Luis Fernando Medina
Assistant Professor, Department of Political Science The University of Chicago. 5828 S. University Avenue (Pick 507) Chicago, IL. 60637. Ph: (773) 702-2941. Email: lfmedina@uchicago.edu. Fax: (773) 702-1689

SUMMARY
In recent years, several Colombian scholars have studied this country’s civil conflict through the lenses of economic reasoning and state-of-the-art statistical testing. Their analyses place most of the explanatory burden of the conflict on the existence of lootable resources and organized crime, not on any specifi c socio-economic factor of Colombia’s reality ("objective causes"). This paper criticizes their claims while accepting their criteria and methods. In particular, it contends that their formal approach to civil conflicts is fl awed, a fl aw that carries over to the type of empirical tests conducted and that there are several gaps in the reasoning from statistical results to the overall interpretation of the conflict.
Key Words: Civil War, Colombia, theories, empirical test.

RESUMEN
En los últimos años, varios académicos colombianos han estudiado el confl icto civil de este país a través de la lente del razonamiento económico y el examen estadístico de punta. Sus análisis ubican la mayor parte de la carga explicativa del confl icto en la existencia de recursos saqueables y crimen organizado, en vez de atribuirla a cualquier otro factor socioeconómico de la realidad colombiana ("causas objetivas"). Este trabajo critica sus posiciones aunque acepta sus criterios y métodos. En particular, sostiene que su enfoque formal hacia los confl ictos civiles es impreciso, imprecisión que se debe al tipo de exámenes empíricos.
Palabras Clave: Guerra Civil, Colombia, Teorias, prueba empírica.

1. INRODUCTION
Hegel famously likened philosophical reflection to Minerva’s owl, soaring at dusk. With this he meant that full, reasoned understanding of historical processes was only possible once they were over. Of course, Hegel is excused for not having Colombia in mind, a country that was barely born by the time of his death. Otherwise, it might have occurred to him that, whatever truth his remark may hold, intellectuals, those epitomes of human curiosity, could hardly wait to analyze a civil war whose "dusk" is still yet to arrive, even 40 years after it began. Colombia’s civil war may go down in history as one of the best documented and diagnosed civil conflicts in record. This is no cause for celebration, but a sad comment on the country’s ordeal: the war has been so long and painful that it has given Colombia’s intellectuals the reasons and the time to measure, study and discuss it with the professionalism intellectuals in other latitudes reserve for historical events.

Not only the war has been long enough to be analyzed as it unfolds, it has been long enough to see intellectual orthodoxies rise and fall, not a mean feat of endurance when we consider the examples of other intellectual orthodoxies such as keynesianism and monetarism. The last decade has witnessed the rise of a new orthodoxy among academics, one that is purported to refute and displace the "objective-causes" theory launched in the 80’s by a set of social scientists convened by the government to study the civil conflict. The new orthodoxy has not yet been christened with any catchy name but, in the interest of brevity, and for reasons that will become apparent, I will call it here the "supply-side" theory of the civil war.

While these academic exercises make a valuable contribution to our democratic disagreements, they can never take its place: hence the danger of orthodoxies. The supply-side theory is acquiring the undeserved status of a fact, putting down other perspectives that were encompassed in the objective-causes school. Left unchecked, the supply side theory is likely to cross the line that separates analysis from dogma, theory from political weapon.

This paper offers a polemic against the supply-side views on Colombia’s civil war. It is by no means the first, or the most comprehensive polemic. My more modest goal is to offer an internal criticism of the supply-side school, a criticism that accepts this school’s conceptual foundations and standards of relevance but that disagrees with its conduct of the analysis, the conclusions it reaches and the implications it draws. In the interest of coherence, I will proceed "downstream," that is, I will discuss first the theory, then the data and then the recommendations. After offering a brief history of the emergence of the supply-side paradigm in Colombia (Section 2), I will claim that its analytical foundations are shaky as a result of an incorrect formal treatment of the problem of civil wars (Section 3). Then, I will argue that the empirical tests that supply-siders have offered remain inconclusive and are consistent with many other interpretations of the conflict (Section 4). While up to that point of the argument I accept the criteria for empirical testing offered by supply-siders, in Section 5 I will question the relevance of such criteria. Section 6 takes issue with the policy implications that supply-siders draw from their analysis, arguing that they are, for the most part a nonsequitur. Section 7 summarizes and offers some concluding remarks.

2. THE RISE OF "SUPPLY-SIDE" EXPLANATIONS
What is often referred to as the objective-causes school is rather a large, not necessarily cohesive, body of scholarship that flourished during the 60’s and 70’sin Colombia and that came of age with the famous "violentologist’s report" (officially known as the report of the Comisión de Estudios Sobre la Violencia) of 1987. In spite of a large diversity within members of this school, they shared the belief that the ultimate causes of Colombia’s guerrilla struggle was to be found in a set of social pathologies, be they economic inequality, social exclusion, political constraints, etc. Since, presumably, these pathologies were objective, that is, not a figment of the guerrillas’ imagination, such belief earned this school its name.

Although its recent critics label it as an orthodoxy, perhaps to underscore their own combativeness, the objective-causes school, in spite of its prominence and the serious intellectual work of its proponents, was anything but. At the time, the orthodox position, the one that carried the day in shaping policy, was, as we shall see, closer to the "new" supply-side theory: the official language up until the early 80’s would refer to guerrilla fighters as "bandits," "thugs" and "agents of foreign doctrines" wasting no time in any discussion of the putative objective causes. It was only in the 80’s, once the notion of a peace settlement with the guerrillas gained enough legitimacy to become a deliberate government policy, and even then at great political cost for the Administration, that the objective-causes theory acquired currency outside the rarefied environment of the academia. In a telling sign of the political moment, the same set of ideas that had previously given intellectual ammunition to the opposition groups, was now enlisted to give a veneer of academic respectability to the government’s decisions.

It was to be a short-lived achievement. For starters, by the time the commission reported, the heady days of Betancur’s peace process were over. From the denouement of Betancur’s peace talks the Barco Administration concluded that the body politic was not ready for high-profile, all-encompassing, negotiations with the guerrilla, lead-ing to and accompanied by grandiose reform schemes. Whatever one believes about the strategy pursued by the Barco Administration, its peace policy hardly seemed a brain-child of the commission. This is much more true of the Gaviria Administration, where some prominent supply-siders held important posts. For all its flaws the "objective causes" school can fairly be exonerated of one of the charges supply-siders level against it: prolonging the conflict through a misguided diagnose. The commission never had the impact over policy that would make it liable to such a charge.

After its brief political apex passed, the objective-causes school became the target of academic attacks from different scholars that, only for convenience, I group here under the rubric of supply-siders. In spite of some differences among themselves, they shared several characteristics. Whereas the objective-causes scholars were usually sociologists, historians or political scientists, trained in Continental traditions (French or German), and working in public universities, supply-siders were economists, trained in the Anglo-Saxon tradition, and working in private universities. As such, they emphasized the rigor of formal reasoning as the source of theoretical propositions and of econometric techniques as the standard to test them.

Operating within this mindframe, they established some findings that, in their view, definitively refuted the objective-causes school:

1. Only a small fraction (between 6 and 12%) of the variance in homicides across municipalities in Colombia is explained by socio-economic conditions. (Sánchez and Nuñez, 2001).
2. The time-series of homicides in Colombia does not track the time series of in-equality, poverty, social exclusion or any other "objective cause." Periods of im-provement in social indicators are often accompanied by periods of increase in homicides.(Montenegro and Posada, 2001).
3. It is not the poorest regions of the country the ones afflicted by the highest homi-cide rate, but instead those municipalities with brisk economic growth. (Mon-tenegro and Posada, 1995).
4. "Lootable" resources are one of the best explanatory variables for the location of irregular armies. (Montenegro and Posada, 2001).
5. Poor "state capacity," especially of the judiciary and law-enforcement branches, is highly correlated with high homicide rates. (Rubio, 1997).

These findings led supply-siders to conclude that violence in Colombia was not the result of some generalized socio-political malaise, but rather of organized crime. As drug barons acquired their formidable force in the 80’s, so the argument goes, their criminal activities fueled Colombia’s conflict through several mechanisms. First, they financed handsomely both guerrillas and paramilitary groups. Second, they overbur-dened an already weak law-enforcement apparatus, rendering it unable to check the irregular armies. Third, they created areas of the country awash in "lootable resources" thus giving the incentives for these same irregular armies to capture the surplus.

In their diagnosis, supply-siders focus on the resource base that turns factions in a civil conflict into viable armies, affording a tiny role, or none at all, to the circumstances that could presumably lead individuals to want to confront the State in the first place. This emphasis on the supply of resources needed for rebellion warrants, I think, the label of "supply-siders."

As if the econometric evidence were not enough, supply-siders can count on other confirming evidence to support their views. No one questions, for instance, that the FARC controls today substantial parts of the coca-producing complex in Colombia and that it extorts vast resources from both citizens (through kidnaping) and the State (through crafty schemes of intimidation and corrupt bargains with municipal councils). (Rubio, 2002).

Much of the intellectual support for supply-siders comes from state-of-the-art schol-arship in the area of civil wars. Collier (2000) redefined the field by criticizing the "grievance-based" theories and tilting the balance in favor of the "greed-based" the-ories. In his analysis, the notion that social grievances lead to civil wars is logically flawed and empirically invalidated. It is logically flawed because any connection be-tween a grievance and the decision to act on it is blocked by the free-rider problem. Would-be rebels find that grievance redress is a public good and prefer to let the bur-den fall on others. It is empirically invalidated because grievances have poor predictive power in explaining civil wars (2).

Instead, since any insurgent group needs resources to survive as a challenger to the state, the access to resources is better at explaining the existence of insurgency. Hence the title of Collier’s paper: "Rebellion as a Quasi-criminal Activity."

Likewise, Fearon and Laitin (2003) have concluded that grievances have little ex-planatory power but, unlike Collier, place the emphasis more on "state capacity" rather than on the existence of lootable resources. At any rate, both lines of study coincide in some objective factors that make countries more vulnerable to insurgencies, being "rough terrain" a prominent one among them.

This set of assertions led many supply-siders to conclude that, instead of trying to reach a political settlement with irregular armies that are more interested in their own purse than in that of the poor they claim to represent, Colombia should try to reestablish the rule of law, focusing its resources in the judiciary and the military. Clearly, in recent years supply-side theories have accomplished something the objective-causes school could only dream about: see their analyses quoted, often verbatim by government officials and their conclusions elevated to the status of official policy.

Although far from complete, this summary describes the three main components of the supply-side school: a. a theoretical model of rebellions, b. aset of empirical tests based on the observable behavior of irregular armies and c. aset of policy prescriptions. In what follows, I will discuss each of these components in this same order.

3. INSURGENCIES, PUBLIC GOODS AND COLLECTIVE ACTION
Although he is in the venerable company of Olson (1965) and Tullock (1971), Collier is wrong in his argument about free-riding. Grievance-redress is not a public good. One of the key properties of a public good, often overlooked by the followers of Olson’s collective-action theory, is that it could be provided by one agent, if only he had all the resources available. Private shipping companies may build a lighthouse even if the result benefits other companies. Grievance-redress is not like that. Redressing a grievance requires institutional change, something no individual can provide unilaterally. If a citizen has a grievance against, say, the property rights of another individual over a piece of land, no effort of his part, no matter how much resources he has will be enough to unilaterally change such property right. He can violate it, for sure, through forced eviction. But, unless a sizable group, at least the relevant members of the com-munity, join him in repudiating such property right, it will stand. A property right is a right because it results from a special coordination of beliefs among members of a society.

In a way, Collier has it backwards. It is not that rebellion is a quasi-criminal activity (the title of his now classic paper), but that crime is a quasi-rebellious activity. The criminal, unable to create the political facts that will tilt the institutional balance in his favor, which is what the rebel does, is reduced to unilateral action, always risking the punishment reserved for those who contravene, without transforming, the recognized laws of a polity.

This conceptual confusion carries over to the choice of analytical tools and, from there, to flawed empirical tests. As Taylor (1987) has pointed out, collective action is not a public-goods problem; it is not a situation where an individual’s optimal choices are independent of those of other actors. Collective action situations resemble coordination games with multiple equilibria, as opposed to "Prisoners’ Dilemmas" that only have unique equilibrium (and a dominance-solvable one at that) (3).

This seemingly arcane point about the number of equilibria in a game has several profound implications for the theory’s empirical tests. By definition, a model with multiple equilibria generates many possible outcomes. Thus, when we test its predictions empirically, we must be prepared to find weak statistical correlations. In other words, the data may seem to suggest that we should reject the model, even though the model is, in fact, correct. An example from economics may be useful to illustrate this problem.

Consider a textbook market for a simple commodity, corn. Under the standard assumptions of supply and demand, there will be a unique price-quantity equilibrium for this market. If we believe that different economies, call them "cities," face the same supply and demand functions, we can collect the data of prices and quantities and test their correlation with other exogenous variables. For instance, since we know that wheat is a substitute of corn we can conjecture that when wheat becomes cheaper, say because of an unusually large harvest, this will depress the demand for corn and, with it, its price. This is a testable conjecture and, if our assumptions are right, we are likely to see that in a cross-sectional analysis, cities with abundant wheat will have cheap corn, and cities with scarce wheat will have expensive corn.

But now imagine that, through some freak of economic laws, the market for corn has multiple equilibria. Although rare in simple markets such as that of corn, multiple equilibria can arise in many other contexts, economic and political. For simplicity, assume that there are only two possible equilibria, one with low prices and one with high prices. Any city, regardless of the size of its wheat harvest, can be in one of these equilibria, depending on the specifics of its "equilibrium-selection" process. It would now become possible to observe very expensive corn in cities with large wheat reserves and cheap corn in cities that hardly produce any wheat: the cross-sectional estimates would display no correlation, or a very weak one, between wheat harvests and corn prices. In that context, supply-siders would rightly refrain from concluding that economic theory is wrong and that there is no connection between the price of a good and the availability of its substitutes. They would, instead, look for different testing procedures. For instance, we could still pick up a connection between wheat harvests and corn prices if we looked at the time-series estimates for each city, instead of bundling all the data in one single analysis (4). After all, the existence of multiple equilibria means that, in a way, each city is unique and that it makes little sense to try to prove or disprove highly general theories comparing these inherently different cases.

Likewise, the collective action problems behind grievance-redress have multiple equilibria. As such, we are unlikely to find impressive cross-sectional correlations between grievances and civil war even if grievances cause civil wars, in much the same way as large wheat harvests cause low corn prices. Notice that I am not saying that this proves grievance-based models. I am simply saying that poor cross-sectional correlations do not refute them.

Intuitively, there is no reason to believe that there is a cross-sectional "law" linking grievances and conflicts. Supply-siders in Colombia keep pointing out that Colombia’s socio-economic indicators are not as bad as those of other countries in the region that do not have guerrillas (Ecuador being an oft-quoted example). Certainly this is an interesting observation but has little value as proof of anything. In particular, it does not refute a grievance-based explanation of Colombia’s conflict. There might be many reasons that lead Colombians to revolt at situations that Ecuadorans accept placidly; that is each country’s problem. Colombian peasants are under not more of an obligation than Ecuadorans to conform to laws of social sciences when deciding what to do about their plight.

4. THE TREACHEROUS ROAD FROM ANALYSIS TO TESTS
The logical confusion between public-goods and coordination games may not be that far-reaching because supply-siders have resorted to other types of arguments and empirical tests. In this section I turn to some of those tests devised by supply-siders to evaluate the arguments from believers in the objective-causes. For the most part, they are deeply flawed but the objective-causes school deserves some blame for this. With a hostile view of the type of empirical testing common in economics, scholars of the objective-causes school never spelled out their theories’ testable implications in terms economists could recognize. Whether they should have done so or not, depends on one’s views on the value of econometric testing. But if I launch a theory and then neglect to tell my potential critics what they should expect from it, how they should test it and what should and should not count as evidence for it, I have no right to cry foul if those same critics design their own tests, in their own terms. Drawing from their training as economists, supply-siders have designed tests for the objective-causes theory that consistently refute it. Those tests, however, are uninformative because they do not speak to the major issues raised by the objective-causes school. I will not claim that my views on the matter represent the "true" theory of objective causes: this is, after all, a school known for its internal diversity. What I will claim is that there are reasonable statements that establish a connection between Colombia’s socioeconomic conditions and its conflict and that these statements have not been refuted by the tests of supply-siders.

Even if we grant that cross-sectional analysis across countries is uninformative, supply-siders could argue, the objective-causes theory has also been refuted by the time-series analysis. Furthermore, one still has to deal with arguments about cross-sectional analysis within the same country. As mentioned above, homicides in Colombia have increased in periods in which poverty, inequality, social spending and political inclusion have decreased (Montenegro and Posada, 2001) and, across municipalities, display weak, even perverse correlations with socioeconomic conditions.

There is something inadequate about this fixation with homicide rates. Doubtless, people are getting killed in Colombia at appalling rates and any contribution social sciences can make to ameliorate this is welcome. But if the point is to analyze the conflict between the government and the irregular armies, homicides only constitute a small part of the picture. Let’s not forget that only a small fraction of all the homicides in Colombia are directly linked (albeit a tricky definitional exercise) to the armed conflict. Guerrilla groups everywhere murder people, to be sure, but that is not all they do. They also procure resources, try to control the local politics of many areas, precisely to avoid having to resort so much to murder and even try to win "hearts and minds." It is a well-established regularity of civil wars (e.g. Kalyvas (2004)) that areas firmly under the control of an army tend to be the safest ones. If right now the FARC were engaged in a wildly successful propaganda campaign, attracting underground sympathizers by the thousands, all of them willing to risk their lives for the FARC, most of us, including supply-siders, would consider this to be an increase in the intensity of the conflict. But the homicide charts would not help us to articulate our concerns.

Taken out of context, the data on homicides are useless as indicators of the conflict. The objective-causes school could be right under many possible scenarios, all of them compatible with these data. One could imagine that, as a response to the courageous struggle of selfless romantics, Colombia’s government decided to put an end to most of the country’s injustices, only to see those romantics being slaughtered by the paramilitary. Alternatively, one could imagine that, under the impetus of a benign, reformist government, Colombia’s social indicators improved vastly during the 70’s and 80’sand so, the population withdrew support from the guerrillas who then turned to more violent tactics. Who is killing whom, when, where, why? Regression analyses are silent to such subtleties.

Not only homicides may be the wrong variable to look at in assessing the evolution of the conflict, the findings about them are far less impressive than what supply-siders claim. For example, Sánchez and Núñez (2001) establish that only 6% of the variance in homicide rates across municipalities is explained by socio-economic conditions. But then they add, somewhat anticlimactically, that almost all the remaining variance, more than 90%, is explained by the presence of guerrillas or paramilitaries.

It is hard to know what to make out of this finding. If we interpret this to mean that socio-economic conditions do not explain violence, then we have to conclude that habits of sexual behavior and drug-consumption do not explain AIDS because all of the variance in AIDS is accounted for by the presence of the HIV virus in the subjects’ bloodstream. The academic production of the objective-causes school has been so large that I am acquainted with only a tiny fraction of it. But I would be astonished if any of its members ever suggested that inequality could kill people by itself, without the intervention of any human agent, be it a guerrilla or paramilitary fighter. This is an instance of a test that does not prove anything because it fails to capture the essence of the argument it was meant to test.

Intra-country, cross-sectional analysis is another instance of such reckless testing, also motivated by the unfortunate focus on homicide rates. The poorest municipalities may not be the country’s most violent, granted. But the objective-cause diagnosis never ruled this out. Conceivably, the poorest municipalities could not be very violent be-cause they are already overwhelmingly supporting the FARC which enjoys a monopoly that allows it not to resort to murder. Even if this is incorrect, and I believe it largely is, the idea that a group must operate only where its constituency is located is crass. Charitable organizations such as the Salvation Army routinely deploy their members in affluent parts of the city, precisely because there is where the money is. Agricultural lobbies locate their headquarters in their country’s capital, as close as possible to Congress, because that is the most strategic location to further their goals. Armed insurgencies need resources; in that Collier is right. Whatever their goals, constituency and audience, smart armed groups will invest heavily in resource-rich areas.

The results on lootable resources are no more airtight. Colombia seems at first glance a poster-child for the lootable-resources theory of civil wars, that is, the notion that the ultimate cause of an insurgency is that the existence of such resources gives incentives and opportunities to irregular armies. Homicide rates in Colombia track nicely, both in terms of time-series and cross-sections, the presence of rents from the coca trade, oil, cattle growing and other such activities.

But it is not clear what makes a resource "lootable." Illegal drugs and diamonds are often adduced as the quintessential example. But if there is something inherent to these commodities that makes them lootable, it is hard to understand why they stop being so once they reach the US territory. Every day millions of dollars worth of diamonds and cocaine, the latter without enjoying any protection from the law, cross the US from coast to coast, often through very rough terrain, and no guerrilla has tried to predate them. (To be sure, the cocaine trade in the US generates high homicide rates, but no insurgency whatsoever.)

One could refine the theory to say that resources are not lootable by virtue of some physical property but by the fact that they can be funneled into illegal activities. But then the theory looses all its explanatory power because, by definition, the moment some resource enters the accounts of a guerrilla group, it has been illegally used. The FARC could set a very profitable real-estate operation in Manhattan and funnel its profits to purchase arms, not unlike what reportedly, Ossama Bin Laden did when he used revenue from his family’s construction concerns to jump-start Al-Qaeda. To remain consistent with the theory, we would have to consider real-estate as another lootable resource. In that case, it would no longer make sense to talk about countries with abundant lootable resources and others without them; there would only be countries with large GDP and small GDP.

A variant of the lootable-resource argument attributes the presence or absence of looting to "state capacity." According to this argument, weak states, unlike strong ones, cannot prevent predation of different resources. But it is hard to give a non-circular definition of a weak state. If the concept of state capacity is to serve as a key to explain rebellions, we must be able to ascertain the weakness of a state without knowing if it is challenged by rebels. Thus far, this has been an elusive task. How can we determine how strong would Colombia’s state be without the insurgency?

All else been equal, rich guerrillas are more likely to grow and succeed than poor ones. This much is clear, and Colombia is a country where it has become relatively easy for guerrilla groups to amass vast resources. If, in light of this, we jump to the conclusion that socioeconomic phenomena are not behind the operation of these groups we must also be willing to tell an obese patient that, since she leads a sedentary lifestyle, her overeating has nothing to do with her problem. Supply-siders have a point: there are logistic, financial and geographic circumstances that benefit insurgent groups. No one disputes this, least of all the insurgents themselves. But supply-siders are distinctive in their insistence that resources tell all the story, rarely entertaining the possibility that both supply and demand, resources and grievances, explain Colombia’s conflict. It is hard to see what is the scientific purpose served by such single-mindedness in studying a phenomenon that most likely results from very complex processes.

5. THE LIMITS OF EMPIRICISM
Not only the empirical strategies designed by supply-siders to refute the objective-causes theories suffer several shortcomings, it is not even clear that data analysis alone can serve the purpose of understanding Colombia’s conflict. I do not deny that empirical analysis, informed by econometric techniques, is a powerful tool in social sciences, but it is no substitute for conceptual analysis.

Supply-siders have been very successful in getting across their view that the FARC are not a politically motivated group, but are instead a cartel dedicated to resource-extraction. This view, however, is not a conclusion of their analysis; it is an assumption they make that cannot be proven by any amount of regressions.

If anything, the quantitative patterns suggests that the FARC have a national political agenda, one that requires them to operate throughout all the country, to gather as much resources as possible to survive in the long haul. The FARC run a huge money-making operation. But this is not evidence that they do not have other goals. The best universities in the US have large endowments invested in many different sectors. For example, Stanford University owns the fancy Stanford Shopping Center and derives income as landlord of those great beacons of knowledge Nordstrom and Victoria’s Secret. No one would infer from this that Stanford is not interested in scientific research and that it has, instead, become an apparel business. I do not mean to compare the noble goals of Stanford University with those of the FARC, but one is left wondering why notions that are common-sensical in the study of legal organizations, elude the grasp of supply-siders when they think about the FARC.

Money can corrupt organizations as much as it corrupts individuals. There are many reasons to believe that this is happening to the FARC. But, whatever those reasons are, they have nothing to do with the statistical correlations between homicides and rents. If the FARC were pristine spokesmen for the poor, they would still be trying to get their hands at the many resources Colombia’s underground economy has to offer and, by necessity, this would put them at the center of not a few homicides: that happens to any group, ideological or not, that deals in such business lines.

Supply-siders give the impression of hoping that some magic number will reveal the nature of the problem; that some statistic, be it homicide rates or hectares of coca holds the key to understand what Colombia’s ordeal is. But social phenomena are not only quantitative, part of their essence is hidden behind the numbers.

Nothing in the number of people gathered in a public square, shouting jubilantly, tells us if it is a political gathering or the celebration of a glorious football victory. Nothing in the time-series of homicide rates in coca-growing towns in Colombia tells us if they result from personal grouches, from fights over unreported coca income or from executions to silence informants of the other side. Recent scholarship has established (Kalyvas, 2004) that much of the violence that occurs in civil wars, not only in Colombia but elsewhere, is only tenuously related, if at all, to grand, ideological motives. Still this does not mean that civil wars are just a bunch of killings. It rather means that civil wars are very complex processes that, among many other things, allow unusual amounts of killings to take place.

If the data cannot tell us what is the nature of the FARC, how can we know? I suggest we apply to this the same non-quantitative technique we would use to distinguish a political rally from a football celebration: read the signs. Once we lift our gaze from the data, we realize that there are many signs that the FARC are, indeed, a political group.

It is very hard for an armed group not to be involved in politics. If a group of well-trained men hang around a village’s main square, toting machine guns, without anybody being able to challenge them, this is already a display of military power that can easily become political power. Some locals will fear them, others may want seek them for "help" with their own conflicts, taking care of troublesome neighbors, intimidating unruly workers or arrogant patrons. Such a group, even if it does not espouse any particular ideology, is a force to be reckoned with in local politics. The Mafia in Southern Italy has historically been a powerful armed group not known for issuing political statements of any kind. But it is not possible to understand Southern Italian politics without its presence. It helped streamline political machines thus tilting the balance in favor of some parties at the expense of others. Just the fact of having guns, lots of them, makes irregular armies political groups.

It would take an especially thick set of methodological blinders to ignore that Colombia’s political landscape has been affected by the guerrillas and the paramilitaries. We do not need regression analysis to conclude that the targeted killing of more than 3000 activists of one political party (the Patriotic Union), is the kind of event that alters the balance of all the political forces in the country, including those sectors who would have otherwise made coalitions with it. We will never know how would Colombia’s politics look now if the Patriotic Union had been allowed to operate undisturbed. But it is certain that it would look very different. Those who have turned Colombia into the most dangerous place on Earth for trade-union leaders, know something that robust estimators cannot tell us: that the climate of labor relations, the balance of power between all the actors involved in them, depend on how many union leaders are killed, threatened or intimidated. Correlation tests over figures of homicide do not inform us about the message each homicide of a human rights activist, a school teacher, a municipal council member sends to other human rights activists, school teachers and municipal council members.

Colombians are rightly appalled at the way the irregular armies have targeted civilians. But this is not a rarity. It is part of what civil wars are. A civil war, especially if it is political, is a war for the control of the civilian population and few tools of control match threats and assassinations.

If next week the FARC launch a military campaign that kills hundreds of soldiers in the elite corps and leaves Bogota without access to the rest of the country for three or four days, one could reasonably predict that the price of the dollar would shoot up as many Colombians scramble to acquire an easily movable asset. In fact, this already happened in 1999 at a much smaller scale than the one of this imagined scenario. I can only think of one explanation: most Colombians understand that, unlikely as it is, if for some reason the FARC were to win the war and topple the government, the entire political, social and economic system of the country would be shaken down to its roots; in other words, it would mean a revolutionary change.

6. FROM ’IS ’TO ’OUGHT’?
Supply-side theories of Colombia’s civil war are an analytical exercise, presumably with purely descriptive goals. But supply-siders have been prompt to draw policy conclusions from it. In that step, they no longer rely solely on economic theory but also on their own ideological conceptions. There is nothing wrong about this; social scientists always bring their own ideological views to bear on their subjects of study. But a healthy discussion of these matters requires us to be clear on where analysis stops and judgment begins.

Montenegro and Posada (2001) offer what appear to be a catalog of the prescriptive implications supply-siders derive from their analysis. In resting their case, they spell out their conclusions while indicting the objective-causes school on several grounds. Although such indictment is not by itself a set of recommendations, it is well worth discussing because it sets the tone of said recommendations and provides analytical clues to assess them.

Their main accusation against the objective-causes school is that it has created a "guilt complex" in the government as it deals with the guerrilla, a complex that, they argue has several negative effects. First, it makes the guerrilla appear as legitimate in the eyes of the government’s negotiators. Second, it makes the negotiations loose focus as they range into substantive reforms instead of the specifics of demobilization and legalization of the guerrilla. Third, it creates the false illusion that the peace process is tasked with ushering in some type of ideal society. Fourth, it undermines the morale of the law enforcement community.

There is no evidence of such a guilt complex or of any of its alleged consequences. There is no evidence that Colombia’s law enforcement community believes in the objective-causes theory or has so much as allowed it to affect its own morale. I cannot recall a single instance of a guerrilla fighter acquitted of charges by a judge on grounds of the objective-causes theory. Colombia has granted amnesties in the past. But there is a large difference between the discretionary actions of a misguided, lenient judge and an amnesty that results from a political decision, processed through a sovereign Congress. In the latter case, which is the one Colombians are familiar with, there is no lapse in morale but simply a political redefinition of the conflict, something that embattled democratic countries keep doing constantly.

The objective-causes theory does not imply that a peace process must be an entire reconstruction of society. To be sure, some of its proponents might have believed that at some point, but it is not a logical consequence and, it is becoming increasingly clear to all Colombians that, however the conflict is solved, the resulting society will be mired dealing with its problematic legacy: no "ideal" society will result from any peace process, regardless of the conflict’s causes.

When they turn to the focus that peace negotiations should have, the accusations of supply-siders lead into their own normative conclusions. Instead of wide-ranging negotiations, that involve economic, political and social issues, supply-siders would prefer negotiations circumscribed to bringing the guerrilla back into legal politics, or perhaps even no negotiation at all, which is the approach of Uribe’s Administration that, apparently counts with the endorsement of several supply-siders.

This is a respectable point of view. A supply-sider could make, with impeccable logic, the following speech: "As a Colombian citizen, I do not want peace at any cost. I believe in the legitimacy of my government and my Constitution and am not willing to compromise its core principles in negotiating with illegal armed groups. If this requires the country to remain at war for many years to come, so be it. I prefer a horrific war than an ignoble peace."

The Colombian Constitution has many things worth fighting for. It is safe to surmise that most Colombians consider their government, warts and all, more legitimate than any of the irregular armies operating in the country and that those same Colombians would hate to see their basic democratic freedoms sacrificed for the sake of some peace agreement. But beyond that point, supply-siders run into what seems to me a normative inconsistency.

Colombia’s government derives whatever legitimacy it has in virtue of being the result of constitutional and legal practices that command and deserve the consent of most Colombians. These principles cannot be easily enumerated, but some of them are beyond dispute. For instance, Colombia’s Constitution establishes a representative government, with separation of powers, with mechanisms to prevent abuses of authority, with guarantees for dissenters, political rights for all citizens and so on. But, as is often the case, the more specific those principles and practices, the less we can attribute to them the legitimacy of the government and the Constitution. For example, my allegiance to the Constitution does not depend on its specific clauses regulating the Central Bank, or the transfers to municipalities. As far as I am concerned, if the Constitutional Assembly of 1991 had decided on some other arrangement on these matters, or even had remained silent about them I would still regard this Constitution as "mine," as worthy of my support. Other aspects of Colombia’s society are not necessarily sources of legitimacy, or at least not for many citizens. The Constitution says nothing about how to deal with settlers in the coca region, or what should be the pattern of land tenure, or how should the country exploit its natural resources and even if it did, few of us would consider those rules as the pillars of its legitimacy. Interestingly, supply-siders tend to be intransigent in specifics such as these but flexible when entertaining the possibility of demobilization and amnesty for the irregular armies, something that touches not less than the Penal Code. I find it odd that while supply-siders often consider fair game in a negotiation general legal and constitutional principles such as the Penal Code and the laws on political parties they regard economic issues as land tenure or settlers in the agricultural frontier as entirely off limits. In the interest of consistency, if the former are negotiable, so should be the latter and if the latter are non-negotiable, so should be the former.

But, whatever the oddities of this normative view, what matters for the present purposes is that, contrary to the claims of supply-siders, it not the logical conclusion of an empirical analysis. There is no connection between the causes of an insurgency and the strategies to solve it. There is no logical reason to negotiate with a group because it has a grievance. Arguably, the Confederate States of America had a grievance against the North during the Civil War, whom they saw as infringing their rights as states. But this did not prevent the North from dealing with them through military means, all the way up until total defeat. Likewise, the FARC in the 60’s emerged, arguably, as the result of some grievances over land tenure but this did not stop the Valencia Administration from bombing its headquarters. As citizens, we can conclude, if we so decide, that the best way to deal with an insurgency is by force even if it is motivated by a grievance. Maybe we do not find its grievance legitimate, or we think that the solutions they want, and the only ones they will settle for, will be terribly ruinous for the rest of society. The Israeli historian Benny Morris, for instance, documented several atrocities perpetrated by the Israeli army against Palestinians in 1948 but he has become notorious for saying that they do not call for any redress and that, if anything, Israel should have ethnically cleansed the West Bank at that point. He believes that Palestinians have an understandable grievance but, in his view, the price for addressing such grievance would be a transformation of the State of Israel such that he finds it unbearable. (Morris, 2004) The "ur-" supply-siders in the 60’s and 70’sin Colombia successfully urged the government to take a military stance against the guerrillas, not because they thought that Colombia was perfect, but because they believed that any concessions to the guerrilla would destroy the fabric of all they held good and dear.

By the same token, it is not true that "non-grievance" groups must always be dealt with by force. Controversial as it might have been, many Colombians in the early 90’sthought it was a good idea to offer concessions to Pablo Escobar, an actor that few would regard as the paladin of the downtrodden. Colombians even made him a concession they would have not considered for the FARC: a constitutional reform. As I write these lines, the Uribe Administration is engaged in a negotiation process with the paramilitary, a move that seems to count with vast opinion support. Once again, the paramilitary do not fit the prototype of a group with a grievance against the government. If anything, they have always thought of themselves as allies of the government.

Not only the facts dug out by supply-siders do not lend logical support to their recommendations, they are not even enough to make them more attractive. This is an echo of a motive brought out forcefully by Goldberger (1979) in the econometric debates around hereditability of IQ. Suppose, for instance, that 98% of the variance in homicide rates across municipalities in Colombia could be explained by differences in terrain and that only 2% could be explained by, say, levels of schooling. If Colombia’s government wanted to reduce homicide rates, a worthy goal in one of the world leader’s in this pathology, should it conclude that increased schooling is not worth it? To answer this question, variance explained is a useless criterion. No policy can change rough terrain, whereas increased schooling can be cheap in a Third World country. In deciding policies, sensible governments look at the cost of affecting a variable, not at its role in explaining variances in a regression.

Few would doubt that Colombia has serious problems of law-enforcement. With more and better-trained judges and police forces, the country could capture and prosecute more criminals. But this does not immediately mean that the solution to the country’s woes is to increase spending on judges and policemen, without any type of negotiation strategy with those involved in illegal activities. Resources are scarce. Imagine, for the sake of argument, that some whimsical billionaire is willing to give huge amounts of money to Colombia, at no charge but with two conditions. First it should be earmarked to the relocation of peasants from the coca region into the country’s agricultural heartland. Second, Colombia would loose those funds if it increased its military budget beyond what it currently is. Should the country accept the gift? It is impossible to tell at this level of generality. It would depend on how much money it is, how much land could be purchased with it, how many peasants could be brought back to the legal agriculture, and so on. But under some scenarios it would be worth it. It could reduce the burdens over Colombia’s law-enforcement community, thus bringing down homicide rates beyond what could be accomplished with whatever increase in the military budget the country can afford. Knowing the amount of variance in crime explained by "state capacity" tells us nothing about the appropriate policy to reduce violence because it tells us nothing about relative costs and benefits.

The pace of any conceivable peace process does not depend on the diagnosis of the conflict’s root causes. Let’s take the example of the negotiations during the Pastrana Administration. We now know they failed for a host of complex reasons. Was the alleged guilt complex one of them? Hardly. Imagine an alternative universe in which the government’s negotiation team had been entirely free of any kind of guilt complex (something that assumes, dubiously, that the real-life negotiation team was a bunch of "objective-causers"). It is hard to see how this would have changed the situation for the better, how this would have led the FARC to lay down its arms, facing the prospect of much reduced concessions from the government. It makes more sense to assume that, whatever the failings of that peace process, they had much more to do with the political and military facts on the ground than with some hard-to-pin feature buried in the psyche of the members of the Administration.

No amount of empirical analysis will tell us how to successfully solve the conflict because the notion of success is political, not analytical. There are no peace processes that are analytically correct or incorrect but peace terms that are acceptable or un-acceptable. No matter what supply-siders say today about the "true" nature of the conflict, whatever will work, will work.

The guilt complex argument assumes that legitimacy can be established or refuted by a set of regressions. Whatever legitimacy Colombia’s government has does not depend on the statistical evidence marshaled by supply-siders. To fix ideas, imagine that tomorrow the government shuts down Congress, establishes censure of the media, fires all the majors and dismisses all the municipal councils in the country and replaces them with military officers, arrests and condemns after a mock trial a dozen of opponents and publicly executes them in Bogota’s Bolivar Square. This would have no statistical effect, not even over the homicide rates because the executions would not count as homicide. But I, for one, would seriously downgrade my views about the legitimacy of the government and I can imagine a handful of Colombians doing the same. The willingness of Colombians to rally behind their government, to endure the hardships of civil war, to kill and die for the defense of fundamental principles of governance depends on their encounters with said government, on their perception of how their society fosters their life, goals and dignity, not on a few statistics, whatever their R-squared index.

Even the image of "Colombia" concurring to negotiations with the FARC is misleading. In any peace process the parties involved show divisions. If some day we have successful peace talks with the FARC, some Colombians will find points of convergence with some members of the guerrilla. The "us" and "them" will become blurred; such is the nature of peace talks, especially successful ones. Not only legitimacy is not purely quantitative, it is not even static, it is not even determined once and for all before any peace negotiations.

7. CONCLUDING REMARKS
The supply-side school has brought to the study of Colombia’s conflict the promise of combining rigorous theory and systematic empirical testing to develop solutions to the country’s ordeal. It has been debatable from the start whether this combination of ingredients is desirable or feasible when it comes to the study of civil wars. In this paper I have stayed away from that debate; I have no principled objection against the use of formal analysis or statistical methods. Instead, I have argued that, by their own criteria, supply-siders have not prosecuted their case convincingly.

Much of the theoretical basis of the supply-side school consists of a specific type of rational-choice modeling of civil wars that borrows from the public-goods approach to collective action problems. In spite of its remarkable influence, this paradigm is not hegemonic within rational-choice theory; strategic models of collective action in the spirit of Thomas Schelling’s tipping games are a well-known alternative (Schelling, 1978; Wood, 2002). Furthermore, it is not even plausible; it misrepresents beyond recognition the fundamental coordination problems of a rebellion. The resulting theory leads to unique, deterministic predictions, inadequate for the statistical study of a phenomenon that, by its very nature, is fraught with the uncertainty of multiple possible outcomes of aggregate behavior. Arguably, the public-goods approach has advanced faster than others in developing formal models of insurgency, something that explains why empirically-minded scholars were so quick to embrace it. But advancing in the wrong direction is not necessarily progress.

State-of-the-art econometric techniques are another trademark of the supply-side school. But, if anything, the undeniable skill of the statistical analyses in this body of scholarship has been something of a mixed blessing. By directing the attention of researchers toward measurable quantities such as murder rates, this commitment to econometric testing has come at the expense of an understanding of the context for such quantities. Thus, while it may be that the data speak for themselves, they speak only about themselves: murder rates inform us only about murder rates, not about their political implications and how they fit in the larger strategies of all the parties involved in the war, hectares of coca leaf inform us only about the size of illegal crops, not about the underlying social processes that create them, figures on prosecution inform us only about the judiciary’s case-load, not about the challenges and possibilities of the institutional changes needed to tackle it.

Supply-siders pride themselves in offering a clear diagnosis of Colombia’s war, the better to terminate it. Even if, contrary to my reservations, their diagnosis was correct, this medical analogy, a leit motiv of supply-siders, does not carry over to the treatment stage. Although we may be tempted to believe that a civil war is a disease of the body politic, it is unlike diseases in medicine in that its cure does not come from an external treatment; there is nothing external to society. The decision to terminate war, be it by making peace or by fighting it till the bitter end, is a political decision that no set of experts, however well-informed and well-meaning can take on behalf of a country. Any proposed "solution" to the conflict makes sense only against the backdrop of a political, not technical, definition of what the conflict is, what is negotiable and what is not. This is something that no statistical analysis, however impeccable it may be, can provide.

As citizens, supply-siders have been articulate participants in Colombia’s public debate and represent widely held views about what ought to be done in the country. They deserve praise for expressing publicly and forcefully their ideas. But little of value is accomplished when they present as inexorable results of some hard data what are no more (and no less) than their own cherished ideological positions. Instead, we are all better served if all sides of the debate acknowledge clearly the limitations of their methods and theories. After all, although in thinking about the conflict, history did not give Colombian social scientists the luxury of the hegelian owl, no one wants to be reduced to the role of the parrot, repeating mindlessly the words of some master.

COMMENTARIES
1. I wish to express my thanks to Alvaro Camacho, Carlos Guevara-Mann, Stathis Kalyvas, Ian Lustick, Carlo Nasi, and Frank Safford and seminar participants in the University of Chicago and in the Annual Conference of the Latin American Studies Association for their comments, criticisms and suggestions. The remaining errors, my sole responsibility, stand as an indicator of the likely result that would have obtained without their help.

2. In a separate, but related piece, I criticize at length Collier’s formal analysis and strategy for empirical testing. (See Medina (2005).)

3. The technical, game-theoretic details of this argument are not essential for the general reader. I have dealt extensively with them, however, in my book Medina (2004).

4. In fact, this is akin to the recent finding of Trejo (2004) in his study of the Zapatista insurgency. Whereas socio-economic conditions do not show up significantly in cross-sectional analysis as explanations for levels of conflict, they have a clear effect in time-series analysis, within each locality.

REFERENCES

1. COLLIER, Paul. 2000. "Rebellion as a Quasi-Criminal Activity." Journal of Conflict Resolution 44(6):838-52.         [ Links ]

2. FEARON, James and David Laitin. 2003. "Ethnicity, Insurgency and Civil War." American Political Science Review 97(1):91-106.         [ Links ]

3. GOLDBERGER, Arthur S. 1979. "Heritability." Economica, New Series 46(184):327-347.         [ Links ]

4. KALYVAS, Stathis. 2004. "The Logic of Violence in Civil Wars." Book Manuscript.         [ Links ]

5. MEDINA, Luis Fernando. 2004. "Formalizing Common Sense in the Theory of Collective Action." Book Manuscript.         [ Links ]

6. MEDINA, Luis Fernando. 2005. "Reconsidering the Link Between Grievance and Conflict: A Game-theoretic Perspective." Under review at the American Journal of Political Science.         [ Links ]

7. MONTENEGRO, Armando and Carlos Esteban Posada. 1995. "Criminalidad en Colombia." Coyuntura Económica XXV(1).         [ Links ]

8. MONTENEGRO, Armando and Carlos Esteban Posada. 2001. La Violencia en Colombia. Alfaomega.         [ Links ]

9. MORRIS, Benny. 2004. "Pandora in Zion." New Left Review (26):327-347.         [ Links ]

10. OLSON, Mancur. 1965. The Logic of Collective Action. Cambridge, MA.: Harvard University Press.         [ Links ]

11. RUBIO, Mauricio. 1997. "La Justicia en Una Sociedad Violenta: los Agentes Armados y la Justicia Penal en Colombia.", Documento CEDE, Universidad de los Andes.         [ Links ]

12. RUBIO, Mauricio. 2002. "Conflicto y Finanzas Públicas Municipales en Colombia.", Documento CEDE, Universidad de los Andes.         [ Links ]

13. SANCHEZ, Fabio and Jairo Núñez. 2001. "Determinantes del Crimen Violento en un País Altamente Violento: El Caso de Colombia." Documento CEDE, Universidad de los Andes.         [ Links ]

14. SCHELLING, Thomas. 1978. Micromotives and Macrobehavior. New York: Norton.         [ Links ]

15. TAYLOR, Michael. 1987. The Possibility of Cooperation. Cambridge: Cambridge University Press.         [ Links ]

16. TREJO, Guillermo. 2004. Indigenous Insurgency: Protest, Rebellion and the Politicization of Ethnicity in 20th Century, Mexico, PhD thesis, University of Chicago.         [ Links ]

17. TULLOCK, Gordon. 1971. "The Paradox of Revolution" Public Choice 11:89-99.         [ Links ]

18. WOOD, Elisabeth. 2002. The Logic of Insurgent Collective Action: Defiance and Agency in Rural El Salvador. Cambridge: Cambridge University Press.        [ Links ]