Institutions in Turbulent Settings
Published on: Mar 26, 2004

This paper critiques some applications of the neoinstitutionalist program (NI) to the study of Latin American and Andean polities, and tries to develop some aspects of an alternative framework.

The critique develops on two levels. First, the "turbulent" institutions of several Latin American and Andean countries highlight some of the shortcomings of NI tout court. Since turbulent settings are the norm, rather than the exception, both theoretically and empirically there is a need in each case to explain the specific sense in which institutions can be taken as an independent variable. Second, the variant of NI most frequently applied to Latin America--with its heavy and almost always implicit normative and theoretical assumptions2--is deeply flawed and fails to address the very core of political conflict and change in these countries. I hope that the context will indicate at which level my criticism is developing. I contend that the shortcomings of standard NI models can be overcome with a kind of political analysis that "brings society back in" and incorporates learning in a nontrivial way. This discussion is linked to the "philosophy of history" of NI (path dependency), sensibility to initial conditions and the transition from laminarity to turbulence.

In the first part of the essay, I flesh out some basic definitions. In the second, I contrast how institutions work in stable (or "laminar") and turbulent settings. In the third, I stress the importance of nontrivial learning from a Schumpeterian perspective. In the fourth, I show that path dependency is only part of the history. Each section leaves some unanswered and (hopefully) interesting questions.

1. Definitions

I understand NI in a very conventional way: a theory that takes institutions, broadly understood, as a relatively fixed set of incentives that explain differential social outcomes (as in North and Thomas´s rendering of "the rise of the Western World", 1976). How broadly understood is open to debate. A typical gambit of NI as applied to Latin American contexts is to resort to so-called "informal institutions" when the proposed correlation between institutions proper and outcomes fails to show up. Thus, the concept of institutions would contain all the rules of the political game. The problem of fancifully broad definitions of institutions is that they are all-encompassing in nature, identifying institutions with any stable pattern of human interaction. This is an open door for circular argumentation, tautology, and programmatic degeneracy (in Lakatos's [1970] jargon). Specifically, the notion of "informal institutions" is open to two criticisms. First, contrary to institutions proper, the so-called informal ones are identified by an external observer without the conscious acceptance of the protagonists of the interaction. It is an analytic device, at another level of reality than explicit rules. How can you demonstrate empirically that the set of informal rules that supposedly were observed actually regulate a given universe of interactions? Second, informal rules have a very vigorous (and unstudied) life, below and above formal institutions. Below: conventions, for example, are potent devices for coordination-solving, that largely live in a world previous to explicit rule-establishment (Lewis, 1986) and that certainly do not require a two person interaction.3 Above: meta-agreements are not necessarily well behaved. In countries with weak polities, it can be the case that the only valid rule is the (informal) motto "the rule is that no rule will be observed". This kind of (paradoxical) order does not orient actors in their every day life--and so the typical inference of the NI program (such and such set of institutions generates such and such social outcomes) is out of place. Thus, it is better to stick to something like the more restrictive but sensible definition of Eggertsson: "Let us define institutions as sets of rules governing interpersonal relations, noting that we are talking about formal and organizational practices" (Eggertsson 1990, 70); see also Tsebelis (1990).

Which is the explanatory power of institutions? In a world of (semi)rational agents, stories of failure and success have to be accounted for: Why do people (sometimes) take the wrong paths? Because they are sent the wrong signals, which encourage suboptimal behavior. And why do wrong signals occur? Mainly because of unspecified property rights, imperfect information and transaction costs. NI offers microfoundations where "organic culturalism" (à la Putnam) does not.4 Suboptimal outcomes can be explained by poor institutional design--which, in turn, is maintained by coalitions that benefit from it. Transaction costs are the prototype of such a mechanism. Market and state failures create social forces that help to maintain those very failures. From positive feedback and the decisive role of institutional design, path dependency follows. This provides for a theory of social change.

Note that the microfoundations provided by NI make sense if and only if institutions constitute a system of incentives that explains the behavior of the agents. On the other hand, NI does not entail hard-nosed rationalism; it should suffice that institutional signaling be the most salient feature of the incentive landscape, the other incentives playing the--perhaps important but logically subordinate--role of noise.

2. The Neoinstitutionalist Framework in Turbulent Settings: Does It Work?

Does NI provide an adequate toolkit for understanding conflict and change in countries with high levels of instability and where "noise" is stronger than institutional signaling? I think not. I will not dispute that in such countries "institutions matter"--an assertion that is trivially true (Harris, Hunter and Lewis 1995)--nor that a broad family of social and political problems can be captured by the rich conceptual framework of NI. I start instead with the defense of the (once again, very conventional) claim that such matters as contemporary political change in Colombia, Ecuador or Venezuela cannot be accounted for by standard NI.5 I believe there is a fundamental difference, from the point of view of the role and status of institutions in social and political life, between those turbulent Andean cases and the core capitalist countries.

a. My main defense of the conventional claim is based on the fact that in a very important sense countries like Colombia, Perú or Venezuela fit too well the theories that constitute the formal backbone of NI. To see this, let us use the powerful dichotomy between "policy politics" (playing the game) and "institutional politics" (debating the rules of the game), common to many variants of social choice theory. Buchanan (1986) argued that in all developed democracies (of his time), policy politics was the dominant practice, because basic institutions were taken for granted. This is perplexing, because if institutions are decisive for outcomes of efficiency and distributional problems, they should be a permanent object of political strife among (even boundedly) rational agents. The Tullock question ("Why so much stability?") has no easy answer; in this regard, I can do no better than quote Eggertsson in extenso:

If institutions are somewhat chosen as we want to argue [rationally], then we are back to the disequilibrium outcomes of majority-rule voting, and the choice among institutions will not lead to stable or equilibrium institutions. And it does not help to argue that the choice of institutions is prescribed by higher rules, written or unwritten constitutions, because this only pushes the argument one step back, requiring us to predict unstable constitutions. . . . Just as in the case of voting outcomes, empirical observations tell us that the institutional structures in democratic countries are relatively stable, that they tend to be [in] equilibrium. . . . Shepsle argues that political institutions are ex ante agreements about cooperation among politicians. According to this view, institutions can be seen as a capital structure designed to produce a flow of stable policy outcomes, and institutional change is a form of investment. One of the costs of institutional change is the uncertainity about which outcomes the new regime will produce. Uncertainty implies that a given structure may ex ante be associated with a set of structure-induced equilibrium points. Ex post, this uncertainty is gradually reduced as the operational qualities of a new institutional structure become known. Finally, Shepsle argues that this uncertainty about the impact of structural change on equilibrium outcomes is enough reason to stabilize institutions and prevent continuous institutional change. He maintains, thus, that the calculations of agents in decisions involving policy choices are qualitatively different from calculations regarding institutional change (Eggertsson 1990, 71-72).

Stability is a major anomaly, for which auxiliary arguments had to be developed. In contrast, in the Andean area we have fully normal cases. The Andean constitutional wave of the 80s and 90s was one of the two biggest in the contemporary world, together with the post-socialist Central-East European. Colombia issued one new constitution and several important reforms related, among other matters, to property rights and the judiciary; Ecuador produced two constitutions (1979 and 1998), and a couple of major reforms in the meantime; Perú also enacted two constitutions; and Venezuela and Bolivia one each. In all these countries, "writing a new charter" has been the main motive of politics for long periods, a phenomenon backed by a strong historical tradition. Furthermore, large-scale institutional change is the main objective of everyday political squabbling: decentralization; electoral, judicial and congressional reform; and so on. The legislative inflation in the Andes is monstrous; not even specialists can follow comfortably the unending stream of change in the rules of game in practically all of the basic areas of life.6

In normal Andean countries, then, institutional politics is dominant (and policy politics is rather poor). But this creates a real problem for NI. First, the basic rules of the game are not stable. Rather, they are the main object of contention and change very fast. Now, "unstable institutions" is rather an oxymoronic expression--whatever sense one can give to it, it affords only weak independent explanatory power to institutions. It also calls into question evolutionary arguments, one of the best alternatives to hyperrationality assumptions (Axelrod 1986; Young 2001), because the slow pressure against maladaptations does not have time to unfold (I will return to this). Thus, we have to ask in what sense we are speaking about a system of incentives that actually affects agents´ behavior. If, as is the case in countries like Colombia or Ecuador, agents are conscious of the velocity of institutional change, their expectations are not strongly linked to any clearly specified (present) system of incentives. More subtly, we cannot maintain simultaneously both the theory and the explanations for the anomaly; both are stated on a general plane, and each one will hold for all cases or for none.

The obvious empirical question is: In such a setting, do institutional arrangements act as independent variables? And, then: How do agents respond to this family of noisy and fluid incentives?

b. Although in the above sense turbulent countries are too normal,7 in another sense they are odd. Instability, war and violence give rise to self-sustaining patterns of human interaction, which in turn generate explanatory problems for the concept of rational calculus and thus of "systems of incentives."

In Colombia, for example, war and elections have coexisted for a very long time--more or less twenty years, if we fix the start of the present wave of armed conflict in the 1982-84 period.8 Violence has become an everyday component of political action, making it a high risk activity. This changes the menu of options for politicians, as well as their mindset. It also creates a rationality problem.

Due to war, Colombian politicians have the option--often the need--of switching between institutional systems: the jurisdiction of the state or the jurisdiction of the warlords. It is important to stress that warlords are involved in electoral politics in their territories, so the modal politician will participate in different and contradictory institutional worlds. This is another type of institutional politics (with changes in space, not in time). Agents can choose among competing rules of the game, indicating what kind of game they prefer in a given time period, but they pay the price of prohibitively high risk taking.

Colombian politicians of all political parties and families are presently taking very high risks of being kidnapped or killed.9 Oddly, there has been a strong increase both in risk and political participation (in the sense that there are many more candidates and lists) in Colombia in the last twenty years.10 Why? Whatever the answer, I would argue that at the limit, when one is playing Russian roulette (i.e., when the loss of one's life is one of the prizes), using the principle of revealed preferences is simply not sound. But in what sense then are we speaking of utility functions and systems of incentives? Risk can be coped with under the incentive system up to a certain point--after that, incommensurability appears.

c. Turbulent countries tend to be weak and vulnerable nation-states. A fundamental part of their decision making and political life is transnational. War, narcotrafficking, decentralization, economic adjustment, to name just some of the dominant motifs in the Colombian case, involve extended webs of national and foreign actors. This would not be a major problem except that transnational governance systems, and their corresponding distributional problems, have generally escaped the gaze of NI analysis--in part, indeed, because they are institutionally mis-specified,11 in part because the mechanism of "micromotives and macrobehavior" (Schelling 1978) implies a change of scale: when one moves from the national to the global, the unit of analysis moves from individuals to states and organizations.

On the one hand we have, then, crucial global-ridden processes--that is, narcotrafficking, development models, technological change--and, on the other hand, a national institutional framework.12 The result is that, empirically, one can observe a strong link between the "small" everyday practices of local politicians and the "big" arrays of transnational phenomena, but NI gives no way of capturing it.13 When one poses simple questions--such as why democracies are unstable in the Andean area or why the Colombian political system has changed in the specific sense it has--this tension becomes particularly uncomfortable. In the new wave of institutional studies on Latin America, novel and interesting aspects of political life were analyzed successfully, but a quaint dichotomy took center stage: adequate problem specification versus strong methods. One can have one of these goods, but not both simultaneously. Geddes (1994), for example, in her neat and intelligent analysis of state reform and the trajectory of bureaucracy-technocracy in Brazil, overlooks the role of international financial institutions. This description of a "purely national dilemma" defies credibility. An even more extreme example concerns the interpretation of political change in Colombia: analysts have focused on the niceties of electoral legislation, or more generally on institutional design, forgetting about such small details as narcotrafficking, war or the changes caused by television in political life (e.g., Nielson and Shugart 1999). Old institutionalism was not free of such difficulties, as denounced by Eckstein (2000), who criticized it for its focus on the small print of electoral design and its failure to give adequate context to the fall of the Weimar Republic.14 This draws our attention to the problem of metrics. Which is the set of institutions relevant for the specific problem at hand? Can, for example, the electoral fragmentation of several Andean polities be explained simply by wrong electoral rules, or are other, more distant, institutions relevant?15

In brief, society (with its "long networks" of political action) has to be brought back in. This applies, if the previous arguments are correct, especially to normal countries, where the institutional framework is unstable.

Conflicts around basic institutional design, the breakdown and creation of coalitions and national--transnational agendas--this is the very matter of everyday politics in "states in crisis". Is it at all possible to speak about it, displaying at least part of NI rigor and using some of its methodological tools?

3. Innovation and Learning

The question merits a positive answer; there are now interesting programmatic reflections (from Hedström and Swedberg [1998] to McAdam, Tarrow and Tilly, [1997]) as well as appealing empirical works that address many of these challenges. Since the adaptation of the framework of "the vast majority of analyses produced by political economists" (using Hall's [1997] expression) to these concerns will necessarily be incremental and piecemeal, I will offer a different, Schumpeterian twist.

Schumpeter's name in political science is bound to the so-called elitist perspective, a contested though fertile view of electoral competition. But here I use another aspect of Schumpeter´s work: his notion of entrepreneurship and innovation. My belief is that studying political innovation allows the researcher to trace the links of the chain that go from transnational processes to small local coalitions and conflicts. It may also allow for a better understanding of the nature of political change and a better fit to empirical data than standard recountings.

The intuition of viewing the politician or social leader as an entrepreneur is already well established and has been particularly successful in the study of social movements (see, for example, McAdam, McCarthy and Zald 1996). However, the typical definition of entrepreneur in the social movement literature is as a resource mobilizer, failing thus to address what Schumpeter considered the distinctive nature of entrepreneurship--innovation.16 Perhaps because social movements tend to be short lived, the study of the repertoires of contention has not led to the analysis of innovation and its long-lasting effects on organisms and systems. More than in risk taking or mobilizing resources, entrepreneurs (by Schumpeter´s definition) are engaged in innovation, defined as "technological change in the production of commodities already in use, the opening up of new markets or of new sources of supply, Taylorization of work, improved handling of material, the setting up of new business organizations . . . in short, any 'doing things differently' in the realm of economic life" (Schumpeter 1939, 84).

What causes innovation? Endogenous change and exogenous shocks.17 Schumpeter focused on the former, actually considering the latter of little interest to economics. Whatever the merits of his reasoning, both types of forces seem crucial to the study of political systems. In politics, exogenous shocks occur frequently and are of indisputable interest: the way in which wars, changes in tastes and new technologies give rise to ways of "doing things differently" are not well understood, yet they are crucial for the interpretation of political change.

The import of applying the Schumpeterian view of innovation is that we can study explicitly the interaction between processes of innovation and "exogenous shocks" caused by distant drivers of political change. We can do this without abandoning the basic tools that give NI its analytical force, particularly some notion of rationality and informational economy (signals, incentives and constraints)--that is, microfoundations.18 In other words, our agents will remain basically the same (for example, politicians who want to win elections) but now they are open to many different incentives and constraints. Exogenous shocks exist, so the system and its environment are moving simultaneously. The social landscape will be the vector that results from aggregated microinteractions, institutions and exogenous forces.

It is important to stress that the idea of innovation also allows for the study of both moments of interaction between institutions and agents: how institutions restrict agents and orient them in specific directions, and how agents, through small changes, transgressions and adaptations perturb and finally transform institutions.19 In this sense, as addressed specifically by Schumpeter, the study of innovation is evolutionary. But if narrow rationalism is replaced by an evolutionary perspective, we have a "syncopated evolution," because the environment changes very fast, creating juxtaposed layers of adaptative practices. We do not just have a society of limited, myopic agents struggling in (sometimes very) noisy environments. Evolution can be imperfect and allow for the survival of maladaptations to previous incentive systems, because exogenous shocks can accumulate, truncating the evolutionary process. One of the typical results of this kind of evolution is thus "mixed types" and heterogeneous coalitions. Keeping track of the exogenous shocks, their unfolding and their effects on the political system, is consequently an antidote for the extremely acritical adoption of the modern-backward dichotomy that is so evident in several NI analyses.

For example, elsewhere I have shown (Gutiérrez 2001) that to explain the type of relations between narcotraffickers and politicians in Colombia in the last twenty years I needed two dimensions: a principal-agent model that accounted for the contractual conflicts between criminals; and a new insertion of Colombia in the international system, marked by the 1991 Constitution, that changed the role of the state vis-à-vis illegality. On both dimensions the characteristic of social interaction was strong ambiguity and a large amount of noise. The 1991 Constitution was considered in its time a modernizing landmark, and in fact it provided a wealth of institutional resources to expel organized criminals from political life; but at the same time it gave in to the main demand of narcotraffickers (a ban on extradition for Colombians).

Schumpeter´s analysis made learning the driving force of change, though focusing only on imitation. The "new ways of doing things" spread in waves, with early imitators replicating successful practices. But entrants keep imitating even after the marginal benefits of the innovation have reached zero, while other practitioners simply can´t catch the beat of the new rhythms. Thus, a typical result of Schumpeterian evolution is that learning entails overreacting. Herd effects and congestion lead innovation-prone rational agents to suboptimal behavior. Agglomeration around successful devices results in catastrophes and organizational destruction. Schumpeter could identify the specific (market) mechanisms behind organizational breakdown and catastrophes.

We are not near doing the same, because in politics there is no equivalent of markets, but I suggest this is a fundamental task. There has been a big bang in the political system in the Andes, but we do not know why it took place or what are the reasons for the differential outcomes (an organizational earthquake in Perú, Ecuador and Venezuela; relative stability in Bolivia; change with important organizational invariants in Colombia). Indisputably, there is some kind of relation between the diverse ways the agents adapted to rather similar processes. Several interesting questions spring from this simple statement: Which concrete mechanisms explain the differential reactions? When and why do the differential reactions imply divergence or convergence of outcomes?

Innovation and learning are the core of technical change. Note that, as in economics, in politics the latter expression has two meanings: the introduction of new devices; and the development of new forms of organization, discourses and practices. Both of them are very important. Television, for example, has had a very serious impact on Latin American political systems, giving an advantage to individuals over organizations20--a circumstance that can hardly be overcome by changing electoral statutes.21 On the other hand, there is a rich menu of purely "soft" technological innovations that trigger long-range political change. The following is a list of basic innovations, with illustrations of the decisions and processes involved:

Finding new financial resources. With the growing importance of organized crime in political life, a major decision is whether or not to accept illegal funds, and, if so, how to do it and how to justify (or deny) it publicly. The diverse ways in which these decisions are made--and rebuked by adversaries--produce a specific technology of public debate over the legal-illegal and formal-informal divide (which, by the way, is crucial in the contemporary world, not only for states in crisis).

Finding new languages, political discourses and symbols. In the Andean area, the second half of the 1980s and the 1990s saw the upsurge of the so-called antipolíticos, who sought to capture the votes of the citizens by staging an involved public performance of denial ("elect us because we are not politicians"). This complicated ritual produced a new technology of political symbolism.

Finding new ideas and interest aggregation--articulation forms. New ideas are decisive in the political experience. In this regard, Hall has made two extremely important points. First,

[T]he vast majority of analyses produced by political economists take the same general form, which is to say that they identify a fixed set of variables, whether composed of interests, institutions or ideas, given exogenously to the process of political conflict, and then show how these structure the situation so as to produce the relevant outcomes. This kind of analysis can have real value, but what it misses is the extent to which the outcomes may be created via processes of political conflict and not generated entirely by the antecedents of that conflict.22 (Hall 1997, 197)

Second, ideas are a basic dimension of the political, because "politics is not only a contest for power. It is also a struggle for the interpretation of interests . . . politics is more open than most political economists see it" (ibid.; see also Hall 1992). Ideas, and ideals, migrate, suffer counterintuitive adaptations and are articulated (to use another Hall category) with others in an evolutionary fashion.

Finding new dimensions of political practice. Shall parties resort to violence or not? Will they change significantly their traditional repertoire of political action? Will they publicly change their ideology?

Highlighting innovations allows one to exhibit the remarkable, and often neglected, technical content of political struggle and change. The technical is not only a way of presenting interests, it is a way of building them (Hall, 1997). At the same time, it shows some stark contrasts with standard NI. From the NI perspective, learning is "transparent." Indeed, the pathbreaking analyses of Akerlof (1970), Stiglitz and others (e.g., Kreps [1990]) develop exquisite models that enhance our understanding of informational problems, and offer potent tools to spell them out. However, their translation into political terms remains doubtful. For example, principal--agent structures have been applied to the people--government relation, losing the fact that "the people" is not an actor but a space crossed by cleavages and fractures. Moreover, NI political scientists have ignored systematically the extremely simple--and intelligent--observation of Hirschman: politics is also (fundamentally) about speaking, and thus we need a theory of voice and signaling.

Learning through technical innovation suggests a different picture. First, agents create and adapt along many different dimensions, and are frequently struck by endogenous waves of innovation and/or exogenous shocks. This has two types of consequences. On the static side, the concept of dimensions of evaluation precedes that of systems of incentives. The best example I know of the dramatic changes in the model when the dimensions of evaluation are methodically and explicitly introduced is the classical critique by Hirschman (1970) of the Schumpeterian-Downsian model of elections. On the dynamic side, as stressed above, this evolution is punctuated by frequent large-scale changes in the environment, a factor that limits the power of the basic mechanism of the slow and gradual elimination of unfit agents.

Second, learning advances in waves, so that even a movement toward Pareto-optimal situations can cause catastrophic organizational mortality. There is a clear analogy, then, between herding behind a successful innovator and the classical collective action dilemmas. A polity that learns well can end in a state of constant disarray. Third, signals are not transparent--they have to be read. This argument can be developed in several stages. To start with, innovators can successfully override institutional considerations. Institutions themselves can be inconsistent: to take a typical contemporary situation, for example, answering simultaneously to a national and an international constituency with contradictory interests and concerns. Additionally, agents are exposed to lumpy stimuli--not one signal from one institution, but an institutional score, if I may, which the agent has to learn to interpret and play in front of different audiences. In turbulent settings, where institutions are unstable, short lived and inconsistent, and coexist with other very strong systems of incentives, the ambiguity and polisemy of institutional signals reach a point where the (sunken) assumptions of the comunicational transparency of contractual incentives is untenable.

4. Whither Path Dependency?

The previous discussion is intimately related to the notions of equilibrium and change. To simplify: while neoclassical models predict convergence independently of historical contingency, NI establishes divergence and irreversibility as core concepts that entail path dependency (single versus multiple equilibria outcomes). Taken together, both perspectives beg many questions. The first one is the level of resolution of the explanatory mechanisms. How "small" can an event be to trigger a "big" change? For example, Eckstein (2000) argues that the analysis of the Weimar republic by old institutionalists was wrong because they highlighted the finesse of electoral legislation, forgetting the huge historical tragedy behind Nazism. The criticism is intuitively appealing, but at the same time one would want an argument that addressed the very real fact that there need not be congruence between the size of the "cause" and the size of the "effect."

The second question is related to the fact that the nature of change can change . Suppose there are two types of systems, those oriented toward an outcome (convergent) and those with several degrees of freedom (path dependent). It is clear that path-dependent systems can in a given moment become convergent--actually, it can even happen that a convergent system becomes path dependent if, for example, it is subjected to a strong enough exogenous shock. This means that "original sin accounts" as those discussed at the beginning of this paper fail because they ignore the second order processes of change. This sounds much less involved when grounded in the experience of the Andean countries. Was the "third wave" of democratization a genuine wave (toward a convergent system), or rather a turn in a cycle of regime change related to the form of insertion of these countries in the global market? Did these democracies depend for their existence on national assets or on international constraints? And what kind of counterfactual could be built by removing one or another constraint?

The intuition that there is an internal dynamic of the system, different from its observed path, might be important in two senses. First, the basic concepts of equilibrium in economics and game theory (i.e., Nash equilibrium) coincide tautologically with the notion of stability. Once the system arrives at a state, the task of the analyst is to explain why it is actually an equilibrium given the nature of the agents. By definition, there is no out-of-equilibrium outcome (Eggertsson notes this difficulty [1990]). However, with the simple idea of exogenous shocks and perturbations, there is the theoretical possibility of stability without equilibrium, when a system is continually perturbed up to a critical point, and thus of self-organizing structures in far-from-equilibrium situations. Political systems in States in Crisis--especially those that show some kind of stable institutional life--would correspond to this description. Second, very different trajectories exposed to similar perturbations may have the same end point. The Andean countries, for example, show similar patterns of political problems, despite their very different traditions--the same can be said about South and Central European nations. Convergence takes place because of many factors, one of the most neglected being that agents learn about what is happening in their neighboring nations. Once again the technical domain--innovation--is a driving force for learning: in the collapse of the Central and East European centrally planned systems, the "round table" technique used for the first time in Poland was used successfully in very different contexts; in the Andean countries the wave of antipolítica was propagated through explicit appropriation of the motives of electorally successful leaders. Space still counts--perhaps it counts more than ever. All this highlights that historicism does not entail path dependency. Systems can be absorbent (a single-ending state, as in neoclassical models), periodic (cyclical) or none of these (path dependent)--and large-scale historical changes can imply "second-order change" passing from one type of system to another.

This takes us back to the "big effects"/"small causes" motive. If we take path dependency seriously, its meaning is very near the "sensibility to initial conditions" concept. When dynamic systems show sensibility to initial conditions, given the trajectory of two distinct "particles," very small differences in the starting point can entail enormous differences in the outcome. In historical analysis, however, this type of analysis is ridden with difficulties. How can one determine the "starting point" of the trajectory? Seldom is this question posed explicitly. Will one accept the theological concept that all future generations are determined by the (original) sin(s) of their great-great-grandfathers? Putnam (1994) has shown an iron consistency in this regard, and goes as far back as possible to find the reasons for differential outcomes--a real and explicit concern for "time zero" in the historical trajectory. Though his effort is basically flawed (Putzel, 1997; Tarrow, 1996), it clearly and honestly reveals that path dependency and sensibility to initial conditions are kindred notions. However, if original-sin determinism is introduced into the analytical framework, in what sense do institutions (or culture, for that matter) count? Institutions would be only the demiurge that expresses the (perhaps tiny) differences at time zero.

All this has implications for the very foundations of an evolutionary perspective. In "abnormal" (stable) situations, institutions constitute a general framework and outcomes can be seen as the result of the interactions of individuals.23 The "micromotives and macrobehavior" mechanism works well. Thus, stable worlds are transparent in yet another sense: given the rules, outcomes can be studied as if they were the result of aggregated microdecisions. But vulnerable and unstable (turbulent) systems are not transparent, because agents are "shooting at a moving target"--the environment changes faster than the system and the boundaries between the system and the environment are ill defined.

5. Conclusions

I very much agree with Rubinstein´s (2000) assertion that the abstract analysis of the interaction of rational agents has an independent value in itself. However, as soon as a theory, or threads of it, is used as a tool for empirical statements about the state of the world, the categories of the theory should be carefully evaluated to see if they capture the basic content of the system under study. What are the conditions for institutions to be the basic "system of incentives" in a social world?

NI literature applied to Latin America has plainly neglected this question. When systems of incentives are taken for granted, the interest of agents can be fixed deductively. The basic question then becomes cui bono? (Elster 1997): "Who is the beneficiary of this rule of the game"? Modernizers are, and should be, interested in liberalization (Diamond et al., 1997), clientelists in statism, and so on. If what I have being saying about "mixed types" is true, the cui bono question, however attractive, is very nearly the death of good empirical research for countries like those in the Andean region.

A focus on learning and innovation would help to alter the analytical landscape (interests, by the way, are also learned and discovered) and take society back in. On the other hand, it is a good antidote against narrow (economic, cultural, original-sin) determinism. Innovation and learning are embedded in sequences of political-socioeconomic-military structures and events. I believe these Hirschmanian sequences give a better understanding than explaining politics through economics, explaining politics through politics or explaining politics through time--zero events.

Contrary to Schumpeter´s framework for economics, in politics exogenous shocks are also of analytic interest. I have used the expression "exogenous shocks" here not as "international" nor as "uncalled for," but as "part of the environment, not of the system." Political change cannot be understood without taking into account the open-ended nature of political systems (Hall) and the exogenous shocks to which they are exposed, especially in turbulent settings. Exogenous shocks also stress the contingent and nontransparent nature of institutional signaling, a fundamental aspect of empirical political conflict in Andean countries.

Endnotes

1 Researcher, Instituto de Estudios Políticos y Relaciones Internacionales, Universidad Nacional de Colombia. This paper was sponsored by the London School of Economics - DFID Crisis States Program. I wish to thank James Putzel, Eric Hershberg, Juanita Villaveces, Juan Camilo Cárdenas and Paul Price for their extremely valuable input.

2 See, for example, the excellent review by Ames (1999).

3 I can develop conventions to protect myself from my weakness of will, Elster (1984).

4 I would note, however, that Northian and Putnamian tales have in common being narratives of an original sin, a pretty strong symptom of their obliteration of power asymmetry motives. This aside means that I do NOT take sensibility to initial conditions as given, a subject to which I will return.

5 I hope to do so by not-too-conventional means.

6 One among many possible examples: during the first three years of the Andrés Pastrana (1998-2002) administration in Colombia, five important tax reforms took place. Even the rules that affect property, that apparently cozy haven of stability, are flexible.

7 And indeed they are simply more than stable ones.

8 1982 was the year in which the main guerrilla force (Fuerzas Armadas Revolucionarias de Colombia-FARC) declared itself the People´s Army. Collier and Hoeffler give 1984 as the initial date of war in Colombia (1998).

9 An ominous symptom of this fact is that insurance companies withdrew coverage for Colombian mayors.

10 Violence affects the political parties differentially. Controlling by size, proportionally more left-wing than tradional party members are killed; but all suffer massive bloodletting.

11 For example, Aoki's (2001) interesting and comprehensive work doesn´t treat the subject.

12 This mismatch between national institutional frameworks and patterns of power distribution appeared long before the present wave of globalization; indeed, it is at the heart of the "rise of the Western world."

13 Certainly, this is a motive explicitly posed by the Crisis States Program.

14 Eckstein, though, seems to take for granted that "big" objects always have "big" causes, a point of view that can´t be shared.

15 And, once again, a meta-institutional problem arises here: the constant change in the rules of game -and not one set of rules valid at a given period-can decisively shape the nature of political conflict. See the excellent Fleischer (1996) for the successive waves of electoral reformism in authoritarian Brazil.

16 McCarthy (1996) comes close to this idea, but then drifts away.

17 It is important to take into account that for Schumpeter, these expressions don´t correspond to the national-international dichotomy, but rather to "internal or external to the economic system." I adopt this usage, replacing "economic" with "political."

18 On the other hand, endogenous innovation will not be successful unless it is robust relative to external shocks, especially if these are strong and repeated.

19 Here the existence of transnational forces is critical: agents can shortcircuit institutions resorting to transnational coalitions.

20 Unlike the majority of countries in Latin America, parties in Colombia were institutionalized long before the introduction of TV. In countries in which both came more or less together, like Brazil, the impact must have been stronger (Mainwaring 1999).

21 As soon as the majority of voters starts to entertain the notion that individuals are better than "machines," stringent statutes that give an edge to parties over ambitious politicians don´t have a chance to survive. Ecuador is a good example of this.

22 That is, the assumption of the exogenous character of institutions can be wrong.

23 Of course groups and organizations enter the analysis, but they in turn are a result of the interaction between individuals.

References

Akerlof, G. 1970. "The market for lemons: quality uncertainty and the market mechanism." Review of Economic Studies, no. 54:345-64.

Ames, Barry. 1999. "Approaches to the study of institutions in Latin American politics." Latin American Research Review 34, no .1:221-27.

Aoki, Masahiko. 2001. Toward a Comparative Institutional Analysis. Cambridge, Mass.: MIT Press.

Axelrod, Robert. 1986. La evolución de la cooperación. Madrid: Alianza.

Buchanan, James. 1986. Liberty, Market and the State. Political Economy in the 1980s. New York University Press, N.Y.

Collier, Paul, Hoeffler Anke. 1998. "On economic causes of civil wars," Oxford Economic Papers no. 50 pp. 563-573.

Diamond, Larry, Plattner Marc, Yu-han Chu, Hung-mao Tien. 1997. Consolidating the Third Wave Democracies. John Hopkins University Press, Baltimore and London.

Eckstein, Harry. 2000. "Unfinished business." Comparative Political Studies 33 (6/7):505-35.

Eggertsson, Thráin. 1990. Economic Behavior and Institutions. Cambridge, England; New York: Cambridge University Press.

Elster, Jon. 1997. Egonomics. Barcelona: Gedisa.

______. 1992. El cambio tecnológico. Investigaciones sobre la racionalidad y la transformación social. Barcelona: Gedisa.

______. 1984. Ulises y las sirenas. Estudios sobre racionalidad e irracionalidad. Fondo de Cultura Económica, México

Fleischer, David. 1996. "Las consecuencias del sistema electoral brasileño: partidos políticos, poder legislativo y gobernabilidad," Cuadernos de Capel-IIDH no. 39

Geddes, Barbara. 1994. Politician´s Dilemma: Building State Capacity in Latin America. Berkeley: University of California Press.

Gutiérrez, Francisco. 2001. "Organized crime and the political system in Colombia." www.nd.edu/~kellog

Hall, Peter. 1997. "The role of interests, institutions and ideas in the comparative political economy of the industrialized nations." Pp. 174-207 in Comparative Politics. Rationality, Culture and Structure, edited by Mark Lichbach and Alan Zuckerman. Cambridge, England; New York: Cambridge University Press.

______. 1992. "The movement from Keynesianism to monetarism: Institutional analysis and British economic policy in the 1970s." Pp. 90-113 in Structuring Politics. Historical Institutionalism in Comparative Analysis," edited by Sven Steinmo, Kathleen Thelen, and Frank Longstreth. Cambridge,England; New York: Cambridge University Press.

Harris, John, Janet Hunter; and Colin Lewis. 1995. The New Institutional Economics and Third World Development. London; New York: Routledge.

Hedström, Peter, and Richard Swedberg. 1998. Social Mechanisms: An Analytic Approach to Social Theory. Cambridge, England; New York: Cambridge University Press.

Hirschman, Albert. 1970. Exit, Voice and Loyalty: Responses to Decline in Firms, Organizations, and States. Cambridge, Mass: Harvard University Press.

Kreps, David. 1990. A Course in Microeconomic Theory. Princeton, NJ: Princeton University Press.

Lakatos, Imre. 1970. "Falsification and the methodology of scientific research programmes." Pp. 91-195 in Criticism and the Growth of Knowledge, edited by Imre Lakatos and Alan Musgrave. Cambridge, England: Cambridge University Press.

Mainwaring, Scott. 1999. Rethinking Party Systems in the Third Wave of Democratization. The Case of Brazil. Stanford, Calif.: Stanford University Press.

McAdam, Doug; John McCarthy; and Mayer Zald, eds. 1996. Comparative Perspectives on Social Movements. Political Opportunities, Mobilizing Structures and Cultural Framings. Cambridge, England; New York: Cambridge University Press.

McCarthy, John. 1996. "Constraints and opportunities in adopting, adapting and inventing." Pp. 141-52 in Comparative Perspectives on Social Movements. Political Opportunities, Mobilizing Structures and Cultural Framings, edited by Doug McAdam, John McCarthy, and Mayer Zald. Cambridge, England; New York: Cambridge University Press.

Nielson, Daniel, Matthew Shugart. 1999. "Constitutional change in Colombia. Policy adjustment through institutional reform." Comparative Political Studies 32, no. 3:313-42.

North, Douglass, Thomas Robert. 1976. The Rise of the Western World: a New Economic History. Cambridge: Cambridge University Press.

Putnam, Robert, Robert Leonardi, Rafaella Nanetti. 1994. Making Democracy Work. Princeton University Press

Putzel, James. 1997. "Accounting for the dark side of social capital," Journal of International Development vol. 9 no .7 pp. 939-949

Rubinstein, Ariel. 2000. Economics and Language: Five Essays. New York: Cambridge University Press.

Schelling, Thomas. 1978. Micromotives and Macrobehavior. New York: Norton.

Schumpeter, Joseph. 1934. Theory of Economic Development. An inquiry into Profits, Capital, Credit, Interest and the Business Cycle. Cambridge, Mass.: Harvard University Press.

______. 1939. Business Cycles. A Theoretical, Historical, and Statistical Analysis of the Capitalist Process. New York; London: McGraw-Hill Book Co., Inc.

Tarrow, Sidney 1996. "Making social science work across space and time: a critical reflection on Robert Putnam´s ´Making democracy work,` American Political Science Review vol. 90 no. 2.

Tsebelis, George. 1990. Nested Games. Rational Choice in Comparative Politics. Berkeley: University of California Press.

Young, Peyton. 2001. Individual Strategy and Social Structure. An Evolutionary Theory of Institutions. Princeton, NJ: Princeton University Press.