar & Badel, Martha (1998). “Los costos económicos de la criminalidad y la
violencia en Colombia: 1991-1996”. Archivos de Macroeconomía, 76: 1-49.
Vásquez, Teófilo et al. (2011). Una vieja guerra en un nuevo contexto. Conflicto y
territorio en el sur de Colombia. Bogotá: CINEP, ODECOFI, Colciencias, Universidad
Javeriana.
Whitworth, Adam (2012). “Local inequality and crime: Exploring how variation in the
scale of inequality measures affects relatinoships between inequality and crime”. Urban
studies, 50(4): 725-741.
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
151
!
Yaffe, Lilian (2011). “Conflicto armado en Colombia: análisis de las causas económicas,
sociales e institucionales de la oposición violenta”. Revista CS, 8:187-208.
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
152
!
ANNEXES
Table 1. Conflict Regressions (dependent variable LFARC)
Variable
Model 1
Model 2
Model 3
Model 4
C
1,06(0,28)***
3,68(2,09)***
-21,93(1,25)***
-13,28(5,65)*
LCOCA
0,18(0,04)***
0,36(0,07)***
0,42(0,06)***
0,43(0,06)***
LRECTRIPC
-0,84(0,41)***
-2,28(0,43)***
-1,93(0,47)***
LPIBRPC
2,04(0,34)***
1,36(0,42)***
MINERO
-16,67(9,03)*
MINERO*LPIBRPC
1,09(0,57)*
N
288
96
96
96
R2
0,07
0,27
0,47
0,51
F
22,47***
17,1***
28,04***
19,08**
DW
1,79
2,27
2,17
2,25
Source. Calculations prepared by the authors. *p<0,10; **p<0,05; ***p< 0,01
Table 2. Conflict Regressions (dependent variable LFARC)
Variable
Model 5
Model 6
Model 7
Model 8
C
1,06(0,28)***
2,10(0,28)***
1,11(0,39)***
0,03(0,34)
LCOCA
0,18(0,04)***
0,19(0,03)***
0,17(0,03)***
0,06(0,03)**
LGINIT
2,64(0,36)***
3,13(0,38)***
1,63(0,32)***
LDISTANCIA
0,22(0,06)***
0,10(0,16)**
JUSTICIA
-0,57(0,14)***
LFFMM
0,66(0,06)***
N
288
246
246
245
R2
0,07
0,24
0,27
0,58
F
22,47***
38,48***
31,16***
64,91**
DW
1,79
1,70
1,83
1,97
Source. Calculations prepared by the authors. *p<0,10; **p<0,05; ***p< 0,01
Tabla 3. Conflict Regressions (Logit) (dependent variable FARCD)
Variable
Model 9
Model 10
Model 11
Model 12
C
0,64(0,21)***
0,55(0,58)
-0,18(0,71)
-0,79(0,84)
DCOCA
0,79(0,26)***
1,11(0,44)**
1,46(0,49)***
2,07(0,62)***
RECTRIPC
-0,005(0,003)*
-0,02(0,006)***
-0,01(0,006)**
PIBRPC
0,00(0,00)***
0,00(0,00)
MINERO
2,01(1,74)
MINERO*PIBRPC
0,00(0,00)
N
372
124
124
124
R2 (Mc Fadden)
0,02
0,08
0,20
0,29
LR
8,72***
12,35***
31,13***
45,58***
Source. Calculations prepared by the authors. *p<0,10; **p<0,05; ***p< 0,01
JANUS.NET, e-journal of International Relations
e-ISSN: 1647-7251
Vol. 9, Nº. 1 (May-October 2018), pp. 137-154
Quantitative determinants of the FARC-EP guerrilla violence in Colombia
Jerónimo Rios Sierra, Camilo Vargas, Paula Bula, Amalia Novoa Hoyos
!
153
!
Table 4. Conflict Regressions (Logit) (dependent variable FARCD)
Variable
Model 13
Model 14
Model 15
Model 16
C
0,64(0,21)***
-2,00(0,83)**
-2,46(0,89)***
0,02(1,05)
DCOCA
0,79(0,26)***
1,31(0,31)***
1,29(0,31)***
0,40(0,39)
GINIT
3,61(1,07)***
4,13(1,12)***
0,65(1,36)
DISTANCIA
0,0001(0,00)
-0,00(0,00)*
JUSTICIA
-1,13(0,38)***
FFMM
0,07(0,01)***
N
372
323
323
323
R2
0,02
0,08
0,08
0,35
LR
8,72***
24,5***
26,65***
64,91**
Source. Calculations prepared by the authors. *p<0,10; **p<0,05; ***p< 0,01
Table 5. Conflict Regressions (Tobit) (dependent variable LFARC)
Variable
Model 17
Model 18
Model 19
Model 20
C
0,62(0,21)***
4,07(2,77)
-30,66(6,70)***
-15,60(8,55)*
LCOCA
0,21(0,03)***
0,33(0,07)***
0,38(0,07)***
0,41(0,07)***
LRECTRIPC
-0,96(0,55)*
-3,03(0,65)***
-2,15(0,67)***
LPIBRPC
2,82(0,51)***
1,55(0,62)**
MINERO
-11,01(11,84)
MINERO*LPIBRPC
0,80(0,75)
N
360
120
120
120
Criterio de Akaike
3,65
3,44
3,21
3,13
Source. Calculations prepared by the authors. *p<0,10; **p<0,05; ***p< 0,01
Table 6. Conflict Regressions (Tobit) (dependent variable LFARC)
Variable
Model 21
Model 22
Model 23
Model 24
C
0,62(0,21)***
1,92(0,23)***
1,07(0,43)**
-0,54(0,35)
LCOCA
0,21(0,03)***
0,24(0,03)***
0,24(0,03)***
0,04(0,02)
LGINIT
3,67(0,47)***
4,07(0,50)***
1,75(0,38)***
LDISTANCIA
0,17(0,08)**
0,07(0,05)
JUSTICIA
-0,71(0,14)***
LFFMM
0,91(0,06)***
N
360
312
312
312
Criterio de Akaike
3,65
3,45
3,44
2,76
Source. Calculations prepared by the authors. *p<0,10; **p<0,05; ***p< 0,01