Threat and Repression: The Non-Linear Relationship Between Government and Opposition Violence SCOTT

An understanding of the causes of political repression has continually eluded researchers for the past decade. We argue that much of this can be tied to the theoretical specifications of the models employed. We developed a decision-theoretic model that predicts the level of repression used by governments to suppress political opposition. We believe that analysis of repression needs to include the political contexts in which states operate. In particular, we theorize and find that the nature of the threat posed by an opposition group influences the impact of both the domestic and international factors on the governmentís decision to repress. We argue that the international and domestic costs associated with a given level of government repression are best represented by separate, non-linear functions of the level of demand made by a dissident opposition group. From this model we deduce an equilibrium level of repression for any given demand; we then empirically test these predictions against original data generated from 18 Latin American countries, 1977-1986. We find that as the nature of the threat posed by an opposition group moves from minor to extreme, the marginal increment of government repression decreases. Analyses of these data support our theoretical propositions, and suggest that both non-linear approaches and the inclusion of opposition group demands provide a useful tool for steadying state repression. 1. Repression and Threat The violent repression of political opposition has been a permanent fixture of the modern world. Although more recent times have not seen the banishing of violent methods of political control, they have witnessed a more concerted effort both to understand the causes of political repression, and to impose international standards for the treatment of political opposition. In terms of international convention at least, the violent repression of all dissent is no longer acceptable. But neither attempts to constrain political violence, nor efforts to understand the causal mechanisms that lead some governments to choose repression while others choose political integration, have been entirely successful. Despite only moderate successes, we believe that a better understanding of the systematic causes of political repression will lead to an increased ability to minimize the degree to which governments resort to violence against their own people. In this article we argue that a government's choice to repress opposing political groups is a function of domestic and international costs. These costs, in turn, vary nonlinearly with the nature of the opposition group's demand. Assuming that the particular form of political control adopted by a government is a result of conscious policy decisions, we attempt to contribute to the study of political repression on three fronts. First we add theoretical rigor to this area of research by introducing a decision-theoretic model of political repression. Our model not only adopts the assumptions of a rational, utility maximizing political leadership, but also formalizes the international and domestic constraints that they face when choosing between the political integration and the political repression of opposition groups. We believe that two reasons for the limited success of previous studies are their failure to control for the violent nature of political dissent, and their reliance on linear approaches. Despite the intuitive appeal of nonlinearity and a sensitivity to opposition forces, these innovations have yet to be sufficiently incorporated into either theoretical or empirical analyses. Second, we subject to empirical scrutiny a formally derived hypothesis that predicts the specific conditions under which international and domestic constraints will impinge on the decision maker's choice to repress. We test this prediction against data on eighteen Latin American countries between 1977 and 1986. Third, we explicitly attempt to address a lacuna in the literature on repression. As Lichbach stated, future research on repression should focus on the ongoing process of challenge and response by dissidents and regimes (Lichbach, 1995, p. 259). This study specifically deals with this interaction and the corresponding levels of political repression that results from it. 2. Why Do States Repress? One of the consistent, and key, results in the literature on political repression is that democratic regimes are less repressive than their authoritarian counterparts (Poe & Tate, 1994; Henderson, 1991; Rummel, 1995). These results have been further supported by the literature on the democratic peace in inter-state behavior (see Morgan & Campbell, 1991; Hagan, 1994; Bueno de Mesquita & Siverson, 1995). Although the range of indicators used varies considerably across studies, most of these studies use some attributes of regime type. Building on this work, we disaggregate the concept of regime, and identify critical domestic political elements and mechanisms that we think influence the decision to repress. As we will argue below, we capture the role of domestic politics through the extent of the demands and domestic pressures faced by the ruling elite, variables that capture more micro-level political processes, and are therefore more likely to fluctuate than regime type. This is particularly important for analyzing government repression in regions such as Latin America which, while largely undemocratic, experience serious changes in the domestic political scene. We believe that the results of our analysis largely support the arguments regarding regime types and repression. While these studies of regime type identify a critical aspect of domestic politics, probably the most thoroughly researched aspect of political repression involves the nexus of US foreign aid and the levels of repression in recipient countries. These studies, however, have produced mixed results in both causal directions (Regan, 1995a). On the one hand, there is evidence that aid levels are responsive to repressive practices (Poe, 1991, 1992; Cingranelli & Pasquarello, 1985), though these results have been questioned by others (Stohl, Carleton & Johnson, 1984; Mitchell & McCormick, 1988; McCormick & Mitchell, 1989). On the other hand, there is reason to believe that aid should be an effective tool in manipulating levels of repression (US House of Representatives Committee on Foreign Affairs, 1979), though again, evidence suggests otherwise (Muller, 1985; Regan, 1995b). We believe that there are two important reasons that help explain the ambiguous findings on repression and foreign aid. First, most of these (and other studies of repression, (e.g.; Henderson, 1991, 1993; Park 1987) employ linear analyses without theoretical grounds to expect a linear relationship. Some scholars have recently argued that the factors that influence state violence manifest through non-linear mechanisms (Rummel, 1995; see also Gurr, 1994, 1989). Lichbach has argued that The returns to collective violence are likely to be highly nonlinear throughout the range of violence (Lichbach, 1995, p. 60). These analyses challenge the implicit assumption of linear relationships between the varied independent variables and repression that is so widespread in the literature. If, as others have found and we argue below, the relationships between the factors that influence repression and observed violence are nonlinear, than the assumption of linearity helps to explain these ambiguous results. Much more important, however, is the exclusion of domestic politics. Regardless of their findings, studies of the impact of US foreign aid on government repression tend either to ignore domestic political factors or specify domestic variables as separate, independent effects. We believe that domestic political factors influence the effect of international variables, such as foreign aid, to influence government repression. Thus, in our view, international variables, such as aid, are important factors in explaining government repression, but domestic politics plays a critical interactive role in influencing the effect of these international variables on the decision to repress. 3. Theoretical Argument We view the level of repression as a choice made by the ruling elite that incorporates a range of options, from facilitating political integration to violent suppression. We formally define how critical factors of concern to the elite affect the costs and benefits that influence the elite's choice of political response. In order to construct this model, we first create a decision-theoretic framework. 3.1 A Decision-Theoretic Framework In developing a theoretical framework we work from a rational choice perspective. We posit that a government's decision to repress one's citizenry is a function of the perceived costs and benefits of choosing a violently repressive strategy over other less violent approaches. Implicit in this formulation is the assumption that the level of repression is a purposeful decision taken by a central decision maker. And while one may challenge the notion that any decision maker consciously chooses a particular level of repression, or even that the scope and intensity of repression are controlled from some central location, we believe that this assumption is warranted. Not only does this set of assumptions allow us to draw useful generalizations about political behavior (Riker, 1991), but we feel that to some extent they reflect the referent world we seek to understand. Much of the research in this area, moreover, implicitly adopts rational choice assumptions (e.g.; Schoultz, 1981; Muller, 1985; Pion-Berlin, 1988), and on the rare occasion has done so explicitly (Lichbach, 1987; DeNardo, 1985). The central authority need not be the one planning and prosecuting each specific repressive act, but what we assume is that this leader has the ability to control those agencies or actors who are the most aggressive persecutors. For example, when the level of political violence waged against opposition forces in South Africa or El Salvador got particularly bloody, both the people on the ground in those countries and the international community called on Presidents De Klerk and Cristiani, respectively, to put a halt to the political abuse. We assume, furthermore, that repression is not a dichotomous choice, where a government either does or does not engage a repressive strategy, but rather can be thought of as a continuous outcome. A leader may choose to move incrementally along the coercive path, expecting that a "little more repression" might serve both to placate domestic coalitions and sufficiently threaten the political opposition. Similarly, the government might choose to decrease repression in order to increase international or domestic support. Therefore, at equilibrium, decision makers choose the level of violence that maximizes their expected payoffs. 3.2 Opposition Demands and Elite Repression We also posit that the decision to opt for a repressive strategy is rooted in a challenge to the ruling coalition or the status quo. This challenge, or the demands placed by the opposition group on the ruling coalition, is the attempt to move from the status quo to which the ruling coalition must respond. As such, there is a range of options available to the regime, from acquiescence to the demands of the opposition to the initiation of a civil war aimed at eliminating opposing factions. In the first instance, the regime would make a complete concession to the opposition; in the latter, it would refuse to make any concession. Seen in this context, the decision to repress can be viewed as a weighing of the various costs and benefits associated with different strategies, with the adopted strategy being the one that affords the greatest expected utility. This decision calculus can be expressed formally in terms of the nature of the opposition groups' demand and the extent of the repression. For example, if we let `D` be the demand placed on the ruling coalition, then this ruling group has to respond with some amount of repression `R` such that they maximize their expected payoff. We can say that the decision to repress is f (D, R), where the function f is increasing in ëDí and decreasing in ëRí (the more severe the demand the greater the concession for any given level of repression; and the greater the repression the lower the concession for any given demand). In order to understand fully the decision to repress dissent violently, it is not enough to argue simply that a government chooses the level of repression that maximizes expected utility. The international and domestic constraints faced by any decision maker will to a large degree shape the necessary cost and benefit calculations associated with the decision process. By formally specifying the utility function that guides this process, we are better able to understand the forces that come to bear on decision makers during times of domestic tension. In specifying this utility function we make a number of empirical assumptions: 1) a regime does not choose to repress its citizens in a political vacuum, 2) the international community will impose costs when the level of political abuse goes beyond some threshold of toleration, and 3) the domestic economic elite will influence decision making by imposing costs if the government's attempts at control are either insufficient or excessive. If there were no serious challenges to the status quo, the ruling elite would not risk its dominant position by violently repressing minor political outbursts from a marginalized segment of the population. Non-violent methods of quieting the opposition would suffice. That is, the government only benefits from the potential outcome of repression; the act of repressing itself provides no utility, and in certain political contexts (which we specify) repression is quite costly. We have to begin, then, with a credible challenge from some segment of the populace which presents an issue that portends to alter the status quo ante (Gurr, 1986). This challenge is generally rooted in either political or economic ideology (Pion-Berlin, 1984), though it would probably be difficult to disentangle these two factors. A political challenge, for instance, could be in the form of efforts to bring down a military -or some other form of authoritarian -regime, while economic threats to the status quo could be the result of rising expectations, relative deprivation or land tenure patterns (e.g.; Gurr, 1971; Tanter & Midlarsky, 1967; Midlarsky, 1988). Given this political or ideological chasm, the opposition presents a challenge to the ruling coalition. Rather than running counter to any of the above-mentioned frameworks of why men rebel, this suggests that rebellion is preceded by a period of challenges and counter-challenges. The underlying conditions for a full-blown rebellion might be in place, but efforts are undertaken to achieve a mutually agreeable outcome short of rebellious activities. Efforts to resolve the dispute short of rebellion might involve varying degrees of repression on the part of the government. If we think in terms of Cartesian space, with axes reflecting political and social characteristics of the government and opposition, it is the spatial distance between the status quo and the demands made by the opposition group that will determine the benefits that accrue to the regime in power by standing firm or acquiescing. When the gulf between the demands of the opposition and the current situation is wide, the potential benefits to remaining resolute are high. When this distance is relatively small, the potential benefits to be gained in the dispute are few. We essentially posit that this relationship between repression and demands is linear. All things being equal, in the former instance the probability of the violent suppression of political expression is quite high; in the latter some form of negotiated settlement may prevail. Depending on this spatial distance between demand and status quo, the government has to choose an ideal strategy from the range of options available, but this range of options is not without its own set of constraints. The structure of the costs associated with various strategies form the basis of our second set of empirical assumptions. 3.3 Domestic and International Constraints on Repression We group the government's costs and benefits into two general categories: 1) those associated with external constraints, and 2) those that need to be borne internally. In neither case, we posit, is the relationship between the costs incurred by a particular strategy a linear function of the level of repression. Presumably, if the strategy adopted to address the challenge is particularly reprehensible to the international community, the government may have to suffer the consequences of international pressure (e.g.; Cingranelli & Pasquarello, 1985; McCormick & Mitchell, 1989; McCormick & Mitchell, 1988; Poe, 1992). The international community, however, may be quite content to tolerate levels of repression consistent with the tactics and demands of the opposition, though the level of toleration quickly diminishes as regime brutality becomes excessive. This could be seen to some degree in the response by the United States to repression in El Salvador. The US tolerated a fairly high level of repression by the ruling coalition in response to the quite violent tactics of the FMLN (Farabundo Martí National Liberation Front), but when the government forces resorted to killing nuns and Jesuit priests, the US immediately clamped down with cutoffs in various forms of aid. The degree of toleration, therefore, is not only a function of the amount of repression, but also related to the nature of the challenge to the ruling coalition. Internationally derived costs, therefore, will be small when repression appears to be consistent with, or less than the nature of the demand. As repression begins to exceed the level of response deemed appropriate by the international community for the given demand, the costs rise steeply until the repressive regime is completely ostracized diplomatically and economically. A plot of these costs would approximate a logistic function, asymptotic at the upper and lower bounds, with a steeply sloping curve in between. The inflection point, furthermore, shifts with the level of opposition demand (see Figure 1). These external costs can be quite visible, such as economic sanctions and restrictions on trade, manipulations of foreign aid, or support for an insurgency. Conversely, if the dispute is handled democratically or addressed through institutional channels, aid, trade, and praise may flow toward the regime. (Figure 1 about here)Internal constituencies also act as constraints on the decision maker. As with internationally imposed costs, domestic costs are tied to both the extent of the repression and the demands of the challenger. We believe that interstate actors react most strongly to extremes (Gartner, 1993). These internal costs are both political and institutional, and will in part be determined by the strength of existing institutions that act as countervailing forces within society. An independent judiciary, a strong economic sector, or a democratic form of government may all contribute to the cost of any strategy adopted by the political leadership (e.g.; Prichard, 1988; Henderson, 1991). For instance, a strong and independent legislative body may hamper the executive charged with resolving a dispute with an opposition group. Likewise, a judiciary that takes an active stance against the use of arbitrary detention or torture will increase the costs of pursuing such a course of action. The physical resources required to repress dissent also figures into the political calculation, and their diversion can facilitate the coalescing or shifting of coalitions (Organski, 1965). On the other hand, these same domestic actors may impose costs on the government for not taking aggressive action to suppress opposition. The potential instability that results from a response that is not well calibrated to meet the challenge will lead to a withering of support for the government from the political, economic, and institutional elites. Recent events in Mexico and Peru are poignant examples of the role played by internal domestic constituencies. In both instances, confidence in the governments' ability to ensure political and economic stability resulted in the flow of money out of the country, as well as a drastic slow down in foreign investments coming into the country. Unable to cope with the economic uncertainty caused by lack of investor confidence, Presidents Fujimori in Peru and Zedillo in Mexico increased the intensity of efforts to repress the Shining Path and Zapatista movements. In Peru at least, these policies of increased repression have restored the confidence of the economic elite; the case of Mexico is still uncertain at this juncture. It is the potentially powerful economic elite that needs to be assured of the stability of the government; without that assurance they cannot have confidence in future investments. Therefore, levels of repression appropriate to the challenge are tolerated -even sanctioned -even though the repression may appear threatening to this segment of society. What is more threatening is the potential loss of control. But when the level of repression is either insufficient to maintain stability or excessively brutal, and therefore threatens stability, the government will incur domestic costs. Under strong demands by the opposition, the cost of not employing some coercion will be high, though as the level of coercion begins to mirror the opponent's challenge, the domestic costs decline rather precipitously. On the other hand if the repressive tactics are considered overly severe relative to the demand, the domestic costs begin to increase. Too much repression may leave the opposition little alternative to revolt; too little repression for a given demand may present the opposition with too many incentives to press for further demands (Gurr, 1971; Lichbach, 1987). This cost function associated with internal factors will approximate that of a U-shaped, convex function (as if generated by the function X2 for all real X). This is depicted graphically in Figure 2. (Figure 2 about here) It might seem reasonable to suspect that the shape of the domestic cost curve would differ between a democratic and an authoritarian form of government, yet closer inspection sheds doubt on this assertion. This is not to say that the two types of regimes are equally repressive, but rather that the domestic costs from adopting a particular strategy to deal with political dissent may be similar across regime types. If the response by the government is insufficient to meet the challenge, then the economic elite reacts to the threat to the institutions and the economy; when the response is excessive there is an additional cost imposed from the institutions. Highly democratic countries generally do not repress as much as those less democratic varieties, not because the potential costs are different, but rather because the extent of the challenges to the status quo in highly democratic countries is generally insufficient to make the benefits of violent repression outweigh the costs. Two recent events in Russia serve as a useful contemporary example. As the leader of a country struggling to institute democratic procedures, President Yeltsin faced a severe challenge to his rule from a group of national legislators in 1992. The threat posed to the stability of the government by the renegade legislators was sufficiently strong to incur the violent response of the Yeltsin faction. His strategy to quell the opposition, while quite extreme, was apparently consistent with the extent of the demand, and appears to have been sanctioned by his domestic constituency and international allies. Domestically, Yeltsin was faced with the potential of a loss of support if he did not act decisively, while internationally there was almost universal support for his actions because the opposition left him few alternatives. Currently, however, Yeltsin faces another challenge to his authority from the Republic of Chechnia, to which he is again responding with deadly force. But if Yeltsin generally received support for his actions against the national legislature, he has been all but condemned for his response against the Chechnians. International financial sources talk of restricting investments and loans, while the domestic coalitions denouncing his policies are quite visible and vocal. Comparatively speaking Yeltsin is paying a higher price for his use of force in Chechnia than he did in Moscow, with this disparity in costs linked to the seriousness of the challenge posed by the respective dissenting groups. The legislative revolt posed a national threat and therefore internal and external constraints were moderated. However, the more local threat in Chechnia appears to be met with an exaggerated level of force relative to the character of the threat posed by the Chechnian opposition. The international and domestic communities, it appears, are responding in kind. Expressing these relationships formally we postulate that the government's objective is to: where Ci reflects the international costs, Cd reflects domestic costs, R is the level of repression, and D is the nature of the oppositionís demands. We argue that Ci and Cd are a function of R and D, such that Ci(D, R) = ci(R-D) and Cd(D, R) = cd(R-D). Furthermore, we assume that Cd is a quadratic expression, and that when cd(R-D = 0) then cd=0. Finally, we postulate that Ci is a logistic functionEU=1 OVER {1+demand SUP 3 OVER e SUP {.7Repression}}+ .015((Demand Repression) + (Repression Demand) SUP 2) , and that when ci(R-D =O) then ci=0. Put simply, when the order of magnitude of repression equals that of demand, the state faces neither domestic nor international costs. The basic logic of the model involves two phenomena. First, for each given demand, there is one incentive to repress up to the level of that demand (domestic benefits), but