January 1, 2005
by MAURICIO RUBIO *
This paper studies the factors associated with the presence of illegal armed groups – guerrillas, paramilitaries and drug traffickers – in the municipalities of Colombia. Statistical analyzes show three kinds of results. Some of these tend to corroborate explanations that are commonly accepted in this country, others defy deeply rooted beliefs, and finally some rather curious associations emerge. The availability of energy resources appears to be a crucial factor of attraction for guerrillas and paramilitaries, but not for drug traffickers. The conflict also seems to affect both the municipalities’ ability to impose taxes, as well as the latter’s composition. Against conventional wisdom, drug traffickers appear as the group having the closest ties to the political sphere. This group shows the greatest capacity to influence civil society manifestations, has the most corrosive effect on elections, and has the greatest influence on the composition of municipal expenditures. A very popular explanation of violence in Colombia posits a strong and positive association between economic conditions, the so-called objective causes, and the armed conflict. The data does not support the arguments that poverty is responsible for it. On the other hand, the data suggest that the age of the population, a purely demographic factor, is determinant. The effect of other social variables is more vague and ambiguous. Curiously, the existence of communications infrastructure is the indicator of State presence that best explains the conflict’s geography. The backwardness and administrative disorganization in public expenditures also appears to be related to the influence of armed actors. However, on an almost anecdotal level, an office for municipal planning is positively associated with the influence of drug trafficking.
There is abundant testimonial evidence in Colombia about the complex ties between guerrillas, paramilitaries and drug traffickers, the so-called “armed groups” in the Colombian conflict, and local politicians. This paper attempts to identify the mechanisms through which a peculiar symbiosis arises between illegal armed organizations and political activity as shown by available indicators of political activity and public finance at the municipal level. Using simple statistical procedures –logit regression- we seek to detect those political and fiscal variables that help discriminate municipalities in which the armed groups are present, from those that are free of them.
The paper is divided in two sections. The first one briefly reviews the literature available in Colombia about the relationship between the armed conflict and public administration at the municipal level. The results of the statistical analyzes are summed up in the second part.
A – CONFLICT, POLITICS AND MUNICIPAL PUBLIC FINANCE
In Colombia, the literature about the relationship between armed groups on the one hand, and municipal public finance and the administration of justice on the other, shows a relatively wide consensus around the idea of a growing interference on the part of the former on the latter . This interference is extended to other aspects of local politics. There also seems to be agreement that the administrative decentralization process that was consolidated with the new Constitution of 1991 was a turning point in this process. The main feature of it was a fiscal, administrative and political decentralization, which granted a greater degree of autonomy and decision-making capacity to the municipalities and regions. In the political field, the most important change was the popular election of mayors and governors that previously were appointed.
During the 90s, a new national dynamic took shape in Colombia, around the municipalities as its core. The armed organizations have played a leading role in this new phase, strengthening themselves regionally thus generating an important expansion process. Municipalities are particularly appealing to illegal organizations. It is the municipality that plans and implements the local budget, which is financed by government transfers as well as by its own income. Moreover, while it is the seat of local and regional politics, it lacks autonomy with respect to public order and the administration of justice. In effect, the Colombian State decentralized the fiscal and political-administrative functions, but maintained centralization in the fields of public order and justice.
Municipalities receive large transfers from the central government but have little or no accountability which makes them attractive to armed organizations as a potential source of financing. Their administrative decentralization, combined with their weakness in matters of justice and public security, permit the illegal groups to profit from municipal income, by means of the so-called armed clientelism, which is “nothing more than the private appropriation of public goods through the threat of weapons” (Rangel 1997, p. 56).
In Colombia, copious evidence exists on the appropriation of public resources by armed groups . On several occasions, government officials have made public statements about this issue. However, the lack of systematic information has been the main obstacle for an accurate assessment of the amount of public resources appropriated by such organizations over the years.
It seems reasonable to assume, according to various testimonies, that this process is not limited to a very small number of municipalities. The president of the Colombian Federation of Municipalities has suggested that, by the late 90s, around 25% of the country’s towns were forced to share their resources .
Testimonial evidence suggests that in some localities there exist virtual co-governments between the municipal authorities and the armed groups. In some cases, this action is limited to diverting public resources to the detriment of the population; in others, it seems that this collusion centers on administrative tasks aimed at municipal development. This behavior is not exclusive to the guerrillas: both the paramilitary groups and the drug-traffickers, who often are considered apolitical, have realized how crucial local influence can be to protect and promote their interests.
At the international level, empirical analysis of conflicts, based on cross-sectional international data, shows that armed groups prefer regions with good economic potential (Collier, 2000). The Colombian case corroborates this to a certain extent. The armed groups’ territorial drive during the nineties were characterized by an obvious expansion toward “zones of economic boom derived from the exploitation and transport of oil, coal, gold, bananas and, lately (mid 90s), in the coffee-growing zones where, until fairly recently, their presence was inconceivable” (Rangel 1997, p. 56).
It is increasingly evident that the illegal armed organizations interfere with local politics. They do so in an active manner when they introduce their own candidates forcing others to resign, and in a passive way when they permit elections to take place but exercise armed pressure while elected officials are in office. It has also been stated that pressure has been applied to departmental assemblies in order to elevate some villages to the category of municipalities so that they may be eligible to receive the funds that come from direct transfers from the national government (Rangel, 1997; Bolivar , 1999).
In spite of the great media coverage of the interference by drug traffickers on national politics, the relationship between this illegal activity and the local power structure has received less attention by analysts. Thoumi (1994) points out as a Colombian peculiarity that facilitated the consolidation of drug-trafficking, the fact that it was sufficient to corrupt local authorities in order to be able to carry out unlawful activities. He also mentions the tendency on the part of drug-dealers to create populist political movements and to play the role of traditional patrons with local clienteles. In addition, some analyses have centered on the disproportionate political power wielded by the magnitude of the drug-dealers’ fortunes, mostly in land, in relation to local economies .
B – WHAT THE MUNICIPAL DATA SHOW
Simple statistical procedures were used to analyze the complex interaction between the Colombian conflict, public administration and local politics. A relatively wide range of demographic, economic, social, institutional, political and judicial indicators (DANE 1998; FS, 1988; BR 1998) was used to detect those variables that better contribute to discriminate municipalities under the influence of the three main armed groups: guerrilla, paramilitary and narco traffickers. In this way, it was possible to identify the relationship of available indicators of public finance and municipal political activity with armed conflict. The work undertaken was exploratory and mostly inductive. It did not pretend only to test theories previously presented in the academic literature but also to point out some statistically significant associations that need both theoretical thinking and further empirical research.
First, a general a priori hypothesis was tested: the one that posits that social and economic conditions of the municipalities are the main cause of conflict. The same logit model was used to explain why the three different armed groups are present in some municipalities and absent in others. The left hand side variable used for each group –whether or not its presence has been detected in a municipality- was a dummy variable from military sources. This information is available for the guerrillas for several periods since the late eighties and the variable used was continuous presence from 1987 to 1998. For the two other groups the indicator available was about their presence in 1997. This was the only exercise for which there was no ambiguity in terms of causality: left hand side variables are clearly exogenous.
Two more exercises were carried out with the same methodology, in spite of the fact that for some variables in the right hand side there might be a two-way causality. So these models have to be interpreted simply as a step beyond simple correlations. The issue of causality was not dealt with econometrically but analytically, and needs additional information and further research about the causal links. However, the exercises were useful in pointing out which variables are strongly associated with conflict. They also allowed to illustrate that each armed group interacts differently with political and fiscal conditions in municipalities. In some way, a second general hypothesis about the political motivation of the three groups was tested.
Table 2 shows the list of right hand side variables finally used in the analysis and their correlation matrix.
The criterion for including a variable in the models was that it significantly contributed to explain the presence of any one of the three armed groups in the municipalities. Although right hand variables are not completely independent, multicolinearity problems were not serious. The results will be presented in a stepwise manner not only to ease the discussion but also to show that coefficients were not very sensible to the number of variables included in the models.
1 - THE OBJECTIVE CAUSES OF THE CONFLICT
Few explanations for conflict in Colombia have received more unconditional support as the so-called objective causes of violence. The idea that poverty, inequality and, in general, the unfavorable social and economic conditions of the population constitute a favorable environment for violence, is undoubtedly one of the most popular and widely accepted explanations of the conflict. In fact, these have been essential elements of the official analysis  and constitute not only the core of several peace processes but even of plan of the Pastrana (1998-2002) administration. This vision of the conflict is shared by politicians, technocrats and judicial officers and it is also the explanation given by intellectuals, journalists and a large part of the public opinion. The guerrillas, the drug-traffickers, even the paramilitary groups agree with this kind of explanation. Not even the State’s security agencies seem to disagree. It is therefore a necessary first step in the analysis.
There are four basic factors associated with the so-called objective causes, that contribute to distinguish those municipalities in which the guerrillas have continuously been present since 1987 from those which have never been much influenced or only sporadically (see Table 2): poverty (Unsatisfied Basic Needs Index, FS, 1998), an inequality index (Gini coefficient based on Household Quality of Life Index, DNP, 1998) the number of inhabitants in the municipality (population) and the average age of the population (a_age) (DANE, 1998).
A first striking fact is the negative effect of the poverty indicator. In contrast to the most conventional diagnosis, what one observes is that poverty in the municipalities, as measured by the proportion of the population having unsatisfied basic needs (UBN) is not positively correlated with the presence of guerrilla groups. On the contrary, guerrillas prefer to settle in the country’s less impoverished municipalities. This result has been arrived at in several earlier contributions (Echandía, 1999; Vélez, 2000; Sarmiento, 1998).
A second observation is that the inequality index has an effect contrary to that of poverty and tends to corroborate the current view: inequality does indeed seem to contribute to the presence of the guerrillas in towns. At first sight, this result seems paradoxical; poverty does not encourage the conflict, yet inequality does . However, this finding tends to support a common observation in cross-sectional studies on violence. At the international level, several results are consistent with this line of thought (Fajnzylber et al, 1998, 1999). Previous work on Colombia also shows this odd negative relationship between violence and poverty, and a positive one between violence and inequality (Sarmiento, 1998).
The degree of urbanization of the municipality, as measured by its population size, also seems to be a factor that appeals to the guerrillas . This finding, which has also been suggested in previous works (Echandía, 1999), clashes with the idea of guerrilla movements as being basically rural.
Perhaps the most striking element in light of traditional theories on the country’s conflict is that the factor that contributes the most to distinguish those municipalities under guerrilla influence is the average age of the population. Although it is rarely, if ever, mentioned in the course of national debate on the conflict, this result is not surprising. It is the local expression of one of the few almost universal characteristics of violence in any of its forms: it mainly involves the youth. When contemplating territorial expansion opportunities, the guerrillas seem to take into account the possibility of having access to groups of young people, and even children, susceptible to recruitment to a far greater degree than factors such as poverty, inequality or the town’s degree of urbanization.
The traditional objective causes invoked to explain the conflict are even less relevant to explain the presence of paramilitary groups in the localities than for guerrillas. Neither poverty, nor the number of inhabitants, are useful to determine which municipalities are under the influence of paramilitary groups. Two factors that help to explain the presence of paramilitary groups, but do not show a statistically significant effect in the case of the guerrillas, can be cited (see Table 2). One is the length of time that has passed since a particular municipality was founded (DNP, 1998): the older it is, the lower the probability to have paramilitary influence. The second factor is another purely demographic one: the proportion of men in the rural population (pmen_rural) (DANE, 1998).
Finally, the presence of drug-traffickers has even less to do with conditions of poverty or inequality in the municipalities. However, two demographic elements –the size of the population and the average age of the inhabitants (a_age)- appear again as factors that help explain the influence of drug trafficking in a particular town (see Table 2).
Perhaps the main conclusion so far is the strong effect that the average age of the population has on the indicators of presence of the various armed groups. Thus the most crucial risk factor in the regional dimension of the Colombian conflict is a population that is composed predominantly of young people. This factor is much more important than inequality. Poverty on the contrary appears to act as a repellent.
One popular extension of the theory of objective causes of the conflict involves the notion regions that have been unattended and neglected by the Colombian State present the most favorable conditions for the consolidation of the presence of armed actors.
Interestingly, it is found that only one of the available indicators of State presence  is significantly related to guerrilla, paramilitary or narco influence (see Table 2): the number of telecommunications offices and post offices (telecom) (FS, 1998). It is not easy to explain the negative association that is observed between the capacity of a municipality to communicate with the rest of the country and the probability of a continuous presence of armed groups in such a locality. For whatever reason, this element is not part of current discussion in the country, according to which the prevailing concept of State presence is more oriented towards the supply of social services, or law enforcement, rather than concerns about telecommunications services and integration with the rest of the country.
One comment can be made about the exercise applied to the paramilitary groups. Most of the available institutional indicators show, individually, a closer and negative association than the one observed for the guerrillas. When analyzing the simultaneous effect of available variables, only the number of telecommunications offices helps explain the presence of paramilitary groups.
On the other hand, the presence of drug traffickers is the one that is less readily explained by institutional indicators. However, the same indicator for access to telecommunications helps to explain the presence of drug trafficking.
1.2 – ARMED CONFLICT AND POLITICAL PARTICIPATION
According to a variant of the theory of objective causes of the conflict in Colombia, the conflict is fed by deficiencies in the channels for democratic political participation and, in particular, by the difficulties in voicing popular unrest or social protest in a spontaneous and pacific manner. It is frequently proposed that regions in which political participation is difficult turn out to be more propitious to guerrilla presence. To test this hypothesis, the same methodology can be used: to detect which of the available indicators of political participation, in addition to demographic, socioeconomic factors and institutional variables, help explain the presence of armed groups in the municipalities.
With respect to guerrillas, two political variables have are related to that influence: the number of political demonstrations  in rural areas (protest_rural) and, on the other hand, the number of associations with other purposes (as opposed to charity, social, economic, political or cultural purposes) (na_other) (FS, 1998). These two variables show a positive association with the influence of insurgent groups (see Table 3).
Regarding paramilitary groups, the impact of available variables of political participation is less meaningful than for the guerrillas.
For the third category of armed groups, the drug traffickers, it is surprising to find that several of the indicators for political participation are related with their presence in the towns with coefficients that are more significant than for guerrillas or paramilitary groups (see Table 3). Also, this association does not appear with the sign that one would expect. In what could be the expression of some kind of perverse social capital, several of the indicators for the ability to establish associations and social networks in the municipalities are positively related to the presence of drug traffickers. This result challenges traditional visions of social capital, according to which associations and networks should prevent illegal activities, not promote them. The existence of several types of civil society associations, particularly those having cultural purposes (na_cultural), and even those of an explicitly political nature (na_political) (FS, 1998), shows a positive and statistically significant correlation with the presence of drug-trafficking groups. This relation can be interpreted in two ways: either that, as for any type of business, drug traffickers tend to locate in communities having some minimum levels of coordinated activity and social infrastructure, or that in the localities where these groups settle they act as organizers, promoters, or sponsors, of civil associations of various types.
Another important point with reference to the drug trafficking organizations is that, in a manner more significant than in the case of the insurgents, or paramilitary groups, their presence is negatively associated with electoral politics, and particularly with participation –number of voters as a proportion of adult population- in the elections of mayors (vote_mayor) (FS, 1998). Although for all armed organizations one observes a perverse relationship between their influence and participation in the elections, in the case of the drug traffickers this effect is very significant. Another indicator of political participation that appears to be positively associated with the presence of drug trafficking groups is the number of protests that take place in urban areas of municipalities (protest_urban) (FS, 1998).
What one can conclude from the above is the precariousness of the notion, widely accepted in Colombia, that political status can and must be acknowledged only for one of the armed groups – the guerrillas. Available data indeed suggest that in regard to all illegal organizations –guerrillas, paramilitary and drug traffickers- there is an association between their influence and the indicators for political participation. In addition to the undeniable vocation for power shared by all these groups, and expressed through means of threats, corruption or alliances with the political class, leaders, or local public officials, data also suggest the existence of links with associations or organizations of the civil society, or with the activities or expressions of these associations.
The most surprising fact about these links between the political activity of communities and the presence of armed groups is that (i) the sign of the relationship is contrary to that posited by the theory of objective causes, or the most widely held notion of social capital, and that (ii) generally, the closest and most significant link between political variables and the actions of the various armed groups is observed for the drug-trafficking groups.
With respect to the first issue, available data show that for all three groups there is a positive association between their presence and one or more indicators of social protest. These results challenge the idealized notion of guerrilla groups as the only alternative for strikes, political marches and community protests. On the contrary, it is precisely in those towns displaying the most activity in terms of social unrest that the presence of anyone of the armed elements is most likely. If one were to argue that this happens because the guerrillas are the element that to a certain extent incite the inhabitants of the municipalities to express their dissatisfaction, one would have to apply the same reasoning to the paramilitary groups and drug traffickers. The presence of any of these groups, and not just the guerrillas, is positively associated with the indicators for social protest. Another way to explain this association would be that the armed elements find more of a breeding ground for consolidating their presence in these municipalities by means of a greater anti-establishment tendency. Therefore, the influence of the guerrillas or of the other armed groups, rather than being the sole alternative to the non-violent exercise of politics would appear as a continuation and extension of the protest for which there would be a greater vocation in certain localities.
A second point that must be stressed is that, while drug traffickers could be thought of as a group without a political agenda, data suggest that they in fact have the closest links with political action in the communities. Two of the available indicators for social capital –the number of associations with cultural purposes and the number of politically oriented organizations- show a positive and statistically significant association with this group’s influence. Furthermore, just as in the case of the guerrillas, or the paramilitary groups, it appears that there is also a positive and strong association between their presence and the number of political movements. Drug trafficking seems to act as a catalyst for social protest, either by facilitating or promoting it, or because these groups find it convenient to establish themselves in localities where protest can arise more easily. Lastly, and this is no minor point, it appears to be the group with the greatest capacity to damage electoral politics.
Another institutional variable that looks related to conflict has to do with the judiciary. Perhaps one of the most analyzed relationships in the field of economics of crime, at both theoretical and empirical levels, is the impact of the justice system on criminal behavior. The empirical validation of the deterrence hypothesis for armed presence in the country’s municipalities runs into several difficulties. The first is that Colombia lacks a good quantitative indicator of the capacity for military response, the State’s main tool for repelling such a presence. The second is that, among the available judicial mechanisms, it is not clear which are used, or should be used, to deter this presence. A third difficulty, both theoretical and practical, involves the possibility that in some municipalities judicial decisions are completely under the control of the armed groups.
Among the limited set of indicators of performance that can be constructed on the basis of Colombian judicial information, one may look at the proportion of preliminary homicide investigations that become formal proceedings, that is, the fraction of cases that lead to a formal investigation and go to trial. It has been argued that this is the most, probably only, reliable index of judiciary performance in Colombia (Levitt and Rubio, 2005). This variable (deterrence) (Calculated with data from DANE, 1998) looks related, with statistically significant coefficients, to the presence of all armed groups (see Table 3). In addition, the association of the performance of the criminal justice system and the presence of armed groups is always negative. It must be remembered that the interpretation of this result for Colombia admits the possibility of a two-way causality. Poor performance in homicide investigations would favor the consolidation of armed groups in a particular town. Conversely, the presence of organized armed groups would also make it more difficult for prosecutors to investigate and clarify murders.
2 – CONFLICT AND MUNICIPAL PUBLIC FINANCE
2.1 – TAX REVENUE
It is not clear what should be the expected effect of tax collection on the influence of armed organizations, as relationships in both senses can be conceived . Nor can we propose a solid hypothesis for the possible association between non-tax income and the presence of armed groups.
Breaking down municipal income into its different components –tax, non tax, transfers (BR, 1998)- does not contribute to explain the presence of guerillas. Another way to analyze the relationship between the presence of armed groups and tax income is to introduce dichotomous variables associated with different categories of tax income that indicate if various types of resources exist or not in the municipality, rather than looking at the per capita amounts of income. This exercise shows the existence of a positive and statistically significant association between armed groups’ presence and tax revenues from oil and gas pipelines (y_oilgas) (see Table 4). It is worth mentioning that this effect is significant only when it is considered as a dummy variable. It is the availability of some resources, and not their amount, that has a significant impact on the probability of guerrilla influence.
This result simply throws light on the relatively well documented situation that in Colombia the insurgent movements are attracted by towns that are rich in natural resources and susceptible to the imposition of forced contributions or extortion. This conclusion also concurs with international evidence (Collier, 2000).
The same exercise for paramilitary groups leads to a similar conclusion. Neither total per capita fiscal income, nor its breakdown into broad categories, show coefficients that are significantly different from zero. In contrast to the guerrillas, none of the current income items help to distinguish those municipalities under paramilitary influence.
On the other hand, the composition of tax revenue, and in particular the share of the “impuesto the industria y comercio” (IIC), a type of sales tax, in total revenue does help identify the municipalities in which paramilitary groups are present (y_IC) (BR, 1998). The negative association observed between the importance of IIC tax as a source of revenue and the chances of paramilitary presence is interesting and can have several explanations. In particular, the fact that, unlike the guerrillas, the presence of paramilitary groups is associated with the tax structure of the municipalities may be interpreted as an indication of the higher dependence of such groups on local and private sources of resources for financing. One could infer that on a nationwide basis paramilitary groups appear to be less coordinated than guerrillas since their presence seems less supported through the transfer of resources among different regions of the country.
The analysis of the effect of municipal tax revenue on the presence of drug-trafficking leads to results that are similar to those found for the other two groups: there is a relative independence between the former and the latter. Unlike paramilitary groups, the municipal tax structure does not help to explain the presence of drug-trafficking groups.
Several succinct observations can be made on the relationship between municipal tax revenue and the regional dynamics of the Colombian conflict. First, armed groups do not seem to make territorial decisions based on the overall volume of municipal tax resources. Neither tax revenue nor non-tax income, expressed in per capita terms, appear to significantly alter the expansion strategies of such groups. Despite this, transfers from the central government to local administrations do attract guerrillas, and some very specific income items also seem to have an impact on the geography of the conflict. The existence of tax resources from the transportation of energy products is the component of the local tax income that shows a greater capacity to predict the presence, above all, of the guerrillas.
A second indirect observation is that the guerrillas, not the paramilitary, appear to have the ability to get resources in one place in the country and transfer them to another in order to support their fronts a fact that could be reflecting a more centralized organization.
2.2 – CURRENT EXPENSES AND ARMED PRESENCE
Current public expenditure per inhabitant in the municipalities presents an enormous variance. The composition of public consumption, and in particular the relationship between general expenditures and the municipal employee’s payroll, also shows great regional differences. This first breakdown of expenditure does not contribute to identify those municipalities that have a guerrilla presence.
Another option is to break down expenditure into the various municipal offices –around twenty- to which it may be allocated. Two expenditure categories appear strongly associated with guerrilla presence. The first is the one channeled through an office of accounting supervision –the “Oficina de Control Interno” (OCI). The second is the expenditure that goes to a technical advisory office called the “Municipal Unit of Technical Agricultural Assistance” or UMATA (BR, 1998).
The negative relationship between these two types of municipal expenditure and the presence of guerrillas appears so strong that it does not depend on the manner in which the variable related to such expenditure is specified. Whether it is expressed in monetary units per capita, or as a proportion of the total expenditure, or as a dummy variable associated with the existence of these expenditures (e_control) (see Table 4), a negative and statistically significant relationship is found between this variable and guerrilla presence. A peculiar aspect of this association is that while the nature of the two offices is radically different, they have a common subjacent element that could be driving a negative association with guerrilla presence.
A safe way to interpret this result is to look at it as an index of the degree of isolation of the municipal administration. In the same way as the existence of telecommunication offices in the municipalities was interpreted as reflecting the capacity of the inhabitants to keep in contact with the rest of the country, local offices such as UMATA or the OCI could be interpreted as the difference between an isolated municipal administration and one which is closely linked with the rest of the country. For example, one could assume that, regardless of its main function, the existence of such offices represents a better technology for public administration in the municipality. In short, this variable could be taken as an indicator of less backwardness and, above all, of less isolation of local governments.
This indicator’s relevance could involve supervision and control of municipal public expenditure. These results suggest that the presence of guerrillas seems to be favored by the disorder and lack of accounting control over operating expenses in municipal administrations.
As concerning paramilitary groups, aggregate current expense items do not help explain their presence. In contrast to what was stated for the guerrillas, breaking down public consumption according to the offices to which they are intended does not help identify those localities under paramilitary influence. In a way, these results corroborate earlier findings: paramilitary groups appear to be less related than the guerrillas to municipal finance. This is consistent with the scant testimonial evidence of the armed clientelism carried out by paramilitary groups, at least up until the time of the data used in this paper were gathered, late 90s.
When expenses are broken down according to the offices to which they are remitted, several results regarding drug trafficking groups are worth mentioning. First, there is a positive and very significant relationship between drug trafficking and per capita expenditure in two types of municipal offices: on the one hand, the municipal council and the Secretariat of Government, and on the other, the planning office. It must be noted that the first two are precisely those municipal offices involved in local politics. In fact, if these components are added up and are included as one explanatory variable that could be termed political expenses in the municipalities (e_political) (see Table 4), a positive relationship with the presence of drug-traffickers is found. It is interesting to note that it is precisely this armed group -traditionally considered business driven and relatively aloof from political and proselytizing activity- whose influence appears to be mostly associated with electoral politics in the municipalities. This result corroborates the results found with indicators of political activity –associations, mayoral elections- which appear more closely linked to the presence of drug-dealers than to other armed groups.
It would not be very convincing to interpret this coefficient by proposing that greater expenses in local politics act as a magnet for drug traffickers. It is more reasonable to argue –in line with what was suggested for the relationship between these and the indicators for social capital - that drug traffickers act as facilitating factors, and even as sponsors, for political proselytizing in the municipalities.
Outlay of the planning office is another item of public expenditure (e_planning) (see Table 4) that is positively associated with the presence of drug trafficking. This effect is so clear that even the dummy variable associated with this type of spending -in other words, whether a planning office exists or not- predicts the presence of drug trafficking. How could one explain this association between the formal planning of expenditures and investment and the presence of drug trafficking? A first option is that a minimum physical infrastructure is required for drug trafficking activities. The absence of a planning office in a municipality would be a sign of a lack of required minimum standards, which in turn would affect drug traffickers´ decisions regarding location. Looking at the causality the other way around, one might consider that drug trafficking resources generate a greater demand for specific infrastructure –for example public utilities or transportation- that in turn impose on the municipalities a need to plan such investment. The fact that the presence of drug-trafficking does not appear to have a perceptible impact on municipal tax revenue does not imply that resources outside the tax flow do not exert pressure on investment expenditure: a warehouse for drug trafficking activities or even a house for a drug trafficker built in a municipality involves a greater demand for roads, water or sewage, regardless of the origin of the resources with which it was constructed.
2.4 – PUBLIC SECTOR INVESTMENT AND THE PRESENCE OF ARMED GROUPS.
Municipalities that benefit from the programs of the National Plan for Rehabilitation (NPR) are highly associated with guerrilla and paramilitary presence (k_NPR) (see Table 4). The NPR was a public investment program for high-conflict regions that was set up during the Virgilio Barco Administration (1986-1990), so it seems reasonable to suggest that the continuous presence of the guerrillas in a locality has been influential for the allocation of NPR resources to localities.
Unlike total operating expenses, that do not help explain the influence of armed groups in the municipalities, total payments for capital investment, per capita terms (k_total), appear to be associated with the guerilla presence and to have a statistically significant effect (see Table 4).
This positive relationship between public sector capital investment and the presence of guerrillas, while statistically significant for the total amount, is not relevant with any of the disaggregated items of capital investment. In other words, the presence of guerrillas appears to be related to the total amount available for capital investment independently of the way in which these funds are allocated.
It is not easy to give a simple explanation for this association. In principle, it could be argued that larger resources for capital investment are an incentive for guerrilla presence, since the may want to influence the allocation of funds. This would be consistent with the idea of armed clientelism. At the same time, one may think of causality the other way around. The country’s most problematic regions, and particularly those under the influence of the guerrilla groups, have received greater inflows of capital investment. One might argue that both effects mutually reinforce each other.
The guerrilla presence also seems to be affected by some types of investment. The proportion of investment in road building and maintenance (k_roads) shows a statistically significant effect, while investment in rural electrification (k_electrical) (BR, 1998) shows a negative one. One way to interpret the positive coefficient for capital investment in construction and maintenance of public works, an important sector that takes up an average of 10% of capital expenditures and can be as high as 50% in some municipalities, is based again on the idea of armed clientelism. It is through public works that the guerrillas’ power to influence capital investment decisions in the towns may be seen to be strongest. The negative coefficient for investment in electrification can be associated with the frequent bombings of power infrastructure on the part of the guerrillas, which would make their presence in a particular area a factor that can negatively affect the allocation of resources in that area.
In contrast to what has been observed for guerrillas, neither the level of public investment per capita nor its allocation help to explain the pattern of territorial influence of the paramilitary groups.
The third armed group, drug traffickers, shows quite peculiar links with public investment in the municipalities. First, and in contrast with guerrillas, there is negative, and statistically significant relationship, with the total amount invested per capita (k_total). Also, certain specific kind of capital expenditures show a positive and statistically significant relation with drug trafficking: investment in “physical education, recreation, sports and culture,” in “municipal equipment,” and, lastly, in electrification programs. A significant association is found for some non-basic infrastructure, expenditures like soccer fields (k_sport). One can posit that the presence of drug trafficking is an incentive for certain non-basic infrastructure investment expenditures instead of arguing that towns having a good athletic team and a preference for sports are more attractive to this type of illicit activity. Testimonial evidence in support of the former interpretation is not scarce (Salazar, 2001). This result corroborates previously stated ideas according to which drug traffickers act as promoters or patrons of certain kinds of activities specifically related to leisure time, sports and even culture.
In Colombia there has been a long intellectual tradition that makes a categorical distinction between politically motivated groups and common criminals, such as narco traffickers. At the positive or explanatory level, it is understood that political criminals differ from their ordinary brethren basically in their intentions and not in their actions. It is thought that the common criminal is inspired to seek personal financial reward, while the political delinquent is socially and altruistically motivated . In a second, normative, dimension, the theory suggests that only ordinary criminals should be penalized and that the rebel must be accorded a privileged treatment: in this case, above all, negotiation should be sought . For different reasons, it is considered that a criminal sanction against rebels is, not only unworkable, but that it can even be counterproductive.
The critique to this distinction between political and common crime can be summarized in two points. The first is its excessive adherence to 19th century modes of thought and the failure to incorporate recent theoretical developments about organized crime. The second, more pertinent, point is its obvious divergence from the contemporary conditions found in the country, that highlight serious discrepancies with the ideal type classification schemes that have been proposed. Municipal data shows that guerrilla presence in the municipalities does not completely fit the exclusively political model and, on the other hand, that narco traffickers do have a strong influence in local politics. Several results challenge the traditional views about both groups and suggest that the line that separates them is increasingly tenuous and confused. In short, these results seem to corroborate the assertion made more than a decade ago in the sense that drug trafficking, to a greater extent than guerrillas or paramilitary groups, has “infiltrated the remotest corners of the society, its politics, economy and even culture and sports.” (Krauthausen and Sarmiento, 1991).
BR (1998). “Banco de datos de Finanzas Públicas Municipales”. Bogotá, Banco de la República. Data was obtained directly from Banco de la República.
Bolívar, Ingrid J. (1999), “Descentralizar el Conflicto”, Bogotá: Revista Cien Días del CINEP, Vol. 10, No. 45, (Jul-Nov)
Collier, Paul (2000), “Economic Causes of Civil Conflict and Their Implications for Policy” in Crocker, Chester A., Fen Osler Hampson and Pamela Aall eds. (2000) Turbulent Peace: The Challenges of Managing International Conflict, Washington, D.C.: United States Institute of Peace Press
DANE (1998). “Estadísticas de Justicia” and “Censo nacional de Población y de Vivienda”. Bogotá: Departamento Nacional de Estadística. Data was obtained directly from DANE.
DNP (1998). “Indice de desigualdad a partir del Indice de Calidad de Vida (ICV)”. Data was obtained directly from DANE. Methodology for the index can be downloaded www.dnp.gov.co/ArchivosWeb/Plan_Colombia/articulos_paz_desarrollo/articulo10_violencia_equidad/articulo10.doc
Echandía, Camilo (1996), “Principales Tendencias en la Evolución Reciente de la Guerrilla y la Violencia en Colombia”, Presidencia de la República. Consejería para la Paz. Observatorio de la Violencia
Echandía, Camilo (1999), El Conflicto Armado y las Manifestaciones de Violencia en las Regiones de Colombia. Bogotá: Biblioteca para la Paz. Presidencia, Observatorio de la Violencia
Fajnzylber Pablo, Daniel Lederman and Norman Loayza (1998). Determinants of Crime Rates in Latin America and the World. An Empirical Assesment. Washington: World Bank Latin America and Caribbean Studies.
Fajnzylber Pablo, Daniel Lederman and Norman Loayza (1999). Inequality and violent crime. Washington: World Bank
FS (1998), Municipios y regiones de Colombia: una mirada desde la sociedad civil, Bogotá : Fundación Social. Data was obtained directly from Fundación Social and Departamento Nacional de Planeación.
Krauthausen C y Sarmiento L F (1991). Cocaína & Co. : un mercado ilegal por dentro. Bogotá: Tercer Mundo
Levitt, Steve and Mauricio Rubio (2005), “Crime and crime prevention in Colombia” en Alesina, Alberto Ed (2005). Institutional Reforms : The Case of Colombia, Cambridge: The MIT Press
Rangel, Alfredo (1997), “El poder local: objetivo actual de la guerrilla”, publicado en Descentralización y Orden Público, Bogotá: Fescol y Milenio
Rubio, Mauricio (2002) “Conflicto Armado y Finanzas Públicas Municipales en Colombia” – Documento CEDE 2002-17. Downloadable at
Salazar, Alonso (2001). La parábola de Pablo. Auge y caída de un gran capo del narcotráfico. Bogotá: Planeta
Sarmiento, Alfredo (1998). "La Violencia y las variables sociales" en DNP (1998) La Paz: El desafío para el desarrollo. Bogotá: TM Editores, Departamento Nacional de Planeación
Thoumi, Francisco (1994). Economía Política y Narcotráfico. Bogotá : Tercer Mundo.
Vélez, María Alejandra (2000). “FARC-ELN Evolución y Expansión Territorial” Documento CEDE 2000-08. Bogotá: Universidad de los Andes. www.mindefensa.gov.co/conflicto_armado/subversion/historiasubversion.pdf
* Universidad Externado de Colombia. This work was carried out as part of a project funded by the Fundación para la Promoción de la Investigación, la Ciencia y la Tecnología of the Colombian Central Bank (Banco de la República). It was carried out at the Universidad del Rosario’s ’s School of Economics in Bogotá. Daniel Vaughan was the research assistant.
 See for example Correa, Néstor Raúl (1997), “Nexos descentralización – Orden público”, in Fescol (1998) Descentralización y Orden Público, Bogotá: Fescol y Milenio or Castro, Jaime (1998), Descentralizar para Pacificar, Bogotá: Planeta Colombiana Editorial.
 See Peñate, Andres (1991) "The Elenos" in Arauca:Politics an oil in a Colombian province", Oxford University: M. Phil Thesis or Deas, Malcolm and María Victoria Llorente (1999). Reconocer la Guerra para Cosntruír la Paz. Bogotá: Uniandes, CEREC, Norma
 “Los Alcaldes del País están Solos” (The Mayors of the Country are Alone), El Tiempo newspaper, Tuesday September 14, 1999, page 9A.
 See Thoumi (1994) as well as Rocha, Ricardo (2000), La Economía Colombiana Tras 25 años de Narcotráfico, (The Colombian Economy after 25 years of Drug-Trafficking). Publisher: Siglo del Hombre Editores, and Reyes, Alejandro (1997), "Compra de Tierras por narcotraficantes", (purchase of lands by drug-traffickers) in: Francisco Thoumi, et al, Drogas Ilícitas en Colombia: su impacto económico, político y social, (Illicit drugs in Colombia: their economic, political and social impact) United Nations Development Programme.
 As in the “Analysis of the economic and social structure has to focus on overcoming the objective causes of violence: poverty and unequal distribution of income” “Peace Policies for Change” National Government Plan, published in the newspaper El Espectador, January 11, 1999.
 The close relationship between poverty and inequality indexes does not explain this result. First, the linear correlation between poverty and inequality is quite misleading because there is enough variance of both variables around the mean given the other. For example, for a Gini index of 0.15, the range for the poverty index is from 10% to 95%. On the other hand, when one variable is included and the other is not the coefficients keed their sign: negative for poverty and positive for inequality.
 Population density, on the other hand, does not show any significant association with armed groups presence.
 The following indicators of state presence, all from FS (1998), were also tested: total number of Public Functionaries, Police Stations, Judicial Courts, Public Notary Offices, Caja Agraria (Agrarian Bank) Offices, teachers, Public Libraries, Firemen, Jails, Tax Collecting Offices, Hospitals, First-Aid Stations. None of them was useful to discriminate municipalities with presence of any of the armed groups.
 This variable refers to the number of political demonstrations of any kind –marches, strikes- that took place in the rural area of the municipality during the period 1990-1996. (FS, 1998)
 On the one hand, it is reasonable to state that municipalities in which the population has a greater capacity to pay taxes may be more attractive to the armed organizations, as this implies greater possibilities for extorsion or racket. On the other hand, a greater tax income may be taken as an indication of a greater consolidation of State authority, which in principle could diminish the probability for the presence of armed organizations.
 Orozco, Iván (1992). Combatientes, Rebeldes y Terroristas. Guerra y Derecho en Colombia. Bogotá: IEPRI, Temis. makes use of the ideas of Gustav Radbruch, the early 20th century German criminologist.
 See for example Comisión de Estudios sobre la Violencia (1995). Colombia : Violencia y Democracia. 4a Edición. Bogotá : IEPRI, Universidad Nacional, Colciencias.