Do mutually reinforcing cleavages harm democracy? Inequalities between ethnic groups and autocratization

ABSTRACTDo mutually reinforcing cleavages harm democracy? Evidence from specific cases suggests that autocratization can be related to the predicament of ethnic groups, if ethnicity is politicized and involves resource distribution. However, we know little about whether this is a cause of autocratization more broadly. The article demonstrates that, with increasing inequalities between ethnic groups, a country experiences a decline in its level of democracy and higher propensity to start autocratizing. The analysis thus advances previous contributions, focusing on individual inequalities and power-sharing institutions as explanations of democratization or democratic quality, in two ways. First, isolating autocratization as downturns in democracy levels and the onsets of related timespans (autocratization episodes), and comparing the impact of (economic, political, and social) types of inequalities between ethnic groups. Second, adopting a global sample of (democratic and non-democratic) countries since 1981, with an original data collection integrating expert surveys with survey data. Quantitative evidence confirms most expectations, particularly on economic inequalities between ethnic groups, and – although less precisely – economic, political and social dimensions combined. The findings have important implications for political regime and ethnic studies, showing that preventing the mutual reinforcement of sociocultural and economic cleavages is key to stabilize democracy.KEYWORDS: Autocratizationethnicityinequalitydemocracyethnic groups AcknowledgmentsI thank Luca Tomini, Jean-Benoît Pilet and Daniel Bochsler, for their precious feedback on the paper. I also thank the participants of the 2022 ECPR General Conference panel “Authoritarianism”, 2022 SISP Conference panel “Regime convergence” and 2022 AuPSA Political Science Day panel “Democratization and Autocratization” for their useful comments. Moreover, I am grateful to Lasse Egendal Leipziger, Christian Houle, Frances Stewart and Nils-Christian Bormann, who shared manuscripts, replication data and ideas that helped me progress this research. Other readers of previous versions of the paper and my research project that I kindly thank are Felix Wiebrecht, Andreas Juon, Andrea Vaccaro, Andrea Cassani, Svend-Erik Skaaning, Seda Gürkan, Matthijs Bogaards, Gianni Del Panta and Licia Cianetti. Finally, I wish to thank the two anonymous reviewers for their excellent suggestions and generous support.Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 Boese et al., “State of the World 2021”; Wiebrecht et al., “State of the World 2022.”2 Houle, “Ethnic Inequality”; and Stewart, “Horizontal Inequalities.”3 Bochsler and Juon, “Power-Sharing”; Juon and Bochsler, “The Two Faces.”4 Coppedge et al., “V-Dem Codebook V11.”5 Vogt et al., “Integrating Data.”6 Baldwin and Huber, “Economic versus Cultural Differences”; Houle, “Ethnic Inequality.”7 Leipziger, “Measuring Ethnic Inequality”; Vaccaro, “Ethnic Dominance and Exclusion.”8 Vogt et al., “Integrating Data.”9 Linz and Stepan, The Breakdown of Democratic Regimes; Berg-Schlosser and Mitchell, Authoritarianism and Democracy; Capoccia, Defending Democracy.10 Carothers, “The End of the Transition Paradigm.”11 Linz and Stepan, Problems.12 Bogaards, “How to Classify Hybrid Regimes?”; Levitsky and Way, Competitive Authoritarianism; Schedler, The Politics of Uncertainty.13 Diamond and Morlino, Assessing the Quality of Democracy; Lijphart, Patterns of Democracy.14 Croissant and Haynes, “Democratic Regression in Asia.”15 Diamond, “Democratic Regression.”16 Gerschewski, “Erosion or Decay?”17 Bogaards, “De-Democratization in Hungary.”18 Bermeo, “On Democratic Backsliding”; Waldner and Lust, “Unwelcome Change”; Jee, Lueders, and Myrick, “Towards a Unified Approach.”19 Cianetti and Hanley, “The End of the Backsliding Paradigm.”20 Lührmann and Lindberg, “A Third Wave of Autocratization.”21 Cassani and Tomini, Autocratization.22 Cassani and Tomini, “Reversing Regimes.”23 Tomini and Wagemann, “Varieties”; Diskin, Diskin, and Hazan, “Why Democracies Collapse.”24 Teorell, Determinants; Coppedge et al., Why Democracies Develop and Decline.25 Bogaards, “Where to Draw the Line?”26 Mechkova, Lührmann, and Lindberg, “The Accountability Sequence.”27 Maerz et al., “A Framework.” Pelke and Croissant, “Conceptualizing and Measuring.”28 Boese et al., “How Democracies Prevail.”29 Lijphart, Democracy in Plural Societies.30 This does not amount to saying that in these contexts autocratization is per se associated with ethnicity, when the latter is politically relevant: Cianetti, The Quality of Divided Democracies.31 Boix and Stokes, “Endogenous Democratization”; Acemoglu and Robinson, Economic Origins; Haggard and Kaufman, The Political Economy; and Scheve and Stasavage, “Wealth Inequality.”32 Knutsen et al., “Economic Development and Democracy”; Abdulai and Crawford, Research Handbook.33 Guelke, Politics.34 Wrigley-Field, “US Racial Inequality.”35 Woo-Mora, “Unveiling the Cosmic Race.”36 Günay and Dzihic, “Decoding the Authoritarian Code”; Vachudova, “Ethnopopulism”; Kapidžić and Stojarová, Illiberal Politics.37 Widmalm, Routledge Handbook of Autocratization.38 Maizland, “Myanmar’s Troubled History.”39 Yusupova, “How Does the Politics of Fear in Russia Work?”40 Onwuzuruigbo, “Researching Ethnic Conflicts in Nigeria”; World Bank Group, Overcoming Poverty and Inequality in South Africa; Benjaminsen and Ba, “Why Do Pastoralists in Mali Join Jihadist Groups?”; Aalen, “The Revolutionary Democracy of Ethiopia.”41 Stewart, Horizontal Inequalities and Conflict. Chandra defines ethnicity: “a subset of categories in which descent-based attributes are necessary for membership” (Chandra, “What Is an Ethnic Party?” 154).42 In the definition of cleavage, Kriesi mentions its “structural base, political values of the groups involved, and their political articulation” (Kriesi, “The Transformation,” 165). The article concentrates on the structural base.43 Measuring the probability that two random individuals belong to different groups: Alesina et al., “Fractionalization.”44 Bochsler et al., “Exchange”; Marquardt and Herrera, “Ethnicity as a Variable”; cf. Hartzell and Hoddie, “The Art of the Possible.”45 Contra Gerring, Hoffman, and Zarecki, “The Diverse Effects.”46 Tilly, Democracy.47 Chandra, Why Ethnic Parties Succeed; Lipset and Rokkan, Party Systems and Voter Alignments.48 Hillesund et al., “Horizontal Inequality.”49 Houle, “Ethnic Inequality”; Stewart, “Horizontal Inequalities.”50 Stewart, Brown, and Mancini, “Monitoring and Measuring,” 11.51 Cederman, Weidmann, and Gleditsch, “Horizontal Inequalities.”52 Baldwin and Huber, “Economic versus Cultural Differences.”53 Kuhn and Weidmann, “Unequal We Fight.”54 Horowitz, Ethnic Groups in Conflict; Snyder, From Voting to Violence. This paper affirms that it is not ethnicity per se that is problematic for democracy, but ethnic inequality.55 Houle, “Ethnic Inequality.”56 Leipziger, “Measuring Ethnic Inequality.”57 Lijphart, Patterns of Democracy.58 Cederman, Hug, and Wucherpfennig, Sharing Power, Securing Peace?; Bochsler and Juon, “Power-Sharing”; Juon and Bochsler, “The Two Faces”; Hartzell and Hoddie, “The Art of the Possible”; Bormann, “Ethnic Power-Sharing.”59 Loizides, The Politics of Majority Nationalism; Schedler, “An Ambiguous Tool.”60 Waldner and Lust, “Unwelcome Change.”61 Stewart, Brown, and Mancini, “Monitoring and Measuring,” 11.62 Merkel et al., Defekte Demokratie.63 Coppedge et al., “V-Dem Codebook V11.”64 Maerz et al., “A Framework.”65 Coppedge et al., “V-Dem Codebook V11.” The EDI does not include indicators related to inequalities or ethnic groups: therefore, there is no risk of circularity between the explanatory variables and the outcome.66 Teorell, Determinants.67 Differently from a fixed-effect analysis on the EDI.68 Alesina, Michalopoulos, and Papaioannou, “Ethnic Inequality”; cf. Cederman, Weidmann, and Bormann, “Triangulating Horizontal Inequality.”69 Baldwin and Huber, “Economic versus Cultural Differences”; Houle, “Ethnic Inequality.” A similar approach has been implemented by Juon, “Inclusion, Recognition” to study satisfaction with the government and perceived ethnic discrimination.70 WVS: https://www.worldvaluessurvey.org/wvs.jsp (last access: March 2023).71 CSES: https://cses.org/data-download/cses-integrated-module-dataset-imd/ (last access: March 2023).72 Latinobarometro: https://www.latinobarometro.org (last access: March 2023).73 The AmericasBarometer by the LAPOP Lab: www.vanderbilt.edu/lapop (last access: March 2023).74 Afrobarometer: https://www.afrobarometer.org (last access: March 2023).75 Asian Barometer: https://www.asianbarometer.org/ (last access: March 2023).76 Vogt et al., “Integrating Data.”77 For Sub-Saharan Africa, I compared ethnic group lists with the help of the R package developed by Müller-Crepon, Pengl, and Bormann, “Linking Ethnic Data from Africa (Leda).”78 The Afrobarometer does not contain an indicator on respondent’s income. Following Houle (“Ethnic Inequality”), I created a variable based on respondent’s ownership of one or more assets, such as a bicycle, car, telephone, etc. (ranging between 0 and the maximum number of assets).79 Cederman, Weidmann, and Gleditsch, “Horizontal Inequalities.”80 Juon, “Minorities Overlooked.”81 Another country-level source, the “Power sharing in the world’s states” dataset (Strøm et al., “Inclusion, Dispersion, and Constraint.”), lacks coding criteria on its ethnic group list.82 Cederman and Girardin, “Beyond Fractionalization.”83 The EPR categorization might be seen related to political, rather than social characteristics. However, our indicator on political inequality considers formal institutions, while that on social inequality refers to informal practices. Juon (“Minorities Overlooked”) demonstrates how the two are not necessarily correlated.84 Ibid., 177. Cederman and Girardin compare the results of the fractionalization index for a country composed by two groups (Group A with 30% and Group B with 70% of the population), identical whether it is the majority or the minority that is excluded, while the nstar indicator varies between 0.072 if the group in power is the majority, to 0.843 if it is the minority. Admittedly, the measurement of social inequality between ethnic groups remains tentative. A more appropriate indicator should consider the specific services allocated to ethnic groups (such as schools or hospitals) depending on residence or identity affiliations. However, such a measure does not exist for a large-N analysis. I proxy it with the EPR indicator which, being based on informal practices, can signal a different service allocation depending on ethnicity. Further research should improve this measurement.85 Shoup, “Ethnic Polarization.”86 Table A.3 reports the (either negligible or insignificant) correlations of the main independent variables.87 Following the set-up of Houle, “Ethnic Inequality”; and Cederman, Weidmann, and Gleditsch, “Horizontal Inequalities” for the analysis at the ethnic-group level and the most common approach in democratization research with country-year data.88 The CPSD dataset only contains observations until 2016 and for countries with formal constitutions.89 With the first indicator rescaled from 0 to 1.90 Models on social inequality and the overall index do not include the group size variable, already computed in the social inequality indicator.91 Houle, “Ethnic Inequality.”92 They reached significance with standard errors clustered at the group level. However, this might risk ignoring spatial autocorrelation of ethnic groups in the same country.93 Bochsler and Juon, “Power-Sharing”; and Juon and Bochsler, “The Two Faces.”94 Cf. Lührmann and Lindberg, “A Third Wave of Autocratization” for how autocratization affects populous countries.95 Beck and Katz, “Random Coefficient Models.”96 Autocratization episode onsets are 71 out of more than 2500 observations in the ethnic-group-level models. Cf. Boese et al., “How Democracies Prevail.”97 Beck, Katz, and Tucker, “Taking Time Seriously.”98 Coppedge et al., Why Democracies Develop and Decline.99 Rovny, “Antidote to Backsliding.”Additional informationNotes on contributorsGuido PanzanoGuido Panzano is PhD candidate in Political Science at the Université libre de Bruxelles (ULB) and works at the Centre d'Etude de la Vie Politique (CEVIPOL) as Fonds de la recherche scientifique (FNRS) Research Fellow. His research interests are democratization and autocratization as well as ethnic issues.