Why Elections: Autocrats Incentive for Electoral Authoritarianism. Hisashi Kadoya. Abstract

Transcription

1 Why Elections: Autocrats Incentive for Electoral Authoritarianism Hisashi Kadoya Abstract Which factors explain the expansion of Electoral Authoritarianism? What kind of authoritarian regime decides to adopt election? The objective of this paper is to answer these questions. Focusing on the incentive of an authoritarian leader to introduce elections, I argue that leaders who do not monopolize power want elections to gain benefits from the election. Thus, elections are likely to occur under authoritarian regimes in which such leaders rule. In the analysis that is presented in this study, the leader s decision power monopoly is important in addition to forces from outside the government. This paper contributes to the studies of elections that take place under authoritarian regimes by constructing a theory that focuses on the leader s electoral incentives and testing the hypothesis induced in the theory. Waseda University, Tokyo (however85@gmail.com) 1

2 1.Introduction Since the Cold War ended, authoritarian regimes holding presidential or parliamentary elections, which are competitive, at least minimally, have become more prominent, despite the fact that the authoritarian incumbent may be defeated. In the 1990s, over half of all authoritarian regimes exhibited electoral authoritarianism (EA) (Howard and Roessler 2009; Kricheli and Magaloni 2010; Schedler 2013). Which factors explain the expansion of EA? What kind of authoritarian regime decides to adopt election? The objective of this paper is to answer these questions. Recently, studies in comparative politics have provided much knowledge about this political phenomenon. Some previous studies have explained the introduction of elections in authoritarian regimes as the result of forces outside the government. For example, domestic anti-government movements and external pressure have been considered to be independent variables (Bratton and van de Walle 1997; Levitsky and Way 2010). Conversely, other studies have depicted that elections possess various functions that lead to stabilized authoritarian regimes (Blaydes 2011; Geddes 2005; Magaloni 2006). Thus, introducing elections without coercion from outside the government is a possible phenomenon. This paper focuses on the incentive of an authoritarian leader to introduce elections. I argue that the degree of monopoly of the political decision power is an essential element in explaining EA. Leaders who do not monopolize power want elections to gain benefits from the election. Thus, elections are likely to occur under authoritarian regimes in which such leaders rule. In the analysis that is presented in this study, the leader s decision power monopoly is important in addition to forces from outside the government. I aim to contribute to the studies of elections that take place under authoritarian regimes by constructing a theory that focuses on the leader s electoral incentives and testing the hypothesis induced in the theory. The paper is organized as follows. Section 2 provides a literature review about the authoritarian regime and elections. The following section develops the theory of electoral introduction and derives implications for quantitative analysis. In the fourth section, after introducing the data, analytical methods, and variables, I test my hypothesis. I end with a conclusion that highlights the main findings and offers an agenda for future research. 2. Previous Literature 2

3 In this section, I review previous literature that discusses elections under authoritarian regimes. Previous studies are roughly divided into two camps: elections as an independent variable and research regarding elections as a dependent variable. Each study references the introduction of elections under authoritarian regimes. First, I discuss research that indicates that elections play an important role in the fate of authoritarian regimes; further, I discuss research that studies factors affecting the introduction of elections under authoritarian regimes. Finally, I point out the problems of these studies. 2.1 Elections Under Authoritarianism Abundant research has been conducted regarding the influence of election on the fate of the system. Several studies have given attention to the role of election under authoritarianism. However, such studies have not yet reached a conclusion. Schdler (2013: 143) says that On the one hand, the literature on the political economy of dictatorship has been emphasizing the regime-sustaining value of authoritarian elections. On the other hand, comparative studies of democratization by elections have been stressing their regime-subverting potential, because elections tend either to destabilize authoritarian regimes or to democratize them. Until recently, these strands of theoretical inquiry and empirical analysis have been developing in a state of peaceful coexistence and mutual ignorance (italics by Schedler 2013). 1 Literature linking elections to authoritarian regimes fails to emphasize electoral opportunities for defiance. These previous studies highlight that elections provide opportunities for regime change (Schedler 2013). Huntington (1991) has produced a pioneering study. He depicts the consequences of elections held during the "third wave of democratization." According to him, authoritarian leaders who aim to restore legitimacy through elections lost in the election, and these results caused the unexpected fall of the regime. Bance and Wolchik (2006, 2009, 2010) enumerate elements of the election that cause the collapse of the authoritarian regime while targeting former communist countries. According to them, the following elements are effective: (1) formation of a unified opposition; (2) efforts to increase voter registration and turnout and to improve the quality of voter lists; (3) efforts to focus campaign debate on the costs of the incumbent regime and the benefits of participation to opposition members and citizens at large; (4) utilization of international and domestic election monitoring as well as the media and public-opinion polls, where possible; and 5) preparations for public protests if incumbents or their anointed successors try to steal the election (Bunce and Wolchik 2006: 6). They suggest that it is not 1 A study also exists that claims that elections have both the effect of maintaining the regime and the effect of collapse, both of which are not conflicting (Knutsen et al. 2017). 3

4 necessarily limited to voting itself that causes the breakdown of a regime. Even if the incumbent wins the election, mass uprisings after the election may collapse the authoritarian regime. Tucker (2007) s research is an example that focuses on such post-electoral regime failure. In previous literature linking elections to maintaining regimes, studies argue that the establishment of elections as a means by which dictators hold onto power (Gandhi and Lust-Okar 2009: 404). Several functions of elections, which serve as tools, have been pointed out by previous research. For example, Geddes (2005) gave attention to the signaling function of election results. If incumbent wins in a landslide victory in an election, the result gives a strong signal of regime invincibility and support from voters. 2 This strong signal deters reckless challenges by the opposition. Some studies focus on information gathering function (Magaloni 2006). Although authoritarian leaders need to gather public opinions about their rule, citizens living under authoritarianism may be reluctant to reveal their true preferences for fear of repression (Wintrobe 2000). Holding some extent of free and competitive elections means the leader can allow citizens to reveal their true preferences. 3 By looking at the results, distinguishing between areas where the support is strong and not as strong is possible. Paying attention to the areas of weak support prevents defiance in advance (Geddes 2005). The information obtained through the election is not limited to public opinion on the regime. Blaydes (2011) points out that information about the elite loyalty gained through the election is useful for sustaining a regime. The elite, who invest plenty of effort to mobilize voters in the electoral competition, are likely to be faithful to the regime. Blaydes says that the election aids authoritarian survival by regularizing intra-elite competition, while at the same time outsourcing the cost of political mobilization and redistribution to the rent-seeking elite (Blaydes 2011: 51). Contrary to several studies that have examined the influence of the election on the fate of authoritarian regimes, few studies have investigated electoral introduction under authoritarianism (Kim 2016). Researchers have argued that elections have been an institution demanded not by internal authoritarian actors but by outside actors. The demand of democratic countries or domestic anti-government actors are principle factors. 2 Debate exists as to the necessity of a fair result. Simpser (2013) says that it is important to be able to produce victory and that victory can be achieved through fraud (Simpser 2013). 3 It seems to be evidence that many of the elections after World War II allowed secret voting with the expectation of functioning to make people in the elections express real preference (Geddes 2005: 19). 4

5 The first actors to demand elections are democratic countries. Democratic countries, including the United States, sometimes interfere with the internal affairs of authoritarian regimes under democratization promotion policies (Escriba-Folch and Wright 2015). To survive politically, an authoritarian leader may decide to introduce elections. Democratic countries evoke elections using various foreign policies. 4 One typical policy is aid. 5 Joseph (1997) argues that most of the electoral introduction in the Sub-Saharan Africa region can be explained by the receiving of aid. The threat exists that democratic regimes may sanction authoritarian regimes if the request of holding elections is refused (Dietrich and Wright 2014). The second actors demanding elections are anti-government forces within the authoritarian countries. According to Bratton and van de Walle (1997), several movements occurred in many Sub-Saharan African countries following the serious economic crisis of the late 1980s. In order to suppress protests, leaders were forced to introduce elections. Kim (2016) also points out the importance of such movements. According to Kim (2016), appealing to a non-violent movement rather than a violent movement to introduce elections is more effective. 2.2 Problems of Previous Literature All preceding research has problems explaining the introduction of elections in authoritarian regimes. In this section, an overview of independent elections is the independent variable, and preceding research is the dependent variable. The problem with discussing the introduction of elections from research that regards elections as an independent variable is in inferring the reasons for the introduction from the functions of the election. Indeed, the election may function as a maintenance tool. However, actors in the government do not necessarily introduce the election in the hope of fulfilling this function. Even if an election was forced to adopt, it may have become familiar with elections as repeatedly implemented, and it can be used as a tool of governance. Whether room actually exists for actors to introduce elections must be analyzed separately from the elections functions (Brancati 2014). The problem with discussing the introduction of elections using research that regards elections as dependent variables is that they consider elections as being imposed from outside the government. There may be an incentive for introducing elections. As has been pointed out by previous research, 4 For more information on what kind of foreign policy is used, see Escriba-Folch and Wright (2015: 9) and others. 5 Various kinds of aid exist, but I do not consider these differences. 5

6 functions exist that can be used as tools in elections. Room exists for the consideration that not only imposition from the outside is occurring but also for the possibility that elections are introduced as spontaneously desired. In order to overcome these problems, the next section creates a theory based on incentives for authoritarian leaders to introduce elections. Then, I demonstrate that the introduction of elections can be explained by whether or not the leader monopolizes power, and then I derive a hypothesis. 3. Theory and Hypothesis In this section, I construct a theory in which the introduction of elections is decided by the leader s power monopoly, and then I derive a hypothesis. First, I assume the political goal of the authoritarian leader, who is the most important actor under authoritarianism. Under the authoritarian rule, the notion that elections will be introduced without a decision from the leader is unlikely. Next, I note that it is necessary for the leader to make his or her regime a monolith in order to achieve his or her goal and refer to the concept of the leader's power monopoly. Then, I discuss the usefulness of the election for leaders who do not monopolize power and present a hypothesis for empirical analysis. 3.1 Goals of the Authoritarian Regime Leader The leader's goal under the authoritarian regime is to defend their political position (Bueno de Mesquita et al. 2003). To achieve this, they are required to maintain the authoritarian regime and hold tight onto the leadership position. For leaders to achieve their goals, they must ask the inner-elites for cooperation in governance. Innerelites can be the largest threat to leaders. According to previous studies, leaders lose their positions mostly due to coups from the inner-elites (Aksoy et al. 2015; Svolik 2012). Additionally, sometimes the elite even causes the collapse of the regime itself. To get the cooperation of the inner-elites, leaders must demonstrate that their rule is useful for the elite to hold a privileged position. If cooperating with the leader is not effective at protecting their status, the elites may cause a coup to change leaders Monolithic Regime and Methods of Achieving It Leaders want to prevent elites from challenging them. If the inner-elites need the leader s rule to protect their privileged position, leaders are not threatened by the elite. In this situation, the 6 In this paper, the means used by the elite to expel leaders is called coup (Aksoy et al. 2015). 6

7 government is monolithic under his or her rule. Two methods exist to make a regime into a monolith. One is appealing the leader s great power. The great power urges inner-elites to rally together with the leader The Soviet Union under the rule of Stalin would be an example. A communist party elite would believe that it is a necessary condition that Stalin s rule continue in order to preserve its own position. Under this circumstance, the authoritarian regime is expected to be a monolith without any efforts from leaders. The other method is introducing an election. If outside actors challenge government, inner-elite have to unite in order to surmount the challenge from outside the government. Even when the inner-elite does not recognize that the current leader is essential to the survival of the regime, the elites must concentrate on overcoming challenges from the outside in order to preserve their own position. They are supposed to cooperate with the leader. The regime is expected to become a monolith under the current leader in order to overcome the regime crisis. One example of such a crisis is an electoral challenge. Electoral introduction enables the leader to avoid an interior fissure and maintain their position. To make a regime into a monolith, two other methods may be utilized. One includes diverting resources to the elite. This introduces a large problem in terms of credibility. Usually, leaders have greater power than the elite, so leaders can go back on their promises to distribute resources to the elite. Thus, the leader has to tie his or her own hands to secure the credibility of the promise (Boix 2013; Magaloni 2006, 2008; Svolik 2012 etc.). This cannot overcome the credibility problem. The second way is by constructing a democratic institution, such as political parties. Literature has pointed out that power sharing between leaders and elites will be realized through political parties (Svolik 2012). In other words, power-shared authoritarianism creates a monolithic regime. In fact, authoritarian regimes with political parties are empirically known to survive longer than other types of authoritarian regimes (Geddes 1999, 2003; Svolik 2012). This seems, at first glance, to be evidence that party-controlled authoritarian regimes are monolithic. However, being a party-controlled regime is not a sufficient condition. Even if the existence of a party makes it possible for a regime to last long, this does not guarantee the survival of the leader himself/herself. Under the party-controlled regime, the leader's head may be replaced. Parties enable the maintaining of not just the leader s position but also the regime. Thus, to make a regime into a monolith and continue the leader s political survival, rallying the power of the leader or introducing the threat (election) from outside the government is still necessary. 7

8 3.3 Leader and Power Monopoly According to decision-power monopolies, a leader under an authoritarian regime can be categorized as one of two types. One is a leader who monopolizes power in the government (power monopoly (PM) leader), and the other is a leader who does not monopolize power (power restriction (PR) leader). The action taken by the leader to create a monolithic regime depends on whether a leader is PM or PR. On the one hand, under the PM leader, the elite will spontaneously rally because they have the incentive to do so from the elite. The survival of a regime with a leader who monopolizes power means that the success of the regime is maintained by the performance of the present leader. As a result, the elite should recognize that rallying under the leader is the most effective choice to preserve their own privilege. Thus, supported by the voluntary cooperation of the elite, the leader does not need to bother introducing an election to make the regime into a monolith. On the other hand, under the PR leader, believing that the elite gather together voluntarily is difficult because the elite do not have the incentive to do so. The survival of a regime with a leader who does not monopolize power means that the success of the regime is not maintained by the performance of the present leader. As a result, the elite should recognize that rallying under the leader is not the effective choice to preserve their own privilege. Thus, if a leader cannot obtain voluntary cooperation, he or she will need to introduce elections to make the regime into a monolith. Figure 1 summarizes the above discussion. YES (PM Leader) Method Appealing Power Power Monopoly NO (PR Leader) Introducing Election Figure 1 Leader s Power Monopoly and Methods Indeed, achieving a monolith is not the only benefit that an election brings. As previous studies have shown, elections are a useful tool for rule. Thus, it seems that even PM leaders have an incentive to introduce elections. However, elections are accompanied by two costs. The first cost is electoral uncertainty. As pointed out by previous research, authoritarian regimes may collapse by electoral turnover. Of course, as part of the electoral competition under authoritarianism, the ruling party (incumbent) is in a very advantageous position (Bunce and Wolchik 2011; Greene 2007; Levitsky and 8

9 Way 2010). However, Huntington sounds a warning bell that the possibility of failure due to the introduction of elections exists (Huntington 1991). The second cost is the entry of new actors. Holding an election means that outside actors, such as citizens and anti-government individuals, acquire the right to political participation. Thus, with the electoral introduction, the number of actors who leaders must keep in mind in order to maintain their position is expected to increase. These two costs inhibit rock solid governance. Through the election, leaders suffer a great cost. Election is a double-edged sword for leaders. If a leader monopolizes power, no need remains to use elections to make the regime into a monolith. Little incentive exists to introduce elections, and doing so can cause leaders to suffer a great cost. Contrary to the PM leader, it is challenging for the PR leader to make the regime into a monolith without elections. Thus, the PR leader must have an incentive to introduce an election, even if it brings with it a cost. The following hypothesis is derived from this difference: H: Under the rule of the PR leader, elections are likely to be introduced. In the next section, I conduct an empirical analysis to test this hypothesis. 4. Empirical Analysis In this section, I analyze the hypothesis derived in the previous section. First, I refer to data and methods used for empirical analysis. Next, the variables (independent, dependent, and control) are described in detail. Then, I provide the results. Based on the results of the analysis, the leader's power monopoly clearly influences the introduction of elections. 4.1 Data and Method To identify the analytical scope, I rely on GWF Autocratic Regimes data (Geddes et al. 2014). The authoritarian regimes to be analyzed are those that existed from 1946 to However, the regimes classified as military regimes in GWF are excluded from this analysis. The reasons are as follows. First, military leaders may have incentives that are different from leaders of other authoritarian regime types. In a military regime, maintaining the regime is not an obvious goal. Sometimes retiring from the political arena and returning to the barracks is rational (Finer 1985; Geddes 1999; Geddes et al. 2014; Nordlinger 1977). Thus, in military rule, it cannot be assumed that a leader wishes to maintain their position. Second, the meanings of elections may differ. In military rule, elections are sometimes 9

10 used as a means of transferring power to civilian rule. This election is not done to maintain the leader s position and make the regime into a monolithic. Thus, it is possible that these elections may have atypical features. Based on these two reasons, explaining the electoral introduction under military regimes based on the incentives of leaders is inappropriate. As an empirical method, I adopt panel-logistic regression analysis using a dummy variable as the dependent variable for the time series cross section data (TSCS). When conducting this analysis, it is considered that time dependency should be dealt with like event history analysis (Beck et al. 1998). Thus, the duration of the regime, the squared term of the duration, and the cubed term of the duration are controlled in the model, respectively (Carter and Signarino 2010). Some regimes analyzed in this paper have invariant dependent variables throughout the period, that is, they either fall or are censored without introducing elections or fall or are censored in a year. Thus, estimation using the fixed effect method may exclude many observations and cause bias to occur (Beck and Katz 2001). In this paper, I control characteristics of each regime using one of the following methods. The first method is that, in addition to controlling regional dummies, 7 I use standard error clustered in each unit. The second method is to intimate the fixed effect method by inputting the mean value for each independent variable and control variable (Escriba-Folch et al. 2015). 4.2 Independent Variable This paper theorizes that the power monopoly of the leader influences the introduction of elections. In this paper, I measure the degree of power monopoly by the degree of personalism variable (personal2) created by Magaloni et al. (2013), and I use it as an independent variable. This variable is converted based on the PolityIV s executive constraint variable (XCONST) (Marshall and Jaggers 2002). 8 Thus, I first describe XCONST, and then, I introduce personal2. XCONST is one of the components of the POLITY index that is frequently referred to in distinguishing democratic and authoritarian regimes. This variable evaluates the extent of institutionalized 7 Regional dummies are Europe (reference category), Latin America, Sub-Saharan Africa, Asia, the Middle East, and North Africa (Wimmer et al. 2009). In Wimmer et al. (2009), although Western Europe and Eastern Europe are separated, they were combined into Europe to avoid small observations. 8 Lai and Sltaer (2006) s study can be cited as one that measures power monopoly degree by XCONST. As a similar variable, a variable (gwf_personal) exists that was created by Geddes (Geddes 1999, 2003; Geddes et al. 2014). However, since this variable does not focus on power monopoly level, it is not always the case that the degree of power is high, even though Geddes regards the regime as personal. Thus, it is inappropriate as an independent variable in this paper. 10

11 constraints on the decision-making powers of chief executives, whether individuals or collectivities (Marshall et al. 2015: 23). The magnitude of institutionalized constraints to the leader's decision is measured on seven levels, from 1 to 7. The smaller the number, the smaller the constraint, and the larger the number, the larger the constraint. A leader whose XCONST is 1 experiences no regular limitations on the executive's actions (Marshall et al. 2015: 24). Conversely, a leader whose XCONST is 7 is constrained by accountability groups that have effective authority equal to or greater than the executive in most areas of activity (Marshall et al. 2015: 24). It can be said that the magnitude of institutional constraint on the leader illustrates the extent of monopoly of power of the leader as it is. The personal2 variable to operationalize the power monopoly in this paper is that the degree of personalism is 2 only when XCONST is 1, the degree of personalism is 1 when XCONST is 2 to 4, and the degree of personalism is 0 when XCONST is 5 to 7. 9 In this paper, personal2 rather than XCONST is set as the independent variable for the following two reasons. Firstly, several data points exist that can be used for hypothesis testing. Personal2 fills the missing value of XCONST. 10 XCONST disappears when it is considered to measure a transitional period to democracy. In verifying the hypothesis of this paper, being able to observe the power monopoly during this period is important. Elections that are introduced are regarded as transition periods in many cases, but some regimes rule without elections despite being part of a transitional regime. Whether elections are introduced in the transition period may be explained by the leader's power monopoly. Therefore, if XCONST, which excludes the transitional period, is set as the independent variable, it decreases the data points that can be used for hypothesis testing. 11 Secondly, it is in accordance with the theory of this paper. Accurately, the theory focuses not on the degree of the leader s monopoly of power but on whether the leader monopolizes power. With the personal2 variable, this can be classified as a PM leader when the personalism level is 2 and a PR leader when the personalism level is 1. In addition, it is a power quite restricted (PQR) leader when the personal level 0. If XCONST is used, it is assumed that elections are likely to be introduce linearly as XCONST increases from 2 to 4. From the viewpoint of power monopoly, however, it does not matter whether the constraint is 2 or 4. Thus, the personal2 variable better conforms to the theory Since personal2 has data only after 1950, during the period from 1946 to 1949, XCONST is converted and supplemented according to Magaloni et al. (2013). 10 Personal2 fills in the missing value with that value if XCONST is constant for a whole period in a regime. Otherwise, it is filled in by the average value immediately before and immediately after the missing value (Magaloni et al. 2013: 4). 11 Handling of missing values in Magaloni et al. (2013) may not be appropriate. Analysis is also conducted with the missing of XCONST left. Except for one model, the independent variable is still significant. See Appendix 2 for results. 12 I also analyze that the PR leader and the PQR leader are treated as the same value. However, the difference in results is small. See Appendix 3 for results. 11

12 Table 1 shows the relationship between personal2 and the election experience of the authoritarian regimes. The relationship that as personal2 becomes lower, the regime is likelier to experience elections is obvious. However, two points should be retained in this relationship. First, it is possible that the election experience prevents the leader's monopoly of power, that is, it may be due to reverse causality. Second, it is possible that this is a spurious correlation. In order to confirm this relationship, I conduct a quantitative analysis with personal2 (t-1) as an independent variable while controlling the variables described later. From Table 1, a few leaders exist whose personal2 is 0. Theoretically, however, it is doubtful whether these leaders have the decision power to introduce elections. Such a leader may be forced to introduce elections for some reason that the theory of this paper does not suppose. I also carry out analysis excluding these PQR leaders. Table 1 Relationship Between personal2 and Election Experience personal2 Inexperienced Experienced Total 2 (PM Leader) , % 27.18% 100% 1 (PR Leader) 986 1,065 2, % 51.93% 100% 0 (PQR Leader) % 78.13% 100% Total 2,008 1,648 3, % 45.08% 100% 4.3 Dependent Variable The dependent variable is an election year dummy. This variable is a dummy variable that takes on 1 if an election is held in that year. As mentioned above, I focus only on elections that meet minimum competitive requirements (competitive elections). This uses the National Elections Across Democracy and Autocracy Dataset (NELDA) (Hyde and Marinov 2011). 13 In the analysis of this paper, the authoritarian regime that introduces the election once is excluded in that year. This is because one regime avoids holding multiple elections. If it includes the second and subsequent elections in the analysis, the estimation results may be distorted due to the following 13 NELDA considers elections that satisfy all of the following three requirements as competitive elections: the opposition is legal (NELDA 3), two or more political parties are legal (NELDA 4) and voters have options (NELDA 5). 12

13 possibilities. First, the possibility exists that the election has path dependency. In many countries, elections are to be repeated once every several years. Thus, the implementation of the election may be considered natural and be difficult to abolish. Factors distinct from the first election (electoral introduction) may affect the second and subsequent elections. Second, the possibility exists that the election itself has the potential to affect power monopoly. This paper is interested in institutionalized constraints on leader decision making. It can be said that the election is one of the typical institutionalized constraints. If second and subsequent elections are included, the study may exhibit reverse causality. Combined with the above possibilities, when the first election is carried out, the degree of power monopoly decreases, and the second election is carried out by path-dependence. Thus, if observation is not finished in the first election (electoral introduction), the study could possibly demonstrate a distorted result in a favorable direction. In order to avoid this, the authoritarian regime that introduces the election once is excluded that year. 4.4 Control Variable As for the control variable, several variables that the literature pointed out as influencing regime collapse and introduction of elections are thrown into analysis. See Appendix 1 for details on control variables. First, as a variable for measuring demands from outside the government, I use the logconverted aid amount and the number of social movements using a three-year average. Social movement refers to the sum of the number of strikes, riots and anti-government demonstrations coded by Banks (2016) (Bueno de Mesquita and Smith 2010; Higashijima 2015; Przeworski 2009). I control GDP per capita (log) and growth rate. Several studies exist on the effects of regime collapse as well as on democratization (Boix 2011; Boix and Stokes 2003; Lipset 1959; Przeworski et al. 2000, etc.). If abundant resources exist that the authoritarian regime can use, it becomes easy to co-opt the political opposition and maintain the regime. Additionally, using resources, incumbents can fight elections advantageously. The regime that is endowed with ample resources should not consider the risk of being defeated by an election. Thus, it is expected that elections are more likely to be introduced, as the resources are larger. The size of resources is controlled by logarithmically converted per capita oil and gas (Smith 2004; Ross 2001, 2012; Wright et al. 2015). I control the degree of ethnic fractionization. High ethnic fractionization may facilitate the collapse of the regime (Przeworski et al. 2000). All control variables thus far are used in the previous year's values. I control after the Cold War. The reality exists that EA regimes have expanded since the Cold War (Howard and Roessler 2009; Kricheli and Magaloni 2010; Schedler 2013). I introduce a post-cold War dummy variable that takes on 1 if it is after Furthermore, it regulates the election experience of the previous regime. If the former regime experienced elections, it is expected that elections will 13

14 likely be introduced in the next regime. Thus, if the former regime is a democracy or EA regime, I control the variable, which takes on 1. It also controls colonial experience, which elections are considered likely to introduce, depending on the character of the former colonial country. For example, in Sub-Saharan Africa, elections were held in countries that were under the control of Britain and France (Hermet 1978: 11). Thus, with reference to Teorell (2010), I control British/French colonial dummies. Finally, I control the subtypes of the authoritarian regime 14 because it is known that a difference in the average lifetime exists depending on the subtype. Several studies have pointed out, in particular, that the party-controlled authoritarian regime is durable (Brownlee 2007; Geddes 1999; Smith 2005). This is because the foundation of the regime is solid. If a solid foundation exists, it affects not only the stabilization of the regime but also electoral competence. 4.5 Results Results are shown in Table 3. If the coefficient of each variable is positive, its variable facilitates electoral introduction, and if it is negative, its variable restrains electoral introduction. Model 1 examines the relationship between power monopoly and electoral introduction without control variables. Model 2 controls demand by outside government variables (aid and social movements). Model 3 controls all variables. Model 4 includes all the control variables after excluding leaders whose personal2 value is 0. For Models 5 and 6, the same analysis used for Models 3 and 4 was carried out by controlling unit average. In Models 5 and 6, variables that take constant values over the entire period in a certain regime are excluded. This difference is statistically significant Prior to reporting the coefficient of the independent variable, I refer to the coefficient of some control variables. First, it is revealed that as aid increases, the election is more likely to be introduced. This effect is statistically significant at the 5% level in all models. The increase in social movements also facilitates the electoral introduction statistically significantly in Models 3 and 4. However, in Models 5 and 6, which controlled the unit average number of social movements, the effect was not statistically significant. In other words, it seems that increasing or decreasing social movements within a regime does not affect the introduction of elections. Although, for the most part, it is likely that a great demand from outside the government tends to direct the authoritarian regime toward the introduction of elections. 14 An authoritarian regime is classified as a subtype of party, personal (reference category), military or monarchy (Geddes et al. 2014). In this paper, each regime is classified into one of the three types due to the exclusion of military regimes. 14

15 The post-cold War dummy raises the probability of introducing elections significantly in all models. It seems that a change in international politics has prompted the introduction of elections. The election experience of the previous regime and the British and French colony dummy do not have a significant effect on the introduction of elections. In other words, even if the former regime had experienced elections, it is not necessary to ensure the orthodoxy of governance through elections in that regime. Subtypes of the authoritarian regime do not have a significant effect on election introduction probability. Whether it is a party-controlled regime does not affect the probability of election introduction. Next, we examine the effect of the leader s power monopoly as measured by personal2, which is an independent variable in this paper. From Model 1, which is without a control variable, to Model 6, which controls the unit average value of all variables, power monopoly is statistically significant negatively at the level of 5% or less for each model. This shows the result that the election is difficult to introduce as the leader dominates decision making power. In other words, elections are more likely to be introduced under the PR leader than the PM leader. Finally, I show the marginal effect of changing from a PM leader to a PR leader. In order to clarify the difference between PM and PR, the result of Model 4, excluding the leader whose personal2 value is 0, is used. When the average value is taken for all the control variables, the probability of electoral introduction under the PM leader is 1.7%. With the PR leader, the probability of electoral introduction raises to 4.5%. This difference is statistically significant at the 5% level. All other things being equal, if the leader cannot monopolize power, electoral introduction is over 2.5 times easier. 15

16 Table 2 Estimation Results (1) (2) (3) (4) (5) (6) Regime All All All excl. PQR All excl. PQR Degree of Power Monopoly *** *** ** ** ** *** (0.255) (0.227) (0.321) (0.308) (0.673) (0.668) Aid (log) 0.334** 0.314** 0.510*** 0.475** (0.136) (0.137) (0.185) (0.217) Movements 0.264*** 0.274** (0.101) (0.122) (0.147) (0.161) Oil and Gas (log) (0.071) (0.077) (0.089) (0.211) (0.275) GDP per capita (log) (0.272) (0.297) (0.356) (0.684) (0.694) GDP Growth (0.013) (0.012) (0.012) (0.011) (0.012) Ethnic Fractionazation (0.675) (0.614) (0.725) (56.98) (75.64) Post Cold War 1.626*** 1.118*** 1.318*** 1.298** 1.355*** (0.341) (0.393) (0.366) (0.513) (0.314) Previous Regime (0.328) (0.308) (0.328) UK/French Colony (0.422) (0.428) (0.412) Party Regime (0.295) (0.280) (0.324) Morarchy Regime (0.772) (0.999) (1.042) Latin America 1.565** (0.708) (0.655) (0.780) Sub-Saharan Africa *** *** (0.577) (0.642) (0.705) Asia ** ** (0.656) (0.813) (0.927) MENA *** ** (0.756) (0.698) (0.743) Constant *** *** ** (0.320) (2.242) (2.314) (2.730) (2.242) (2.368) Duration^1, 2, 3 NO YES YES YES YES YES Unit Average NO NO NO NO YES YES N 2,119 1,728 1,324 1,281 1,324 1,281 N Regime Robust standard error in the parentheses. *** p<0.01, ** p<0.05, * p<0.1 16

17 5. Conclusion In this paper, I sought answers for the question of what kind of authoritarian regime introduces elections. I theorized that a leader who does not monopolize power has an incentive to introduce elections to make the administration a monolith and hypothesized that, under this leader, an authoritarian regime is likely to introduce elections. The results of the statistical analysis supported this hypothesis. This paper revealed that elections are not only a concession to demands from outside the government but also a phenomenon that can be voluntarily introduced by leaders. The discussion developed in this paper has an important implication. It is the provision that the cause of electoral introduction can be classified. The election under the authoritarian regime is what is requested from outside the government and what is desired by the leader. Classification of elections due to these introduction causes may suggest a solution to the question of why some elections collapse regimes, whereas other elections maintain regimes. This solution may be a helpful clue to the unresolved controversy of how the election affects the fate of the authoritarian regime. Due to this solution, this paper can contribute to electoral research under authoritarian regimes. However, this paper leaves the following two problems to be solved. First, this paper does not distinguish the type of election. I consider that an election is introduced when either a presidential or parliamentary election is held. In many cases, both types of elections are introduced at the same time. For example, Zambia introduced presidential and parliamentary elections simultaneously in However, authoritarian regimes like the one in Jordan, which only introduced parliamentary elections, also exist. The effect of power monopoly may differ between the introduction of presidential elections and the introduction of parliamentary elections. Analyzing while considering the type of election is necessary. Second, this paper does not consider the inner-elites attitude about the election. In this paper, I assume that the inner-elite does not resist the electoral introduction. However, for the elite, extra competition is imposed by the electoral introduction. Elections may threaten their political privileges. If so, it is possible that the inner-elite opposes the introduction of elections. Constructing a theory incorporating not only the leader but also the attitude of inner-elite is necessary. 17

18 Appendix 1 Descriptive Statistics Variable N Mean S.D Min Max Source Degree of Power Monopoly Magaloni et al. Aid (log) WDI Movements Banks Oil and Gas (log) Ross GDP per capita (log) Maddison GDP Growth Maddison Ethnic Fractionazation EPR Post Cold War GWF Previous Regime NELDA UK/French Colony Teorell Party Regime GWF Morarchy Regime GWF Personal Regime* GWF Latin America EPR Sub-Saharan Africa EPR Asia EPR MENA EPR Europe* EPR *Reference Category. Banks: Banks, A, S., Wilson, K, A. (2015). Cross-National Time-Series Data Archive. Databanks International. Jerusalem, Israel. EPR3: Ethnic Power Relations 3.0. GWF: GWF Autocratic Data Sets. Maddison: The Maddison-Project, 2013 version. Magaloni et al.: Autocracies of the World Dataset NELDA: National Elections Across Democracy and Autocracy Polity IV: Polity IV project. Ross: Ross, M. (2011), "Replication data for: Oil and Gas Production and Value, ", Harvard Dataverse, V4. Teorell: Teorell, J. (2010). UCDP: UCDP/PRIO Armed Conflict Dataset. WDI: World Development Indicator. 18

19 Appendix 2 Estimation Results (Leaving Missing Value of XCONST) (1) (2) (3) (4) (5) (6) Regime All All All excl. PQR All excl. PQR Degree of Power Monopoly *** *** ** ** ** (0.308) (0.262) (0.317) (0.351) (0.848) (0.811) Aid (log) 0.407*** 0.431** 0.412* (0.155) (0.176) (0.218) (0.221) Movements 0.359** 0.464** (0.171) (0.190) (0.223) (0.240) Oil and Gas (log) (0.078) (0.086) (0.095) (0.261) (0.271) GDP per capita (log) (0.306) (0.338) (0.377) (0.806) (0.891) GDP Growth (0.013) (0.013) (0.012) (0.010) (0.010) Ethnic Fractionazation (0.779) (0.723) (0.795) (60.48) (65.39) Post Cold War 1.564*** 1.065*** 1.278*** 1.270*** 1.353*** (0.336) (0.373) (0.434) (0.355) (0.369) Previous Regime (0.327) (0.321) (0.358) UK/French Colony (0.465) (0.451) (0.489) Party Regime (0.316) (0.297) (0.373) Morarchy Regime (0.838) (1.377) (1.516) Latin America (0.847) (0.685) (0.811) Sub-Saharan Africa * *** *** (0.601) (0.624) (0.738) Asia *** *** (0.689) (0.812) (0.969) MENA *** *** (0.771) (0.783) (0.828) Constant *** * *** (0.387) (2.515) (2.686) (2.794) (2.507) (2.555) Duration^1, 2, 3 NO YES YES YES YES YES Unit Average NO NO NO NO YES YES N 2,019 1,634 1,244 1,218 1,244 1,218 N Regime Robust standard error in the parentheses. *** p<0.01, ** p<0.05, * p<0.1 19

20 Appendix 3 Estimation Results (PR/PQR Leader Dummy) (1) (2) (3) (4) Regime All All All All PR/PQR Leader *** *** * ** (0.361) (0.307) (0.466) (0.893) Aid (log) 0.343*** 0.532*** 20 (0.133) (0.185) Movements 0.274** (0.107) (0.166) Oil and Gas (log) (0.071) (0.077) (0.293) GDP per capita (log) (0.266) (0.295) (0.791) GDP Growth (0.012) (0.012) (0.012) Ethnic Fractionazation (0.700) (0.686) (73.48) Post Cold War 1.762*** 1.209** (0.360) (0.538) (2.227) Previous Regime (0.328) (0.347) UK/French Colony (0.443) (0.545) Party Regime (0.315) (0.285) Morarchy Regime (0.828) (1.010) Latin America 1.404* (0.729) (0.675) Sub-Saharan Africa *** (0.580) (0.779) Asia ** (0.682) (0.897) MENA ** (0.776) (0.922) Constant *** (0.204) (2.144) (2.362) (0) Duration^1, 2, 3 NO YES YES YES Unit Average NO NO NO YES N 2,119 1,728 1,324 1,324 N Regime Robust standard error in the parentheses. *** p<0.01, ** p<0.05, * p<0.1

21 References Acemoglu, D., & Robinson, J. A. (2005). Economic origins of dictatorship and democracy. Cambridge University Press. Aksoy, D., Carter, D. B., & Wright, J. (2015). Terrorism and the Fate of Dictators. World Politics, 67(03), Ansell, B. W., & Samuels, D. J. (2014). Inequality and Democratization. Cambridge University Press. Beck, N., & Katz, J. N. (2001). Throwing out the baby with the bath water: A comment on Green, Kim, and Yoon. International Organization, 55(02), Beck, N., Katz, J. N., & Tucker, R. (1998). Taking time seriously: Time-series-cross-section analysis with a binary dependent variable. American Journal of Political Science, Blaydes, L. (2011). Elections and distributive politics in Mubarak s Egypt. Cambridge University Press. Boix, C. (2003). Democracy and redistribution. Cambridge University Press.. (2011). Democracy, development, and the international system. American Political Science Review, 105(04), Boix, C., & Stokes, S. C. (2003). Endogenous democratization. World politics, 55(04), Brancati, D. (2014). Democratic authoritarianism: origins and effects. Annual Review of Political Science, 17, Bratton, M., & Van de Walle, N. (1997). Democratic experiments in Africa: Regime transitions in comparative perspective. Cambridge University Press. Brownlee, J. (2007). Authoritarianism in an Age of Democratization. Cambridge University Press.. (2009). Portents of pluralism: How hybrid regimes affect democratic transitions. American Journal of Political Science, 53(3), Bueno de Mesquita, B., & Smith, A. (2009). Political survival and endogenous institutional change. Comparative Political Studies, 42(2), Bunce, V., & Wolchik, S. L. (2006). Favorable conditions and electoral revolutions. Journal of Democracy, 17(4), (2009). Opposition versus dictators: explaining divergent election outcomes in post- Communist Europe and Eurasia. In Lindberg, S. I. (2009). Democratization by elections: a new mode of transition. Johns Hopkins University Press, (2011). Defeating authoritarian leaders in postcommunist countries. Cambridge University Press. 21

22 Carter, D. B., & Signorino, C. S. (2010). Back to the Future: Modeling Time Dependence in Binary Data. Political Analysis, 18, Cederman, L. E., Wimmer, A., & Min, B. (2010). Why do ethnic groups rebel? New data and analysis. World Politics, 62(01), Cox, G. W. (2009). Authoritarian elections and leadership succession, Dietrich, S., & Wright, J. (2014). Foreign aid allocation tactics and democratic change in Africa. The Journal of Politics, 77(1), Escribà-Folch, A. (2013). Accountable for what? Regime types, performance, and the fate of outgoing dictators, Democratization, 20(1), Escriba-Folch, A., Meseguer, C., & Wright, J. (2015). Remittances and democratization. International Studies Quarterly, 59(3), Escribà-Folch, A., & Wright, J. (2015). Foreign pressure and the politics of autocratic survival. Oxford University Press. Gandhi, J. (2008). Political institutions under dictatorship. Cambridge University Press. Gandhi, J., & Lust-Okar, E. (2009). Elections Under Authoritarianism. Annual Review of Political Science, 12, Geddes, B. (1999). What do we know about democratization after twenty years? Annual review of political science, 2(1), (2003). Paradigms and sand castles: Theory building and research design in comparative politics. University of Michigan Press.. (2005). Why Parties and Elections in Authoritarian Regimes? Annual Meeting of the American Political Science Association, (March), Geddes, B., Frantz, E., & Wright, J. G. (2014). Military rule. Annual Review of Political Science, 17, Geddes, B., Wright, J., & Frantz, E. (2014). Autocratic breakdown and regime transitions: A new data set. Perspectives on Politics, 12(02), Hermet, G. (1978). Elections without choice. Springer. Huntington, S. P. (1991). The third wave: Democratization in the late twentieth century. University of Oklahoma press. Hyde, S. D., & Marinov, N. (2011). Which elections can be lost? Political Analysis, 20, Joseph, R. (1997). Democratization in Africa after 1989: Comparative and theoretical perspectives. Comparative politics, Kim, N. K. (2016). Anti-regime Uprisings and the Emergence of Electoral Authoritarianism. Political Research Quarterly. Knutsen, C. H., Nygård, H. M., & Wig, T. (2017). Autocratic Elections: Stabilizing Tool or Force for Change? World Politics, 69(1),