MOTIVATION

As for inequality, part of the motivation for studying polarization is ethical. Unlike inequality, however, the ethical motivation comes from the view that distances and dif­ferences across groups—as opposed to across individuals for inequality—are normatively undesirable.

Much of the motivation for the study of polarization also owes to the view that it is closely linked to the “generation of tensions, to the possibilities of articulated rebellion and revolt, and to the existence of social unrest in general” (Esteban and Ray, 1994, p. 820). Linking group formation and the eruption of social unrest has a long history. Aristotle wrote (2350 years ago) that “it is manifest that the best political community is formed by citizens of the middle class, and that those states are likely to be well- administered, in which the middle class is large... where the middle class is large, there are least likely to be factions and dissension” (Aristotle-350).

To take just another prominent historical example, the Marxist critique of Hegel’s philosophy (see, for instance, O’Malley and Blunden, 1970) has forcefully argued that the emergence of two distinct social classes, the bourgeoisie and the proletariat, leads pro­gressively to class struggles as societies industrialize. More generally, humanity’s history is often described as a series of group struggles, those groups being typically fairly well defined according to socioeconomic characteristics, interests, and statuses.

The modern formal conceptualization of economic polarization along such antago­nistic lines owes much to Esteban and Ray (1994)—ER. Polarization is defined as the grouping of the population into significantly sized clusters such that each cluster has members with similar attributes, and different clusters have members with dissimilar ones.

Consider first the case of social polarization. Ethnicity (broadly interpreted to include religious, racial, and linguistic identities) is commonly perceived as an important source of it; ethnic groups are seen as firmly bounded, inclined toward ethnocentrism, hostile to outsiders, and as exhibiting a belief that one’s ethnic group is centrally important. This can position ethnicity at the center of politics and development in many societies—it is indeed frequent to observe political parties split along ethnic lines (Horowitz, 1985). A social norm of equality further makes the subjugation of ethnic minorities illegitimate and spurs ethnic groups to compare their standing in society against that of other groups. With various technological and social developments, associated political divisions and resultant conflicts may also be increasingly common (Glaeser and Ward, 2006).

The economic consequences of ethnic identities and of (potential or actual) conflict can be numerous (Montalvo and Reynal-Querol, 2005a). Trust and trade may be restricted to individuals of the same ethnic group; public infrastructure may be ethnically biased; government transfers may disproportionately favor some ethnic groups, and so on.

Underlying the listed social tensions also clearly lies a diversity of economic statuses and interests. It is in fact often argued that sociopolitical markers of tension are just proxies for more fundamental economic determinants of polarization (Creamer, 2007). The evi­dence in McCarty et al. (2006) suggests, for instance, that the increase in the divergence of economic interests of the major constituencies of the two major American political parties may have led to an increase of polarization in American politics.

Theoretical demonstrations of the role of polarization in politics and governance can take a variety of forms. Rent seeking on the part of the different groups is one of them. Montalvo and Reynal-Querol (2005b) present a simple pure contest game in which agents seek rent by spending resources in favor of a preferred group outcome. The utility distances across the groups are set to a constant, and group sizes are assumed to be equal. The level of resources spent by agents affects the probability of success in the contest; that probability is equal to the share of each group in the total resources allocated to the con­test. Because of the interaction between group sizes and group probabilities of contest success, the total resources allocated to the contest are shown to be a function of the social polarization index shown in Equation (5.33).

Polarization can also be modeled to lead to conflict in contexts in which distances across groups matter. A brief formalization of this takes the following form. Individuals in a given group derive greater utility from outcomes preferred by groups that are “closer” to them. The probability of a group implementing its preferred (ideal) outcome at the expense of other groups depends on the resources spent by the group in taking control. The sum of these resources also determines the importance of group conflict.

Outcomes of such a game have been studied in Esteban and Ray (1999). Increases in the utility distance between any pairs of groups lead to increases in social conflict. Con­flict is always maximized on the symmetric bimodal distribution of the population. But there are many nonlinearities. For instance, a merger of two groups in a society with at least three social groups can increase or decrease conflict depending on the size of the merged groups as well as on the distribution of the population across the nonmerged groups. A movement away from a symmetric distribution of three equally sized groups to a symmetric two-group distribution initially decreases conflicts before it eventually increases it.

A richer framework for studying both the occurrence and the intensity of conflicts and the effect of polarization and fractionalization in each case is found in Esteban and Ray (2008). In a highly polarized society, conflict is expensive, so that its appearance is rare. But when conflict occurs, it is intense. Less-polarized societies, where conflict is less costly, witness a higher frequency of social unrest but of a more moderate intensity. Frequency and intensity can therefore be negatively correlated; it may be that the overall importance of conflict (frequency times intensity) may be reached at intermediate levels of polarization. The occurrence and the intensity of conflicts also depend crucially on the nature of the political system. For all these reasons, the precise overall relationship between polarization and conflict ends up being complex and nonlinear, even in rela­tively well-structured theoretical settings.

Group distances and group identity affect differently inequality, social polarization, and income polarization. Group cohesion and group distances can also influence dif­ferently the size of group conflict in an environment in which control over resources is partly determined by group action. The most elaborate modeling of the linkages between conflict and features of group distance and identity is found in Esteban and Ray (2011b).

The game consists of fighting over a budget, a fraction of which is used for a public good and the rest of which is used for a private good. Each group has a most-preferred composition of public goods. An exogenous fraction of the budget is used for the public good; the precise allocation of that public good is endogenously determined by the iden­tity of the winning group. A degree of group cohesion captures whether group payoffs, and not individual ones, are maximized. Group cohesion can be interpreted, for instance, as a degree of altruism or as an indicator of group leadership.

Group members choose to make contributions to the resources used by their group to increase the group’s probability to win the game, seize control over the government’s budget, and choose the allocation of the public good. In equilibrium, members of any given group make the same contribution; in equilibrium, the groups might be contrib­uting different per capita levels of effort, but it is shown that this is not expected to affect significantly the value of aggregate conflict intensity. An approximately linear relation­ship is then established in Esteban and Ray (2011b) between equilibrium conflict inten­sity, the Gini index, fractionalization, and polarization over differences in public good utilities. Four interesting polar cases emerge.

First, when all goods are private, utility distances across different group choices of public good allocations do not matter. This is because utility distances matter only when the allo­cation of public goods is important to groups. Only group sizes (and not group distances) drive conflict over the allocation of private goods. Distance-based indicators of divergences such as inequality and polarization do not predict conflict; it is only those indicators that are based solely on group size divergences (namely, fractionalization) that matter.

Second, when all goods are public, distances over preferences become salient, and the distance-influenced measures of polarization and inequality dominate fractionalization (which is invariant to group distances) as determinants of conflict. The degree to which polarization dominates inequality as a determinant of conflict depends on the strength of group cohesion; the greater the degree of group cohesion, the greater the conflict­determining importance of polarization as a measure influenced both by group sizes and by group distances.

Third, the relative importance of polarization and fractionalization depends on the importance of public goods. If public goods dominate the allocation of government spending, then conflict rests entirely on the value of winning the public goods allocation game. The importance of polarization (which takes into account the utility distances across groups) then dominates the importance of fractionalization as a determinant of conflict because it alone takes into account public good utility distances across groups and therefore the value of competing for the precise allocation of that good.

Last, inequality has a negligible role when group cohesion is important and, more important, when population size is large (a large population makes individual action less effective). More generally, conflict arises only if population size is small (in which case inequality across individuals is important because there are few interacting individuals and individual action is thus important) or if there is group cohesion (in which case it is polarization and fractionalization—instead of inequality—that play an important role).

5.3.