Variation in subnational electoral authoritarianism: evidence from the Russian Federation

Despite the burgeoning comparative literature on electoral authoritarian regimes, fewer studies have accounted for the emergence of hegemonic and competitive authoritarian regimes at the subnational level. This article examines the variation in subnational electoral authoritarianism with data from the Russian Federation. First, the article shows that by using a comparative regime classification most Russian subnational cases can be classified as electoral authoritarian between 1991 and 2005. Yet, there are considerable differences in competitiveness between the electoral authoritarian regimes. The article accounts for this variation by drawing on both comparative electoral authoritarianism literature as well as more context-specific explanations. Statistical analysis on 192 subnational electoral authoritarian cases shows that the determinants of Russian subnational authoritarian stability are rather similar to those found in cross-national studies. Subnational (non)competitiveness in Russia appears to be related to the structure of the regional economy and natural resource rents, and to a lesser degree to the specific Russian federal context. Authoritarian “know how” also plays a role in authoritarian regime building. The findings of the article contribute to the literature on electoral authoritarianism, subnational democratization and Russian subnational politics.


Introduction
At the end of the Cold War, many authoritarian regimes adopted formally democratic institutions and began holding multiparty elections. However, in many cases this did not result in full democratization, but in the formation of "electoral authoritarian" regimes, where elections are the main means of access to power, but do not function as an accountability mechanism between the rulers and the citizens. 1 Democratization has also remained spatially uneven in many federal states, and outright authoritarian regimes have emerged at the subnational level.
Despite the burgeoning comparative literature on electoral authoritarian regimes, 2 less is known about the determinants of variation in subnational electoral authoritarianism, and the dynamics of subnational hegemonic and competitive authoritarian regimes. 3 What accounts for variation in subnational electoral authoritarian regime competitiveness? How well do theories on electoral authoritarianism, which are mostly developed to explain national level cases, explain subnational dynamics?
This article probes the dynamics of subnational electoral authoritarianism with data from the post-Cold War Russian Federation. Russian 89 "regions" held several rounds of executive ("gubernatorial") elections between 1991 and 2005. 4 However, the majority of the cases did not lead to democratization. Many incumbents won gubernatorial elections with more than 80% of the votes, while others remained in office by slim margins after manipulated elections. This article first adapts Marc Morje Howard's and Philip Roessler's 5 global post-Cold War regime classification to Russian data, and categorizes Russian subnational regimes using the Moscow Carnegie Center's Index of Democracy scores and author-collected electoral data. The results show that a clear majority of Russian subnational regimes until 2005 can be classified as electoral authoritarian systems. Most of the cases were competitive authoritarian regimes, with competitive but not fully free and fair elections. Hegemonic authoritarian regimes, with non-contested elections, form the second largest group, while electoral democratic cases make up the smallest regime cluster.
While most of the Russian subnational regimes can be classified as electoral authoritarian, there are striking differences in electoral competitiveness between the hegemonic and competitive authoritarian regime types. The mean margin of victory in hegemonic authoritarian regimes was 75.8%, whereas it was 24.5% in the competitive authoritarian cases. The second part of the article examines the reasons for this variation. The article tests both explanations drawn from comparative authoritarianism literature, focusing especially on the role of state organizational capacity, 6 and explanations that take into account the specificities of Russian federalism. The theories are tested on a set of 192 Russian subnational electoral authoritarian cases. The findings suggest that the determinants of subnational electoral authoritarianism in Russia were rather similar to national level cases, and that context specific variables, apart from minority ethnicity, play a lesser role. Natural resource rents have a strong association with more electoral hegemony at the Russian subnational level. Economic concentration also appears to be related to hegemonic authoritarianism, although the effect is less robust. Also, the results suggest that some regional elites "learned" how to curtail electoral competition in successive elections. In contrast, the emergence of subnational electoral authoritarianism in Russia appears to be unrelated to either the changing federal context or asymmetries in Russian federalism. Titular republic status per se is not associated with electoral hegemony. However, the effect of minority ethnicity appears to be conditioned by region type, and is stronger in titular republics.
The variation in Russian subnational democratization has been much studied. 7 Existing quantitative studies have produced important findings on the effect of democratic diffusion, 8 socio-economic inequality, 9 as well as socio-economic and context specific variables 10 on Russian subnational democratization, which this study builds on. However, unlike previous works, this article draws explicitly on theories on electoral authoritarianism to account for Russian subnational regime variation. 11 As shown, the majority of the Russian subnational regimes can be classified as electoral authoritarian, and thus we should employ the electoral authoritarian "theoretical toolset" systematically to conceptualize and explain Russian subnational regime variation. Recent literature has begun to take steps in this direction. For example, Petr Panov and Cameron Ross classify Russian regions between 2007 and 2012 into hegemonic and competitive authoritarian cases based on the level of contestation in regional legislative elections. 12 This article goes further than this by both classifying Russian regional regimes and testing the determinants of the subnational variation with new data.
The article proceeds as follows. Section 2 outlines the conceptual differences between electoral democratic and electoral authoritarian regimes, and Section 3 then categorizes Russian subnational political regimes by adapting a global regime classification to Russian data. The article then proceeds to explain the variation in electoral authoritarian regime competitiveness. Section 4 develops hypotheses for statistical testing based on both comparative literature on electoral authoritarianism as well as more context-specific explanations. Sections 5 and 6 present the data and the results of the statistical analysis. The final section concludes.
Conceptualizing subnational democracy and non-democratic regimes Political regimes have two dimensions -the form of access to power (the mechanism by which the principal governing offices are filled) and the form of exercise of power. 13 Post-Cold War political regimes can be categorized into five types based on their mode of "access to power". 14 First, regimes differ in how they select the executive. Electoral regimes fill key government posts by popular elections, whereas "closed" authoritarian regimes, such as contemporary China, do not. However, not all electoral regimes are democracies, and the conceptual "boundary" between democracies and electoral authoritarian regimes is determined by the integrity of elections. 15 Representative democracy is conceptualized here with a "minimalist" definition 16 -elections are seen as a means of delegation via the "democratic method". 17 Electoral manipulation undermines democratic delegation between the principals (voters) and their agents (politicians) in several ways. 18 First, electoral manipulation hinders the ability of voters to screen the candidates with adequate information. 19 Second, if the incumbents can manipulate elections once in power, the potential of losing office no longer constrains them between the electoral cycles. Non-contested elections are also problematic in terms of democratic delegation, as they do not provide voters with a free supply of alternative candidates. 20 Electoral democratic regimes, such as the contemporary Philippines, 21 are not "perfect" democracies, but the electoral process should present a reasonable chance for the opposition to come to power and for different preferences of the electorate to be represented. Despite being "minimalist", this definition is actually rather demanding -and should be, as electoral corruption undermines the legitimacy of electoral institutions and thereby democracy itself. 22 Electoral democracies are often differentiated from liberal democracies, which hold free and fair elections and guarantee the rule of law and full civic rights to all the citizens. 23 Electoral authoritarian regimes can be divided into hegemonic and competitive, 24 which are differentiated by varying levels of uncertainty that the elections generate. Hegemonic electoral authoritarian regimes, such as post-2004 Russia, 25 hold multiparty and multicandidate elections, but the electoral results (and, subsequently, the government) are dominated by one political group or party, and there is little electoral uncertainty. Hegemonic electoral outcomes are often a result of credible opposition challengers choosing not to run against the incumbent. 26 However, it is important to note that the resulting hegemony reflects strategic choices by the opposition, and not that opposition parties are not able to campaign (as in "closed" autocracies). Competitive electoral authoritarian systems manifest more electoral competition and uncertainty, and often very underhanded electoral tactics. Contemporary Armenia can be considered an example of a competitive authoritarian regime, where elections are competitive, but undermined by unfair electoral practices. 27 Russian subnational regime categorization This article categorizes Russian subnational political regimes between 1991 and 2005 into three electoral regime types: electoral democratic, competitive, and hegemonic authoritarian regimes. There were no "closed" authoritarian cases in the Russian regions, as the subnational executives were selected through popular elections. 28 It is also unlikely that "liberal democratic" regimes would have developed at the subnational level in a country that was classified as electoral democratic/competitive authoritarian under Yeltsin and during Putin's first term. 29 This article adapts Howard and Roessler's two-tier electoral regime categorization scheme to Russian data. Howard and Roessler first distinguish electoral democratic regimes from electoral authoritarian regimes by a cut-point in the Freedom House and Polity scores. 30 To demarcate competitive and hegemonic authoritarian regimes they then use a (fairly intuitive) cut-point of 70% of the votes for one party/candidate. 31 It is important to note that this threshold is used to mark differences within the electoral authoritarian regimes only after they have been delineated from electoral democratic regimes, and not simply to operationalize democratic/non-democratic regimes as such.
To operationalize Russian subnational electoral democratic and authoritarian regimes this article uses the Index of Democracy, which is an expert rating on Russian regional democratization compiled by the Moscow Carnegie Center scholars Nikolay Petrov and Alexey Titkov. It includes 10 indicators, which are measured on a 1-5 scale (1 being the least democratic, 5 the most). The index covers 88 regions, and excludes Chechnya. 32 Despite some limitations, it is by far the most comprehensive Russian subnational democracy measure, and is widely used in the literature. 33 However, four of the sub-categories of the index (political regime, local self-administration, privatization, and corruption) are more closely related to the "exercise of power" dimension of political regimes than to the "access to power" dimension. For a more accurate operationalization of the latter dimension, six sub-scores of the index (openness of political life, free and fair elections, political pluralism, media freedom, civil society, and elite relations) were combined into an "Access Index" using the publicly available scores for 2000 and 2004. 34 This makes the highest possible score 30. 35 Establishing cut-points in continuous indices is always contentious. 36 The authors suggest two possible ones in the original, 40 (out of the total 50) for "democratic leaders" and 37 for "moderately democratic" regions. 37 Accordingly, the cut-point for electoral democratic regimes was set to 24 in the Access Index (this is equivalent to the original 40 cut-point). Regimes were coded from one election to the next. Thus elections in a region which received 24 or higher in the Access Index were categorized as electoral democratic between 1991 and 2005.
The "boundary" between competitive and hegemonic authoritarian regimeselectoral competition -is more straightforward to operationalize. Cross-national studies use both executive and legislative election data. However, until 2003 the Russian subnational legislatures were largely non-partisan, 38 and therefore legislative party shares are not good indicators of political hegemony. Instead, author-collected data on 241 Russian gubernatorial elections between 1991 and 2005 from 87 regions was used. 39 Electoral authoritarian regimes were classified on the basis of the "current" election results, until the following elections. Cases that scored less than 24 in the Access Index, and held elections in which the winner received more than 70% of the votes in the first round were coded as hegemonic authoritarian. Competitive authoritarian regimes were coded as those with a score of less than 24 and less than 70% of the votes for the winner in the first round. 40 For example, the regime in Murmansk Oblast (18 in the Access Index) was coded competitive authoritarian after the founding 1996 elections, which were tightly contested between E. Komarov and Yuriy Evdokimov. The regime was then coded hegemonic authoritarian after the 2000 and 2004 elections, which the incumbent Evdokimov won with 87% and 77% of the votes, respectively. Table 1 summarizes both the conceptual "boundaries" between the regime types and their operationalization, and presents examples of Russian subnational cases.
Applying the comparative regime categorization to Russian data reveals that over 90% of the Russian subnational political regimes between 1991 and 2005 can be classified as electoral authoritarian. Electoral democratic regimes make up only 9.5% of the total. Competitive authoritarian regimes form the largest group (67.6%), while hegemonic authoritarian regimes make up 22.8% of the total. 41 The regime categories here are also pretty similar to those found by Petr Panov and Cameron Ross. Over 80% of the regions (13 out of 16) categorized as hegemonic authoritarian between 2007 and 2012 by Panov and Ross had at least one hegemonic authoritarian case also between 1991 and 2005. 42

Explaining variation in subnational electoral authoritarianism
Most of the Russian subnational regimes could thus be classified as electoral authoritarian regimes, with multicandidate but not free and fair elections. However, there were striking differences in electoral competitiveness between the electoral authoritarian regimes. The mean margin of victory in hegemonic authoritarian regimes was 75.8%, whereas it was 24.5% in the competitive authoritarian cases. The rest of this article focuses on accounting for this variation. How well do theories on electoral authoritarianism, which are mostly developed to explain national level cases, explain subnational dynamics? What is the role of Russian context-specific factors, such as institutional asymmetries between the regions, in accounting for the variation in authoritarian competitiveness?
Authoritarian regime capacity and electoral (non)competitiveness

State scope and cohesion
Much of the comparative literature on electoral authoritarianism has concentrated on election proximate factors, such as opposition strategies or opposition  (1991 -2005) mobilization 43 to account for authoritarian (in)stability. However, Steven Levitsky and Lucan Way focus on the "levelness of the playing field" between the regime and opposition, and argue that the incumbent regime's control of coercive and economic resources undermines the opposition's ability to contest the elections fairly -even before the elections take place. 44 In Levitsky and Way's account, a state's organizational capacity determines the regime's ability to withstand opposition challenges. State capacity is based both on elite cohesion and the scope of a state's coercive apparatus. These are rooted in certain structural features of the economy, which condition the ability of the regime to monopolize economic resources. Conversely, electoral competition arises from weak authoritarian state capacity. 45 In electoral authoritarian regimes, state bureaucracy such as tax or regulatory agencies are typically employed in "low-intensity coercion" against political opponents. 46 State jobs are also an important form of patronage, and a large state bureaucracy can thus be conducive to electoral authoritarian stability. It could therefore be expected that Russian regions with larger bureaucracies would have greater state control and less contested elections.
Authoritarian political economy is also likely to underpin electoral (non)competitiveness. Economic wealth and the control of economic resources are key sources of authoritarian regime strength. 47 A higher revenue base allows regimes to improve their "electoral prospects" by targeted political business cycles and by distributing patronage and additional welfare benefits to voters. 48 Russian regional wealth and growth rates have diverged considerably since the collapse of the Soviet Union. 49 Differences in regional wealth have been largely shaped by structural legacies from the USSR. While some Russian regions inherited economies based on natural resources and profitable heavy industries (such as the defence sector), other regions were left with uncompetitive light industrial sectors or predominantly agrarian economies after the Soviet collapse. 50 Greater regional wealth could be expected to enable greater distribution of patronage transfers, which promotes elite cohesion, and to build support for the regional government in elections. Economic wealth can also translate into greater state (repressive) capacity, as fiscal strength can enhance the "level of compliance" in the state repressive apparatus. 51 Therefore it can be hypothesized that regions with more favourable economic legacies from the USSR and greater regional wealth would have more electorally hegemonic regimes.
The structure of the economy and "rentier statism" However, it may be the structure of the economy, and not economic wealth per se, which affects authoritarian stability. The structure of the economy shapes the regime's ability to monopolize and control the economic rents. 52 State control of economic resources allows the regime to outspend its competitors, and to deny opposition access to financial support. 53 Resource asymmetries between the state and the opposition are most evident in non-privatized, state-controlled economies. 54 Overall, privatization in Russia was more extensive than, for example, in Belarus, and despite differences in privatization implementation, only a few Russian regions can be characterized as having developed "state owned" economic sectors. 55 However, there were notable differences in the degree of economic concentration between the Russian regions. Some regions' economies were dominated by a single company, such as Severstal in Vologda Oblast, while some regions, such as Sverdlovsk Oblast, had very diverse economies. 56 More concentrated economic sectors could be more easily monopolized by the incumbent regime, undermining the opposition's access to economic resources. It could thus be expected that greater regional economic concentration would be reflected in electorally more hegemonic regimes.
Several cross-national studies have found an association between mineral wealth and authoritarianism. Natural resource rents can allow for lower levels of taxation, and higher spending on patronage and state coercion. 57 Levitsky and Way argue that natural resource rents can be easily monopolized by the regime, which subverts the electoral playing field. 58 There are vast differences between natural resource endowments among the Russian regions. Control over natural resources underpinned movements for greater regional autonomy in Russia in the early 1990s, and these allocations were later formalized into a series of bilateral treaties between the centre and the regions. Natural resource rents should be easier to monopolize by the regime for patronage purposes, engendering greater elite cohesion. It can therefore be expected that greater regional natural resource endowments would be associated with less competitive elections.

Experience in electoral manipulation
Electoral authoritarian regimes are characterized by incumbents' attempts to try to isolate themselves from the "electoral threat". Electoral authoritarian incumbents often come to power in relatively clean founding elections, but then seek to constrain the electoral playing field to avoid electoral defeat in subsequent elections. Several authors have noted that experience in authoritarian elections and their manipulation ("authoritarian know how") is important in building authoritarian regimes. 59 If the "learning hypothesis" is true, electoral competitiveness should decline over time, and we could expect successive electoral cycles to be less competitive than the first elections.

Context-specific explanations
Alternative approaches link the growth of subnational authoritarianism in Russia to the different socio-political environments and asymmetrical federalism in the Russian regions. It can be also expected that changes in Russian federal administration would have shaped subnational political dynamics.
Recent literature has suggested that authoritarian stability can be rooted in economic wealth, especially if the economy is closed and dominated by one (state-controlled) sector. Yet, growing economic wealth can also lead to the creation of an independent economic class and engender greater pressure for political and economic reforms. 60 Higher levels of education could also foster a greater demand for democracy. 61 Following the modernization theory, a plausible alternative explanation for the effect of development would be that wealthier Russian regions and regions with higher levels of education would be more politically pluralistic and would have more competitive elections. While regional wealth is already included in the analysis (although here the expected association is in the opposite direction), the models below also control for the levels of higher education in a region.
Due to the legacies of Soviet institutionalized ethnicity, there were great differences in the degree of political and institution-building autonomy between the Russian regions in the 1990s. Despite the 1993 constitution formally treating all of the "subjects of the federation" equally, the Russian Federation remained de facto asymmetrical. The titular republics were able to draft their own constitutions, and to determine their systems of government, while President Yeltsin appointed chief executives (governors) to all of the other regions apart from the two federal cities, Moscow and St Petersburg, until 1996 (with some exceptions). Many authors have argued that the institutional autonomy would have allowed the titular republic elites to build more authoritarian political regimes. 62 Therefore the models below control for titular republic status.
Russian regions are nested in a federation, and thus the changing federal context is likely to influence regional political developments. 63 Russian regions enjoyed considerable internal autonomy due to central state weakness in the 1990s. The Putin administration sought to recentralize the federal system and reign in some of (aconstitutional) asymmetries in the institutional and political powers of the regions from 2000 onwards. The changing federal context could have affected the ability of the regional elites to curtail electoral competition. Therefore the models control for the changing federal context between the Yeltsin and Putin administrations.
The Russian Federation is a multi-ethnic state and the Russian regions vary widely in the share of the ethnic (non-Russian) population. Levitsky and Way have suggested that state cohesion can be rooted in "shared ethnic identity", especially in cases where the ethnic cleavage is highly salient. 64 Minority ethnic ties could thus provide a source of elite cohesion in the Russian regions. Henry Hale has also suggested that "gubernatorial political machines" targeted ethnic minority areas in electoral mobilization, especially in the titular republics. 65 Therefore the models control for the minority ethnic population in a region, and also examine the interaction between titular republic status and minority ethnicity.

Operationalization and data
The analysis examines the determinants of electoral authoritarian regime competitiveness in the Russian regions, and thus includes only electoral authoritarian cases (hegemonic and competitive authoritarian regimes), which were separated from the electoral democratic cases above. 66 The dependent variable, electoral competitiveness, is operationalized by the margin of victory (that is, the difference in vote shares between the winner and the runner up in the first round). Margin of victory provides a "simple and intuitive" measure of electoral competitiveness and is widely used in the extant literature. 67 Regarding the independent variables, the size of the regional bureaucracy is operationalized by the number of workers in state and local administration agencies per region (Bur), which comes from Goskomstat. 68 The measure stays remarkably stable over time (the bivariate correlation between two adjacent years between 2000 and 2005 is consistently over 0.99), and so the models are run with data from 2000. The economic wealth that the region inherited from the Soviet Union is operationalized by gross regional production (per capita) in 1995 (the earliest measure provided by Goskomstat), which is logged (GRP95pc). The share in the overall regional production by firms which have a dominant position in the market (Dominant firm) is used to operationalize regional economic concentration. These data come from Goskomstat's Regiony Rossii yearbooks. 69 Unfortunately, the indicator covers only 62 regions (thus data are missing in a potentially non-random way), and can only be used in a supplementary model. Natural resource rents are operationalized by the share of natural resources (oil, gas, coal, and metals) in the regional industrial output (Nat res), which is calculated from Goskomstat data. The rank order of the consecutive elections in each region is used to test the hypothesis that elections would become less competitive over time (for example, the second election held in a region is coded "2") (El Cycle). The titular republic status is operationalized by a dummy variable (coded "1" for Republic). The differences in higher education levels are operationalized by the share of students in higher education per 10,000 people (Education) from Goskomstat (the adjacent years are highly correlated so data from 1999/2000 are used). The change in the federal context is operationalized by a dummy variable (coded "0" for the Yeltsin administration until 1999, and "1" for the Putin administration after 2000) (Fed Context). The data on the share of non-Russian population (Non Rus) in a region come from the 2002 Russian census (the share of the ethnically Russian population was subtracted from 100). 70 Most of the autonomous okrugs (AOs) were merged with their "parent" region after 2005, and Goskomstat does not provide economic or education data for all of them. Thus the models below do not include observations from the AOs (nor from the Jewish Autonomous Oblast). This actually makes for a more correct set of cases, as the AOs cannot be considered fully equivalent units to the rest of the regions. The AOs had an ambiguous constitutional status, and were not fully politically and economically separate from the "parent" regions. Data on Chechnya is excluded due to the military conflicts, and Dagestan did not hold gubernatorial elections during the time of the analysis. This results in 192 observations. Table 2 provides descriptive statistics of all the variables. Table 3 presents estimates from several linear (OLS) models on electoral competitiveness, with standard errors clustered by region. 71 Model 1 is a full model with all the explanatory variables and the controls, and Models 2 -6 present several alternative specifications. The size of the regional bureaucracy appears not to be associated with authoritarian competitiveness, as the coefficient for Bureaucracy is not significant in this model or in any other specifications. 72 In contrast, natural resource rents have a strong effect on electoral contestation. Natural resource rents (Nat res) have a positive, substantively large, and highly statistically significant association with lower levels of electoral competition (that is, higher margins) throughout the models. Based on Model 1, it can be estimated that as the natural resource concentration increases by one standard deviation, the winning margin grows with almost nine percentage points. The coefficient for regional wealth (GRP95pc) is also significant at the 0.05 level in most models. The sign is negative, suggesting that, once natural resource wealth is controlled for, poorer regions conduct less competitive elections, which runs counter to the hypothesis that greater fiscal capacity would translate into more electoral hegemony. The coefficient for the consecutive elections variable (El cycle) is positive, and significant at the 0.1 level in Model 1, supporting the "learning effect" hypothesis. Apart from Education, none of the coefficients for control variables are statistically significant, suggesting that the emergence of hegemonic authoritarian regimes is unrelated to both the federal context and titular republic status. Greater higher education levels appear to be associated with reduced political pluralism, contrary to the prediction of the modernization theory. The coefficients of the federal (Fed context) and regional context (Republic) variables do not attain statistical significance in any of the models. Nor is the coefficient for the share of the non-Russian population (Non Rus) significant here.

Empirical analysis and discussion
The results of Model 2, with a categorical electoral cycle variable, support the learning effect hypothesis further, that is, that more experience in running elections would have allowed regional elites to "learn" how to curtail electoral competition. The coefficient for the second and third elections held in a given region is positive and statistically significant, indicating that these were less competitive than the first electoral cycle. (Only a handful of regions held four elections, so not much can be inferred from the fourth cycle results.) Model 3 tests the effect of economic concentration with the variable measuring the share of the dominant firm's production in the regional economy. Since the indicator only contains data on 62 regions (in a possibly non-random way) it was not included in the main models. The measure is positively correlated with natural resources (r = 0.58), and thus the latter variable is dropped from this model. The positive and significant coefficient indicates that regions with concentrated economies held less contested elections, in line with predictions from the electoral authoritarianism literature. 73 Models 4 -6 test the effects of titular republic status and ethnicity on electoral competitiveness. Model 4 interacts the titular republic dummy with the share of the non-Russian population to test Hale's hypothesis on ethnic mobilization in titular republics. The interaction term is positive and significant at the 0.1 level, suggesting that the effect of ethnicity would be stronger in the titular republics. However, neither the titular republic dummy nor the ethnicity variable are significant when included in the models on their own, although the coefficients are in the expected direction (Models 5 and 6). Thus, higher levels of the non-Russian population seem to have an effect on more hegemonic elections, but only in the context of titular republics.
For a robustness check, the models were also run with the full sample, that is with the electoral democratic regimes included (the results are presented in Table 4 in the online appendix). Most results remain substantively similar, but the coefficients for Dominant firm and the interaction term between titular republic status and ethnicity are no longer significant.
The findings above suggest that the determinants of Russian subnational electoral authoritarianism are rather similar to those in cross-national settings, and that factors specific to the Russian federal context play a lesser role. Natural resource wealth appears to be as harmful for subnational democracy as at the cross-national level. The association between natural resource rents and more curtailed political competition is in line with the findings of several subnational 74 and cross-national studies. 75 Although the precise causal mechanisms linking resource rents with more hegemonic regimes are not possible to test here, a case study on Tatarstan has suggested that resource rents were used effectively for patronage purposes. 76 Economic concentration also appears to play a role in hegemonic authoritarian regime formation, although the results were less robust for sample changes and should be interpreted with some caution due to missing data. Also, the results suggest that some regional elites "learned" how to curtail electoral competition in successive elections. Not surprisingly, authoritarian "know how" and "state crafting" plays a role in subnational electoral authoritarianism as well as national level cases. In contrast, greater wealth or fiscal capacity did not translate to more stable authoritarianism, as electoral contestation declined in poorer rather than wealthier regions (when natural resource wealth was controlled for). However, it is important to note that the effect of fiscal federalism was not tested here. It may be that fiscal transfers "prop up" authoritarian regimes in poorer regions. 77 Interestingly, the results suggest that once other regional characteristics were controlled for in the statistical analysis, asymmetries in Russian federalism per se were not associated with more electorally hegemonic regimes. While democratic norms were clearly violated in many titular republics in the 1990s, 78 this did not necessarily lead to less electorally competitive regimes in all cases. Politically monopolistic regimes developed in many titular republics (such as Tatarstan and Mordoviya), but elections in other titular republics (such as in Mariy El) remained intensely contested. Once the gubernatorial elections were introduced to regions other than republics in 1996, hegemonic authoritarian regimes emerged in many non-titular regions, such as in Novgorod, Orel, and Kemerovo Oblasts. 79 These results are in line with earlier findings by Bryon Moraski and William Reisinger. 80 Indeed, it should be emphasized that the titular republics were a highly heterogeneous group in terms of economic wealth, ethnic make-up, and institutional developments, and thus it is not surprising that their regime trajectories also vary. 81 However, there seems to be something particular about minority ethnicity in the context of titular republics that promoted the emergence of more hegemonic regimes in the Russian regions. The results showed that the effect of minority ethnicity on electoral contestation was stronger in titular republics than in other regions. 82 These results support Hale's theory on ethnic minority mobilization in titular republics, and are in line with case study evidence. 83 Conclusions This article has examined the variation in subnational electoral authoritarian regime competitiveness with data from the Russian Federation. The dynamics of Russian subnational electoral authoritarianism appear quite similar to national level cases. The article found considerable support for the existence of a natural "resource curse" in the Russian regions. Economic concentration and authoritarian "know how" were also associated with authoritarian regime stability. Contextspecific factors, apart from minority ethnicity, appeared to play a lesser role in the development of Russian subnational authoritarianism. All in all, theories that were initially developed to account for national level authoritarian stability appear also well-suited to the examination of subnational authoritarianism. This article has taken first steps at trying to explain variations in subnational electoral authoritarianism in Russia. However, much further work is still needed to fully probe the complex ways that the federal setting affects subnational dynamics. 84 Theories of electoral authoritarianism stress the role of state resources that the incumbents are able to utilize in tilting the "electoral playing field" against the opposition. Yet the control over state coercion, natural resources, and fiscal federalism differ between and within federations, and this also affects the resources available to subnational incumbents. Thus further work is required to fully probe the (constantly changing) resources that were available to Russian subnational incumbents, especially in the realm of fiscal federalism and state coercion.
Gubernatorial elections were re-instituted in Russia in 2012. The current elections are not directly comparable to those held until 2005, as the governors' power is more limited, and candidate entry is more restricted than previously. 85 Thus far the gubernatorial elections have not challenged the regime's grip on power in the Russian regions. multiple candidates (the uncontestedness was due to opposition boycotts or biased registration rules), and thus are not clear cases of "closed" authoritarian regimes. See for example Sharafutdinova on the 1991 and 1996 presidential elections in Tatarstan, Sharafutdinova,Political Consequences,. Nor does the Index of Democracy data allow for separating between electoral democracies and liberal democracies. 30. Either 2/better in Freedom House or 6/higher in Polity. 31. For similar cut-points see, for example, Brownlee, "Portents of Pluralism." 32. The ratings are available at the Nezavisimiy institut sotsial'noy politiki website at http://atlas.socpol.ru/indexes/index_democr.shtml. The ratings for 2005 were kindly provided by Alexey Titkov. Karachaevo-Cherkessiya and Udmurtiya Republics enter the data set in 1999 and 2000, respectively, when these republics switched from parliamentary systems to executive elections. Regional gubernatorial election data comes from Vybory glav ispolnitel'noy vlasti sub 'ektov Rossiiskoi Federatsii 1995, McFaul, M. and N. Petrov (1998, Politicheskiy al'manakh Rossii 1997, Vybory v organy gosudarstvennoi vlasti sub'ektov Rossiiskoi Federatsii. 1997-2000, and the website of Nezavisimiy institut vyborov. 40. Note that this operationalization does not distinguish whether it was the incumbent or an opposition candidate who won the elections hegemonically. These cases were rare, which supports the intuition behind the operationalization (non-incumbents won in Kursk Oblast 1996, Lipetsk Oblast 1998and North Ossetiya Republic 1998 Kursk Oblast the incumbent was not allowed to participate in the elections). 41. Regime lists can be found the online appendix. 42. See Panov and Ross, "Patterns of Electoral Contestation," 398. 43. See, for example, Bunce and Wolchik, "Oppositions versus Dictators"; Howard and Roessler, "Liberalizing Electoral Outcomes"; Tucker, "Enough! Electoral Fraud." 44. Levitsky and Way, Competitive Authoritarianism. See also Howard, Roessler, and Sallam, "The Structure of Power." 45. See also Way, "Authoritarian State Building."