Decentralization(MioCar2009PRQ)

7/28/2019 Decentralization(MioCar2009PRQ)

1/16

Political Research Quarterly

Volume XX Number XMonth XXXX xx-xx

2009 University of Utah10.1177/1065912909338462

http://prq.sagepub.comhosted at

http://online.sagepub.com

1

Does Political Decentralization Exacerbate

or Ameliorate Ethno-political Mobilization?

A Test of Contesting Propositions

Dan MiodownikHebrew University Jerusalem

Britt CartriteAlma College, Michigan

This article presents the results of an experiment that attempted the reconciliation of opposite expectations regardingthe effectiveness of political decentralization on ethno-political mobilization. An agent-based model was run thou-sands of times to explore the effect of decentralization. The experiments suggest that the impact is nonlinear: weakand medium levels of decentralization increase the likelihood of ethno-political mobilization, while strong decentral-ization decreases it. The explanation derives from how minority control of political institutions affects the dynamicof minority identity ascription and the realization of the goal or the frustration of ethnic members seeking more com-plete political dominance of the regional ideational space.

Keywords: political decentralization; ethno-political mobilization; agent-based modeling

In August 2000, then French interior minister andleading leftist politician Jean-Pierre Chvenementabruptly resigned from the government in protest overa proposed agreement extending limited legislativecapacities to the regional Corsican assembly begin-ning in 2004. While Socialist Prime Minister LionelJospin, a close personal friend of Chvenement,raised the Corsican issue as an attempt to end morethan two decades of separatist violence on the island,Chvenement, among others, argued that accommo-dating the Corsicans would only stimulate additionaldemands from Corsicans as well as other regionalgroups in France, such as the Alsatians, Bretons, andBasques (all of whom have active ethno-regionalistparties), thereby threatening the future of the republic

itself. In the end, Jean-Pierre Raffarin was elected andreplaced Jospin in 2003; while the extension of limitedautonomy to Corsica was subsequently withdrawn, theRaffarin government proposed to merge the two dpar-tements of Corsica into a single administrative entityfor the whole region. Then incoming interior ministerNicolas Sarkozy remarked on July 5 of that year, as anapparent justification of the measure, Sardinia, justnine kilometers away, has enjoyed autonomy since1948 and has no independence seekers (Deutsche-Welle 2003). However, Corsicans narrowly rejected

the proposal in a referendum the following day.

The issue of how best to address ethno-politicaldemands is hardly unique to France: ethno-politicalmovements seem to be on the rise everywhere.Political organizations primarily seeking to shape andpromote the economic, cultural, and political agendasassociated with a substate region and its residents(typically an ethnic or some other kind of minority)have left their mark since the Second World War onthe political and social life of advanced and less-devel-oped states (Horowitz 1985; Tiryakian and Rogowski1985), stable democracies (e.g., Keating 1998;Newman 1996), and states undergoing transition virtu-ally anywhere across the globe: in the Balkans, CentralAsia, Africa, the Middle East, and the former Soviet

Dan Miodownik, Assistant Professor of Political Science andInternational Relations, Hebrew University Jerusalem; e-mail:[email protected].

Britt Cartrite, Assistant Professor of Political Science, AlmaCollege, Michigan; e-mail: [email protected].

Authors Note: The authors are thankful for the useful commentsof the anonymous reviewers who read the article as well as com-ments from Ian Lustick, Roy Eidelson, Ben Eidelson, Lilach Nir,Ifat Maoz, Maurits van der Veen, Rudra Sil, David Rousseau, andAmal Ahmed. Support for this project was provided by theSolomon Asch Center for the Study of Ethnopolitical Conflicts(formerly at the University of Pennsylvania) and the Israeli

Science Foundation.

Political Research Quarterly OnlineFirst, published on June 24, 2009 as doi:10.1177/1065912909338462


2/16

2 Political Research Quarterly

Union (Gurr 1993; Hale 2000; Laitin 1991; Laitin andFearon 2003; Roeder 1991). Long-standing demands,typically when followed by the emergence of move-ments engaged in violent resistance, represent sim-mering crises requiring state responses. As the Frenchexample demonstrates, many central governments are

concerned that political decentralization, the devolu-tion of power to regions in which the minority popula-tion resides, may encourage local elites and politiciansto raise demands to set the autonomy bar higher. Onthe other hand, it is also plausible that regionalempowerment may satisfy ethno-political movements,relieving pressure on the state.

The debate and disagreement among French policymakers reflect to some extent concerns about possibletrajectories of ethno-political mobilization and aboveall the fear that such mobilization would spiral out of

control, leading to violence and ending with secession.The central government, under such circumstances,would then be forced to invest significant and poten-tially increasingly ineffective resources to regainstability. From the states point of view, options forrestoring stability, either through violent repression,costly peaceful concessions, or acceptance of thepartition of the state and secession, all lead to poten-tial political and/or economic disempowerment (anddelegitimization; see Brancati 2006 and Lustick,Miodownik, and Eidelson 2004 for reviews of thesepositions).

Unfortunately for policy makers, the large andgrowing literature on ethno-political mobilizationdevotes significant attention to the circumstancesdriving people to raise demands, to assert theirrights for self-determination, and to adopt varioussometimes more and other times less peacefulstrategies to attain their goals, yet it fails to provide aclear evaluation of the effectiveness of politicalarrangements intended to accommodate differingagendas. The absence of scholarly consensus under-mines attempts to develop appropriate responses to

the demands of ethno-political and self-determinationmovements. This perhaps is surprising given the wealthof quantitative and qualitative analyses evaluatingethno-political mobilization, although issues regard-ing the relatively small number of cases, data collec-tion and coding issues, and meaningful comparabilityof cases across the globe, while not unique to studiesof ethno-political mobilization, may explain the fail-ure to reach consensus within the field.

In this article, we develop an agent-based model,based on but not exclusive to the postindustrial het-

erogeneous democracies of Western Europe, that

serves as a rigorous experimental tool for explor-ing and testing expectations drawn from competingtheories about the impact of political decentraliza-tion on ethno-political mobilization to gain furtherinsight into these dynamics. In particular, we seekto problematize the impact of decentralization on the

dynamics of identity ascription and ethno-politicalmobilization that occur at times between putativemembers of an ethnic group. Our analysis shows thatin general, the impact of decentralization on ethno-political mobilization has a nonlinear effect: moderatelevels of decentralization encourage ethno-politicalmobilization, while higher levels decrease it.

Political Decentralization

and Ethno-political Mobilization

The universe of ethno-political movementsthegoals they seek to attain and the strategies they adoptfor that purposeis quite diverse. It includes politi-cal organizations seeking secession and outrightindependence (e.g., Corsicans in France, Basques inSpain), territorial readjustments or the uniting of aregional population with its homeland (e.g., Germansin Alto Adige in Italy, Catholics in Northern Ireland),protection of cultural uniqueness and the allocationof political rights based on such communal affilia-tions (e.g., Danes in Schleswig-Holstein in Germany,Swedes in land in Finland), and political leverageto shield and promote regional economic interests(e.g., Northern Italy, Scotland). Regardless of theparticular goals they strive to achieve, ethno-politicalmovements share two significant elements: (1) theambition to alter the structure of relations betweenthe state and one or a group of substate regions (DeWinter 1998) by negotiating (peacefully or not) forthe transfer of more responsibilities and decision-making power from the center to the area in whichthe group resides and (2) the strategy of mobilizing

public support by (re)evoking and enlisting, amongother methods, local identities as part of their struggleto demarcate a political space (Miodownik andCartrite 2006). In other words, ethno-political move-ments seek and hope to gain public support for theirgoals by appealing to and nurturing group identifica-tion and erecting and politicizing boundaries betweena group residing within a demarcated political spaceand the rest of the state.

State responses to ethno-political mobilizationcover a strikingly broad range, even among such

ostensibly similar states as the advanced democracies


3/16

Miodownik, Cartrite / Political Decentralization 3

of Western Europe. The Belgian state, initially config-ured on the Jacobin unitary model, came under grow-ing pressures in the 1950s and 1960s to establishformal institutions reflecting the increasingly salientlinguistic divide; the failure of these reforms to satisfy,in particular, widening Flemish demands resulted in

the de facto and, with the establishment of the newconstitution in 1992, de jure institutionalization of ahighly decentralized federalism. Finland, establishedin 1919 with a considerable Swedish-speaking minor-ity, established local self-rule for the land Islandswith considerably fewer accommodations extended toSwedish speakers on the mainland. Denmark estab-lished constitutional measures that allowed for thecreation and independence of the Kingdom of Icelandin 1919 and extended home rule to Faroe and Greenlandfollowing World War II. The United Kingdom experi-

mented with varying degrees of devolution, particu-larly for Northern Ireland, with Scotland now havingits own parliament and Wales a dedicated executive.Democratization in Spain established, in the 1977constitution, a two-track process for the devolutionof competencies to the seventeen autonomous com-munities: one for the three historical nationalities(the autonomous Basque, Catalan, and Galician com-munities, and another for the remaining regions);legislation limiting the range of powers the communi-ties may claim were determined to be unconstitutional,and, as a result, the Spanish state continues to asym-metrically devolve powers to the various communi-ties. And in France, where resistance to any institutionalrecognition of ethnic minorities has been a feature ofthe political culture since well prior to the FrenchRevolution, limited reforms allowing for education inlocal languages and some (often crosscutting) institu-tional frameworks have been undertaken. Thus, therange of ethnic groups (including their levels ofpolitical activism and demands) interacts with a rangeof states with varying political cultures and strategies(and tolerance) for institutional accommodation of

ethnic minorities, even in what may be understood asa cluster of relatively similar states.

Ethno-political mobilization requires the satisfac-tion of two conditions, although the literature tendsnot to frame the analysis in this manner (for a com-parable analysis regarding the emergence of ChristianDemocratic parties, see Kalyvas 1996). First, thesocial cleavage separating the minority group fromthe majority must be salient and accepted by putativemembers of the former as dividing the polity intoseparate groups. For minority groups, particularly in

contexts in which the state actively or passively

promotes the homogenization of the population, thecontinued salience of a social cleavage may itselfrepresent an ongoing challenge. While the most obvi-ous examples involve attempts to preserve or revivelocal languages, the reification of a minority identitycannot be taken as unproblematic, and in fact, the

preservation of a local identity may represent one ofthe more prominent goals of ethno-political move-ments, particularly in the early stages of ethno-politicalactivism (Cartrite 2003). Studies that focus on diver-gent linguistic, historic, ethnic, and/or religious back-grounds may be understood as examinations of thefactors that contribute to or inhibit the ongoing reifi-cation of a social cleavage (De Winter 1998; Rokkanand Urwin 1983). Other studies focus on the impactof economic grievance (e.g., Hechter 1975; Nielson1980; Ragin 1979) or the availability of economic

resources (e.g., Giordano 2000; Gourevitch 1979;Harvie 1994) as significant factors related, at least inpart, to the reification and maintenance of minoritydistinctiveness. However, it is equally clear that thereare likely no specific types of social cleavage that areeither necessary or sufficient to drive ethno-politicalmobilization (Fearon and van Houten 2002; vanHouten 2000).

The second necessary condition for ethno-politicalmobilization requires that members of the minoritygroup accept the linkage between the existence of thesocial cleavage and the right to some degree ofpolitical accommodation derived from that groupsdistinctiveness; secondary analyses of this dimensioninvolve the varying levels of support for a range ofaccommodations, from limited governmental supportin reifying the minority identity to the formation ofan independent polity exclusive to the minoritygroup, among members of the minority group. In thisrespect, specialized political institutions representboth a perceived right by some in the group and amechanism for ethnic entrepreneurs to maintain oreven strengthen the satisfaction of these two criteria.

Some scholars have traced the development of politi-cal rights of minority groups broadly (e.g., JacksonPreece 1998; Alcock 2000), but the emergence at thesupranational level of normative arguments and insti-tutions for the protection of minority identities likelyrepresents at most a facilitating condition for the suc-cessful linkage between group distinctiveness andpolitical rights in the minds of individual members ofa minority group.

However, contributions to the literature stronglydisagree on the relationship between political decen-

tralization (e.g., devolution and transfer of power,


4/16


limited self-governance, and federalization) and theemergence, and more importantly the appeasement,of ethno-political contention. One position holds thatpolitical decentralization is related to the emergenceand exacerbation of mobilization (e.g., Bunce 1999;Hale 2000; Kymlicka 1998; Rogowski 1985; Treisman

1997). Devolution of power and the creation ofregional self-governing institutions are seen from thispoint of view as generating opportunities that localpoliticians use to demand the expansion of the author-ity already granted to the regional institutions. In thatcontext, ethno-political movements mobilize publicsupport and pressure the central government to agreeto widen the jurisdiction of regional institutions.

Other scholars argue, however, that political decen-tralization diminishes the likelihood of mobilization(e.g., Gurr 2000; Kaufman 1996; Lijphart 1994;

Stepan 1999; Tsebelis 1990). Holders of this positionconclude that political institutions, which facilitatethe expression of local grievances and are capable ofarticulating immediate responses to collective needs,decrease the motivations of regional and ethnic entre-preneurs to raise the contention bar by increasing thelikelihood of loyalty to the state (Hirschman 1970).Proponents of this position advocate carefullydesigned institutions guaranteeing that (1) significantpolitical representation, (2) participation in policyand decision making, and (3) the transfer of powerand responsibilities to elected regional assemblies andexecutives will appease regional grievances and con-sequently reduce the likelihood of and support forethno-political movements (e.g., Brass 1991; Lijphart1977; McGarry and OLeary 1993; Sartori 1997).

The empirical record of the relationship betweendecentralization and ethno-political mobilization isremarkably murky (Hechter and Okamoto 2001).First, inferences regarding the impact of decentraliza-tion may vary by the regional focus of the researchersor the methodological technique they adopt. Studentsof the former Soviet Union, for example, tend to find

a strong association between decentralization and dif-ferent aspects of mobilization including secession(e.g., Bunce 1999; Roeder 1991; Snyder 2000;Treisman 1997). Others, using large-Ncross-nationalstatistical analyses, tend to assert that decentraliza-tion decreases ethno-political mobilization (Gurr2000; Hechter 2000). Such empirical disagree-ments may be associated at least in part with theconventional practice of focusing on the nationallevel even when the political phenomenon unfoldsbelow the level of the state (Snyder 2001). Indeed,

some scholars suggest that the substate level is the

most appropriate level for gathering data and study-ing phenomena such as ethno-political contentionand other types of civil violence, although data col-lection at that level may be highly problematic (vanHouten 2000; Buhaug and Rd 2006).

Yet even recent studies using meticulously con-

ceived substate-level data sets fail to generate a con-sensus on the impact of decentralization on theorigins and trajectories of ethno-political mobiliza-tion. Brancati (2006), for example, stated that politi-cal decentralization is likely to decrease antiregimemobilization and intercommunal conflict, but shealso concluded that decentralization encourages sup-port for ethno-regional parties, thereby indirectly exac-erbating ethnic conflict and secessionism. Similarly,van Houten (2000) found regional assertiveness to behigher when regional parties control a higher share of

the overall regional vote. Jolly (2005) found (counterto his initial intuition and Brancatis 2006 empiricalfindings) that both the likelihood that a regional partywill enter electoral competition and the extent of sup-port it receives increase at lower levels of decentral-ization up to a certain threshold and decline at highlevels of self-governance. Last, both van Houten andSorens (2005) found no significant associationbetween political decentralization (i.e., the territorialarrangement of the state or regional self-governance)and either regional assertiveness or the support forsecessionist parties.

We propose that the explanation for this lack ofconvergence regarding the impact of devolving polit-ical power on ethno-political mobilization likely stemsfrom two general oversights in the literature. Thefirst, alluded to above, lies in the failure to appreciatethe varying influence of factors, such as devolution,on ethno-political mobilization when affecting thetwo necessary conditions within a heterogeneouspopulation. While we agree with the broader litera-ture that identity formation precedes or coincideswith ethno-political mobilization for any individual,

putative members of the group may vary in terms oftheir acceptance of these two conditions both overtime and across the population. Although there is aclear tendency to attribute to whole groups the politi-cal positions or tactics of a few (e.g., Basques areseparatists or terrorists), this shortcut often beliessignificant disagreement among the putative mem-bers of a group regarding the existence and sociopo-litical significance of particular cultural or ethnicdifferences. In fact, much of the groundwork for ethno-political mobilization, particularly in contexts in

which the state seeks to downplay cultural distinctions


5/16


as politically irrelevant, requires active efforts to per-suade individuals that, indeed, differences betweenthe minority and the majority do exist and that, infact, these differences delineate a group boundary.Thus, part of the difficulty faced by the literatureresults from the assumption that the group is a single

entity, assuming away the potentially significant dif-ferential impacts of causal factors when mediatedthrough a population that may be in disagreementabout either the existence of a social cleavage or thepolitical importance of such a cleavage.

The second potential oversight that might accountfor the continued disagreement in the literature canalso be located below the group level. As a socialphenomenon, ethno-political mobilization likelyentails not merely the processing of (perhaps limited)information by atomized heterogeneous individuals

but may also be a function of the information sharingthat occurs between these heterogeneous individuals.Because the information-sharing patterns (i.e., who issharing information with whom) likely cannot befully specified for any reasonably sized group, theliterature implicitly assumes a simple processing ofinformation that at best merely represents the modalimpact of those factors in a group; however, thisassumption may at best identify only the modal out-come, with some configurations capable of generat-ing unexpected outcomes based on how informationflows through the group. To fully account for theimpact of decentralization on ethno-political mobili-zation, we must therefore take into account the flowof information about the ethnic identity through theputative population and how decentralization mayshift, disrupt, or exacerbate that flow.

Thus, we contend that at least part of the explana-tion of ethno-political mobilization must take intoaccount that the politicization of ethnic minorityidentity requires individuals to both accept the cul-tural cleavage(s) as indicative of group distinctive-ness and link that distinctiveness to political activism.

The meeting of these two conditions requires thatindividuals interact with others to evaluate thesalience of the group identity as well as determine therelative local predominance (or lack thereof) of thatidentity. Individuals embracing a particular identitybut frustrated by the encounter with other identities(particularly those associated with the hegemonicstate) may become politically active to realize somerelief from this subordinate status. While this positiondoes not exclude the importance of cultural and eco-nomic factors, it does suggest that even when those

variables are held constant, there may be significant

(and difficult-to-measure) variation in the onset ofethno-political mobilization. And as we contend, thedegree of control of political institutions by membersof the minority group has a demonstrable impact onboth ethnic identity ascription in a putative popula-tion and its potential politicization.

A Simulation of a Multiregional

and Multicultural State

To take seriously the implications of decentral-ization on ethno-political mobilization in light oftheir potential impacts on the flow of informationand ethnic identity ascription and politicizationbetween putative members of the group, we deploya dynamic agent-based model, a computer-assisted

methodology enabling researchers to design, exper-iment with, and investigate artificial worlds inhab-ited by agents that interact with each other followingsimple prespecified rules derived from well-theorizedand empirically established social mechanisms.Indeed, we offer that given the difficulties of col-lecting empirical data from putative members of anyethnic group, the deployment of computer simula-tion represents the best approach to exploring thepotential significance of intragroup heterogeneity.A growing body of literature has already takenadvantage of this methodology to study phenomenarelated to ethno-politics, including mobilization(e.g., Bhavnani and Backer 2000; Epstein, Steinbruner,and Parker 2001; Lustick and Miodownik 2002),identity diffusion (e.g., Lustick 2000; Rousseau andvan der Veen 2005), and ethnocentrism, ethnic con-flict, and secessionism (e.g., Hammond and Axelrod2006; Cederman 2002; Lustick, Miodownik, andEidelson 2004; Bhavnani and Miodownik 2009).This methodology is a particularly powerful way toexperimentally develop, evaluate, and test theoriesand to enrich understandings of complex social pro-

cesses in that it allows for literally thousands ofcases to be established from which one can drawmeaningful conclusions. Careful design and opera-tional definition of artificial worlds permit schol-ars to use such research laboratories to evaluaterelative contributions of alternative theoreticalexplanations. This way, one can undertake compli-cated and complex thought experiments that arevery difficult to conduct in the real world (for gen-eral introductions, see Axelrod 1997; Cederman2001; Epstein and Axtel 1996; Lustick 2000; Macy

and Willer 2002).


6/16


To conduct the exercises, we developed an agent-based model designed with PS-I (Lustick 2002). Themodel shares some similarities with other simulationsdeveloped to explore related social phenomena,including the clustering and consolidation of opinion(Latan and Nowak 1997), enduring political dis-

agreement (Huckfeldt, Jonson, and Sprague 2004),cultural dissemination (Axelrod 1997), and the emer-gence of collective and shared identities (Lustick2000; Rousseau and van der Veen 2005). The mostimportant dimension that sets apart this simulationfrom its aforementioned predecessors is that it isexplicitly designed as a nonabstract space, one thatresembles, to some extent, a real-world political con-struct. The model takes the form of an ensemble: areplicable simulation of a specific kind of problem(Lustick and Miodownik 2009). The building blocks

of this type of model are not simply algorithmicmechanisms, although they may be included in themodel; rather, they are used to operationalize theo-retical positions with enough empirical corroborationto attract interest. The rules guiding agents interactionsare based explicitly on empirically corroborated theo-ries in ways that are consistent with their premises.Thus, while our model does not presume to be anaccurate representation of any specific empirical con-text and, therefore, is not designed to and is incapableof predicting the future of any actual state, it does cre-ate multiple analytic opportunities to explore therange of possible and likely variation in trajectoriesthat may be more or less conducive to ethno-politicalmobilization, facilitating the probabilistic estimate ofthe impact of decentralization on such phenomena. Inmany respects, this project represents an exercise ingenerative (Epstein 2008) theory building, as we seekto operationalize subgroup dynamics in an attempt toaddress at least some of the contradictory findings inthe extant literature.

Model Description1

The literature exploring the relationship betweendecentralization and ethno-political mobilization hasgenerated contradictory expectations: decentraliza-tion will exacerbate ethno-political mobilization, anddecentralization will ameliorate ethno-political mobi-lization. We contend that part of this divergence is afunction of the inability of these studies to take intoaccount the possibility that political activism is, at best,only partially a linear outcome of resources, interest

aggregation, and/or institutional control. Rather, we

hypothesize that part, and indeed much, of the effectis a function of many individual interactions underthose conditions, as ideas and interests articulatethrough a population, and that some of the variationin the impact of decentralization on ethno-politicalmobilization is attributable to the unknown, and

indeed unknowable, distribution of heterogeneousindividuals in the population.

Our purpose, therefore, in developing our agent-based model (Ethniland) is to attempt to generate thedivergent outcomes in the literature as a function ofvarying both the distribution of individuals in thepopulation and the degree of decentralization toevaluate the apparent divergent expectations of theempirical literature. We simulate a number of popula-tions for each level of decentralization to establishthe propensity, for any given level of decentralization,

for ethno-political mobilization; this allows us tooperationalize our assumption that ethno-politicalmobilization is a probabilistic result at any level ofdecentralization as a result of the a priori unknowableconfiguration of the population. We then compare theprobability for the occurrence (but not magnitude) ofethno-political mobilization across multiple levels ofdecentralization to generate a more robust under-standing of the impact of levels of decentralization.

Our model, in many respects, resembles an ethni-cally heterogeneous democratic society in whichideas and interests are allowed to flow through thepopulation without external constraint. We allow forthe occurrence of socially influential agents indepen-dent of the state apparatus in addition to bureaucratswith official sanction of their position. We includeuniversal information for all agents that in somesense represents governmental and media influencesas additional information for agents in the model.And we allow for the control of the state to be respon-sive to local interests rather than fixed by the politicalcenter. While there may be comparable dynamicsfound in nondemocratic plural societies, the model

here is perhaps most readily understood in thoseterms, perhaps best exemplified by multiethnic WesternEuropean states.

Ethniland represents a powerful model to explorethe impact of decentralization on ethno-politicalmobilization. Furthermore, careful experimentaldesign allows us to explore a range of highly detailedstylized cases in a more rigorous way than even verycareful small-N studies allow. Finally, computersimulation experimental design allows for experi-mental replication by later researchers. As a result, a

thoroughly detailed description of the model and


7/16


experimental design involve a degree of technicalspecificity that many readers may find cumbersome.With this in mind, we leave most of the finer techni-cal points to the technical appendix,2 which itself isdetailed enough to facilitate experimental replication.Here we address the basic workings of the model and

the overall experimental design with an eye towardreadability and brevity, providing, we hope, enoughdetails for the reader to follow the subsequent analy-sis of the experimental results.

Basic Agents in Ethniland

Recall that our hypothesis regarding the divergentfindings of the decentralization literature is that suchstudies fail to take into account the impact that infor-mation sharing can have in a group and that remark-

ably similar conditions potentially can generate quitedistinct outcomes purely as a function of who is talk-ing to whom. Our expectation is that the control ofpolitical institutions affects the salience of ethnicidentification and, therefore, the potential for mobili-zation and the probability for the politicization ofidentities. To test such a supposition, our modelrequires that we have (1) agents with multiple identi-ties and opinions, such that they can change theirposition; (2) the capability of agents to receive exter-nal information about those identities, which theninforms their behavior, as well as transmit informa-tion regarding their identification to other agents, in aprocess that iterates over time; (3) the capability ofagents to politicize their identities; and (4) the opera-tionalization of political institutions in the landscapewith the capacity to have those institutions representdifferent ethnic identities or interests. Howeverandthis cannot be stressed too stronglythe model wedevelop must not produce institutional co-optation,ethnic identification, or political mobilization as asimple output of particular inputs; rather, we need amodel in which identification and politicization of

identities are possible, but not inevitable. Absent thiscritical element, we would simply have a computer-ized stylization of preexisting linear expectations.

Ethniland is operationalized as a square lattice gridconsisting of 65 65 (4,225) agents enclosed withina boundary and divided into four quadrants (north-west, northeast, southwest, southeast). In theEthniland landscape there are twenty identities, someof which are given to agents as ethnic identities tiedto a specific quadrant of the landscape; the otheridentities in the landscape represent placeholders to

allow for the emergence of nonethnic political issues

in the landscape rather than having ethnic identityrepresent the only basis for identification and informa-tion sharing. The most commonly occurring agenttype in the landscape (basic) will have, on average,five of the twenty identities as part of its personalrepertoire, one of which will be active and available as

information to other agents. With the exception of theethnic identities, which are seeded in their particulargeographic quadrants, identities are initially randomlydistributed to all basic agents in the landscape.

To allow for the ebb and flow of identities in thelandscape and, therefore, the possibility that someidentities may become more or less prominent andsalient as well as the potential for political alienationand politicization of identities, agents make simpledecisions regarding their identities, seeking, ceterisparibus, to internalize those identities deemed most

popular and to publicly evidence the single most popu-lar identity. To make this determination, an agent willdetermine a weight for each identity based on twosources of information: its Moore neighborhood(itself and its eight adjacent neighbors, unless theagent is on the edge of the landscape) and global biasinformationinformation about the relative attractive-ness of each identity, which is the same for all agents.3

To allow for the disproportionate influence ofpolitical institutions, agents in the landscape have aninfluence level that represents their relative impact inthe identity weight determination of each agent.Basic agents have an influence level of 1, so an agentmaking an identity weight calculation will simplycount how many basic agents in its neighborhood areactivated on each of the twenty identities;4 thus, forexample, an agent activated on identity i and havingonly two basic agent neighbors also activated on iwill calculate an identity weight for i of 3. To thiscalculation will then be added the global bias value;extending this example, if i has a bias value of 1, thefinal weight would be 2. Each agent makes this deter-mination for each identity on every time step in the

simulation.

Agent Behavior

After determining the identity weights, agentshave the ability to adapt in light of this information.Perfectly adaptable agents would be able to changeall identities as determined by the identity weights,generating a landscape that would quickly homoge-nize and not reflect the social phenomena we seek tosimulate. Therefore, and drawing on social psycho-

logical and constructivist identity theories, each


8/16


agent has a threshold that must be exceeded beforechange occurs.5 Underpinning the behavioral rules isthe notion that identities held either individually bypeople or collectively by groups or states are notfixed but rather are potentially open for evaluationand change or may solidify over time.6

Agents in Ethniland can (1) rotate their identity,deactivating on their current identity and activatingon a different identity (a rotate candidate) alreadywithin their repertoire; (2) substitute an identity cur-rently in their repertoire with a preferable one not intheir repertoire (while remaining activated on theircurrent identity); (3) substitute and rotate an identity,which involves discarding the identity with the low-est weight in the repertoire and obtaining a new iden-tity and activating on that identity; or (4) keep theirrepertoire and activation the same.

Rotation, or the swapping of one internalized iden-tity for the public display of another, represents arelatively painless process, requiring merely that anagent shift its public persona as the situation war-rants. Substitution represents a more difficult pros-pect, as an agent has to discard a held identity toacquire its replacement. And rotation and substitutionrepresents the most difficult adaptive move an agentcan make. The introduction of these thresholds,which inhibit but do not preclude agent adaptation,has the effect of allowing for the ongoing ebb andflow of identities in the landscape, although quiteexpectedly, clusters of homogeneity can appear.

Political Institutions and Political

Entrepreneurs

To approximate ethno-political mobilizationbroadly and the impact of decentralization in particu-lar, we modify some basic agents to represent politi-cal entrepreneurs and the state bureaucracy; however,basic agents represent more than 90 percent of allagents in the landscape. These specialized agents

include regime bureaucrats, which are more influen-tial than basic agents (with influence levels rangingfrom 2 to 4) and whose repertoires include identitiesassociated with the state and the northwest region;decentralized bureaucrats, which are similar in mostrespects to regime bureaucrats except that they con-tain within their repertoires the identity associatedwith the region in which they are located (exclusiveto the northeast and southwest regions); and regionalentrepreneurs, which have an influence similar toregime bureaucrats and decentralized bureaucrats but

are not associated with the regime and contain thesoutheast regional identity, the only region where

they are located. Given the higher influence levels ofthe specialized agents, these both tend to be slightlymore resistant to change (given that they weightthemselves more heavily than do basic agents) andtend to act as local anchors of whatever identity theyare activated on as a result of their larger influence.

However, as with basic agents, these specializedagents are allowed to adapt to the information theyreceive, and their influence is never so large as todetermine the identity of the local neighborhood.

Decentralization

The goal of this experiment is to evaluate theimpact of decentralization on, specifically, ethno-political mobilization. In Ethniland, decentralizationis operationalized by increasing opportunities forsemiautonomous self-governance by the regionallydominant group. Specifically, self-governance issimulated by transforming a proportion of the regionalregime bureaucracy into a decentralized bureaucracyby adding the regionally dominant identity in the south-east region into the repertoire of some of the agents inthe bureaucracy and activating it. Given the smallnumber of regime bureaucrats in the landscape andgiven that such agents retain the state and northwestidentities in their repertoires, this represents in ourview a very modest experimental manipulation.

Politicization of Identity

Finally, to operationalize ethno-political mobiliza-tion in Ethniland, we enable the possibility of anagent transforming into a political boundary; that is,an agent can become a border cell, unchanging,ignored by its neighbors, and representing an institu-tionalized boundary between agents differentiatedalong identity lines.

The mechanisms that drive the ethnic politicizationof identities (i.e., the emergence of political bound-

aries) are modeled as similar to those that result inthe appearance of other forms of social boundaries:alienation, nontrivial size, and antagonistic relationswith others. In general, the production of boundar-ies may be the result of processes that take place atthe state, group, and/or individual level. Tilly (1998,10), in his account of the production of inequal-ity, suggested that two mechanisms, exploitationand opportunity hoarding, drive the appearanceof social boundaries. According to this account, theemergence of boundaries is a function of the alien-

ation and exclusion of regions and populations fromenjoying their share in the goods provided by the


9/16


state. However, Tilly stressed, not only do boundar-ies emerge externally (i.e., as a result of a groupsbeing bounded by the center), but they are producedby internal processes. In particular, boundaries aredriven by a sustainable ability to support and enhancenetworks of activities among the members of the

bounded group. From an individual perspective, theemergence of boundaries is related to psychologicalprocesses of self-identification and categorization thatcrystallize in the entrenchments of institutionalizedexpressions of in-group and out-group affiliations(Lamont and Molnr 2002).

The operationalization of these conditions followsclosely the one offered by Lustick, Miodownik, andEidelson (2004) in a study of the effects of powersharing and repression on the magnitude of seces-sionist mobilization and secessionism. Unlike this

above-mentioned study, here we seek to expand theanalysis and devote more explicit attention to therelationship between the institutional representationof minorities and the unfolding of mobilization andthe emergence of demands for further accommoda-tion, regardless of the substance of these demands,rather than focusing on the effects that repression oraccommodation may have on the likelihood that asecessionist region/movement will break away fromthe state. In other words, while the earlier study wasinterested in the effects of repression and accommo-dation on the trajectory of a secessionist movementonce such a movement emerged, here we are explor-ing the likelihood that patterns of institutional accom-modation may encourage or inhibit the emergence ofsuch movements.

Functionally, an agent politicizes its identity (i.e.,transforms into a boundary) when it finds its acti-vated identity to be subordinate (not the most widelysubscribed, dominant, identity in the landscape),salient (at least 10 percent of all agents are activatedon the identity), and oppositional (no more than 20percent of agents activated on the identity also have

the dominant identity in their repertoires). For anygiven time step in which an agent meets these condi-tions, there is a varying 0.1 to 0.3 likelihood that itwill become a border agent and, as a result, no longeradapt; nor is the agent considered by its neighbors inthe identity weight calculation.

Conclusion

In a model like the one described above, with mostidentities randomly seeded and identity activation ran-

domized, we expect to see an initial patchwork ofmany identities across the landscape. Quite quickly,

however, identity clusters appear as a result of thereinforcement of local information and subsequentagent adaptation. Change will persist in the model inresponse to changes in the identity biases, but theimpact of changing biases typically has only a limitedimpact once pockets of homogeneity have emerged.

We expect that the experimental treatment will alterthe flow of information regarding different identities,given the initial activation of some nonbasic agents onthe minority identity. But which identities come topredominate, where the clusters of homogenizationoccur, and if and where the politicization of identitiesemerges cannot be determined (easily) a priori, evenonce the initial landscape is set, given the large numberof agents and the calculations they undertake.

Experimental Design

The Ethniland model represents an attempt to gen-erate robust findings regarding the impact of decen-tralization on the likelihood of the emergence ofsecessionist sentiment. One option would have beento develop a single Ethniland landscape in which wesubsequently varied the level of decentralizationwhile holding all else consistent, maximizing experi-mental control;7 in addition, we could have attemptedto operationalize the institutionalization of a specificpolitical context, testing the effects of decentraliza-

tion on a specific case. In many respects, this approachwould be consistent with other examples of thedeployment of agent-based models as experiments.However, given our understanding of the overlookedimpact of intragroup communication, when com-bined with the unknown (and indeed unknowable)composition of any group prior to a particular politi-cal event (such as the appearance of secessionistsentiment), we determined that relying on any onebaseline landscape for the experiment representedreal risk (and would undermine our ability to test

theories developed from a broad range of cases).Indeed, there is an unknowable likelihood, given howcomplicated the model is, that any one landscapemay unintentionally generate unusual results throughthe interplay of nonbasic agent locations, randomizedinitial repertoire sizes and compositions, bias changes,and other dynamic elements.

However, the design space of possible landscapes,given these numerous elements and the number ofvariations on each one, is far too large to be exploredcomprehensively; the randomization of nonbasic

agent locations in the landscape alone representsmany thousands of possible configurations. In an


10/16


11/16


identity. However, notice that although a larger share ofthe bureaucracy actively mobilizes support for theregion, such identification is not followed by theemergence of separatist elements.

By the end of the experimental run (t = 500), we cansee three distinct outcomes. The No Decentralization

run ends with a significant number of supporters ofthe regional identity. In the Weak Decentralizationscenario, support for the regional identity is quitesubstantial as well. Ethno-political boundaries sepa-rate small enclaves of more stubborn isolationism aswell as areas with more social tension, but by andlarge, the region seems well integrated within thestate. Last, the Strong Decentralization run results ina widespread identification with the regional identitywithout the emergence of ethno-political contention(i.e., no boundary agents). Put differently, the massive

regional identification depicted in the third scenariodoes not seem problematic inasmuch as the identifi-cation with the state and the integration of the regionwith it are concerned.

Statistical Analysis

The statistical analysis of the experiment evaluatesthe impact of decentralization on the occurrence ofpoliticization of a regional identity, in other words, onthe emergence of ethno-political boundaries. We donot explore the relationship between decentralizationand the intensity of contention (the number of bordercells), as the extant literature suggests that there maybe other variables at work than those we have opera-tionalized; furthermore, to evaluate the ebb and flowof support for ethno-political mobilization, we wouldneed to add to the model the possibility of borderagents rejoining the political landscape and of ethno-political mobilization itself, as a dynamic, to spreadthrough the landscape, significantly complicating themodel. We statistically test the experimental resultsusing two logistical models and a marginal effects

analysis, demonstrating that the observed curvilinearrelationship between decentralization and identitypoliticization is statistically robust.

Figure 2 plots the relationship between the level ofdecentralization (the percentage share of the statebureaucracy at t = 0 activated on the regional iden-tity) and the percentage of runs in which boundariesappear. The relationship appears curvilinear. Weakdecentralization (up to 20 percent) did not signifi-cantly affect the occurrence of ethno-political bound-aries. However, stronger decentralization (20 to 40

percent) effected a sharp increase in the percentage ofruns in which political boundaries appeared. In fact,the percentage of iterations with ethno-politicalboundaries was, on average, larger than the NoDecentralization scenario even when 60 percent ofthe regional bureaucracy was set to represent regionalidentities and interest. Interesting to note, the likeli-hood of boundaries begins to fall below the NoDecentralization condition as decentralizationincreases beyond 60 percent, F test combined =25.58,p (linear) < .0001,p (nonlinear) < .0001.

We then tested this pattern using a logistic regres-sion; the results are shown in table 1. Model 1includes a dummy variable, Decentralized Cat, that iscoded 0 for 5,000 landscapes set to iterate withoutdecentralization and 1 for 5,000 with various levelsof decentralization in the southeast region. Model 1controls for other parameters included in the simula-tion: cultural differences, the extent of minority sup-

port at the beginning of each run, global biasesavailable about the regional identity, the length ofhistory (time), and the probability of ethno-politicalboundary appearance (Boundaries Probability). Asindicated by the statistically significant coefficient ofthe decentralization dummy variable in model 1,experimental iterations with minority representationas part of the regional bureaucratic structure were morelikely to result in the emergence of ethno-politicalboundaries or, in other words, in the articulation ofthe region as a distinct political space. According to

Figure 2

Percentage of Runs with Emerging

Borders by Decentralization


12/16


13/16


marginal effects allows us to consider the possibilitythat political decentralization of power may have dif-ferent effects across various conditions. For instance,compare the impact of decentralization under condi-tions of weak minority support, cultural differences,and global cues (the bottom line in figure 4) to its

impact in iterations that began with a large number ofagents supporting the regional identity, strong cul-tural differences between the agents in the southeastregion and the rest of the state, and strong global indi-cations supporting identification with the regionalmovementin other words, all three conditions thatincrease support for mobilization (top line in figure 4).Decentralization of political power seems to have astronger impact under the latter compared to the for-mer experimental setting.

Explanation

The various statistical analyses conducted on thedata generated by the computer simulation are clear:a significant, curvilinear relationship exists betweenthe level of minority control of political institutions(decentralization) and the appearance of boundaries.Understanding how the level of decentralization pro-duces this curvilinear relationship requires a focus onthe micro-mechanisms at work in the model. In otherwords, explaining the causal linkage between decen-tralization and ethno-political contention requires afocus on agent decision making.

Recall that agents change identity activation andidentity subscription in response to limited informa-tion weighing the relative local prevalence (activa-tion) and global attractiveness of each identity (bias).An agent will change its activated identity when onein its repertoire has an identity weight greater thanthe current activated identity plus the rotation thresh-old, or an agent will discard one identity in its reper-toire and subscribe to a new identity when the identityweight on the new identity exceeds the identity weight

of the discarded identity plus the greater threshold forsubstitution. In this way, each agent adapts to achanging landscape by subscribing to and activatingon those identities with the highest aggregate weighton any updating time step.

A basic agent becomes a border agent when fourconditions are met. Its activated identity must be,first, one that is not the identity with the most acti-vated agents in the landscape as a whole but, second,one with at least 10 percent of all agents activated onit; in other words, the identity must be subordinate

but not trivial. Third, the agent must have at least

three agents from whom it receives information acti-vated on other identities. And finally, a probability oftransformation to a border agent must be satisfied.Perfectly adaptive agents (agents without thresholdsthat must be satisfied before change occurs) will con-tinue to change their activation or substitute identities

in their repertoire until they reach some equilibriumwith their local information network in which theyare activated on the identity with the highest weightand have in their repertoire the other most heavilyweighted identities, generating a landscape with iden-tity clusters but relative homogeneity across reperto-ries. However, the introduction of thresholds willinhibit this homogenization to some degree, as agentsmay be activated on or subscribed to identities that,while not having the highest identity weight, none-theless are not so suboptimal as to be discarded.

Institutions (bureaucrats) serve as identity anchorsin the model in that the specialized agents represent-ing political institutions carry a higher influence thanbasic agents; thus, their activated identity will have ahigher weight for all adjacent agents than is other-wise the case, causing those adjacent agents, all otherthings being equal, to tend to subscribe to and acti-vate on the identity of the specialized agent. Theexperiment shows that when a large proportion ofthe specialized agents are activated on the subordi-nate identity, large islands of the identity form andmerge, preventing the isolation of most agents. Atlow levels of decentralization, agents are more ableto adapt to their environment given the paucity ofthese subordinate identity anchors. However, itappears to be the case that at moderate levels ofdecentralization, enough anchors exist to create andsustain numerous clusters of the subordinate identityyet not enough to facilitate the merging of those clus-ters into one large group. Put differently, there is moreopportunity for an agent to feel isolated on a subordi-nate identity (have at least three of its neighbors be ona different identity) from which it cannot change than

at either low or high levels of decentralization.Recall that we argue two conditions must be met for

ethno-political mobilization to occur: the social cleav-age must be salient, and the linkage to political objec-tives must be seen as legitimate. The translation of theeffect of decentralization in the model to real-worldethno-political mobilization is perhaps best under-stood in terms of how salient, or accepted, a particularsocial cleavage is. Low levels of decentralizationallow for other cleavages to penetrate the putativeethnic group, facilitating the shifting of public dis-

course to other bases. High levels of decentralization


14/16


raise and reinforce the salience of the social cleavageto such a degree that most putative members of agroup identity themselves in those terms, so much sothat encounters with other identities are limited.However, at moderate levels of decentralization,enough putative members of a group identify them-

selves with that group to preserve the identity withinthe larger social landscape, but the salience of thatcleavage is not so high as to attract all putative mem-bers, increasing the likelihood of disagreement aboutthe significance of the cleavage and possible isolationof individuals identifying themselves in terms of theminority identity. It is from these isolated individualsthat ethno-political contention emerges.

Conclusion

This article was motivated by the absence of aclear and consistent theoretical foundation appropri-ate for future policy guidance and the lack of mecha-nisms for mitigating policy maker concerns regardingthe rise of ethno-political contention. The literature,we argued, devotes significant attention to the cir-cumstances driving regions, groups, and people toorganize and demand political rights collectively butfails to reach a consensus on the political arrange-ments intended to accommodate such demands. Forevery policy recommendation, it seems, there arestudies showing that it either would work well orwould work terribly.

The article presents the results of a simulatedexperiment that attempted to address opposite expec-tations regarding the effectiveness of political decen-tralization in response to ethno-political mobilization.An agent-based model of a multiregional and multi-cultural state was run thousands of times to explorethe impact of decentralization on the emergence ofpolitical boundaries. These experiments suggest thatthe impact of decentralization is not linear: weak and

medium levels of decentralization increase the likeli-hood of boundary appearance, while strong decentral-ization decreases it. We argue that the explanation ofthis relationship lies below the group level and is afunction of how minority control of political institu-tions affects the dynamic of minority identity ascrip-tion and the realization or frustration of seeking morecomplete dominance of the regional ideational space.

Clearly, politicized ethnic groups remain active, andmany continue to press for decentralization to varyingdegrees: additional decentralization in Catalonia in

2007 and the upcoming Welsh referendum on the

establishment of a parliament similar to that in Scotlandare just two among many examples in Western Europe,with literally dozens of examples to be found aroundthe globe. And national politicians of varying ideo-logical stripes continue to debate the wisdom of decen-tralization. Both scholars and policy makers continue

to confront an increasing range of cases, and advanc-ing the understanding of the interrelationship betweeninstitutional design and ethno-political activism remainsa vitally salient need.

The research presented in this article indicates thatthe risks of transferring power and responsibility tolower local or regional levels of government may beoverstated, although half-hearted reforms may in factproduce the most extreme ethno-political mobiliza-tion not as a function of the realization of politicaldemands (or lack thereof) but in the facilitation of

increasing minority identity salience while simulta-neously increasing the probability of identity con-frontation and frustration. Entrusting regional eliteswith more responsibilities, under most circumstances,seems to be the best and more appropriate way toreconcile/mitigate/regulate regional demands forautonomy. Policy makers may find that accommoda-tion, rather than suppression or control, of the diverseinterests of regions and groups will protect the integ-rity of the state, the political institution entrusted withthe responsibility of promoting the welfare of the pol-ity as a whole.

Perhaps most importantly, this study highlightsthe need to move beyond linear models in under-standing ethno-political mobilization broadly and theimplications of any specific policy. Ethnic groupscomprise heterogeneous individuals, and identifica-tion with a particular group is at least partially afunction of intragroup information sharing, particu-larly in contexts wherein identity ascription is (par-tially) voluntary and where an alternative (i.e.,national) identity is encouraged by the state. Giventhese dynamics, any particular policy must be aimed

not at an inevitable outcome but at shifting the prob-abilities of particular results, given that the composi-tion of any group and the impact of who is speakingto whom on the flow of information cannot be apriori fully specified. As a result, despite the bestefforts of policy makers, sometimes the unlikely out-come will obtain. For scholars to take this seriouslyrequires a reconceptualization of ethnic identity andethno-political mobilization that takes into accountprocesses below the level of the group and the impli-cations of imperfect information on the findings of

empirical research.


15/16


Notes

1. For a full description of the model and experimental algo-rithms, see the supplemental materials available in the onlineversion of the article.

2. The appendix is available in the online version of this arti-cle at http://prq.sagepub.com/supplemental/.

3. Identity biases can change on any time step, although theprobability is small (0.5 to 1.0 percent); this introduces limiteddynamism to the model, allowing for ongoing change over time.Bias values fall in the range from 3 to 3; as a result, even atthe extremes bias, values do not outweigh local neighborhoodinformation.

4. Note that agents do not look inside the repertoires of otheragents; they simply count activated identities.

5. In Ethniland, the values of the various rotation triggerswere set to rotation trigger= 2, rotation and substitution = 7, andsubstitution = 5. These specific trigger values are typically usedin the published literature with PS-I and can be justified by socialpsychological arguments.

6. The rules are not designed as an operational definition ofany one theory of social psychology or of constructivist identitytheory but are consistent with theories in social psychology andrelated fields that explicitly attempt to define and to describeconditions affecting the formation and change of attitudes, opin-ions, and identities (e.g., Asch 1956; Latan 1981; Tajfel andTurner 1986; Turner et al. 1987).

7. Note that the use of seeds for the various randomized ele-ments enables us to re-create exact versions of a single landscape,including the random numbers generated for updating biases anddetermining secession during the model run.

References

Alcock, Anthony E. 2000.A history of the protection of regionalcultural minorities in Europe: From the Edict of Nantes to the

present day.New York: St. Martins.Asch, Solomon E. 1956. Studies of independence and confor-

mity: A minority of one against a unanimous majority.Psychological Monographs 70:1-70.

Axelrod, Robert M. 1997. The complexity of cooperation: Agent-based models of competition and collaboration. Princeton,NJ: Princeton University Press.

Bhavnani, Ravi, and David Backer. 2000. Localized ethnic con-flict and genocide: Accounting for differences in Rwanda andBurundi.Journal of Conflict Resolution 44:283-306.

Bhavnani, Ravi, and Dan Miodownik. 2009. Ethnic polarization,ethnic salience, and civil war.Journal of Conflict Resolution53:30-49.

Brancati, Dawn. 2006. Decentralization: Fueling the fire ordampening the flames of ethnic conflict and secessionism?

International Organization 60:651-85.Brass, Paul R. 1991.Ethnicity and nationalism: Theory and com-

parison.Newbury Park, CA: Sage.Buhang, Halvard, and Jan Ketil Rd. 2006. Local determi-

nants of African civil wars, 1970-2001. Political Geography25:315-35.

Bunce, Valerie. 1999. Subversive institutions: The design and thedestruction of socialism and the state. New York: Cambridge

University Press.

Cartrite, Britt. 2003. Reclaiming their shadow: Ethnopoliticalmobilization in consolidated democracies. PhD diss.,University of Colorado at Boulder.

Cederman, Lars-Erik. 2001. Agent based modeling in politicalscience.Political Methodologist10 (1): 16-22.

. 2002. Endogenizing geopolitical boundaries with agent-based modeling. Proceedings of the National Academy of

Science U.S.A. 99 (3): 7296-303.De Winter, Lieven. 1998. The Volksunie and the dilemmabetween policy success and electoral survival in Flanders. InRegionalist parties in Western Europe, ed. Lieven De Winterand Huri Trsan, 28-50. London: Routledge.

Deutsche-Welle, Current Affairs, 7 July 2003. http://www.dw-world.de/dw/article/0,,913546,00.html

Epstein, Joshua M. 2008. Why model? Journal of ArtificialSocieties and Social Simulation 11 (4). http://jasss.soc.surrey.ac.uk/11/4/12.html (accessed April 11, 2009).

Epstein, Joshua M., and Robert Axtel. 1996. Growing artificialsocieties: Social science from the bottom up. Washington,DC: Brookings Institution.

Epstein, Joshua M., John D. Steinbruner, and Miles T. Parker.2001. Modeling civil violence: An agent-based computationalapproach. Center on Social and Economic Dynamics WorkingPaper No. 20, Brookings Institution, Washington, DC.

Fearon, James, and David Laitin. 2003. Ethnicity, insurgency, andcivil war. American Political Science Review 97 (1): 75-90.

Fearon, James D., and Pieter van Houten. 2002. The politicizationof cultural and economic differences: A return to the theory ofregional autonomy movements. Paper presented at the fifthmeeting of the Laboratory in Comparative Ethnic Processes,Stanford University, CA, May 10-11.

Giordano, Benito. 2000. Italian regionalism or PadanianregionalismThe political project of the Lega Nord in Italianpolitics.Political Geography 19:445-71.

Gourevitch, Peter A. 1979. The re-emergence of peripheralnationalisms: Some comparative speculations on the spatialdistribution of political leadership and economic growth.Comparative Studies in Society and History 21:303-22.

Gurr, Ted R. 1993. Minorities at risk: A global view of ethnopo-litical conflicts. Washington, DC: U.S. Institute of PeacePress.

. 2000. Peoples versus states: Minorities at risk in thenew century. Washington, DC: U.S. Institute of Peace Press.

Hale, Henry E. 2000. The parade of sovereignties: Testing theo-ries of secession in the Soviet setting. British Journal of

Political Science 30 (1): 31-56.Hammond, Ross A., and Robert Axelrod. 2006. The evolution of

ethnocentrism.Journal of Conflict Resolution 50:926-36.Harvie, Christopher. 1994. The rise of regional Europe. London:

Routledge.Hechter, Michael. 1975. Internal colonialism: The Celtic fringe

in British national development 1536-1966. Berkeley:University of California Press.

. 2000. Containing nationalism. New York: OxfordUniversity Press.

Hechter, Michael, and Dina Okamoto. 2001. Political conse-quences of minority group formation. Annual Review of

Political Science 4:189-215.Hirschmann, Albert O. 1970. Exit voice and loyalty: Responses

to decline in firms, organizations, and states. Cambridge,

MA: Harvard University Press.


16/16


Horowitz, Donald L. 1985. Ethnic groups in conflict. Berkeley:University of California Press.

Huckfeldt, Robert, Paul E. Johnson, and John Sprague. 2004.Political disagreement. New York: Cambridge University Press.

Jackson Preece, Jennifer. 1998. National minorities and theEuropean nation-states system. Oxford, UK: Clarendon.

Jolly, Seth K. 2005. Another reason to party? European integration

and regional political parties. Paper presented at the annualmeeting of the Midwest Political Science Association, Chicago,April 6-10.

Kalyvas, Stathis N. 1996. The rise of Christian democracy inEurope. Ithaca, NY: Cornell University Press.

Kaufman, Chaim D. 1996. Possible and impossible solutions toethnic civil wars.International Security 20 (4): 133-75.

Keating, Michael. 1998. The new regionalism in Western Europe:Territorial restructuring and political change. Cheltanham,UK: Edward Elgar.

Kymlicka, Will. 1998. Is federalism a viable alternative to seces-sionism? In Theories of secessionism, ed. Percy B. Lehning,111-50. New York: Routledge.

Laitin, David D. 1991. The national uprising in the Soviet Union.World Politics 44 (1): 139-77.

Lamont, Michle, and Virg Molnr. 2002. The study of boundar-ies in the social sciences. Annual Review of Sociology 28:167-95.

Latan, Bibb. 1981. The psychology of social impact.AmericanPsychologist36:343-56.

Latan, Bibb, and Andrzej Nowak. 1997. Self-organizing socialsystems: Necessary and sufficient conditions for the emergenceof clustering, consolidation and continuing diversity. InProgressin communication sciences. Vol. 13,Persuasion, ed. Franklin J.Boster and George Barnett, 43-74. Norwood, NJ: Ablex.

Lijphart, Arend. 1977. Democracy in plural societies: A com-parative exploration.New Haven, CT: Yale University Press.

. 1994. Electoral systems and party systems: A study oftwenty-seven democracies, 1945-1990. New York: OxfordUniversity Press.

Lustick, Ian S. 2000. Agent-based modeling of collective iden-tity: Testing constructivist theory. Journal of ArtificialSocieties and Social Simulations 3 (1). http://jasss.soc.surrey.ac.uk/3/1/1.html (accessed April 11, 2009).

. 2002. PS-I: A user-friendly agent-based modeling plat-form for testing theories of political identity and politicalstability.Journal of Artificial Societies and Social Simulations5 (3). http://jasss.soc.surrey.ac.uk/5/3/7.html (accessed April11, 2009).

Lustick, Ian S., and Dan Miodownik. 2002. The institutionaliza-

tion of identity: Micro adaptation, macro effects, and collec-tive consequences. Studies in Comparative and International

Development37 (2): 24-53.. 2009. Abstractions, ensembles, and virtualizations:

Simplicity and complexity in agent-based modeling.ComparativePolitics 41:223-44.

Lustick, Ian S., Dan Miodownik, and R. Eidelson. 2004.Secessionism in multicultural states: Does power sharingprevent or encourage it? American Political Science Review98:209-29.

Macy, Michael W., and Robert Willer. 2002. From factors toactors: Computational sociology and agent-based modeling.

Annual Review of Sociology 28:143-66.

McGarry, John, and Brendan OLeary. 1993. The macro-politicalregulation of ethnic conflict. In The politics of ethnic conflictregulation, ed. John McGarry and Brendan OLeary, 1-40.London: Routledge.

Miodownik, Dan, and Britt Cartrite. 2006. Demarking politicalspace: Territoriality and the ethnoregional party family.

Nationalism & Ethnic Politics 12 (1): 53-82.

Newman, Saul. 1996. Ethnoregional conflict in democracies:Mostly ballots, rarely bullets. Westport, CT: Greenwood.Nielson, Francois. 1980. The Flemish movement in Belgium

after world war: A dynamic analysis.American SociologicalReview 45:76-94.

Ragin, Charles C. 1979. Ethnic political mobilization: The Welshcase.American Sociological Review 44:619-35.

Roeder, Philip G. 1991. Soviet federalism and ethnic mobiliza-tion. World Politics 43 (2): 196-232.

Rogowski, Ronald. 1985. Causes and varieties of nationalism. InNew nationalism of the developed West, ed. Edward A.Tiryakian and Ronald Rogowski, 87-108. Boston: Allen andUnwin.

Rokkan, Stein, and Derek W. Urwin. 1983. Economy, territory,identity: Politics of West European peripheries. London: Sage.

Rousseau, David L., and Maurits van der Veen. 2005. The emer-gence of shared identity: An agent-based model of idea diffu-sion.Journal of Conflict Resolution 49:686-712.

Sartori, Giovanni. 1997. Comparative constitutional engineer-ing: An inquiry into structures, incentives, and outcomes.

New York: New York University Press.Snyder, Jack. 2000. From voting to violence: Democratization

and nationalist conflict.New York: Norton.Snyder, Richard. 2001. Scaling down: The subnational compara-

tive method. Studies in Comparative InternationalDevelopment36 (1): 93-110.

Sorens, Jason. 2005. The cross-sectional determinants of seces-

sionism in advanced democracies. Comparative PoliticalStudies 38:304-26.

Stepan, Alfred. 1999. Federalism and democracy: Beyond the USmodel.Journal of Democracy 10 (4): 19-34.

Tajfel, Henri, and John C. Turner. 1986. The social identity the-ory of intergroup behavior. In The psychology of intergrouprelations, ed. Stephen Worchel and William G. Austin, 7-24.Chicago: Nelson-Hall.

Tilly, Charles. 1998.Durable inequality. Berkeley: University ofCalifornia Press.

Tiryakian, Edward A., and Ronald Rogowski, eds. 1985. Newnationalisms of the developed West. Winchester, UK: Allenand Unwin.

Treisman, Daniel S. 1997. Russias ethnic revival: The separat-ist activism of regional leaders in a postcommunist order.World Politics 49 (2): 212-49.

Tsebelis, George. 1990. Elite interaction and constitution build-ing in consociational societies.Journal of Theoretical Politics2 (1): 5-29.

Turner, John C., Michael A. Hogg, Penelope J. Oakes, Stephen D.Reicher, and Margaret S. Wetherell. 1987. Rediscovering the

social group: A self-categorization theory. New York: BasilBlackwell.

van Houten, P. 2000. Regional assertiveness in Western Europe:Political constraints and the role of party competition. PhDdiss., University of Chicago.

Decentralization(MioCar2009PRQ)

Documents

Transcript of Decentralization(MioCar2009PRQ)