Sage Journals: Discover world-class research

Abstract

This paper presents a game-theoretic model where dissidents with heterogeneous abilities and motivations contribute to collective action. A regime demotivates dissidents by preemptively increasing their costs of contributing, using a budget that can be spread across them in any way desired. The regime’s optimal targeting strategy is shown to depend on the (technological) degree of complementarity between dissidents’ contributions. For low complementarity, it is optimal to equalize all dissidents’ strengths (where strength depends both on motivation and ability; tall-poppies strategy). For intermediate complementarity, it is optimal to focus all repression on the most able dissidents (key-player strategy). For high complementarity, it is optimal to focus all repression on the least-motivated dissidents (weakest-link strategy). The range of intermediate complementarities for which the key-player strategy is optimal is larger, the larger heterogeneity in abilities. The paper finds indication for the use of these strategies in concrete examples of preemptive repression.

Keywords

preemptive repression targeted repression collective dissent

Recent studies argue that modern surveillance techniques have drastically reduced regimes’ surveillance costs (e.g., Gohdes 2020; Xu 2021). Informed regimes may no longer wait and see where dissent arises to respond with reactive repression, but shift towards preemptive repression. This may not only occur in autocracies, but also in democracies, with a shift in protest policing away from negotiated management (Howe and Monaghan 2018). Keeping the traditional focus on reactive repression then leads one to underestimate both the level of grievances among the population (Ritter and Conrad 2016), and the overall level of repression itself: the fact of not observing open protests with the ensuing reactive repression may merely be a sign of successful preemptive repression.

This paper seeks to answer the question: at whom will well-informed regimes target preemptive repression, given their limited resources? Answering this question is non-trivial, as regimes use a wide range of targeting strategies across time and space (Hassan et al. 2022). Repression can be targeted at those ideologically most remote from oneself, with the example of “radicals” in Guatemala (Sullivan 2016), or on the contrary at those ideologically closest, with the example of swing voters in Zimbabwe (Robinson and Torvik 2009). In other instances, repression is not based on ideology, but on demographic characteristics such as ethnicity (Rozenas 2020). Repression may be targeted at leaders, such as in the case of Navalny (Ma 2020), or instead spare leaders (Esberg 2021). It may be targeted at intellectuals (Zeng and Eisenman 2018) and members of the elite, and include within-regime purges (Sudduth 2017), or may on the contrary target groups that are already disadvantaged (Rozenas 2020). Repression may hit entire populations collectively and indiscriminately (e.g., curfews, Aksoy et al. 2023), or may on the contrary be highly surgical (Siegel 2011). Repression may have a low intensity and merely constrain dissidents, or on the contrary have high intensity and violently eliminate them (Davenport 2007, 46-47). Violent, high-intensity repression may moreover be applied to an arbitrary dissident among a number of peers (Herreros 2006), in contrast with the blanket application of restrictions. What explains this diversity of targeting strategies?

To answer this question, this paper develops a game-theoretical model where a regime exerts targeted preemptive repression with the purpose of minimizing collective dissent. In my stylized model, dissidents are heterogeneous and differ along two broad dimensions, namely their ability to contribute to collective dissent, and their motivation to do so, where dissidents are assumed to perfectly observe each other’s abilities and motivations. Dissidents individually decide how much to contribute. Targeted repression consists of reducing individual dissidents’ motivations. The determining factors in explaining the regime’s optimal targeting strategy are, first, the degree of complementarity between dissidents’ contributions, and second, the degree of heterogeneity in dissidents’ abilities and motivations. The degree of complementarity reflects the technology of dissent, and measures to what extent the level of collective dissent is determined by the smallest effort among the dissidents, where with the largest degree of complementarity this level if fully determined by the smallest effort. The degree of complementarity and degree of heterogeneity determine to what extent the regime targets dissidents using their ability and/or motivation as a criterion. Holding the degree of heterogeneity fixed, my model predicts that for low complementarity, the regime optimally employs what I call a tall-poppies strategy,¹ where dissidents’ post-repression strengths are equalized (with strength a measure positively determined by both ability and motivation); for intermediate complementarity, the regime employs a key-player strategy, and minimizes the motivations of the most able dissidents, independently of their initial motivations; for large complementarity, the regime employs a weakest-link strategy, and minimizes the motivations of the least-motivated dissidents (where repression is more biased towards high-ability dissidents as the technology deviates more from perfect complementarity). It is shown that the range of degrees of complementarity for which the key-player strategy is optimal is larger, the larger the heterogeneity in the dissidents’ abilities.

Targeted Repression

Following Siegel (2011, 993), I define repression broadly as any actions to deter a population from participating in collective dissent. Existing explanations for how regimes choose to target repression can be grouped in three categories. First, the extent to which repression is targeted has been argued to depend on a regime’s surveillance capacity, or on its costs of investing in surveillance. In the model of Gregory et al. (2011), a regime represses a larger part of the population the lower the quality of its information. Shadmehr and Haschke (2016) present a model where single-party, non-military regimes have the capacity to gain information on the loyalty of citizens and to selectively distribute benefits among them, but not to selectively repress; for military regimes without a party, the opposite is the case. For this reason, in response to a youth bulge, military regimes will repress less, as selective repression of the young then becomes too costly. Single-party non-military regimes will on the contrary repress more, because selective cooptation of the young becomes more costly (for other authors that stress the link between institutions and the ability to gather information, see e.g., Dimitrov and Sassoon 2014, Greitens 2016; Steinert 2023). Yet, the argument that regimes selectively repress when they have sufficient information to do so, cannot explain why well-informed regimes sometimes deliberately ignore information on dissidents’ past behavior, and base the targeting of repression purely on demographic characteristics (Rozenas 2020). Moreover, before we can assess whether or not a regime will invest in gathering information, we need insight in the extent to which regimes selectively repress once they are informed; providing such insight is the focus of this paper.

A second strand in the literature explains targeted repression by referring to the differential costs to regimes of repressing particular individuals or groups. Beiser‐McGrath (2019) presents a simple game between a regime and dissident groups as unitary actors. The regime can either adopt a level of preemptive repression that raises a group’s costs just enough to stop if from challenging, or can instead not repress and put up with the challenge. Whether the government represses and avoids the challenge by a group, or does not repress and puts up with it, depends on the audience costs that repression causes, which in turn are determined by the visibility of repression. Other authors who attribute the targeting of repression to differential repression costs include Kalyvas and Kocher (2007, 191) arguing that civilians form easier targets, and Savun and Gineste (2019) arguing the same about refugees. Lust-Okar (2004, 160) and Svolik (2012, 163) argue that regimes repress radicals rather than moderates, because the cost of coopting radicals is relatively large. Yet, these arguments do not consider that dissidents whom it is equally costly to repress, may still contribute differently to collective dissent, justifying targeting them differentially.

A third category of explanations argues that regimes base the targeting of repression on whom they perceive as the largest threat (Josua and Edel 2021, 595-596). In the model of Ma (2020), a regime can use repression to eliminate candidates from elections. If the regime highly values staying in power, it represses to eliminate all candidates that could beat it in elections. If it does not value staying in power that much, it represses to let a desirable successor win. Similarly, Brathwaite (2014) argues that regimes target preemptive repression at ethnic groups whose co-ethnics have revolted abroad, and Zhu and Zhang (2017) argue that leaders use anticorruption campaigns to eliminate their most threatening rivals. However, regimes may not typically face a single rival or rival group, and successful action against a regime may require contributions from several dissidents or dissident groups (Woo and Conrad 2019). For this reason, my model focuses on targeted repression aimed at minimizing the level of collective dissent.

Collective Dissent, and its Repression

Before setting out the model, I additionally position it in the collective-dissent literature (for a wider perspective, see SI Appendix A), and in scarce literature that considers strategically targeted repression of collective dissent.

In order to allow for fine-tuned optimal repressive strategies, in my model dissidents with heterogeneous motivations and abilities decide how much to contribute from a continuum, and produce an outcome from which they collectively benefit according to a continuous production function, with a parameter allowing for a range of technologies. This contrasts with theories of collective dissent specifically focused on regime change, which assume that contributing is an all-or-nothing decision by homogeneous dissidents, and assume that the outcome is determined by a step function where a fixed positive benefit is only obtained when a threshold of contributing dissidents is achieved. With perfect information, dissidents can either coordinate on an equilibrium where none of them contributes, or where the threshold is achieved. Adding imperfect information about the rewards of collective dissent (Shadmehr and Bernhardt 2011) or about the size of the threshold (Bueno De Mesquita and Shadmehr 2023; Shadmehr 2021) leads to a unique equilibrium, and moreover can determine whether contributions are strategic substitutes or complements. In my model, collective dissent is not a coordination problem, and I assume that dissidents have perfect information about each other’s types and about the type of the government. Whether there is strategic substitutability or complementarity is purely driven by the technology of collective dissent,² and this strategic dependence in turn drives the optimal repressive strategy, which is not typically the focus in the collective-dissent literature. Extensions in SI Appendix F and G suggest that allowing for imperfect dissident information does not change my results.

The two most related models in the broader collective-action literature are the following. In Ray et al. (2007) players collectively produce output according to a similar production function as I assume here, and also have heterogeneous costs, but deviating from these authors I add heterogeneity in dissidents’ abilities, and most of all add a strategic saboteur.³ Sandler (2013) models a terrorist organization as a unitary actor producing output by means of two homogeneous inputs (who could be two individuals), the prices of which a counterterrorism agency can selectively increase. He considers the extreme cases of a linear production function, and a production function with perfect complementarity, and shows that in the former case it is optimal to increase all prices to the same extent (similar to my tall-poppies strategy), whereas in the latter case it is optimal to increase a single price to the full extent (similar to my weakest-link strategy). My model instead considers decentralized collective dissent, allows for a continuum of technologies, and allows for heterogeneity. For these reasons, my tall-poppies and weakest-link strategies detail different characteristics based on which dissidents are targeted. Moreover, I obtain a third optimal strategy in the form of the key-player strategy, which though intuitively appealing is never optimal in Sandler’s model. To derive these strategies, I offer a novel way of representing equilibrium output as a function of dissident characteristics.

To my knowledge, the only theoretical paper combining collective dissent and strategic targeting of repression is Rozenas (2020). In his model with two dissidents, a non-supportive dissident tends to obtain higher randomly drawn expressive benefits from opposing the regime than a supportive dissident. The regime first commits to a scheme that individually punishes each dissident conditional on the future observance of a noisy signal indicating opposition, with the purpose of deterring opposition. Rozenas first shows that, in case dissidents obtain no further collective benefit of opposing the regime, the regime punishes the dissidents equally, independently of whether they are supportive or not. Next, defensive opposition is added, where opposing the regime linearly increases the probability that the regime collapses and that the punishment is not incurred. Punishing a dissident can now make her both less and more likely to oppose. Strategic complementarity between dissidents’ decisions to oppose is obtained because of dissidents’ imperfect information about each other’s expressive benefits. The main result is that the regime punishes the non-supportive dissident more severely. The following intuition explains this result. Repressing the supportive dissident leads her to oppose more often, so that the non-supportive dissident opposes more often as well, causing a negative externality for the regime. Repressing the non-supportive dissident makes her oppose less often, meaning that the supportive dissident opposes less often as well, creating a positive externality for the regime.⁴

However, Rozenas’ model cannot explain why in the typical case where dissidents obtain collective benefits, regimes sometimes still treat opposing and supportive dissidents equally, or even treat supportive dissidents more severely. In my model, interpreting dissidents with a lower ex-ante motivation as supportive, and those with a higher ex-ante motivation as non-supportive dissidents, all of these are possible. My model differs from Rozenas in that the regime does not commit to a punishment scheme as a function of the level of the future contributions of individual dissidents, but instead increases the cost to individual dissidents of making contributions. There is no defensive opposition in my model: the only way in which dissidents can avoid incurring higher costs is by making lower contributions. My assumption of allowing for a continuum of collective-dissent technologies allows for strategic substitutability as well as complementarity between the dissidents’ contributions, and this fact, along with the assumption of heterogeneous abilities and motivations, leads to a diversity of optimal repressive strategies.

The Model

Table 1 lists the symbols used in the model. Treated here is a simplified version of the more general model in SI Appendix C, leading to the same qualitative results. A regime and

n

dissidents (with

n \geq 2

) play a game over two stages. At stage 2 of this game, each dissident simultaneously chooses a contribution

e_{i} \in R_{\geq 0}

, leading to output

Q

. The individual dissident

i

obtains payoff

Q^{α} - \frac{1}{μ_{i}} e_{i}

, which it maximizes with respect to

e_{i}

. The first term is the dissident’s benefit from the output; assuming

0 < α < 1

, larger levels of

Q

benefit the dissident less and less extra (law of diminishing returns).⁵ The second term represents the costs of contributing, which are assumed linear, with constant marginal cost

1 / μ_{i}

(SI Appendix C shows that all results extend when these costs are convex). I refer to

μ_{i}

as dissident

i

’s ex-post motivation, which is inversely related to marginal cost (SI Appendix C shows that it does not matter for the results if heterogeneity in dissidents’ motivation is also determined by heterogeneous benefits from output).

Table 1.

List of Symbols.

Symbol	Description
$α$	Parameter reflecting diminishing marginal returns to collective dissent
$e_{i}$	Contribution to collective dissident by dissident $i$
$γ_{i}$	Strength of dissident $i$ , with $γ_{i} = {w_{i} μ}_{i}$
$m_{i}$	Ex-ante motivation of dissident $i$
$μ_{i} = m_{i} - t_{i}$	Ex-post motivation of dissident $i$
$n$	Number of dissidents
$Q$	Output of collective dissent
$σ$	Degree of complementarity between dissidents’ contributions
$t_{i}$	Level of preemptive repression directed by regime at dissident $i$
$T$	Repressive budget of regime
$w_{i}$	Ability of dissident $i$ (weight of dissident i’s effort in production function)

Output is determined by a function $Q = {[\sum_{l = 1}^{n} {w_{l} e_{l}}^{1 - σ}]}^{\frac{1}{1 - σ}}$ , which is a weighted power mean (Bullen 2003) of the contributions, or a so-called constant elasticity of substitution (CES) production function.⁶ I refer to the weight $w_{i}$ (normalized such that $0 < w_{i} < 1$ and $\sum_{l = 1}^{n} w_{l} = 1$ ) as dissident $i$ ’s ability. Note that when two dissidents with different abilities make the same contribution, this costs them the same, but the contribution of the more able dissident matters more for the output. Also, note that when two differentially motivated dissidents make the same contribution, they contribute equally to output; however, this contribution costs more to the less-motivated dissident.

The parameter $σ$ , with $0 < σ < + \infty$ , measures the degree of complementarity between contributions (Ray et al. 2007). Looking at two extremes, approaching $σ = 0$ , the production function approaches a weighted arithmetic mean $Q = \sum_{l = 1}^{n} w_{l} e_{l}$ of the contributions; approaching $σ = + \infty$ , the production function approaches the minimum function $Q = \min (e_{1}, e_{2}, \dots, e_{n})$ and abilities stop mattering (intuitively, with maximal complementarity, each dissident’s contribution is essential, even if this dissident would seem to play a small role). In general, by the weighted power mean inequality (Bullen 2003), ${[\sum_{l = 1}^{n} {w_{l} e_{l}}^{1 - σ}]}^{\frac{1}{1 - σ}}$ decreases in $σ$ : the higher the degree of complementarity, the more $Q$ is determined by the smallest efforts. I exclude cases with $σ < 0$ , as in this case in equilibrium only one dissident contributes, so that there no longer is collective dissent; moreover, dissidents face a coordination problem in this case, as there are multiple equilibria where only one dissident contributes. A multiplicity of equilibria also exists in the extreme cases $σ = 0$ and $σ = + \infty$ (Cornes 1993), which are also not considered.

At stage 1, the regime decides what level of preemptive repression $t_{i}$ to apply to each dissident $i$ , in such a way that $\sum_{l = 1}^{n} t_{l} \leq T$ , where $T$ is the regime’s budget. I assume that the regime is worse off the higher $Q$ , so that the regime can simply be modeled as solving $\min_{t_{1}, \dots, t_{n}} Q$ such that $\sum_{l = 1}^{n} t_{l} \leq T$ . As SI Appendix H shows, results are not changed qualitatively if repressing one dissident takes up more budget than repressing another one. Denoting by $m_{i}$ $i$ ’s ex-ante motivation, with $m_{i} > 0$ , it is assumed that the regime can reduce a dissident’s motivation linearly, so that her ex-post motivation equals $μ_{i} = m_{i} - t_{i}$ for ${t_{i} \leq m}_{i}$ and 0 for ${t_{i} > m}_{i}$ . Put otherwise, ex-post marginal cost $1 / μ_{i}$ relates to ex-ante marginal cost $1 / m_{i}$ as $1 / μ_{i} = 1 / (m_{i} - t_{i})$ . The regime is therefore able with a finite level or repression to make it prohibitively costly for dissident $i$ to contribute, as ex-post marginal cost becomes infinite for ${t_{i} = m}_{i}$ . As SI Appendix C shows, all results extend as long as ex-post marginal cost is a convex function of repression. As the regime is worse off the larger the output, it will set $t_{i} \leq m_{i}$ , with $\sum_{l = 1}^{n} t_{l} = T$ .

Summarizing, I consider a sequential game where at stage 1 the regime (which observes the abilities and ex-ante motivations of all dissidents) sets a repression level $t_{i} \leq m_{i}$ for each dissident $i$ such that $\sum_{l = 1}^{n} t_{l} = T$ . At stage 2, each of the $n$ dissidents observe the repression levels, and simultaneously choose their contributions, achieving a Nash equilibrium at that stage. At stage 1, the regime sets its preemptive repressive strategy anticipating the Nash equilibrium that is played at stage 2. The equilibrium concept used is therefore the subgame-perfect equilibrium.

Interpretation of the Model

Collective-dissent problems can arise at several levels, such as between the rank-and-file members of a dissident group, between several leaders of such a group, or between several dissident groups. While SI Appendix D shows that the results extend when allowing for collective-dissent problems taking place simultaneously at several levels, the focus in the body of the paper is on a single collective-dissent problem considered in isolation. The $n$ dissidents in the model can thus either be interpreted as all dissidents within a group (either outside the regime, inside the regime or spanning regime and opposition), as $n$ leaders in a single dissident group, or as the individual leaders of $n$ groups who can each mobilize within their own group. The contributions made by dissidents take the form of goods and/or services contributed to collective dissent. Depending on the context, the output that dissidents produce is either interpreted as determining the probability of regime change (Rozenas 2020; Shadmehr 2015), the scope of a regime reshuffle, or the extent to which the regime makes concessions – justifying the assumption that the regime negatively values dissidents’ output. Diminishing returns can either be interpreted as dissidents’ efforts contributing less and less extra to the mentioned factors, or as dissidents valuing increases in these factors less and less.

Repression in the model is preemptive repression, as it is set ex-ante (Dragu and Przeworski 2019; Oztig 2023), rather than reactive repression that is set ex-post.⁷ Preemptive repression in the model consists of increasing the costs of contributing goods or services to output (Tilly 1978, 55), that is of reducing dissidents’ motivations. In terms of classifications of repression specified in the literature, such repression can consist of restrictions, which raise the costs of contributing by for example “increasing the difficulty of engaging in speech, association, and assembly” (Davenport 2007, 47), but also of elimination, where the regime represses a dissident to such an extent that it becomes prohibitively costly to contribute.⁸ SI Appendix E also considers preemptive repression that reduces individual dissidents’ ability to contribute. While any real-world case of preemptive repression may both affect dissidents’ motivations and abilities, my focus is on demotivating repression, as it is this aspect of repression that accounts for a wide set of repressive strategies.

More intensive repression in the model takes up a larger part of the budget’s regime.⁹ This makes sense even for collective repression such as curfews, as enforcing such measures is more costly the larger the number of people to whom they are applied (Davenport 2007, 49), or the more strictly they are imposed. The assumption that the repressive budget is fixed corresponds to a short-run perspective, where the regime cannot change its budget (Bueno De Mesquita and Shadmehr, 2023). Assuming a fixed budget makes sense, given my focus on obtaining insight into the optimal targeting of repression.

In the absence of repression, dissidents may perceive the relative costs of contributing to collective dissent differentially based on ideological, ethnic, or religious distance to the regime. This distance determines a dissident’s ex-ante motivation, which is heterogeneous. Repression decreases the dissident’s motivation by increasing the costs of contributing. When dissidents are interpreted as individuals within a single group, ability may be reflected by age, education, or physical condition. In case each dissident is a leader of one of several dissident groups, the leader’s ability depends on the size of the group that she leads, and on her popularity within her group (as she will then be more able to mobilize within her group).

Complementarity in the model can be determined by the degree of specialization, or of criticality. Specialization refers to the extent to which the dissidents take up differentiated roles in contributing to collective dissent. This may be determined by the extent to which dissidents have diverse skills. For instance, consider dissidents as a group of leaders within one and the same movement. Then if this group is composed of task-oriented and people-oriented leaders (Goldstone 2001, 157-58), these may be expected to coordinate on taking up separate roles in protest mobilization (high complementarity), with people-oriented leaders specializing in inspiring protestors and task-oriented leaders specializing in the practical organization of a protest; whereas this is not the case if the leaders have similar skills (low complementarity). The degree of specialization also depends on the extent to which dissident groups are formal rather than informal, where they are informal at their inception and gradually formalize with more division of labor (Staggenborg 1988). Finally, as stressed in sociology, groups may have overlapping membership (Carroll and Ratner 1996). Consider now the interpretation of the model where dissidents are leaders of separate groups, and where a leader’s contribution to collective dissent consists of mobilizing protest within the group that she leads. If membership overlaps only to a limited extent (e.g., because the population is segregated along ethnic or religious lines), leaders’ mobilization efforts are specialized in that they only reach a small part of the population (high complementarity), whereas this is less the case if membership overlaps.

Criticality refers to the extent to which dissidents’ contributions are critical for achieving any given level of collective dissent. Criticality may be determined by the type of the collective dissent. For instance, consider a petition or any other form of collective voice (Dowding et al. 2000), versus a general strike. For a general strike, each dissident’s contribution (interpreted as the number of days one stops working) is critical in the sense that to make the general strike last an extra day, each dissident needs to stop working for an extra day (high complementarity); for collective voice, output (e.g., in the form of obtained concessions) may increase linearly in dissidents’ efforts in voicing their discontent (low complementarity). Also, criticality may be determined by the overall threat level perceived by the regime. When the regime faces a very large threat, a single dissident making a large contribution already suffices to produce a large output; this is the case when output is as high as the highest contribution (e.g., candidate elimination in Ma (2020), in case several candidates could beat the incumbent). When the regime faces a small threat, high output is only achieved if all dissidents make a large contribution; this is the case when output is as high as the lowest contribution. For instance, considering the collective-dissent problem of several groups, it has been argued that the Chinese regime could only be threatened by joint action of several population groups, such as students and workers (Perry 2018). In between these two extremes, collective dissent is a linear function of the contributions, and regimes face an intermediate threat. Given my focus on collective dissent, where all dissidents contribute in equilibrium, such an intermediate threat is the highest threat level I consider.

Results

All proofs are contained in SI Appendix B. Applying backward induction, I first characterize the outcome of the game between the dissidents after any level of preemptive repression has been set. Proposition 1 characterizes the Nash equilibrium of the dissident game at stage 2, where each dissident chooses its optimal effort given the efforts of all other dissidents. To describe the results, I interpret the multiple of the ex-post motivation and the ability of a dissident $i$ , as her ex-post strength $γ_{i}$ , i.e., $γ_{i} = {w_{i} μ}_{i}$ .¹⁰ As Proposition 1(i) establishes, the equilibrium contribution of a dissident increases in her strength. Also, for large complementarity, the contribution of each dissident increases in the other dissidents’ contributions (strategic complementarity), whereas the opposite is the case for small complementarity (strategic substitutability).

Proposition 1(ii) establishes that the relative equilibrium contribution of any two dissidents is a positive function of their relative strengths, where for minimal complementarity the contribution of the stronger dissident is infinitely larger, and for maximal complementarity the contributions approach each other, so that in the limit dissidents take the same contribution whatever their strengths. Intuitively, for small complementarity, the fact that the strongest dissident sets a higher contribution, leads other dissidents to drastically reduce their contributions, which leads the stronger dissident to set an even higher contribution; in the limit, only the strongest dissident contributes. For large complementarity, as each dissident’s contribution is essential, there is little added value of one dissident contributing more than another.

Finally, Proposition 1(iii) establishes that equilibrium output is a positive function of a weighted power mean of the dissidents’ strengths, with abilities serving as weights. This novel way of presenting equilibrium output will turn out useful to derive the regime’s optimal targeting strategy at stage 1. Looking at some benchmark cases, for minimal complementarity ( $σ \to 0$ ), the weighted power mean approaches a maximum function of the dissidents’ strengths, $\max (γ_{1}, γ_{2}, \dots, γ_{n})$ : given the strong strategic substitutability, output is almost fully determined by the characteristics of the strongest dissident. For maximal complementarity ( $σ \to + \infty$ ), the weighted power mean approaches a harmonic mean of the dissidents’ strengths, ${[\sum_{l = 1}^{n} {w_{l} γ_{l}}^{- 1}]}^{- 1}$ . In the harmonic mean, the output is to a larger extent determined by the lowest strengths: given the strong strategic complementarity between contributions, the fact that the dissident with the smallest strength makes a small contribution, induces other stronger dissidents to make a small contribution as well. Given that strength itself is defined as ability times motivation, and given that abilities serve as weights, because of the specific form of the harmonic mean, equilibrium output in this case only depends on motivations, and not on abilities, that is the weighted power mean approaches ${[\sum_{l = 1}^{n} {μ_{l}}^{- 1}]}^{- 1}$ . Intuitively, if the contribution of each dissident is an essential ingredient in producing output, the weight of each ingredient (as reflected by ability) stops mattering. Finally, for intermediate complementarity ( $σ = 0.5$ ), the weighted power mean takes the form of a simple arithmetic mean of dissidents’ strengths, $\sum_{l = 1}^{n} w_{l} γ_{l}$ . This is the intermediate case where equilibrium output is neither more heavily determined by the largest strengths, nor by the smallest ones.

Proposition 1

In the Nash equilibrium of the stage-2 dissident game:

(i) Equilibrium contribution of any dissident is a positive function of her strength, and a negative (respectively positive) function of output for small (respectively large) complementarity, so that dissidents’ contributions are strategic substitutes (respectively strategic complements) ( ${e_{i}}^{*} = {(α γ_{i})}^{1 / σ} {Q^{*}}^{\frac{α - 1 + σ}{σ}}$ , where $\partial {e_{i}}^{*} / \partial Q^{*} ⋚ 0$ for $σ ⋚ 1 - α$ );

(ii) The relative equilibrium contributions of any two dissidents are a positive function of their relative strengths. For minimal complementarity, the strongest dissident approaches being the only one to contribute; for maximal complementarity, dissidents’ contributions approach each other ( ${e_{i}}^{*} / {e_{j}}^{*} = {(γ_{i} / γ_{j})}^{1 / σ}$ ; when $γ_{i} / γ_{j} > 0$ , ${e_{i}}^{*} / {e_{j}}^{*}$ is a decreasing function of $σ$ , where for $σ \to 0$ , ${e_{i}}^{*}$ is infinitely larger than ${e_{j}}^{*}$ ; for $σ = 1$ , ${e_{i}}^{*} / {e_{j}}^{*} = γ_{i} / γ_{j}$ ; for $σ \to + \infty$ , ${e_{i}}^{*} \to {e_{j}}^{*})$ ;

(iii) Equilibrium output is a positive function of a weighted power mean of the dissidents’ strengths, ranging from the maximum function of the strengths ( $σ \to 0$ ), to an arithmetic mean ( $σ = 0.5$ ), to a harmonic mean ( $σ \to + \infty$ ) ( $Q^{*} = f {{[\sum_{l = 1}^{n} {w_{l} γ_{l}}^{\frac{1 - σ}{σ}}]}^{\frac{σ}{1 - σ}}}$ , with $f (.) = α^{1 / (1 - α)} {(.)}^{1 / (1 - α)}$ ).

Anticipating the outcome for any level of preemptive repression as given by Proposition 1, the regime sets its optimal repressive strategy. In the simplified model in the body of the paper, I focus on the case where $T \leq m_{\min}$ , where $m_{\min}$ denotes the motivational level of the least-motivated dissident. In this case, the repressive budget is relatively small, such that no dissident’s ex-post motivation can ever be reduced to zero. An artifact of the production function of the simplified model is that for $σ \geq 1$ , output can always be reduced to zero by setting one of the contributions to zero. When the budget exceeds one or more of the ex-ante motivations, the regime can then reduce output to zero by setting any of these motivations ex-post to zero, so that there are several optimal targeting strategies. As shown in the generalized model in SI Appendix C, this artifact of the model is avoided when assuming an autonomous, non-strategic part of the contribution. Such a more general model allows for any repressive budget, and leads to qualitatively identical results to the ones obtained below.

Consider first the case where all dissidents are homogeneous (i.e., for each dissident $i$ it is the case that $w_{i} = 1 / n$ , and $m_{1} = m_{2} = \dots = m_{n})$ . Then Proposition 2 establishes that the regime targets each dissident at the same level for small complementarity, and targets one arbitrary dissident with the full budget for large complementarity. This is because the weighted power mean in Proposition 1(iii) is to a larger extent determined by larger strengths for small complementarity, so that the regime avoids any strength being the largest. Intuitively, if one dissident would be made less strong than others, then this is ineffective as it causes a negative externality to the regime, with other dissidents taking over and contributing more. For large complementarity, the weighted power mean is to a larger extent determined by smaller strengths, meaning that the regime can have most impact by weakening one dissident maximally, where because of the assumed homogeneity the regime is indifferent about which dissident to target. Intuitively, targeting one dissident causes a positive externality to the regime because it leads other dissidents to contribute less. It should be noted that the cutoff for optimality of the two repressive strategies in Proposition 2 ( $σ = 0.5$ ) need not coincide with the cutoff for strategic substitutability and complementarity ( $σ = 1 - α$ ; see SI Appendix). This is because the optimal repressive strategy is also determined by the shape of the function relating repression to ex-post motivation. However, the results apply for a wide range of such shapes (see SI Appendix C), and the concept of strategic dependence serves to provide intuitions for the results.

Proposition 2

In the homogeneous game, let $T \leq m_{\min}$ , where $m_{\min}$ denotes the motivational level of the least-motivated dissident. Then in the subgame-perfect equilibrium the regime adopts the following strategies:

(i) for smaller complementarity ( $σ$ such that $0 < σ < 0.5$ ), the regime spreads repression evenly over all dissidents (for every dissident i, $t_{i}^{*} = T / n$ );

(ii) for larger complementarity ( $σ$ such that $0.5 < σ < + \infty$ ), targets a single arbitrary dissident with the full budget.

To analyze the heterogeneous case, consider first the benchmark cases. For minimal complementarity ( $σ \to 0$ ), as output approaches being fully determined by the dissident with the ex-post largest strength, a regime that wants to minimize output will prioritize targeting the ex-ante strongest dissident, but not to the extent that ex-post this dissident is no longer the strongest. Clearly, if the regime would target the ex-ante strongest dissident in such a way that the ex-ante second-strongest dissident becomes the strongest ex-post, then this dissident approaches being the only one to contribute (strong strategic substitutability), so that part of the repression targeted at the ex-ante strongest dissident is wasted. To avoid incurring this negative externality, the regime equalizes the ex-post strengths of the strongest dissidents as much as possible. I refer to this as a tall-poppies strategy.¹¹

For maximal complementarity ( $σ \to + \infty$ ), as output approaches a positive function of ${[\sum_{l = 1}^{n} {μ_{l}}^{- 1}]}^{- 1}$ , given that abilities do not play any role in equilibrium output, they do not play any role in the regime’s optimal targeting strategy either. Given that equilibrium output is disproportionately determined by the motivation of the least-motivated dissident, the regime will maximally focus repression on this dissident, independently of her ability. I refer to this as a weakest-link strategy. Intuitively, the regime can maximally exploit the positive externality it obtains from dissidents reducing contributions when others reduce contributions (strategic complementarity), by targeting the dissident who makes the smallest contribution. In the generalized model in SI Appendix C, for larger repressive budgets, if the repressive budget is large enough to minimize the motivation of the least-motivated dissident, the regime additionally targets the second least-motivated dissident, and so on, so that in this case the weakest-link strategy can still target multiple dissidents.

Finally, for a particular intermediate degree of complementarity ( $σ = 0.5$ ), as output equals a positive function of $\sum_{l = 1}^{n} w_{l} γ_{l}$ , marginally reducing the motivation of a dissident always reduces output to a larger extent the more able the dissident, and this independently of her motivation. It follows that the regime maximally targets the most able dissident. I refer to this as a key-player strategy. Intuitively, as it obtains neither a positive nor a negative externality of targeting a dissident, the regime targets the dissident that weighs most. In the generalized model in SI Appendix C that allows for large repressive budgets, the key-player strategy can again target multiple players.

The three optimal repressive strategies considered are defined in Definition 1, with a definition in analytical terms relegated to SI Appendix A. As I will now show, for the entire range of considered degrees of complementarity, either of these three strategies is the optimal targeting strategy. In order to consider degrees of complementarity beyond the benchmark cases above, both the weakest-link and tall-poppies strategies need to be generalized such that they can be biased in favor of targeting the most able dissidents (Definition 1). That the regime has such a bias is intuitive, since for $σ = 0.5$ the regime exclusively takes ability into account when deciding whom to repress.

Definition 1

In the heterogeneous game, the following targeting strategies are considered:

(i) Tall-poppies strategy: equalize the ex-post strengths of the ex-ante strongest dissidents as much as possible (according to a modified measure of strength ${w_{i} μ_{i}}^{1 - 2 σ}$ );

(ii) Key-player strategy: reduce as much as possible to zero the motivations of the most able dissidents;

(iii) Weakest-link strategy: reduce as much as possible to zero the motivations of the ex-ante least motivated dissidents (where dissidents are ranked according to a modified measure of motivation ${m_{i}}^{2 σ - 1} / w_{i}$ ).

It should be noted that these three strategies are not always mutually exclusive. For instance, if motivation and ability are perfectly negatively correlated, there is no distinction between the key-player strategy and the weakest-link strategy (only the most able, least motivated dissident is targeted). Also, if motivation and ability are perfectly positively correlated and the budget is small, there is no difference between the tall-poppies strategy and the key-player strategy (each time, only the strongest and most able dissident is targeted). However, it suffices that there are at least three dissidents and that the most able, the strongest, and the least-motivated dissident are different individuals, for the three strategies to be distinctive. Following a convention to only consider the regime as employing a tall-poppies or a weakest-link strategy if this strategy cannot at the same time be characterized as a key-player strategy (as makes sense given the intuitive appeal of the key-player strategy), Proposition 3 offers a full characterization of the regime’s optimal targeting strategies in the subgame-perfect equilibrium of the game when the budget is small. It follows that, starting from minimal complementarity, as complementarity increases, the regime decides whom to repress more and more based on ability rather than motivation, until a middle range of complementarities is reached where the most able dissident is targeted, independently of her motivation. However, as complementarity is again increased up from this middle range, the regime focuses repression more and more on the least-motivated dissidents, until for maximal complementarity only motivation counts, independently of ability.

Proposition 3

In the heterogeneous game, let $n \geq 3$ , and let the most able, the strongest, and the least-motivated dissident all be different individuals. Let $T \leq m_{\min}$ , where $m_{\min}$ denotes the motivational level of the least-motivated dissident. Then degrees of complementarity $σ_{1}$ , $σ_{2}$ , exist with $0 <$ $σ_{1} < 0.5 <$ $σ_{2} < + \infty$ such that as a function of complementarity, in the subgame-perfect equilibrium the regime adopts the following strategies (see Definition 1):

(i) for small complementarity ( $σ$ such that $0 < σ < σ_{1}$ ), the tall-poppies strategy;

(ii) for intermediate complementarity ( $σ$ such that $σ_{1} \leq σ \leq σ_{2}$ ), the key-player strategy;

(iii) for large complementarity ( $σ$ such that $σ_{2} < σ < + \infty$ ), the weakest-link strategy.

With the exception of the case of perfect correlation between ability and motivation (for an example, see SI Appendix B), $σ_{1}$ and $σ_{2}$ in Proposition 3 cannot be solved in closed form. One is then left to wonder what constitutes intermediate complementarity. As Proposition 4 shows, for each considered degree of complementarity, a sufficiently heterogeneous vector of abilities can be found such that the key-player strategy becomes optimal. Put otherwise, what is considered as intermediate complementarity in Proposition 3, takes up a larger range the larger heterogeneity between dissidents’ abilities (relative to any heterogeneity in dissidents’ motivations). This is already intuitive from the homogeneous case in Proposition 2: as the two optimal strategies there are limit cases of the tall-poppies and weakest-link strategies, the key-player strategy is never optimal, showing that the optimality of the key-player strategy depends on heterogeneity in abilities.

Proposition 4

In the heterogeneous dissident game, let $T \leq m_{\min}$ , where $m_{\min}$ denotes the motivational level of the least-motivated dissident. Then for any given degree of complementarity $σ$ with $0 < σ < + \infty$ , all else equal the ability of one dissident can always be made large enough compared to the ability of another dissident (without reducing the ability of the latter dissident to zero) so that the key-player strategy becomes optimal.

It should be noted that when differences between strengths are not too large, the regime tends to target fewer dissidents as complementarity increases up from minimal complementarity. For minimal complementarity, the regime applies the tall-poppies strategy to equalize the strengths of several ex-ante strongest dissidents. As can be seen from the modified measure of strength employed by the regime, ${w_{i} m_{i}}^{1 - 2 σ}$ (see Definition 1(i)), the closer $σ$ is to 0.5, the less this measure is reduced by reducing motivation, so that the regime will tend to focus its repression on fewer and fewer dissidents as complementarity increases.

As SI Appendix C shows, all results are qualitatively the same in a model of centralized collective dissent where a leader can decide on the contributions of individual dissidents. This is because the optimal contributions set by such a leader depend in similar ways on dissidents’ motivations and abilities, in spite of the fact that with decentralized collective dissent dissidents contribute less. However, if dissidents have differential benefits from collective dissent, a leader maximizing total payoff will not take these into account when setting contributions, and therefore differential benefits do not matter for the optimal repressive strategy. This contrasts with the decentralized collective-dissent problem, where dissidents contributions do depend on their benefits, which in turn matter for the optimal repressive strategy.

Applications

I now look at possible applications of the three optimal repressive strategies. Key for applying the model is to evaluate for the particular type of collective dissent considered how complementary dissidents’ contributions are, and how heterogeneous dissidents are in their ability to contribute. SI Appendix I shows that the model can be extended to targeted cooptation (as increasing a dissident’s benefit of being coopted raises her opportunity cost of contributing to collective dissent) and provides applications.

Key Player

Considering political parties as unitary actors, the key-player strategy may explain why the Egyptian regime prohibits the Muslim Brotherhood (Johnston 2012). As the largest opposition party, its ability to contribute to collective dissentagainst the regime, is much larger than that of smaller opposition parties. Even if opposition parties’ contributions are complementary, the large degree of heterogeneity caused by the presence of a dominant opposition party makes the key-player strategy optimal to the regime. When it comes to dissidents interpreted as leaders, as pointed out by Buckley et al. (2022), there are many historical examples of repression targeted at prominent regime members who get to be perceived as threats (such as Trotsky in the Soviet Union) or at prominent opposition members (such as Cardinal Romero in El Salvador; Bob and Nepstad 2007). Even if complementarity between opposition members’ contributions is low, when the popularity of prominent figures by far exceeds that of other dissidents, they are much more able to mobilize (Buckley et al. 2022), and following the key-player strategy will be targeted.

With large repressive budgets, the key-player strategy may also lead to the targeting of large numbers of dissidents. An example is found in the Anti-Rightist Campaign of 1950s China (Zeng and Eisenman 2018) targeted at intellectuals, seemingly possessing the highest ability to act against the regime; another example is found in the Cheka’s Red Terror in the early years of the Soviet Union, where repression took place depending on “education, upbringing, origin, or profession” (Rozenas 2020, 1257). In these examples, dissidents were targeted based on their abilities, independently of their motivations, as witnessed by the fact that the victims of the Anti-Rightist Campaign included highly-educated party members with apparently low motivation to act against the regime (Zeng and Eisenman 2018), rather than more motivated non-party members with a low education. Standard explanations for such a repressive strategy include an anti-intellectual ideology within the regime (ibid.), or statistical discrimination (Phelps 1972) by a regime able to observe abilities but not motivations and that therefore casts its net wide (Liu 2022). My model provides an alternative explanation that is only driven by the regime’s purpose to survive, and applies even if the regime does have information on motivations. Key is that heterogeneity in abilities, for example in education, is sufficiently large, in which case the key-player strategy is optimal for all but the lowest and highest degrees of complementarity.

Tall Poppies

In spite of the strong intuitive appeal of the key-player strategy, its effectiveness has been questioned in security studies (Johnston 2012; Ryckman 2020), studying for example counterinsurgency in Iraq and Afghanistan (Long 2014). Specifically, when all preemptive repression is targeted at the dissident who is expected to make the largest contribution (the “leader”), the result may be that another dissident takes over as the leader in doing most of the work (Long 2014), so that leadership targeting resembles a whack-a-mole game. This is what my model predicts when dissidents have similar skills and therefore take up similar roles when contributing to collective dissent, with low complementarity of dissidents’ contributions. In such circumstances, while in the absence of repression the strongest dissident does most of the work, once all repression is focused on this dissident, the second-strongest dissident takes over. Moreover, if the strengths of the two strongest dissidents differ only to a limited extent (low degree of heterogeneity), the level of collective dissent does not decrease that much when adopting the key-player strategy and focusing all repression on the strongest dissident. In line with the tall-poppies strategy, the best way for the regime to reduce collective dissent is then to broadly repress dissidents, even if this means spreading its budget thin, as dissidents that could take over doing most of the work are then also hit.

For another application, let us follow Josua and Edel (2015, 292) in considering restrictions on career opportunities as a form of preemptive repression. Gueorguiev and Schuler (2016) interestingly find that in China and Vietnam officials are less likely to be promoted when they stand out in their number of Internet search queries. These may constitute a good measure for officials’ ability to contribute to within-regime dissent, and possibly also for their motivations to do so. By failing to promote such officials, the regime relegates them to less visible jobs, ensuring that officials’ strengths to contribute to within-regime dissent are equalized. More generally, the tall-poppies strategy may explain the relation identified in social psychology between authoritarianism, and negative attitudes towards high achievers (Feather 1993). In terms of my model, the tall-poppies strategy is effective if the level of collective dissent approaches a linear function of contributions; this makes sense if the threat officials pose to the regime consists of the collective voice that these officials produce, where the pressure this puts on the regime may relate to contributions in a close to linear way, with complementarity between officials’ contributions low. The case for the tall-poppies strategy is moreover strengthened if heterogeneity between officials is relatively low.

Weakest Link

Again referring to the key-player strategy, in spite of the strong intuition behind it, targeted repression is sometimes focused on what appear the least-motivated dissidents, rather than the most able ones. For instance, in the Syrian civil war, the regime has been reported to target moderate insurgents rather than Islamist insurgents (Becker 2015; Shadmehr and Bernhardt 2019), even though the Islamist insurgents seem to constitute the largest threat to the regime in being the most able to overthrow it, and the moderate opposition seems ideologically the least distant from the regime and therefore the least motivated to overthrow it. A standard explanation is that in this way the regime ensured that the opposition was perceived as consisting of religious extremists (Becker 2015). In terms of my model, an alternative explanation is that, given the non-overlapping constituencies of these groups, each group was critical for overthrowing the regime, and that for this reason repression was focused on the weakest link. In a different civil war application, consider insurgency as a collaborative effort between specialized dissidents, with combatants’ contributions consisting of fighting, and civilians’ contributions of (possibly reluctantly) providing resources such as food. Even though combatants may provide the bulk of the contributions in the uprising, the insurgency cannot succeed if civilians do not provide resources. The strategy of “draining the sea to catch the fish” (Valentino et al. 2004) targets the less-motivated civilians, rather than the combatants themselves.

A final application may be found in the targeting of swing voters by the Mugabe regime in Zimbabwe, even though these were ideologically the closest to it (Robinson and Torvik 2009). The point may be again that in rigged elections, only votes against the regime from the entire opposition could overthrow the regime (high complementarity), and that the best strategy was therefore to target the voters that were already least motivated to vote against the regime. In each of these applications, the weakest-link strategy is optimal because each player’s contribution to collective dissent is highly critical, trumping the fact that dissidents are heterogeneous in their ability to contribute.

Discussion

While I have identified examples where regimes appear to apply the optimal repressive strategies predicted, the theory can only be rigorously tested by finding a proxy for the degree of complementarity in large data sets on repressive events, and by checking whether repressive events that are as similar as possible except for the proxy, lead to different targeting strategies as predicted. For an example where such an empirical analysis could be feasible, consider strikes and targeted cooptation. Complementarity may be measured in this context by the tightness of the labor market. In a tight labor market, a strike by a part of the personnel may suffice for a successful outcome (low complementarity), whereas in a slack labor market, only a general strike may achieve success (high complementarity). Indeed, in his treatment of US labor strikes in the 1910s and 1920s, Brown (2000) observes a variety of employer strategies to preempt strikes, including giving benefits to white workers (resembling the key-player strategy), and giving benefits to black workers (resembling the weakest-link strategy). The question to investigate is then whether the use of these strategies depends on the tightness of the labor market as predicted.

An alternative way to test the theory is to derive further predictions from extensions of the model. First, I have shown that depending on the degree of complementarity, the level of repression is conditioned on either dissidents’ motivations, on their abilities, or on both. Yet, motivations are harder to observe (Ritter and Conrad 2016, 87). To what extent will the regime invest in surveillance to observe motivations? My analysis suggests that such investments make sense both for low and high complementarity (as in these cases motivational levels matter for the targeting strategy), but not for intermediate complementarity (as in this case the key-player strategy is optimal and targeting takes place based on abilities only). Moreover, a wider range of complementarities count as intermediate the larger the heterogeneity in abilities (Proposition 4). Thus, the results predict less investment in surveillance when there is large heterogeneity in abilities.

Also, while the model can be applied to both targeted repression and cooptation, I have not looked at the choice between the two (Svolik 2012, 9-10; Shadmehr and Haschke 2016; Xu 2021). There are two reasons to expect that cooptation is used more often than repression for large complementarity, in which case my theory predicts that regimes should target the least motivated. First, if the least motivated are also the ones that have the smallest ideological distance to the regime, then the regime may incur extra costs when repressing these, which are avoided in case of cooptation. Second, cooptation counters the lack of information the regime may have about motivations, by making the least-motivated self-select into cooptation (see SI Appendix I). In contrast, with intermediate and small complementarity, cooptation self-selects the least motivated, which may not be the most able dissidents (as required by the key-player strategy) or the strongest dissidents (as required by the tall-poppies strategy).

Furthermore, I have assumed regimes have a fixed repressive budget. Endogenizing the budget, the question arises how effective budget increases are in reducing collective dissent. As formally treated in SI Appendix J, a budget increase away from zero has relatively large impact for large complementarity when the smallest motivational level is low; intuitively, the weakest-link strategy is very effective when the weakest link is already very weak to begin with. At the same time, a budget increase has relatively large impact for small complementarity when the ability of the most able dissident is very large; intuitively, output is then predominantly determined by this dissident, and reducing her motivation has a large impact. Given the large impact of preemptive repression in both these cases, regimes may then be expected to focus more on preemptive than reactive repression. Also, it follows from the model that for low complementarity, once the regime has equalized all ex-post strengths (tall-poppies strategy), raising the repressive budget leads to a relatively small reduction in the level of collective dissent, because the regime needs to spread its budget thin across all dissidents. Thus, as soon as the regime has expended enough preemptive repression to equalize dissidents’ strengths, it may put more emphasis on reactive repression.

Finally, while my model is focused on optimal repressive strategies, the analysis can be extended to look at what makes dissident groups resilient in the face of strategic preemptive repression, in terms of group heterogeneity and of the technology of collective dissent. In the absence of repression, dissidents are better off with heterogeneous motivations when complementarity is low, and better off with homogeneous motivations when complementarity is high (Ray et al. 2007). In the presence of preemptive repression, with high complementarity repression is very effective when the motivation of the least-motivated dissident is low; the advantage to dissidents of having homogeneous motivations is then further reinforced by the presence of strategic repression. With low complementarity, repression is less effective when dissidents are homogeneous in their strengths; while in the absence of repression it is better to have heterogeneous motivations, the presence of strategic repression creates a countervailing effect where homogeneous groups are more resilient. In general, I conclude that strategic preemptive repression mitigates any advantageous effects of forming heterogeneous groups.

Looking at the resilience of different technologies of collective dissent, in the absence of repression, the level of collective dissent is smaller the larger the degree of complementarity (cf. Cornes 1993). In the presence of preemptive repression, if the motivation of the least-motivated dissident is very low, this effect is only reinforced. However, if the most able dissident is much more able than the other dissidents, preemptive repression can do large damage for lower complementarity; in this case, there is a countervailing effect that makes technologies with lower complementarity less resilient. Given these ambiguous results, the resilience of dissident groups in the face of preemptive repression deserves separate attention.

Supplemental Material

Supplemental Material - Whom to Repress: Tall Poppies, Key Players, and Weakest Links

Supplemental Material for Whom to Repress: Tall Poppies, Key Players, and Weakest Links by Kris De Jaegher in Journal of Conflict Resolution

Footnotes

Acknowledgments

I would like to thank two anonymous referees, the editor, Joyce Delnoij and Emilie Rademakers for helpful comments. Thanks are also due to participants of the internal seminar of Utrecht University School of Economics, and of the 23^rd Annual Conference of the Society for the Advancement of Economic Theory, Santiago de Chile, 2024.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Kris De Jaegher

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Aksoy

Deniz

Menger

Andrew

Tavits

Margit

. 2023. “The Effect of Curfews on Political Preferences.” Journal of Conflict Resolution 67 (1): 3-37.

Becker

Michael

. 2015. “When Terrorists and Target Governments Cooperate: The Case of Syria.” Perspectives on Terrorism 9 (1): 95-103.

Beiser‐McGrath

Janina

. 2019. “Targeting the Motivated? Ethnicity and the Pre‐emptive Use of Government Repression.” Swiss Political Science Review 25 (3): 203-25.

Bob

Clifford

Nepstad

Sharon Erickson

. 2007. “Kill a Leader, Murder a Movement? Leadership and Assassination in Social Movements.” American Behavioral Scientist 50 (10): 1370-94.

Brathwaite

Kirstin J. H.

2014. “Repression and the Spread of Ethnic Conflict in Kurdistan.” Studies in Conflict and Terrorism 37 (6): 473-91.

Brown

Cliff

. 2000. “The Role of Employers in Split Labor Markets: an Event-Structure Analysis of Racial Conflict and AFL Organizing, 1917–1919.” Social Forces 79 (2): 653-81.

Buckley

Noah

Reuter

Ora John

Rochlitz

Michael

Aisin

Anton

. 2022. “Staying Out of Trouble: Criminal Cases Against Russian Mayors.” Comparative Political Studies; forthcoming.

Bueno De Mesquita

Ethan

. 2007. “Politics and the Suboptimal Provision of Counterterror.” International Organization 61 (1): 9-36.

Bueno De Mesquita

Ethan

Shadmehr

Mehdi

. 2023. “Rebel Motivations and Repression.” American Political Science Review 117 (2): 734-50.

10.

Bullen

Peter S.

2003. Handbook of Means and Their Inequalities. Dordrecht: Kluwer.

11.

Carroll

William K.

Ratner

Robert S.

. 1996. “Master Framing and Cross-Movement Networking in Contemporary Social Movements.” The Sociological Quarterly 37 (4): 601-25.

12.

Cornes

Richard

. 1993. “Dyke Maintenance and Other Stories: Some Neglected Types of Public Goods.” The Quarterly Journal of Economics 108 (1): 259-71.

13.

Davenport

Christian

. 2007. State Repression and the Domestic Democratic Peace. Cambridge University Press.

14.

De Jaegher

Kris

. 2022. “Threat of Sabotage as a Driver of Collective Action.” Economic Journal 132 (647): 2339-65.

15.

Dimitrov

Martin K.

Sassoon

Joseph

. 2014. “State Security, Information, and Repression: a Comparison of Communist Bulgaria and Ba’thist Iraq.” Journal of Cold War Studies 16 (2): 3-31.

16.

Dowding

Keith

John

Peter

Mergoupis

Thanos

Van Vugt

Mark

. 2000. “Exit, Voice and Loyalty: Analytic and Empirical Developments.” European Journal of Political Research 37 (4): 469-95.

17.

Dragu

Tiberiu

Przeworski

Adam

. 2019. “Preventive Repression: Two Types of Moral Hazard.” American Political Science Review 113 (1): 77-87.

18.

Esberg

Jane

. 2021. “Anticipating Dissent: The Repression of Politicians in Pinochet’s Chile.” The Journal of Politics 83 (2): 689-705.

19.

Feather

Norman T.

1993. “Authoritarianism and Attitudes Toward High Achievers.” Journal of Personality and Social Psychology 65 (1): 152-64.

20.

Gohdes

Anita R.

2020. “Repression Technology: Internet Accessibility and State Violence.” American Journal of Political Science 64 (3): 488-503.

21.

Goldstone

Jack A.

2001. “Toward a Fourth Generation of Revolutionary Theory.” Annual Review of Political Science 4: 139-87.

22.

Gregory

Paul R.

Schröder

Philipp J. H.

Sonin

Konstantin

. 2011. “Rational Dictators and the Killing of Innocents: Data from Stalin’s Archives.” Journal of Comparative Economics 39 (1): 34-42.

23.

Greitens

Sheena Chestnut

. 2016. Dictators and Their Secret Police: Coercive Institutions and State Violence. Cambridge University Press.

24.

Gueorguiev

Dimitar D.

Schuler

Paul J.

. 2016. “Keeping Your Head Down: Public Profiles and Promotion under Autocracy.” Journal of East Asian Studies 16 (1): 87-116.

25.

Hassan

Mai

Mattingly

Daniel

Nugent

Elizabeth R.

. 2022. “Political Control.” Annual Review of Political Science 25: 155-74.

26.

Herreros

Francisco

. 2006. “‘The Full Weight of the State’: The Logic of Random State-Sanctioned Violence.” Journal of Peace Research 43 (6): 671-689.

27.

Howe

Miles

Monaghan

Jeffrey

. 2018. “Strategic Incapacitation of Indigenous Dissent: Crowd Theories, Risk Management, and Settler Colonial Policing.” The Canadian Journal of Sociology/Cahiers Canadiens de Sociologie 43 (4): 325-48.

28.

Johnston

Patrick B.

2012. “Does Decapitation Work? Assessing the Effectiveness of Leadership Targeting in Counterinsurgency Campaigns.” International Security 36 (4): 47-79.

29.

Josua

Maria

Edel

Mirjam

. 2015. “To Repress or Not to Repress - Regime Survival Strategies in the Arab Spring.” Terrorism and Political Violence 27 (2): 289-309.

30.

Josua

Maria

Edel

Mirjam

. 2021. “The Arab Uprisings and the Return of Repression.” Mediterranean Politics 26 (5): 586-611.

31.

Kalyvas

Stathis N.

Kocher

Matthew Adam

. 2007. “How “Free” is Free Riding in Civil Wars?: Violence, Insurgency, and the Collective Action Problem.” World Politics 59 (2): 177-216.

32.

Liu

Howard

. 2022. “Dissent Networks, State Repression, and Strategic Clemency for Defection.” Journal of Conflict Resolution; forthcoming.

33.

Long

Austin

. 2014. “Whack-a-mole or Coup de Grace? Institutionalization and Leadership Targeting in Iraq and Afghanistan.” Security Studies 23 (3): 471-512.

34.

Lust-Okar

Ellen

. 2004. “Divided They Rule: The Management and Manipulation of Political Opposition.” Comparative Politics 36 (2): 159-79.

35.

Shichao

. 2020. “Candidate Elimination in Competitive Autocracies.” Quarterly Journal of Political Science 15 (1): 105-39.

36.

Mitchell

Terence R.

1982. “Motivation: New Directions for Theory, Research, and Practice.” Academy of Management Review 7 (1): 80-88.

37.

Oztig

Lacin Idil

. 2023. “Big Data-Mediated Repression: a Novel Form of Preemptive Repression in China’s Xinjiang Region.” Contemporary Politics 2023: 1-22.

38.

Perry

Elizabeth J.

2018. “Casting a Chinese ‘Democracy’ Movement: The Roles of Students, Workers, and Entrepreneurs.” In Popular Protest and Political Culture in Modern China, edited by Wasserstrom

Jeffrey N.

Perry

Elizabeth J.

, 74-92. London, UK: Routledge.

39.

Phelps

Edmund S.

1972. “The Statistical Theory of Racism and Sexism.” The American Economic Review 62 (4): 659-61.

40.

Powell

Robert

. 2007. “Defending Against Terrorist Attacks with Limited Resources.” American Political Science Review 101 (3): 527-41.

41.

Ray

Debraj

Baland

Jean-Marie

Dagnelie

Olivier

. 2007. “Inequality and Inefficiency in Joint Projects.” Economic Journal 117 (522): 922-35.

42.

Ritter

Emily Hencken

. 2014. “Policy Disputes, Political Survival, and the Onset and Severity of State Repression.” Journal of Conflict Resolution 58 (1): 143-68.

43.

Ritter

Emily Hencken

Conrad

Courtenay R.

. 2016. “Preventing and Responding to Dissent: The Observational Challenges of Explaining Strategic Repression.” American Political Science Review 110 (1): 85-99.

44.

Robinson

James A.

Torvik

Ragnar

. 2009. “The Real Swing Voter’s Curse.” American Economic Review 99 (2): 310-15.

45.

Rozenas

Arturas

. 2020. “A Theory of Demographically Targeted Repression.” Journal of Conflict Resolution 64 (7-8): 1254-78.

46.

Ryckman

Kirssa Cline

. 2020. “Lasting Peace or Temporary Calm? Rebel Group Decapitation and Civil War Outcomes.” Conflict Management and Peace Science 37 (2): 172-92.

47.

Sandler

Todd

. 2013. “Winning the War on Terror: Supply-Side Perspective.” Defence and Peace Economics 24 (2): 121-32.

48.

Savun

Burcu

Gineste

Christian

. 2019. “From Protection to Persecution: Threat Environment and Refugee Scapegoating.” Journal of Peace Research 56 (1): 88-102.

49.

Shadmehr

Mehdi

. 2015. “Extremism in Revolutionary Movements.” Games and Economic Behavior 93: 97-121.

50.

Shadmehr

Mehdi

. 2021. “Protest Puzzles: Tullock’s Paradox, Hong Kong Experiment, and the Strength of Weak States.” Quarterly Journal of Political Science 16 (3): 245-64.

51.

Shadmehr

Mehdi

Bernhardt

Dan

. 2011. “Collective Action with Uncertain Payoffs: Coordination, Public Signals, and Punishment Dilemmas.” American Political Science Review 105 (4): 829-51.

52.

Shadmehr

Mehdi

Bernhardt

Dan

. 2019. “Vanguards in Revolution.” Games and Economic Behavior 115: 146-66.

53.

Shadmehr

Mehdi

Haschke

Peter

. 2016. “Youth, Revolution, and Repression.” Economic Inquiry 54 (2): 778-93.

54.

Siegel

David A.

2011. “When Does Repression Work? Collective Action in Social Networks.” The Journal of Politics 73 (4): 993-1010.

55.

Slantchev

Branislav L.

Matush

Kelly S.

. 2020. “The Authoritarian Wager: Political Action and the Sudden Collapse of Repression.” Comparative Political Studies 53 (2): 214-52.

56.

Staggenborg

Suzanne

. 1988. “The Consequences of Professionalization and Formalization in the Pro-Choice Movement.” American Sociological Review 53 (4): 585-605.

57.

Steinert

Christoph Valentin

. 2023. “The Impact of Domestic Surveillance on Political Imprisonment: Evidence from the German Democratic Republic.” Journal of Conflict Resolution 67 (1): 38-65.

58.

Sudduth

Jun Koga

. 2017. “Strategic Logic of Elite Purges in Dictatorships.” Comparative Political Studies 50 (13): 1768-1801.

59.

Sullivan

Christopher M.

2016. “Undermining Resistance: Mobilization, Repression, and the Enforcement of Political Order.” Journal of Conflict Resolution 60 (7): 1163-90.

60.

Svolik

Milan W.

2012. The Politics of Authoritarian Rule. Cambridge University Press.

61.

Tanneberg

Dag

. 2020. The Politics of Repression Under Authoritarian Rule. Springer International Publishing.

62.

Tilly

Charles

. 1978. From Mobilization to Revolution. Random House.

63.

Valentino

Benjamin

Paul

Huth

Balch-Lindsay

Dylan

. 2004. “‘Draining the Sea’: Mass Killing and Guerrilla Warfare.” International Organization 58 (2): 375-407.

64.

Woo

Ae Sil

Conrad

Courtenay R.

. 2019. “The Differential Effects of “Democratic” Institutions on Dissent in Dictatorships.” The Journal of Politics 81 (2): 456-70.

65.

. 2021. “To Repress or to Co‐opt? Authoritarian Control in the Age of Digital Surveillance.” American Journal of Political Science 65 (2): 309-25.

66.

Zeng

Zhaojin

Eisenman

Joshua

. 2018. “The Price of Persecution: The Long-Term Effects of the Anti-Rightist Campaign on Economic Performance in Post-Mao China.” World Development 109: 249-60.

67.

Zhu

Jiangnan

Zhang

Dong

. 2017. “Weapons of the Powerful: Authoritarian Elite Competition and Politicized Anticorruption in China.” Comparative Political Studies 50 (9): 1186-220.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.71 MB

Whom to Repress: Tall Poppies,Key Players,and Weakest Links

Abstract

Keywords

Targeted Repression

Collective Dissent, and its Repression

The Model

Interpretation of the Model

Results

Applications

Key Player

Tall Poppies

Weakest Link

Discussion

Supplemental Material

Supplemental Material - Whom to Repress: Tall Poppies, Key Players, and Weakest Links

Footnotes

Acknowledgments

Declaration of Conflicting Interests

Funding

ORCID iD

Data Availability Statement

Supplemental Material

Notes

References

Supplementary Material