Knowledge

Why are institutionalized rules such a salient feature of societies? Why, for example, doesn't each member of a society develop a distinctive rule of behavior for him- or herself through experimentation and induction? In addressing these questions, it is useful to first note the contributions of game theory to our understanding of the complexity of decision-making in strategic situations and hence, indirectly, the role of rules. Game-theoretic models usually assume that players have a complete and closed model and correct common priors. Each player has complete information about the details of the situation, including others' preferences, the magnitude of various parameters, and various causal links. When such information is missing, the players assign the correct prior probabilities to all possible values of the unknown parameters. All players assume that their opponents model the game exactly as they do and they too assign the same correct probability to all unknown parameters.

These assumptions do not capture the world as we know it. Indeed, the complete model is demanding. Interactions among individuals are carried out in a complex environment containing many unobserved features that affect one's decision about how to act. These features include others' preferences and the specific magnitude of various parameters, such as wealth, the time discount factor, and outside opportunities. The computational complexity required to solve even a moderately complicated game is daunting.

Bounded rationality learning models is the game-theoretic response to its own unrealistic assumptions (e.g., Rubinstein 1998). Such investigations have been conducted, by and large, while retaining the assumption (that also prevails in classical game theory) that individuals are not guided by institutionalized, social rules. Learning is done in an individualistic and atomistic manner. Modeling such learning processes, to make them tractable, has involved replacing the assumptions of classical game theory with a new set of questionable assumptions, the central one being that individuals are myopic (Fudenberg and Levin 1998; Marimon 1997; Young 1998). Myopic behavior can imply, however, such unreasonable behavior as not performing a costly experiment no matter how high the resulting expected return might be. When such restrictions were not imposed, however, the resulting analysis became too complicated to provide a convincing account of how individuals in general learn.

In contrast, models that allow for socially transmitted rules are able to substantiate that learning can lead to regularities of behavior—to an equilibrium—without having to impose the restrictive conditions of either classical game theory or bounded rationality learning models. Kalai and Lehrer (1993, 1995) considered learning in a repeated game in which individuals share a cognitive system but all know only their own payoff matrix and discount factor. The players have the same cognitive understanding of the game but no player knows all the relevant parameters of the model. All players, observing the outcomes of the game, can develop only a subjective evaluation of the parameters and others' strategies. This analysis assumes that individuals are subjectively rational and does not impose the restriction that players assume that others are rational. Each one starts with subjective beliefs about the strategies used by each of their opponents and uses these beliefs to compute one's own optimal strategy. As the game unfolds and the players update their beliefs, players develop their own subjective models of the situation.

Analyzing this process of learning, Kalai and Lehrer show how an initial “grain of truth” regarding others' behavior is sufficient for individuals to learn independently how others will play. Specifically, if each player's initial subjective beliefs assign a positive probability to the events that would occur in the play of the game, then learning will lead each player, over time, to be able to appropriately predict the behavior of the others. If this is the case, these individuals will converge in finite time to play a Nash equilibrium in the real game.⁴

While the Kalai and Lehrer argument is intuitive, its technical analysis rests on the assumption that individuals will use Bayesian updating, which is not necessarily how people update their beliefs. But if people do not employ the logic of probability theory that Bayesian updating captures, institutionalized rules specifying others' behavior are arguably even more important in leading to regularities of behavior. Indeed, a sufficient condition for a Nash equilibrium is that individuals, who know only their own action sets and preferences, have an accurate prediction of what others will do. Knowledge (or common knowledge) of the rules of the game is not necessary (Aumann and Brandenburger 1995).

Institutionalized rules only assist in forming, but do not determine, beliefs because retrospective individuals compare outcomes with expectations. Only rules in which expected and actual behaviors correspond to each other will lead to the reproduction of these beliefs as they are confirmed by observed behavior. Hence, it is appropriate to restrict social rules that can establish themselves in a society and be followed—that can become institutionalized—to those that correspond to an equilibrium.

Because institutionalized rules constitute an equilibrium in the responses of individuals to these rules, they not only assist individuals in choosing behavior, but also aggregate, in equilibrium, the dispersed information that each of these individuals has. In other words, these rules both provide individuals with the information they need to make decisions regarding how to act as well as aggregating the information privately held by each of them. Institutionalized rules of behavior aggregate information in a compressed form and direct individuals to play an equilibrium strategy in the underlying game, although the assumptions held in classical game theory do not hold. And because such rules correspond to an equilibrium, individuals are seemingly rule-followers; they follow the rules associated with the social position they occupy.

Just as market prices aggregate the economic agents' private information, game theory reveals similar relationships between institutionalized rules of behavior and each individual's private information. These rules inform each individual about others' expected behavior. But the only rules that can correspond to actual behavior are those in which all individuals, basing their decisions on private information, find it optimal to follow. In an institution, institutionalized rules aggregate the private information of all agents, providing each with sufficient statistics to make an informed decision.⁵

In our analysis of transactions in which individuals are guided by institutionalized rules, it is therefore appropriate to assume that players share common knowledge of the rules of the game. The information compressed in socially transmitted rules enables individuals without knowledge of all the relevant parameters and causal mechanisms, and with limited computational ability, to act in a manner that leads to equilibrium behavior. Once equilibrium is achieved, because individuals do not observe relevant parameters and lack full comprehension of causal relationships, the best they can do is to perceive the world as stationary as long as observations—including those conveyed through others' behavior—do not contradict this perception.

Hence, past behavior would reign despite marginal parametric changes because institutionalized rules enable individuals with limited knowledge and information to choose behavior. And thus, behavioral rules learned in the past are the best predictor of future behavior. As long as others' behaviors (that one neither observes nor understands their causal underpinning) do not reflect a change in the parameters, one would not change one's own behavior either. Conversely, when the parameters that one observes marginally change, one faces the problem of which behavior to follow in the new situation, given the multiplicity of self-enforcing behaviors. Expecting others to continue following the previous institutionalized rules of behavior, it is rational to continue following the past patterns of self-enforcing behavior.

Attention

Past patterns persist also because what one sees, knows, and understands in a given situation also reflects the amount of attention one devotes to the task. Attention is a scarce resource. Institutionalized rules come to the rescue. They enable one to choose behavior in complicated situations while devoting limited attention to decision-making in noninstitutionalized situations.⁶

Coordination

Coordination failure is the third reason why a marginal parameter shift does not necessarily lead to changed behavior. When a situation marginally changes, individuals face the problem of which behavior to follow in the new situation, given the multiplicity of self-enforcing behaviors. Because people do not share the expectations that some new equilibrium behavior will be followed, they are likely to rely on past rules of behavior to guide them and to continue following past patterns of self-enforcing behavior. This would be the case even when there are individuals and organizations with the ability to coordinate on new behavior. There are many reasons why such coordination may fail to transpire even when it is beneficial. Sunk costs associated with coordinating change, free-rider problems, distributional issues, uncertainties, limited understanding of alternatives, and asymmetric information may hinder coordination on new behavior.

Historical Background

The residents of the Venetian lagoon established Venice as a political unit in 697 and residents of Genoa organized themselves into a Commune around 1096. These cities became the two most commercially successful Italian maritime city-states.¹⁰

In both cities, clans and families became the prominent unit of social organization due to the decline of central authority (Hughes 1978). Given this decline, Herlihy (1969, 178) notes, “The corporate or consortial family was better able than the nuclear household to defend its wealth and status,” increasing “family solidarity, at least among the aristocratic classes.”¹¹

The resulting political institutions governed a particular transaction: motivating individuals—members of the cities' strong clans and families—to relinquish decision-making power and resources in return for political order and the economic benefits of collective action. The political organizations of Genoa and Venice were seemingly identical. Both cities were governed by an oligarchy that, by and large, selected their political leaders, and these leaders were subject to the law. At the top of Venice's political system was a Doge and the Ducal Council; Genoa was governed initially by consuls and, after 1194, by one or more executives—called the podestà (power)—and a council of rectors.

The political institutions that prevailed in Venice and Genoa from the late eleventh century were able to support interclan cooperation that initially fostered commercial expansion and political order. Yet the subsequent history of the two cities differs. Venice was able to maintain political order in a changing economic environment and to mobilize resources maintaining its economic prosperity even following the decline of its trade with the Far East. Meanwhile, its members' social attachments to the clan structure seem to have slowly declined. Genoa's experience was the opposite. Political order often broke down contributing to its economic decline. Meanwhile clans became more important social and political units over time.

How can these different histories be accounted for despite the similarity in initial conditions, outside opportunities and the basic political structures of the two cities? Both Genoa and Venice had initially developed political regimes that were sufficiently self-enforcing to sustain economic prosperity. Yet Genoa's institutions were self-undermining while Venice's were self-reinforcing. In developing this argument we consider quasi-parameters such as the wealth of the cities, the strength of the popoli, and the social identities of the clans. Understanding these cities' subsequent histories requires considering how these quasi-parameters changed their values as a result of the implications of the institutional equilibria. Changes in the quasi-parameters in Genoa had the effect of undermining political order, making its institutions sensitive to relatively small exogenous shifts in clans' strength, trading opportunities, and level of external threat. The opposite changes, however, transpired in Venice, whose magistracy was self-reinforcing.

To understand these histories and their long-term implications, we examine more closely these cities' institutions. Although the origins of these two distinct institutions are not the focus of the discussion, they probably reflect the institutional heritage of the Doge, a less unequal initial distribution of interclan military might and wealth, and a series of able leaders who coordinated on and developed elements of Venice's institutions.

Genoa

Initially (from 1096 to 1194), Genoa had elected consuls who functioned as the city's political, administrative, and military leaders.¹³

Early in this period, the relatively high gains from the joint mobilization of resources implied that interclan cooperation was not hindered by interclan rivalry. But because interclan cooperation advanced Genoa's economic prosperity (an endogenous change in a quasi-parameter), it intensified (given beliefs about clan motivations) the competition over political and economic dominance in the city. Fearing that any temporary decline in its relative power would constitute an opportunity that the other clan would take advantage of, clans became engaged in an “arms race” (leading to yet other endogenous changes in quasi-parameters): they bought land that they then fortified to dominate particular quarters, they established patronage networks, and they socialized their members to internalize loyalty to the clans and the norm of revenge to protect clan honor.

A foreign threat constituted an exogenous shift in parameters that sustained interclan cooperation. For a period following 1154, attempts by the Emperor Frederick Barbarossa to regain de facto control over northern Italy weakened the link between political order and mutual deterrence. This particular external threat did not alter beliefs of the clan members of what other clans would do if the threat receded, but it did alter beliefs about the long-lastingness of the external threat, and this change not only lowered each clan's incentive to challenge the other militarily, but also made Genoa's political institutions self-enforcing in a wider range of situations than before. The result was that the Genoese clans mobilized their resources, acquired overseas commercial possessions, and expanded commercially. Through peace, Genoa's economic structure was transformed to one based on long-distance commerce.

Yet were the imperial threat to be weakened, this commercial expansion and structural transformation would undermine interclan mutual deterrence by making it self-enforcing for a smaller range of parameters. The higher level of economic prosperity (a change in a quasi-parameter), with concomitant gains to a clan for controlling the city, in the absence of an external threat, implied a smaller set of parameters for which mutual deterrence would have been self-enforcing.

In 1164 unexpected civil wars in Germany engaged the Emperor's attention. Genoa returned to its pre-1154 situation of low level of external threat. But the quasi-parameter of wealth was now higher than it was before, and with beliefs remaining stable, the previous mutual deterrence equilibrium between the clans was no longer self-enforcing. The commune sank into a lengthy civil war during which various clans gained the upper hand for a time, only to be challenged again when exogenous conditions changed. As noted by a twelfth-century Genoese chronicler, “Civil discords and hateful conspiracies and divisions had risen in the city on account of the mutual envy of the many men who greatly wished to hold office as consuls of the commune” (Annali 1190, vol. II, 219–20). The fighting from 1189 to 1194 was particularly devastating and endangered the existence of the city.

These events in Genoa do not reflect only the influence of a shift in exogenous conditions. Rather, they reflect the fact that endogenous changes—increasing commercialization and prosperity, the clans' past investments in military ability and patronage, and, arguably, the fomenting of individuals' identities as clan members—made Genoa's institution self-enforcing for a smaller set of parameters. The city that was peaceful despite the absence of a threat by an Emperor prior to 1154 became embroiled in a civil war during the Emperor's absence after 1164. An exogenous situation that previously would not have led to the collapse of Genoa's institution now had a devastating effect.

In 1194, the Holy Roman Emperor, now needing the assistance of Genoa's navy, had an interest in ceasing the civil war. By the promise of rewards and threat of war he induced the Genoese clans to agree to alter Genoa's political institutions by introducing a self-enforcing organization that restored interclan mutual deterrence and cooperation. At the center of Genoa's new institution was a non-Genoese podestà who was hired for a year to be Genoa's military leader, judge, and administrator and who was supported by the soldiers and judges he brought with him.

The podestà and his military contingent fostered the clans' ability to cooperate by creating a military balance between them. The podestà's “threat” of assisting the other faction deterred each clan from attempting to control the city. Moreover, because the podestà was to receive a high wage at the end of his successful term, his threat was credible. For if one clan took control of the city, there was no reason for it to reward the podestà. Furthermore, this reward scheme made it in the podestà's interest not to alter fundamentally the balance of power between the factions. Hence he could credibly promise to be impartial and retaliate only against people who broke the law rather than against an entire clan. The selection of an incoming podestà was made by a committee of representatives from the city's various neighborhoods. It was large enough not to be dominated by any particular clan. The podesteria fostered interclan cooperation for a while and, thus, political stability and economic growth. It was a self-enforcing institution: The self-enforcing belief in the futility for any clan to gain political dominance using force deterred clans from trying. The belief that all clans could gain from cooperation without risking their economic position through unexpected military confrontation also motivated cooperation.

Yet, like the consular system, the podesteria was not reinforcing—indeed, it contained the seeds of its own destruction. Specifically, since the podesteria was based on balance of military strength between the clans and each clan wanted to be militarily prepared in case of need, it contained interclan rivalry but did not eliminate it. Each clan was still motivated to strengthen itself militarily vis-à-vis the others, and clan members' main identification was still with their clan and not the city.

The creation of the alberghi and the rise of the popolo as a faction in this period were both further manifestations of nonreinforcement of the institutional equilibrium. Alberghi were clanlike social structures whose purpose was to strengthen consorterial ties among members of various families through a formal contract and by assuming a common surname, usually that of the albergo's most powerful clan (Hughes 1978, 129–30). By the fifteenth century the city's politics and economics were dominated by about 30 alberghi, each containing from 5 to 15 lineages. With the alberghi, battle lines between clans hardened. Furthermore, the attempt of each clan to develop a patronage network and the access of all residents of the city to Genoa's overseas lucrative trade both implied that over time the nonnoble element of the city— the popolo—would acquire the resources, organization, and self-consciousness regarding their common interests to form a political faction that could disrupt the noble-controlled equilibrium.

Despite the peace of the podesteria, Genoa's institutions motivated clans to further invest resources in acquiring military ability aimed at attacking other clans, fortifying their residences, establishing patronage networks (thereby mobilizing the popoli), and indoctrinating their members to internalize the norms of revenge and adopting identities (through the alberghi) as clan members. In the short run, all these changes did not render the podesteria ineffective; it was still self-enforcing. But over time these changes caused Genoa's political structure to become self-enforcing for a smaller range of situations, leading to its eventual demise. In the long run, a podestà could not constrain the balance of power incentives among Genoa's rival clans and the system collapsed.

Venice

The history of Venice during its early days parallels that of Genoa. After an initial period of interclan cooperation, Venice's history was characterized by interclan rivalries aimed at capturing the office of the Doge (Lane 1973; Norwich [1977] 1989). Originally the Doge was a Byzantine official, but shortly after Venice was established in 679, the post became that of an elected monarch. For the next few hundred years, clans fought in Venice for control over the Doge's post. Similar to Genoa, economic cooperation was hindered by the lack of an institution able to contain interclan rivalry.

Changes around the Mediterranean increased the cost of such confrontations. Toward the end of the eleventh century, the decline of Byzantine naval power increased the gains to the Venetians from forming a political institution enabling cooperation. They responded to this opportunity by establishing a new self-enforcing institution. At its center was the belief that each clan would join together to fight against a renegade clan that attempted to gain political dominance over the city and its economic resources. Arguably, this belief and the behavior it entailed fostered a common Venetian identity. In any case, the belief was sustained by a set of rules whose prescribed behavior was made self-enforcing by that belief. The rules limited the Doge's power to distribute economic and political rents, curtailed the clans' ability to influence the outcome of the election of a Doge (or any other officer), established tight administrative control over gains from interclan political cooperation, and allocated these rents among all the important Venetian clans fairly so all had a share in them regardless of clan affiliation. This allocative rule therefore did not give incentives for clans to increase their military strength to plan for interclan military conflict. Since the establishment of these rules began to occur when Byzantine and Muslim naval powers were on the decline and cooperation was most beneficial, Venetians were able to make the most of this opportunity.

Starting in 1032, the Doge's authority was limited through the establishment of advisory councils until it was de facto altered from an elected monarchy to a republican magistracy. In 1172 it was established that a Doge should never act contrary to the advice of his councilors. The selection of the new Doge was entrusted to an official nominating committee to inhibit the ability to use a clan's political machine and popular support to influence the election. The nominating committee itself was selected and formed through an elaborate process that used both lots and delegations. The (partially random) process began in the Great Council, in which all adult nobles were eligible, so that all Venice's main clans had representatives. From this Council a committee of 30 was chosen by lot and the selection proceeded through an additional nine steps of delegation and selection by lot until the proposed candidate for Dogeship was brought before the Venetian assembly for approval. The importance of clans was reduced in this process by requiring that only one family member could be on any committee, and a person could not vote when a relative was being considered. The process itself was designed to reach a quick decision.

Similar, although less elaborate, systems were used for the selection of other officials. Their numbers were relatively large and their time in office relatively short, so that members of many clans could hold an office in a given period of time. Nomination committees for many posts were selected by ballot in the Great Council in a way that gave every person present an equal chance of being on a nomination committee. The conduct of all officials (including the Doge) was subject to scrutiny by committees to prevent unlawful gains.

That each clan had a stake in the implementation of these rules made self-enforceable the belief that each clan would join others to confront any clan that attempted to use military power to gain control over the city. But the rules and the associated beliefs were also reinforcing: They provided clans with few incentives to invest their resources in fortifying their residences or instilling in their members norms of loyalty to the clan rather than the city. By weakening the clans, Venice's republican magistracy over time increased the range of situations in which it was self-enforcing. This institution also prevented the endogenous formation of a political faction among nonnoble elements of the city, the popoli, because the magistracy as an institution did not motivate clans to establish patronage networks that would have channeled rents from political control over Venice's overseas possessions to nonnoble clans.¹⁴

Summary

During the eleventh and twelfth centuries, the self-enforcing institutions that emerged in Venice and Genoa were successful in fostering interclan cooperation. Yet they had distinct long-run implications.¹⁵

Cleavage Structures as Institutions

Cleavage structure is usually thought of not as an institution, but more as a social reality. Yet it fits well within our definition of institutions. Consider first our criterion of an institution as being human-made. While schools of thought a generation ago held that people were given their social identities, it has become increasingly accepted that in fact people construct those identities. In the now-classic study by Thompson (1963), we know that the English “working class” had to be “made” through concerted social, cultural, and political activity. The creation of tribal cleavages in Africa or caste cleavages in India similarly has a strong human-made component (Chandra 2004; Laitin 1986). More important, the relative salience of cleavages—for example, that of religion in western Europe—is the result of social and political activity (Kalyvas 1996).

The elements that make up a cleavage structure include beliefs, organizations, and rules. Take rules. While categories such as “Jewish” and “African-American” and “worker” are constructed, there are usually clear rules concerning who can claim membership in these categories. Jews accept rules of maternal descent; in the United States an ounce of black blood is the rule for African-American membership; and salaried employment in blue-collar work is a rule-of-thumb for someone to qualify as a worker. As for organizations, categories of membership on cleavage structures often have preexisting organizations (e.g., churches, unions), but it is equally the case that organizations emerge to represent groups identified by salient cleavages (e.g., ethnic associations). And, as will shortly become clear, cleavage structures entail beliefs about the social world and how it is constituted.

As per our definition, the system of human-made elements that constitute the cleavage structure is exogenous to each individual whose behavior is influenced. People condition their political behavior in part on their self-identifications across social divisions or cleavages and on the perceived salience of those cleavages in political competition. Yet each individual cannot unilaterally legislate a new hierarchy of cleavages or a new set of categories of membership that define cleavages. Cleavage structures are therefore endogenous to societies but exogenous to any individual member of that society. And, finally, cleavage structures govern a transaction—political support or votes for group favors.

In studying cleavage structures and their implications for social peace and economic growth, many analysts take the ethnic/linguistic fractionalization (ELF) of a country's population to be parametric (Barro 1997; Easterly and Levine 1997). There is a clear plausibility to this—if there are two tribes in a society, one with 55% of the population and the other with 45%, we should expect a different form of conflict than if there were 100 tribes, each with 1% of the population. ELF is one algorithm designed to capture consequential differences in the ethnic arithmetic of a society. If ethnic demography is exogenous to political conflict, then ELF can be taken as one parameter governing the equilibrium supporting a particular societal cleavage structure.

There is a problem, however, in postulating ELF as a parameter in support of a particular cleavage structure. Suppose the equilibrium allows for slow processes of assimilation, or emigration, or pogroms. In these cases, ELF can change in value in the long term. We therefore postulate ELF as a quasi-parameter. We want to see whether changes in ELF, induced by the self-enforcing cleavage structure, reinforce or undermine that cleavage structure. In our paired comparison, we suggest that given parametric shift in ELF, the tribal-based cleavage structure in Nigeria is self-reinforcing but that the nationality-based cleavage structure in Estonia is self-undermining.

Application

The cleavage structure in Estonia and Nigeria shares two institutional elements: the organization of national (tribal) groups and everyday primordialist beliefs that supported this organization. Where institutional elements differ is that in Estonia the majority group has high social status that is recognized by the minorities, while in Nigeria the plurality group does not enjoy such status. The embeddedment of these cleavage structures in new states—states engaged in nation-building as well as management of their national economies—can be considered parametric. But the ELF index (that is, the categories of nation/tribe and the numbers in each category) is modeled here as quasi-parametric. Beliefs in the high status of Estonians give incentives, if voting results consistently favor Estonians, for non-Estonians to assimilate; over time these small changes in identity will affect the value of ELF. In Nigeria, there are no such incentives to assimilate, and therefore ELF is not likely to change in value. Our analysis foresees that in the context of political organization for elections (the transaction that is governed by the institution), a factor exogenous to the cleavage structure in Estonia (unitary political institutions) will transform Estonia's cleavage structure over time. Meanwhile, for elections in Nigeria, without any endogenous change in the quasi-parameter of ELF, shifts in exogenous political and economic institutions (moves from federalism to unitary government or reduction in the importance of oil) would not, because it is self-reinforcing, likely change its tribally based cleavage structure.

We now describe the cleavage structures. In newly independent Nigeria (1960), political parties were regionally based and catered to the needs and aspirations of the majority tribal group in their regions. In the Western Region, the Action Group catered to the interests of the dominant Yoruba population, and Yorubas largely associated themselves as supporters of that party. Within the Yoruba Region, factions within the Action Group represented the interests of subtribes of the Yoruba, associated with different ancestral cities. Similarly the NPC, the party of the Northern Region, catered to Hausa interests; and the NCNC, the party of the Eastern Region, catered to Ibo interests. We can summarize the dominant cleavage structure of newly independent Nigeria as tribally based, with three principal groups dividing the political pie.¹⁷

In independent post-Soviet Estonia (1991), the dominant cleavage was based on nationality, one that pits a nationalist majority of Estonians against a largely disenfranchised but nationally conglomerated population of Russian-speakers (made up of Russians, Ukrainians, Belarusans, and Jews). The majority of Russian-speakers in Estonia are from families that migrated to Estonia after the Soviet occupation of 1940 and do not have command of the Estonian language. Most of these Russian-speakers did not qualify for citizenship in 1991 and therefore could vote only in local elections.

Nigeria and Estonia both have self-enforcing cleavage structures. Political leaders present platforms and candidate lists that reflect the interests of nationality constituencies and voters tend to respond to symbols and messages that speak to them as members of a particular tribal or nationality group. These cleavage structures are sustained by beliefs that have been dubbed “everyday primordialism” (Fearon and Laitin 2000). Primordialism is the belief that ethnic/nationality differences are biologically given and ultimately more important than any other possible identification when it comes to social, political, or economic transactions. Primordial beliefs of this sort are hardly universal. In the two cases here, they were created and sustained under previous political structures. British colonialism ruled “indirectly” through tribal chiefs, who were under salary by the British colonial state. These tribal chiefs were granted levels of authority they rarely achieved in the precolonial period, and Nigerians, to get a hearing from the British overrulers, had to petition through tribal authority structures. Thus colonialism played an important role in delineating tribal boundaries, clarifying tribal cleavages, and generating primordial beliefs.

Stalinist rule in the Soviet Union had a similar structure (Suny 1993). Each nationality was given its own political entity, and people's rights to position and property were dependent in large part on their nationality, as recorded on their passport. Soviet campaigns to change popular beliefs oftentimes only changed the public presentation of selves. Many Soviet citizens acted as if they were atheists but were secret believers; many Soviet citizens portrayed themselves as Stakhanovites who worked ceaselessly for the common good but privately stole from the state to sustain their families. But with the policy of korenizatsiia (cultivating local roots), the Stalinist system succeeded in creating a common knowledge of nationality. People not only accepted their nationality designations as real, but also believed that others believed their nationality to be real. An excellent example of this is in the Central Asian republics, where the Soviets named and codified nationality groups that were new and foreign to those who were assigned new nationalities (Allworth 1990). Yet two generations later, these designations became the bedrock of a set of newly independent national republics. Under British and Soviet overrule, tribal/nationality differences were codified, distributions were made in terms of tribal/national identities, and local political authorities who had monopoly representative rights for their tribal/nationality groups had an interest in cultivating beliefs that the cultures of their constituents were primordially given.

Nigeria

While tribal- and nationality-based cleavage structures have been self-enforcing in both Nigeria and Estonia, only in the Nigerian case can they be considered self-reinforcing. In Nigeria, the cleavage structure is self-reinforcing in that beliefs about the value of one's own tribal group, if anything, increases. Furthermore, institutions exogenous to the cleavage structure lend support to it. Federal institutions were built into the independence constitution. Political distributions were made based on formulae that returned federal funds to the original three regions. In 1967, the Eastern Region (whose leaders were opposed to the formula for the distribution of newly gained oil revenues) attempted secession, but lost a three-year bloody war fought against federal forces. Subsequent to the war, several minority tribes were given their own federal units (then called states). Each of the 12 states got a budget supported in large part on federally collected oil revenues. Since each state got a base allocation to cover the infrastructure of rule, smaller and smaller nationality groups grasped the incentives to demand their own states. By 1996, there were 36 separate states, almost all dominated by a single tribal group. Parallel to the demands for single tribe states, there have been in Nigeria increasing demands for the designation as a new local governing area for smaller (and sub-) tribal groups. For just a single example, in the city of Warri a communal war has pitted the Itsekiris who have controlled the local government against the Ijaws and Urhobos, both of whom want the city divided into separate local governing areas such that each of these groups can have control over a federally dispensed budget.

The federal system, in rewarding groups that can demonstrate a primordial claim to differentiate itself from the group that controls a particular state, supports the tribally based cleavage structure (Suberu 2001). Furthermore, with oil as the only part of the economy that provides substantial revenues, and with the federal authorities controlling those oil revenues, distributions from the center were of increasing relative value. Tribal groups that want more revenues need to claim primordial difference from the rulers of their state in order to be awarded a state of their own. With a state of their own, these groups have a claim on oil revenues for their own purposes. Through oil and federalism—exogenous to the cleavage structure—tribal identification gains increasing relevance, as these identifications are the key to statehood (and thus the value of ELF remains unchanged). More important, because tribal membership is key not only to state recognition but also to tribal group resources, tribe remains the dominant Nigerian cleavage. In this case federalism and oil are exogenous and parametric. But unless there are some endogenous processes undermining tribal membership rules (or promoting assimilation, doubtful given the institutional beliefs about status), the value of ELF as a quasi-parameter will not change, thus making the Nigerian cleavage structure self-reinforcing. This is what we mean by an institution surviving under a wider range of situations. It is the stronger boundaries between tribal groups, reinforced by beliefs about status, that stabilize ELF as a quasi-parameter, thereby making the primordially based cleavage structure in Nigeria self-reinforcing.

Estonia

Although korenizatsiia was abandoned as official policy by the mid-1930s, Estonia (coercively brought into the Union in 1940 based on von Ribbentrop's secret pact with Molotov) under Stalin's Bureau of Nationalities inherited many features of that policy. Estonians were in an institutional sense the “owners” (called “titulars”) of the Estonian Soviet Socialist Republic. All party and state documents, education, and the media in Estonia would be transmitted either in Russian (the language of Russia, the “elder brother” of all titular peoples) or in Estonian. Korenizatsiia institutions sustained the belief throughout the population of everyday primordialism that Estonians were forever and primarily Estonians, even if the Soviet Union was a suprarepublic that made possible the “brotherhood of the peoples.” In the standard country study written by an Estonian, the notion that a Russian could be an Estonian was not even considered (Raun 1987). Nationality boundaries are popularly conceived of as consequential for political loyalties and not subject to change.

With the breakup of the USSR, with titulars in the majority, Estonian parliamentarians declared Estonia to be a unitary state and wrote citizenship laws that effectively disenfranchised (for national elections) the 30% of the population that was Russian-speaking with roots outside of Estonia. At first, these new institutions united the Russian-speaking population into an opposition front, with some leaders articulating a goal of reunification with the Russian Federation. In this sense, the primordial cleavage dividing Estonians from Russian-speakers was self-enforcing.

But the cleavage structure in Estonia is not self-reinforcing. The relatively high status of Estonian language and culture in the eyes of the Russian-speakers is a belief that encourages Russian assimilation into Estonian culture, potentially changing the value of ELF. Exogenous parameters such as the opportunities as citizens of Estonia to take part in European institutions, the unitary political institutions, and the chances to participate in those institutions with the learning of the Estonian language all gave incentives, especially to young Russian-speakers, to assimilate into Estonian society. In the short term, the quasi-parameter of the national demography or ELF of the population is only marginally changing. Estonian is becoming (for some cross-national interactions) the language of internationality communication between Estonians and Russians. At first, only Russian-speakers with a low cost for passing (e.g., those who learn languages easily) become Estonians. In the longer term, there could well be a new generation of Russian-speakers fluent in Estonian. In consequence, the cultural differences between Estonians and (descendants of) Russian-speakers will begin to narrow. Russians can become Estonians, or Russian-Estonians, thereby changing the value of ELF in the longer term. As the country's population through assimilation becomes decreasingly divided by language and nationality, the Russian/Estonian cleavage will decline in the realms for which such identifications are today significant. We can thus project institutional undermining of the dominant cleavage in the longer term, perhaps giving way to a class-based dominant cleavage. In this case, then, the ethnic cleavage structure in Estonia is self-enforcing but not self-reinforcing.