Electoral Violence: Comparing Theory and Reality

Transcription

1 Electoral Violence: Comparing Theory and Reality Gregory Wallsworth September 25, 2015 Abstract Various studies on election violence in Africa have found violence to mar as high as eightypercent of African elections. Using Round 5 Afrobarometer survey data from 35 African countries in 2011, 48 percent of respondents reported some fear of election violence. This paper examines the impact of election violence on voting behavior through the lens of several theoretical models. We find some support for a key assumptions made by these models, election violence is associated with a reduction in voter turnout, particularly among undecided voters. However, we do not find strong support for key predictions made by any of the models. We argue this is due in part to an inability for current data to determine the perpetrator of election violence, and conclude that a variety of explanations for election violence each partially explain its occurrence. 1 Introduction Elections are intended to provide a peaceful method for groups to compete for political power, sometimes directly replacing violent alternatives. However, in situations when political parties have not yet accepted the rules of the game, violence is theorized to be an effective tool to influence election results. (Machado et al. (2011), Collier and Vicente (2012), and Robinson and Torvik (2009) represent a few such theories.) Furthermore, the clear demarcations drawn by political parties, especially when formed by already contentious groups can make conditions ripe for violence to emerge, a subject of interest in its own right (Chandra, 2001; Reilly, 2002). We argue, in a similar vein to the theories we examine, that visible group markers can make violence more effective at manipulating election results, this implies that groups formed along such lines, like ethnicity, are more prone to election violence. Although many theories of election violence exist, they and the effectiveness of electoral violence, remain largely untested (Hickman, 2009). This paper, like others, such as Goldsmith (2014) and Höglund (2009), focuses on election violence, as separate from other types of political violence, and starts with the following broad definition of election violence from (Fischer, 2002): Any random or organized act or threat to intimidate, physically harm, blackmail or abuse a political stakeholder in seeking to determine, delay or to otherwise influence an electoral process. With such a broad definition of electoral violence and so many possible reasons for its occurrence, it is necessary to refine the focus of electoral violence for this paper. We choose to focus on violence as a tactic of voter manipulation, and examine if patterns of electoral violence are consistent with the strategic use of violence, to influence the outcome of an election by the purposeful targeting and exclusion of particular groups of voters. This type of electoral violence has been the subject 1

2 of significant and varied academic work, such as case studies (find some citations), empirical work Hickman (2009), and the two theoretical papers we focus on Chaturvedi (2005), hereafter CH, and Collier and Vicente (2014), hereafter CV. We compare the assumptions and predictions made in these theories, to the patterns observed using Afrobarometer survey data and the Social Conflict in Africa Database (SCAD). Before examining whether observed patterns of electoral violence are consistent with theory, we first examine the key assumptions about how individuals respond to election violence. The analysis in this paper expands on previous country specific work in papers like Bratton (2008), Hickman (2009), and Collier and Vicente (2014) by using a broader sample of countries and examining heterogeneity in response to, and fears of, violence. Understanding how electoral violence influences behavior in elections is a fundamental part of constructing accurate models of election violence. The Afrobarometer survey data allows us to analyze 20 countries using round 4 and 34 countries using the round 5 data. The survey data is used to measure perceptions of violence, political participation, and political affiliation. The measures of violence are compared to more traditional event count data and demonstrate a strong correlation but are clearly not measuring the same phenomenon: events need not occur near an individual for them to fear violence. We argue that one measure in particular, that captures and individual s fear of victimization, Exp Violence, is well suited to analyzing who is impacted by violence. This variable is shows significant deviations even within small geographical areas where all individuals are likely to observe the same occurences of violence. Furthermore, groups which theory predicted to experience greater levels of election violence, such as those belonging to ethnically aligned political parties, exhibit a greater fear of violence with this measure, regardless of observed level of violence. This is among the first papers to use survey measures of violence and we argue this provides several potential advantages over traditional event count measures, especially in circumstances where the fear or threat of violence may be enough to induce the intended behavior, as we will demonstrate is the case for election violence. We find swing voters are the most reactive to political violence, they are the only group who will abstain from voting in the presence of violence even if they do not exhibit a greater personal fear of violence. We also find that direct targeting of violence is a potentially effective strategy to cause absention, as regardless of political affiliation personal fear of violence leads to a reduced likelihood of voting. Finally, we find that members of relatively small political groups and ethnically aligned political parties exhibit the greatest fear of violence in elections. All of these conclusions are consistent with the assumptions in the theoretical models we examine. When we turn to examining the predictions of the models the results are less clear, the key predictions of the models we focus on make stronger arguments for who should perpetrate violence than its occurrence. Taken literally, violence will always occur in both simplified versions of the models we discuss, though this is largely due to the exclusion of alternatives. Nonetheless, we find much stronger support for CV than for CH. A key prediction is that as the fraction of swing voters increases, the occurrence of violence should increase in CV and decrease in CH, we find evidence showing that violence increases with the fraction of swing voters in the electorate. However, the CV model is not the only potential explanation for such a pattern, if violence and swing voters are both correlated with the competitiveness of an election we may expect the same result. This paper makes considerable progress in explaining who reacts to election violence, and how. Given the prevalence of violence in elections, and the considerable debate over how electoral engineering, can be used to help mitigate violence in elections, having a thorough understanding of how violence influences electoral behavior is important if we are to try to mitigate its impact 1. We 1 For a thorough comparison of the two leading theories of electoral engineering, and accompanying examples of 2

3 argue that policies should encourage political parties to form across existing societal divides, and that better data should be gathered on election violence in order to examine if, as is predicted by these theories, division of the electorate is an important determinant of election violence. Furthermore, suggestive evidence supports the idea that even in societies where there may be elections, the use of election violence may be preventing serious opposition from forming. International responses should be harsh when such obvious infractions are observed, or this potentially effective strategy of violence is likely to continue. The structure of this paper is as follows: the following section examines the existing academic literature related to electoral violence. Section III examines the predictions of two theoretical models in detail and presents potential hypotheses. Section IV presents the data, and Section V discusses the empirical results. Section VI concludes and suggests paths for future work. 2 Literature Review Elections play a fundamental role in democratic governance, They [elections] provide legitimacy through direct popular participation, and, in turn, legitimacy creates capacity for effective governance (Brown, 2003). Violence in this process then has the capacity to undermine not only the elections, but also the legitimacy of the government itself. A significant amount of research has examined the use of elections in post-conflict societies; however, as the Afrobarometer data, and previous research show, some violence is quite common in African elections (Lindberg, 2006), (Straus and Taylor, 2009), and (Goldsmith, 2014). Given the importance of elections, and the high degree of suffering violence can cause, it is not difficult to motivate the need to understand electoral violence. 2.1 The Role of Elections In post-conflict societies elections are often intended to trade bullets for ballots ; however, as pointed out in Rapoport and Weinberg (2000), succession is the most turbulent time for any type of government. A great deal of research has considered the consequences of electoral engineering, or how an electoral system can mitigate or encourage the use of violence based on how groups in society end up represented. Violence is not the only dimension affected by the choice of electoral system, as pointed out in (Norris, 1997) raises another important issue: Electoral rules are not neutral. In her paper, Norris compares the advantages and disadvantages of different types of electoral systems. She highlights some now well known facts that Majoritarian systems can manufacture majorities in legislative bodies with only small swings in the percentage of the votes grabbed by parties. A key condition for recurrent civil conflict discussed in Walter (2004) is the lack of a nonviolent method to influence governance. Not understanding the implications of how electoral rules allocate representation could then be a true recipe for disaster, when elections are used in postconflict societies. There are two major strands of thought as to how to best address the issue of underlying social tensions in society, which can be exacerbated by elections. Both of these theories are discussed at length by Reilly (2002). The first, consociationalism, argues all groups should be given a fair voice in governance. The preferred electoral system by its advocates is a proportional representation system which gives groups representation near their respective percentage in society. The alternative to this argues that doing so encourages division along existing ethno-religious divides and does nothing to address the fundamental issues. The alternative argues for a reciprocalvote approach, where parties must gain votes from groups outside their core constituency, and their success and failure see (Reilly, 2002) 3

4 should lead to more moderate candidates. Both approaches have been shown to work, and to fail, in some cases. For a thourough examination of these issues as they relate specifically to African countries, see Lindberg (2005). Given the impact electoral systems have been shown to have on electoral participation, legislative majorities, and even the number of competitive parties in a country, all discussed by Lindberg (2005), it seems reasonable to assume that electoral systems could also have a significant impact on when, where and how violence is used in an election. Although we highlight the potential role this may play, we leave a thorough investigation of this relationship to future work. However, understanding the impact the electoral system may have on the use of violence first requires an understanding of how violence influences elections, we now turn to the existing literature examining the impact of violence. 2.2 The Impact of Violence Elections play several fundamental roles in democracies, and one is their role in conferring legitimacy to a government. An Afrobarometer working paper, Kerr (2009), asks when elections accomplish this in the eyes of those they govern? Kerr argues that, it is most important for elections to be perceived as free and fair. When this is true, he finds constituents are more satisfied with democracy, more supportive of the ruling government, and that members of the losing party are more likely to accept defeat. The final point, is especially salient in regards to elections and violence, as violence is often incited following elections where one party refuses to accept the results, and higher levels of violence are strongly correlated with lower perceptions of free and fair elections. The existing research examining the more direct link between election outcomes, electoral participation, and violence is limited to a few studies analyzing particular countries in detail. The first Hickman (2009), examines the impact of violence on voter turnout and election results in Sri Lanka. He finds violence perpetrated by individuals associated with one political party, leads to a reduction in turnout for the opposition in that district. He argues that the impact is small, did not change the election results, and the use of violence by either side essentially cancelled out in the aggregate. Another paper using the Afrobarometer data from Nigeria, shows that Vote buying is far more effective than violence (Bratton, 2008). He also demonstrates that the most common response to any form of illegitimate campaigning is abstention. These papers fail to show a causal link though, as the investigation in this paper shows this relationship is consistent across a wide array of African countries, time periods, and robust to a variety of alternative explanations. Empirically, Blattman (2009) and Collier and Vicente (2014) provide the most convincing causal evidence of the impact of violence on political participation. The first Blattman (2009), uses variation in exposure to violence in Uganda, which he argues to be exogenous, to estimate the impact of that exposure on political participation. He finds that having been exposed to violence, nearly two decades earlier, makes an individual more likely to be involved in the community and more likely to vote. Another paper, Collier and Vicente (2014), uses random placement of antiviolence campaigns during the 2007 Nigerian election to create exogenous variation. The campaign successfully reduced election violence in the regions of implementation, and they found violence to significantly and negatively reduce voter turnout. The type of violence each group was exposed to likely has a lot to do with how two studies, both with plausibly exogenous variation, can come to completely opposite conclusions. In Blattman (2009), the author leverages the random nature in which young Ugandans were abducted and forced to join the rebel group. They find this past exposure to violence increases political participation. In Collier and Vicente (2014), the violence is during the election campaign, and they find it decreases voter turnout. The difference could be explained by differential long and short run effects. In the 4

5 long run observing violence may make an individual wish to be more involved in politics because of past experiences. However, in the short run, it may make the act of voting far too dangerous or costly to be worth participating in. Data limitations are likely the primary reason that such an important subject has such a limited empirical literature, furthermore given the irregular nature of elections in many countries where election violence is a common problem serious estimation issues can result as discussed in Goldsmith (2014) and Cheibub et al. (2012). However, an expansive theoretical and case-study literature has examined the types and causes of electoral violence. 2.3 Causes and Types of Electoral Violence Understanding the causes of electoral violence requires a solid understanding of what we mean by electoral violence. Broadly speaking, this paper use the definition from Fischer (2002). Another paper Kehailia (2014), published as part of a series of case studies on election violence from practioner experience at International Foundation for Electoral Systems (IFES), (Cyllah et. all 2014), lays out a useful typeology of election violence. This typology classifies election violence into eight types, based on who is involved in the incident. The examples he provides for each type, demonstrate how different types of electoral violence can have fundamentally different causes, and be used for different purposes. Such a typology is useful when studying election violence, especially when considering the underlying causes or intent of election violence. However, even within a single type the causes, goals, and purpose may differ. This paper focuses on election violence where the intent is vote deterrence, specifically strategically manipulating who votes with the intent of influencing the election. Even in such a narrow range as vote deterrence, the purpose of doing so may not always be manipulating the outcome on polling day; some groups may simply have the intent of undermining the electoral process as a whole. Take one of the examples from (Kehailia 2014), where unidentified gunmen opened fire on a polling station in the Democratic Republic of the Congo (DRC), with supporters of both parties present. As we will show, this could be consistent with the models that follow assuming they could infer that the voters in attendance were far more likely to support one party over the other, but actions like this could also have a simpler intention of undermining the electoral process. Nonetheless, focusing on who perpetrates acts of electoral violence and their intended target is an illuminating thought process, and a crucial missing piece in currently available data. No complete survey of the causes of election violence exists at this time, though an excellent summary can be found in Goldsmith (2014). We discuss several potential causes of electoral violence that are beyond the scope of this paper. One commonly recurring message throughout the literature written by election practioners is this so-called culture of impunity, mentioned in Bekoe (2011) and IFE. Bascially, time and time again countries in Africa experience some level of election violence, form committees to investigate the occurences, and make few prosecutions or fail to create independent bodies with the power to enforce punishments following the investigation, setting the stage for violence to occur again. Lack of punishment and a culture of impunity sets the stage for the violence encouraging rhetoric many politicians have used in African elections. Politicians in the 2011 election campaign in the DRC refferred to members of a political group aligned with an ethnic minority as mosqitoes and advocated to spray some insecticide. In Nigeria former president Olusegun Obasanjo said the election was a Do-or-die affair. So often violence is encouraged publicly from leading political figure heads, who use existing ethnic tensions to divide and conquer. A sizeable academic literature has also suggested that election violence often mirrors existing ethnic, religious, and political tensions in society. As stated in Bates (1983), electoral competition arouses ethnic conflict. 5

6 This critical view of elections does not imply elections can t mitigate conflict; however, sometimes unexpected results can relight fires with a catastrophic impact. Take for example the violence Kenyas 2007 elections, where hundreds of people died following each election (Klopp and Kamungi, 2007). The likely cause of such violence is best outlined in (Walter, 2004), where she discusses causes of recurring conflict and argues in some ways that it is twice conditional, not only do you need the traditional structural conditions to exist for conflict to recur, but a group must feel they have no alternative recourse from misrepresentation in government, arguably a contributing factor in both Nigeria, where an unwritten rule of alternating power was violated, and in Kenya. (citation for nigeria, USIP). A final use of electoral violence we will discuss briefly, one which this paper s results suggest is plausible, is government repression of potential competition. This relates to a classical example of rational conflict, the first-strike advantage. This can be summarized by arguing that even if the use of violence is costly now, if it permanently eliminates competition it may be strategically viable. There are countless examples of candidate assasinations and unjust prosecution. One recent examples of such behavior includes the arrest of two opposition candidates before political their political rallys in the Uganda s 2015 elections WSJ, though this particular abuse of state resources was non-violent it is a good example of how state backed political parties can abuse the system in their favor. Nigeria s 2007 elections represent another, more violent, example of the abuse of states resources during an election campaign. The elections resulted in about seventy deaths, which accorrding to Suberu (2007), included assassinations. This sort of manipulation is the subject of several theoretical, which address abuse of the state security apparatus to repress competition in elections. It is easy to see how preventing competition from forming may be one way in which semi-autocratic governments can stay in power via the use of elections (cite a paper on that). Even though electoral violence is distinct from political violence in general, it is clear that it can take on a wide variety of forms, and be brought about for an equally wide variety of reasons. This paper focuses on two models, with clearly competing assumptions and predictions because of the clarity of the predictions in those models. With so many competing theories, it is important to begin to lay out hypotheses which highlight the distinct predictions of each model, and even more important to base these theories on assumptions which can be proved accurate. Before turning to a detailed discussion of these papers, we discuss related theoretical papers on political competition, which incorporate violence. 2.4 Theories Political Competition and Violence One paper of particular interest (Ellman and Wantchekon, 2000) introduces the threat of violence by a third party into a traditional Hoteling model of platform choice among politicians with opposed preferences. This paper is of particular interest because violence does not have to occur, for it to influence the election, either via the party elected or the policies they implement. Another paper (Robinson and Torvik, 2009), focuses on the related question of which group of voters should a party intimidate, undecided voters or opposition supporters. They derive a series of propositions that analyze how competitive the elections are with and without particular groups of voters participating in the election, and basically conclude that swing voters are usually the best target for violence, as long as they care relatively more about monetary issues than core party supporters. Incorporating violence into traditional models of political competition can paint a very different picture of how violence influences the electoral process than the models examined in the following section. It is also difficult to look at electoral violence as a phenomenon separate from other forms of socio-political conflict, because it is so often interrelated. These same caveats must be considered when we turn to analysis, as elections are as often a reaction to past violence, and violence is a 6

7 reaction to the electoral process. An expansive literature is emergening which examines electoral violence; however, the scope of the empirical literature tends to be too narrow and focus on single countries, and that of the theoretical literature too broad, generalizing case study evidence to a broad sample. Both types of papers are important contributions, this study tries to take a broad empirical approach, at the expense of important individual details. 3 Models and Hypotheses This section compares two models of election violence; Chaturvedi (2005), hereafter CH and Collier and Vicente (2014), hereafter CV. The natural starting point for comparing two models empirically is usually analyzing the differing predictions made by the models, and generating falsifiable hypotheses from them. However, with few papers aside from Bratton (2008) and Goldsmith (2014) which examine how violence influences voting behavior, we argue that it is neccesary to start by focusing on a thorough analysis of the models assumptions. Although these models are based on similar fundamentals, they make divergent predictions with respect to when electoral violence occurs. The focal point of the varied predictions are with regards to how the division of the electorate predicts the occurrence of violence and the perpetrating party of the election violence. These differences are driven by different assumptions about who is impacted by election violence, the alternatives included, and the way in which the decision is modeled. We examine these differences in detail in subsequent sections, this section focuses on comparing the assumptions in the models which result in the varied predictions. The first model, CH, sets up a two-party winner-takes-all election where each side attempts to maximize the difference between their expected vote shares. In the second model, CV, the incumbent is assumed to always be capable of winning the election through illegitimate means and each side simply seeks to maximize their share of the vote. In both models, violence is used to influence the election results by changing the composition of the electorate which votes on Election Day. The models differ in who abstains from voting following the use of violence. In CV the assumption made is that if violence occurs, all swing voters would abstain from voting. In contrast to this assumption, CH assumes that violence is directly targeted; each party targets the opponent s core supporters and swing voters are not directly impacted by violence. Each model describes and sets up swing voters in a slightly different fashion, so some alterations to these models will be made to better facilitate comparisons. These assumptions, on how violence influences election results, are crucial in the resulting predictions on when and where it should occur; however, their validity remains largely unexplored. The core assumption in both models is that violence can reduce voter turnout, we turn this into our first hypothesis. H 1. Both Assume: Violence, perceived, threatened, or actually occurring, reduces the likelihood of an individual voting. There are two ways in which violence could influence voting. First, it could be the case that violence is targeted and anyone who is targeted will be less likely to vote as a result of fear and intimidation. Second, violence could simply be used to create havoc in an area and certain groups of people may be more likely to abstain from voting as a result of violence raising the cost to vote. The Afrobarometer data allows us to examine both possibilities, if the presence of violence generally or direct fears of violence influence voting behavior. 7

8 If violence is used as a scare tactic, intended to disenfranchise large groups of swing voters 2, as in the CV model, we would expect to see differential reactions to violence. CV assumes that swing voters, since they have the lowest propensity to vote, will have stronger reactions to violence. This leads to our second hypothesis. H 2. CV Assumes: Individuals without a strong political affiliation should be less likely to vote due to violence. The assumptions made in CH are closer to the first explanation; particular groups of voters are targeted by violence, and therefore cease voting. The CH model assumes that core party supporters are more likely to be targeted by violence in an election, we assume this targeting will translate to core party supporters fearing violence more in elections, this leads to our next hypothesis. H 3. CH Assumes: Individuals with a strong political affiliation should be more likely to fear violence during an election. Both models allude to conditions that would make it easier to use violence to swing the vote. The more accurately it can be inferred who an individual would vote for, the easier it is to direct violence at the relevant population of voters. This is true when attempting to disenfranchise swing voters in the CV model or the oppositions support base in CH. We argue that both models predict ethnically or religiously aligned parties would be more susceptible targets for violent intimidation tactics. The argument is best summarized in CV, Violence aimed at discouraging the turnout of opposing voters may become easier because the allegiance of voters can be readily inferred from their ethnicity. This leads to the following hypothesis: H 4. Both Predict: Individuals with strong ethnic identities, or belonging to ethnically aligned political parties should be more likely to fear violence in elections. Starting with a solid understanding of how people react to the threat of violence during elections is essential if we wish to design empirical models capable of predicting its occurrence. Summarizing the results presented in Section VI, we find slightly stronger support for the CV model. Individuals who personally fear election violence are slightly less likely to vote, and swing voters are particularly responsive to violence even if they do not personally fear becoming a victim. We also find that members of political parties which we argue are ethnically aligned, are more likely to fear violence. Finally, we show evidence that minority parties may be especially prone to being targeted by violence in elections. We argue that the CV model receives greater support because we dont see evidence that core-party supporters are being targeted by violence, and we find evidence that just the occurrence of violence is enough to deter swing voters, a key difference in the assumptions between the two models. We now turn to a more detailed examination of the models. 4 Comparing Simplified Models The preceding section discussed the assumptions made with regards to how individuals react to violence in the two models. This section examines simplified versions of these models focusing on how violence impacts each party s vote share, while removing the alternatives present in the original models. This simplification facilitates a more direct comparison of the models. 2 The rational for this is explained in the following section, the intuition for who wishes to do this relates to who has more core supporters in an election who will continue to vote even when violence occurs 8

9 4.1 Collier and Vicente Model (CV) Each political party, C, the challenger and I, the incumbent, wish to maximize their share of the votes. A unit mass of voters are divided into three groups, C, I, & S. C, is the fraction of voters supporting the challenger; I, is the fraction of voters supporting the incumbent; and S, is the fraction of undecided or swing voters. If neither candidate were to take any actions, a fraction of swing voters, α, would vote for the incumbent, and (1 α) for the challenger. Voters in group C and I always vote for their respective candidate. Each party has the ability to use violence in an election to intimidate swing voters. We assume that violence has a linear impact on voter participation, after violence, of level v, the fraction of voters who continue to vote is (1 v). This implies that if we assume swing voters are targeted, with violence of level v, then (1 v)s of them vote Without loss of generality (WLOG), we focus on results for the incumbent, we can do this because each party faces an opposite but symmetric set of choices. Absent violence, the incumbents vote share is, Including violence, I s vote share is V oteshare = V oteshare = I + αs C + I + S. (1) I + (1 v)αs C + I + (1 v)s Note that violence is not indexed by party, rather the total amount of violence between both parties is what matters. However, we can show only one side will use violence in equilibrium under these assumptions. The derivative, or gain in vote share from violence, after some simplification is: (2) V oteshare v = ((1 α)i αc)s = MB (3) (C + I + (1 v)s) 2 To determine if violence results in a gain in vote share only requires examination of the numerator. The Incumbent will raise their vote share using violence iff, This condition simplifies to, (1 α)i αc > 0. (4) 1 α α > C I. (5) This implies only one candidate has a positive return on violence, this condition demonstrates which party benefits from excluding the average swing voter. Placing each parties initial support on a unit simplex, as in Figure 1, displays clearly how division of the electorate determines the use of violence in this model. The dividing line in the figure, which splits the graph by who perpetrates violence, has a slope of 1 α α, the figure assumes symmetry, or α = 1 2. Assuming a constant marginal cost of violence, γ, the following proposition determines the equilibrium level of violence, Proposition 1. Party I will set v = 1 iff: their vote share increases by excluding the average swing voter, or if 1 α α > C I, (6) 9

10 and the total benefit to excluding all swing voters exceeds the total cost, ((1 α)i αc)s (C + I + S)(C + I) > γ (7) The proof is as follows, if 1 α α > C I, then I gains vote share by disenfranchising swing voters. This implies the opposition party would not use violence because the conditions are symmetric but opposite. The second condition, ((1 α)i αc)s (C+I+S)(C+I) > γ determines if the total benefit exceeds the total cost. This is examined, instead of a more traditional marginal cost, marginal benefit equivalence, because violence has an increasing marginal return, 2 V oteshare > 0, until the point where v = 1. v 2 The increasing returns on violence imply that if any violence is used, it will be used to exclude all swing voters. The left hand side (LHS)of the second condition, compares the change in vote share for the incumbent, with and without swing voters in the election to the right hand side (RHS), γ, the cost of setting v = 1. If both of these conditions hold, the incumbent will use violence. These are essentially the same conditions used to determine the use of violence in the original CV model. The first condition simply demonstrates when I, the incumbent party, benefits from preventing the average swing voter from voting. The second condition asks if the net benefit from using violence exceeds its cost. The original model examines these same two conditions, first by exploring which party would benefit from removing swing voters from the race, and second by examining whether doing so is cost effective. The only time it is not cost effective for the incumbent, assuming they gain from excluding swing voters, is when they have more efficient alternatives such as vote buying. The CV model also assumes that all swing voters are intimidated if any are, although they do not demonstrate this condition could be derived endogenously as a result of the increasing returns to excluding voters, they instead argue that violence is a low cost method of voter exclusion. To analyze the comparative statics in the model, we focus on how violence impacts a party s vote share. We assume that as this impact increases, it becomes more likely that a party may use violence. The impact of violence on a party s vote share increases as S, the fraction of swing voters, increases and generally increases in own party supporters, I as shown above 3. The impact of violence decreases in opposition support, C as set-up above, and in the fraction of swing voters supporting the perpetrating party, α as set-up. The arrows, in Figure (1), demonstrate the direction in which the marginal return to violence increases. As the ratio of support for core supporters diverges from that of swing voters, or as 1 α α gets farther from C I, removing swing voters changes the overall ratio of support more dramatically. Additionally, as the fraction of swing voters decreases, or as we move towards the origin in the figure, the return to disenfranchising swing voters increases simply because they represent a greater fraction of the electorate. The predictions in this simplified version of the CV model, mirror the predictions in the baseline model presented in their paper. 4.2 Chaturvedi Model (CH) Two main differences occur when examining the CH model, first the goal of each party is to maximize their plurality rather than their share of the votes; CH defines plurality as the difference in expected votes. Second, violence is targeted at the oppositions support base. These assumptions can t be migrated to a model identical to that in CH, because both sides would alway set the level of violence at 1, because doing so would completly eliminate competition. Instead and in order to preserve the predictions from the original model, we continue to maximize the plurality. 3 Violence increases in own support if the denominator under the radical is not too small 10

11 C I > C I uses Violence C > I C uses Violence Violence up with Arrows I Figure 1: Violence use and Incumbent Support, CV The composition of the voters remains the same, a unit mass of voters are divided into three groups, C, I, & S. C, is the fraction of voters supporting the challenger; I, is the fraction of voters supporting the incumbent; and S, is the fraction of undecided or swing voters. If neither candidate were to take any actions, a fraction of swing voters, α, would vote for the incumbent 4. In this model violence is indexed by the party using it, again WLOG we focus on the incumbent s decision to use violence. The incumbent seeks to maximize their plurality, difference in vote share. This can be derived, for the incumbent, as follows: Incumbent Votes = (1 v C )I + αs (8) Challenger Votes = (1 v I )C + (1 α)s (9) P lurality = (1 v C )I (1 v I )C + (2α 1)S. (10) We see immediately that the return on violence is proportional to the opponents support base, P lurality v I = C. (11) Incorporating a constant marginal cost, γ, we see that the use of violence would be an all or nothing choice, as long as the oppositions support is greater than γ resources would continue to poor into reducing turnout among opposition core supporters, until they were exhausted. Incorporating a traditional cost function such as 1 2 γv2 would lead to the following 5 : 4 This last assumption is the biggest divergence from the original model, however it leaves the direction of the predictions unchanged. (This is because it essentially makes S exacerbate the use of violence by including it in the marginal cost, should I discuss this more?) 5 We focus on the case of symmetric costs, their is no reason to believe this is the case; but given the inability to determine the perpetrator of the violence in the data, different cost functions would not yield additional testable hypotheses. 11

12 C I > C C uses more Violence C > I I uses more Violence Violence up with Arrows I Figure 2: Violence use and Incumbent Support, CH v I = C γ (12) This simplified model does lose some of the nuanced predictions of the original model, which we will discuss as we make comparisons. However, the core prediction that the allocation of resources to violence increases in the initial support of the rival party remains intact, as does the prediction that violence decreases as the number of swing voters increases, though the rational is different. In the original model this is the result of the opportunity cost of violence rising as the fraction of swing voters increases, here it is the result of the the impact of violence decreasing as the fraction of core supporters decreases. Contrasting this with the CV model, we see that instead of only one side always using violence, both sides always use some violence. This is because as long as the opponent has some supporters in the election, eliminating some of those supporters will result in a greater plurality for the perpetrating party. This prediction could be altered by including a fixed cost to violence, then only if the opponents support was large enough would we see violence used by a political party. The more stark contrast is with regards to who uses more violence, we get the opposite prediction and with how swing voters influence the use of violence. The CH model predicts that the side with a lower base of support has the higher return to violence, and that the level of violence used decreases as the fraction of swing voters increases. The latter prediction is amplified in their original model, as the alternative use of resources in the original model was traditional campaigning targeted at the swing voters. 4.3 Hypotheses from Baseline Models Comparing these models, several hypotheses related to how the electorate is divided in an election can be derived. The validity of the predictions made by these hypotheses rest on the assumptions made about how electoral violence impacts election results, this paper argues that the assumptions 12

13 in the CV model seem most accurate. Nonetheless, an examination of how political division is related to electoral violence is useful even if the models predictions are incorrect. Swing voters see a great deal of attention in models of electoral competition, and are a primary focus of both models of election violence highlighted in this paper. They are also one of the primary distinctions between the two models. The intuition for this difference is also fairly straightforward, in the CV model as the fraction of swing voters increases, eliminating them from the election places greater weight on each party s core supporters, so whichever side has an initial advantage in core supporters stands to gain more by excluding undecided voters 6.The CH model makes the opposite prediction, as the fraction of swing voters increases the return to violence decreases. In the simplified model the intuition for this is a reduction in swing voters implies an increase in core supporters. The original model presents a better intuition for this, as the fraction of swing voters increases, the return to traditional campaigning, the opportunity cost for violence, increases as well, implying violence is a less useful tactic. This leads to our first hypotheses comparing the predictions of the two models: H 5. Observed violence should increase (CV), or decrease (CH) as the fraction of swing voters in the electorate increases. Both models include the fraction of core supporters as an important predictor of violence, and several predictions can be derived. We start by looking at own-party support, and how it relates to who is more likely to use violence in an election. The models make a key departure in this regard; unfortunately the data do not allow this or the following hypothesis to be directly tested because we cannot see who perpetrates the violence. The CV model predicts that only one side uses violence and that side is the one with more core-supporters, assuming that the undecided voters do not lean significantly towards that party. The CH model predicts the opposite; the side with lower core support stands to gain more vote share by using violence. From this we can derive the following hypothesis: H 6. The candidate with a greater CH, (smaller CV) fraction of core supporters is more likely to use violence. The level of own-party core-supporters can also be used to examine the level of violence, at least in the CV model. The simplified version of the CH model makes no prediction from ownparty support on the level of violence. For the CV model the prediction is technically ambiguous, but under reasonable conditions it is increasing. The intuition is that as own party core-support increases, excluding undecided voters and placing more weight on each party s core supporters, becomes a more viable strategy. This leads to the following hypothesis: H 7. CV predicts: The amount of violence used increases in own party support. In a similar fashion we can examine how core supporters from the rival party influence the use of violence in elections. Again, these predictions cannot be tested because it is not possible to accurately determine the perpetrator of the violence with the available data. The CH model makes a clear prediction with regards to opposition support, as opposition support increases the return to violence increases. The CH model also makes a clean prediction, although it is opposite CV; as opposition support increases, the return to violence decreases. This difference is perhaps the key 6 This prediction assumes that the dominant parties predicted fraction of voters among swing voters is less than their fraction of core-supporters, if this does not hold the weaker party would use violence. The resulting prediction that violence increases with the fraction of swing voters would remain unchanged. 13

14 difference in between these models and occurs for the same reason that the models predict opposite sides to use violence. Finally, we derive some predictions which the Afrobarometer data can test. In the CH model, aggregate violence increases as either C or I increase, whereas in the CV model we observe violence increasing as the disparity in these two measures increases. Basically, as own support increases, and opposition support decreases the return to violence rises in the CV model. This yields a unique and counterintuitive prediction, as the difference in vote shares increases violence should increase. This is in stark contrast to most models of political competition; generally close contests will lead to a greater expenditure of resources. The CH model does not refute this prediction; the prediction in the CH model simply mirrors the earlier hypothesis that violence should decrease as swing voters make up a larger fraction of the electorate. Focusing on the new prediction leads to the following hypothesis: H 8. Violence measures should be higher in elections when the major parties have very different levels of core support Overall division of the electorate plays an important role in both models of election violence. The basic predictions outlined here can be seen clearly in Figures (2) and (1). These figures graph division of the electorate on a unit simplex, with incumbent support along the x axis, challenger along the y, and swing voters being represented by omission, swing voters represent a greater fraction of the electorate closer to the origin. The shading depicts which side is predicted to use, or use more, violence. This difference is perhaps the clearest distinction between the models; however, given data limitations this is untestable. The available violence measures do not make a clear distinction as to who the perpetrators of violence are. The arrows give an indication of when violence should be more likely, or when its marginal benefit increase. Figure (1) has violence increasing towards the origin and as distance from the dividing line increases. The former prediction results from an increased fraction of swing voters closer to the origin, and the latter prediction results from a higher return to violence as the division in the electorate becomes more uneven. Figure (2) has lines moving in nearly opposite directions, the arrows parallel to the dividing line are the result of decreased violence use as the fraction of swing voters increases, whereas the lines parallel to the x and y axis come from the increased marginal return to violence as the fraction of opposition core-supporters increases. Testing these predictions directly is quite difficult. The use of violence could lead to changes in which party an individual supports, creating an intrinsic link between cause and effect. It seems even more likely that even if it does not alter the party an individual supports, effective violent intimidation may influence the measurement of political support. Furthermore, these models are meant to look at elections as a whole and relatively strong assumptions are needed for all predictions to remain identical when looking at districts within an election. All of these caveats imply that results from an examination of the impact of electoral division on violence are not causal, and could be the result of a variety of complicating factors. 5 Data Sources The primary data sources used in this paper are the round 4 and 5 Afrobarometer (AB) surveys. These surveys provide a unique look at political sentiment across twenty and thirty-four African countries, respectively. They included a variety of questions related to elections, political participation, and election violence. The variety of countries included creates an interesting opportunity to analyze electoral violence in countries with different electoral systems and varying histories of political violence. 14

15 5.1 Political Affiliation The theories discussed in this paper all use division of the electorate as an important predictor of electoral violence. The AB survey provides two potential measurements of political affiliation, from which the division of the electorate can be estimated. The preferred measure is created from the following questions. The survey first asks, Do you feel close to any particular political party?. If the respondent answered yes, they then asked, Which party is that? Respondents that said No, to the first question are classified as undecided or swing voters. Respondents who provided parties are then placed in one of three groups; those who support the party with the highest number of responses in the country, the second highest number of responses, and all others. The survey asks another related question, If a presidential election were held tomorrow, which party s candidate would you vote for? Respondents are divided into an equivalent set of groups using this measure. Both measures share one particularly problematic weakness, political intimidation may have swayed their reporting behavior in one of two possible ways. Those who are fearful of intimidation may be more likely to report an affiliation with the intimidating party or refuse to answer. Both of these possibilities can be examined. Being that we see the lowest fear of violence reported among the largest parties, it seems unlikely that intimidation is persuading people to alter which party they say they would support. The latter case is a significant possibility, as violence is higher among those categorized as swing voters. This is unfortunate considering that this is also a group hypothesized to be the target of violence by one of the theories. Table 1: Political Affiliation Summary Variable Mean R4 Mean R5 Exp Violence Min Max Individual Level Country Level Support Maj Support Sec Sup Other Undecided Vote Maj Vote Sec Vote Other Undecided Observations Table 2: Agreement Between Measures Would Vote For Supports Majority Second Other Undecided Total Majority 22,463 1, ,786 25,954 Second 1,162 6, ,105 Other ,040 2,043 8,349 Undecided 7,050 2,860 2,658 21,296 33,864 Total 31,288 11, ,769 77,272 Agreement 0.39 Table 1 displays summary statistics on political affiliation across the two survey waves. Slightly 15