Taking the Cue: The Response to US Human Rights Sanctions Against Third Parties Timothy M. Peterson University of South Carolina
[email protected] August 1, 2013 Abstract Although scholars have suggested that sanctions could have an international symbolic effect in which they inform third parties of sender preferences and resolve, studies have not examined whether and when sanctions against one state lead other states to change similar proscribed behavior. In this paper, I examine whether abusive regimes change their respect for physical integrity rights when they witness US human rights sanctions against third parties. Synthesizing contributions from the literatures on sanction effectiveness, reputation, and human rights promotion, I develop a new theory asserting that human rights sanctions can motivate leaders in non-sanctioned states to improve their human rights practices proactivelyor at least to prevent worsened abusewhen they perceive themselves as sufficiently similar to the sanction target. I find support for my expectations in stratified Cox proportional hazards models using data spanning 1976 to 2000.
1
1
Introduction Classic studies suggest that economic sanctions are more than tools of direct coercion;
they hold the potential to change behavior throughout the international system by informing third parties of the sanctioning state’s (sender’s) resolve to punish states acting counter to its preferences (Galtung, 1967, see also Barber 1979; Baldwin 1985; Lindsay 1986). However, subsequent research in the area is more pessimistic. Studies note in particular that major senders, such as the United States, are inconsistent in their use of use sanctions (e.g., Nossal, 1989). These studies conclude that this inconsistency precludes the transmission of clear signals that motivate non-sanctioned states to change policy. However, in light of new research demonstrating the strategic behavior inherent in sanctions episodes, the third-party reputation effect of sanctions deserves reevaluation. Recent studies note that the most effective sanctions need never be imposed because targets will acquiesce to the mere threat of interrupted economic ties (e.g., Drezner, 2003). Furthermore, senders evaluate potential outcomes before deciding whether and how much to make demands against potential targets (Krustev, 2010). In this paper, I take this line of research a step further, examining the strategic behavior of potential targets before any sanction threat has been issued. I explore the conditions in which sanctions against a third party can lead leaders to perceive a likelihood that sanction threats could follow from engaging in a similar proscribed behavior. Assuming that these leaders prefer changing policy to enduring sanctions, I argue that they will change policy before being threatened with sanctions in an attempt to avoid the appearance of weakness associated with backing down to an explicit sanction threat. Applying this argument to the issue of human rights, I hypothesize that leaders witnessing a relevant third-party sanction will accelerate reform intended to improve human rights, or, at the least, attempt to prevent policy changes that would lead to worsening abuse. There are a number of challenges associated with identifying a third-party reputation
2
effect of sanctions, not least of which is isolating the domain of relevance that states consider when witnessing third party sanctions and deciding on their own policy. To isolate a relevant domain in which sanctions could have a wider reputation effect, I focus specifically on human rights sanctions issued by the United States.1 I examine the issue of human rights because previous research highlights the strategic behavior inherent in these practices, and demonstrates that desire for economic ties with large markets like the US can (but does not always) motivate improvement in these practices (e.g., Hafner-Burton, 2005; Cao, Greenhill and Prakash, 2012). Synthesizing the insights from these studies with recent advances in the area of multilateral reputation (Crescenzi, 2007), I uncover when states will change human rights practices proactively when witnessing human rights sanctions against third parties. I demonstrate that this reputation effect occurs when a state witnesses sanctions against a similar third party, with similarity defined in terms of region, regime characteristics, and preexisting political affinity with the US. The finding that human rights sanctions can motivate non-sanctioned states to forestall worsening abuse and even to improve their human rights practices proactively has implications for the literatures on sanctions effectiveness, multilateral reputation, and human rights promotion. First, it suggests the presence of a symbolic impact of sanctions, whereas previous studies have been skeptical that such an impact exists (e.g., Nossal, 1989). Accordingly, my results address an enduring puzzle: why sanctions are used despite the fact that they rarely succeed in changing target behavior. Second, this finding suggests that vicarious learning, previously demonstrated with respect to security policy (Crescenzi, 2007), also functions for coercive economic policy. Whereas some studies are skeptical that reputation matters during crises (Press, 2005), my argument highlights the fact that these effects occur before crises (in this case, sanctions episodes) begin. Finally, my findings highlight a positive impact of human rights sanctions, which previously have been shown to be counter-productive, typi1 Although
human rights sanctions encompass only a small part of coercive behavior, and the US is only one (yet, the most common) sender, this limited scope is useful to preclude unmodeled heterogeneity by sender and issue, which might otherwise bias results. Nonetheless, this project is but a first step towards understanding the reputation effect of economic coercion.
3
cally fostering worse violations of human rights in target states (Wood, 2008; Peksen, 2009). Taken as a whole, my results suggest that the United States faces a tradeoff in its attempt to promote improved human rights practices. Its decision to use economic coercion against abusive states could worsen any such abuse. Yet, the symbolic nature of this policy appears to motivate improved human rights practices throughout the international system.2 I proceed with a discussion of the literatures on strategic sanctions behavior, international reputation, and human rights promotion. Synthesizing arguments from these three literatures, I present a theory that isolates when human rights sanctions against a third party are likely to influence abusive regimes to improve human rights practices, or at least delay worsening abuse. Next, I present my research design, in which I specify Cox regression models examining the duration until human rights practices improve or worsen. I present a statistical analysis of country years spanning 1976 to 2000, using the Political Terror Scale human rights indicator (Wood and Gibney, 2010) and sanctions data from the Threats and Impositions of Sanctions (TIES) project (Morgan, Bapat and Krustev, 2009). I conclude with a discussion of the relevance of my results for scholars and policy-makers, noting that the limited domain of this study nonetheless suggests the wider question of how US foreign policy, including the use of carrots (e.g., foreign aid) and sticks (e.g., military force), influences behavior throughout the international system.
2
Linking Sanction Effectiveness, Reputation, and Human Rights Promotion Scholars have long noted that imposed sanctions rarely convince targets to concede to
sender demands (Galtung, 1967; Baldwin, 1985; Hufbauer et al., 2007; Pape, 1997, 1998; Drury, 1998). However, recent studies demonstrate that observed sanctions appear ineffec2A
recent sanction against the Democratic Republic of Congo regarding human rights abuses committed in pursuit of resource wealth serves to illustrate the contemporary importance of this issue. With this sanction (the implementation of which has been met with considerable resistance), the US could foster improvements in human welfare throughout the international system.
4
tive because those sanctions that would be most successful need not be imposed; targets will acquiesce to sanction threats when the sender holds sufficient economic leverage and the credibility to follow through with its ultimatum (e.g., Drezner, 2003; Nooruddin, 2002). Furthermore, Krustev (2010) demonstrates that the severity of sender demands – and even the decision to make a demand – are endogenous to sender expectations regarding the outcome of a coercion attempt. Krustev addresses a classic puzzle of sanctions research: that higher target costs are not associated with a higher likelihood of target acquiescence to sender demands. The author notes that the sender, inferring that higher target costs provide a greater potential to harm the target, could demand more than it would have if its leverage were less. Research has given less attention to the fact that the potential target’s behavior (i.e., the behavior of a state not yet threatened with sanctions) could likewise be crafted strategically, incorporating the state’s beliefs regarding the likelihood that a sanction threat could follow from its actions. Instead, examination of the target’s strategic behavior has focused on modeling a reputation penalty for targets backing down to economic coercion, and, conversely, a reputation benefit (including possible “insurance against future sanctions”) for targets successfully resisting coercion (Lacy and Niou, 2004, 30). Previous studies have shown that behavior during sanctions episodes depends on each side’s beliefs regarding its opposition’s preferences and resolve (e.g., Morgan and Schwebach, 1997; Lacy and Niou, 2004; Eaton and Engers, 1999). Galtung’s (1967, see also Barber 1979; Baldwin 1985; Lindsay 1986) conception of the expressive, symbolic function of sanctions posits that sender action against one target could also inform other states of the sender’s foreign policy preferences and threat credibility.3 A recent study searches for evidence that this international reputation effect influences target behavior, finding that targets of US sanction threats respond to past decisions by the US whether to back down or impose sanctions against prior resistant targets (Peterson, Forthcoming). Backing down against 3 Lindsay
(1986, 166) points out that senders may use this symbolic effect of sanctions negatively, to avoid creating unfavorable images of themselves. For example, United Kingdom sanctions against Rhodesia arguably were motivated at least in part by a desire to avoid the perception that the UK endorsed the Smith apartheid regime.
5
one resistant target leads future threat targets to resist more often because they view US threats as lacking credibility; conversely, by sanctioning states that resist its threats, the US convinces subsequent targets that its threats are credible, leading them to acquiesce in the threat stage more often. Studies note that a target will update its beliefs about sender threat credibility during a sanctions episode (e.g., Lacy and Niou, 2004; Eaton and Engers, 1999; Hovi, Huseby and Sprinz, 2005). However, it is less clear how states use information about a potential sender before that state makes a demand with an accompanying sanction threat. Leaders who consider engaging in internationally proscribed behavior may look to whether similar behavior led third parties to be sanctioned. Yet, prior studies are pessimistic regarding the prospect that sanctions against third parties lead states to change behavior proactively (Nossal, 1989). Nossal notes that the failure of a third-party reputation effect strong enough to motivate policy change is a consequence of inconsistency. Senders do not reliably follow through with threats, nor do they systematically threaten all states violating norms they advocate. However, it would be difficult to see a reputation effect even when one exists if prospective targets engage in strategic behavior prior to the onset of sanctions episodes. Similar to senders (Krustev, 2010), targets may incorporate reputation effects into their pre-sanction policy-making, which then affects the likelihood that they are threatened. Furthermore, we may fail to observe a reputation effect of sanctions because there is no guarantee that an imposed third-party sanction is relevant to any given state. There has been little systematic research on how information communicated through the imposition of third-party sanctions could affect the beliefs, and therefore the strategic behavior, of leaders deciding whether to engage in some controversial behavior. 2.1
Reputation and Learning
To understand how sanctions against third parties could lead states to change behavior proactively, I connect the strategic sanctions behavior literature to the literature on repu6
tation and learning – and specifically to studies examining third-party reputation effects. Classic studies on reputation introduce mechanisms by which a given action could have widespread consequences throughout the system. For example, Schelling (1967) refers to the “interdependence of commitments,” noting that the credibility of a US deterrent threat in one geographic area is contingent on its behavior in others. To illustrate his argument, Schelling notes that US policy-makers believed that US anti-communist interventions in South Asia could increase the credibility of its threat to intervene against Soviet aggression in West Germany. Indeed, Schelling feared that failure to intervene in South Asia could threaten successful deterrence in Europe. Although Schelling (1967) emphasizes the reputation effect associated with the coercive power of military force, later research suggests a generalizable effect in which states learn from the actions of others. Morrow (1989) demonstrates that the sequential nature of international politics allows states to update beliefs as they witness events, noting that this behavior, often associated with psychological approaches to the study of perception in world politics (e.g., Jervis, 1976), can be understood as a rational process. Levy (1994, 283) defines rational learning as a “a change in beliefs (or the degree of confidence in one’s beliefs) ... as a result of the observation and interpretation of experience.” Levy further distinguishes diagnostic learning, referring to a change in beliefs regarding “the definition of the situation or the preferences, intentions, or relative capabilities of others” from causal learning, regarding changing beliefs in “the consequences of actions.” Noting the potential for learning to function vicariously, Crescenzi (2007) develops a third-party reputation model in which state A learns about some other state B by observing B ’s actions against some third party state C and incorporating the relative similarities between A and C. Crescenzi notes that all interactions between B and C could inform A, but that A will scrutinize all such interactions for relevance. Examining conflict behavior, Crescenzi incorporates information on power similarities and political affinity to identify relative similarity that could trigger vicarious learning. As I note below, relative similarity 7
is also the key concept to define in order to understand a third-party reputation effect of sanctions. It is important to note that skepticism regarding the presence of reputation effects is common. For example, Press (2005) argues that reputations are shed quickly, as states prioritize the current political environment when deciding policy. However, Press focuses on military rather than economic threats. Additionally, Press examines how a state’s reputation from a crisis affects its future adversaries’ behavior during future crises. Yet, it is possible that reputation effects are not evident during crises because potential adversaries are less likely to begin a crisis with a state having a reputation for resolve. In other words, while reputation from cases of immediate deterrence may not matter in future cases of immediate deterrence, it could influence the stability of general deterrence. My argument applies this same logic to sanctions episodes. 2.2
Human Rights Practices as a Strategic Policy Choice
To capture the ability of third-party sanctions to motivate proactive policy change, I focus specifically on the issue of human rights. This issue is crucial to explore given the potential for economic coercion to improve the lives of people living under abusive regimes. Previous studies have shown that sanctions typically lead to worsened abuses in target states (Wood, 2008; Peksen, 2009). However, if sanctions also motivate non-sanctioned states to improve human rights, this policy tool would be less clearly counter-productive. Furthermore, while a third-party reputation effect of sanctions could exist for a variety of issues, human rights practices offer a particularly useful issue with which to examine this phenomenon for two reasons.4 First, human rights practices, and changes therein, are relatively easy to document. Given efforts by scholars to code comprehensively human rights practices by country over time (e.g., Wood and Gibney, 2010; Cingranelli and Richards, 2010), we are able to examine yearly variation in respect for human rights potentially corresponding to variation 4 For
example, US sanctions against Iran over nuclear proliferation (increasingly strict iterations of which continue to be pursued) could factor into a third party’s decision whether to begin pursing nuclear weapons.
8
in sanctioning behavior.5 Second, although the study of human rights is particularly amenable to normative approaches, there has also been a great deal of research noting the strategic behavior evident in human rights practices.6 This contention is illustrated by prior studies aiming to explain why human rights abuses occur. Poe, Tate and Keith (1999, 293) address the question: “Why do some states abuse the dignity of the person? ... If we assume that most political leaders are rational actors, we conclude that they choose to commit abuses of personal integrity rights because they see these inhumane actions as the most effective means to achieve their ends.” The authors note that egregious violations of personal integrity can follow from self-interested behavior, rather than simply from malice. Scholars have used this concept of rational human rights abuse as the foundation for studies on how economic incentives can foster improvement in human rights practices. For example, Hafner-Burton (2005) demonstrates that enforceable human rights clauses in trade agreements are successful in motivating abusive regimes to improve human rights practices in order to maintain preferential access to partner states’ markets. Studies also suggest that trade generally can foster the diffusion of human rights-respecting practices through the “California effect,” wherein domestic groups in human rights-respecting, importing states demand improvements in the practices of their trade partners (Cao, Greenhill and Prakash, 2012).7 Along similar lines, I argue below that some leaders will attempt to improve human rights practices if they perceive that sanctions could follow from continued abuse. Furthermore, I contend that this perception can be prompted by the observation of human rights 5 While
a regime’s nuclear ambitions may be kept secret and the extent of its nuclear weapons may be unclear (see Israel, for example), mass violations of human rights tend to be known at least generally. Even in extreme cases of closed societies (e.g., North Korea or Myanmar), scholars know at least that these regimes are among the worst abusers, even if estimations of abuse severity are rough. Cingranelli and Richards (2010, 421) note that their physical integrity data are coded using ordinal indicators in order to maximize reliability and validity given variation in opportunity to collect information on abuse by state. Similarly, Wood and Gibney (2010, 373) explain that it is difficult and probably not appropriate to attempt to create equivalent values for different classes of abuse. Accordingly, an ordinal indicator is preferable to any indicator that attempts to compare abuses. 6 The sender’s use of sanctions as symbols leading to improved human rights throughout the international system illustrates the “intimate” connection between norms and rationality (see, e.g., Finnemore and Sikkink, 1998). 7 Importantly, this demand for improvement could stem from selfish behavior, as import competitors could stand to lose economically because human rights abuses facilitate lower labor costs for exporters; production is more costly when workers are treated humanely.
9
sanctions against a similar third party.
3
The Reputation Effect of Human Rights Sanctions To uncover the domain in which human rights sanctions can provoke proactive policy
change in non-sanctioned states, I begin with the assumption that leaders seek to remain in power (Bueno de Mesquita et al., 2003). Trade with wealthy states provides economic benefits that leaders seek to maintain. For democratic leaders, trade promotes economic growth that helps secure re-election. For autocratic leaders, trade provides wealth that can be captured directly, or distributed to a selectorate to ensure continued support. Imposed sanctions therefore risk destabilizing leaders by harming the economy. Although classic studies suggest that leaders can insulate themselves – and perhaps even benefit – from sanctions (e.g., Galtung, 1967; Olson, 1979), other studies find that sanctions are costly to leaders, although this impact may vary depending on the political structure of the target state (Marinov, 2005; Allen, 2008; Escriba-Folch and Wright, 2010, but see Licht n.d.). It is this cost that spurs sanctioned leaders to increase repression in order to maintain control (Wood, 2008; Peksen, 2009). For sanctions to have a reputation effect, economic ties between abusive states and known senders, as well as the adjustment costs to the loss thereof (e.g. Crescenzi, 2003; Hirschman, 1945), must be high enough that leaders in these states not yet threatened with sanctions prefer to improve human rights practices – or at least to prevent worsening of abuse – rather than endure lost commerce. One challenge to the identification of a third-party reputation effect of sanctions is that, in many cases, states not facing sanctions could be those where the would-be sender holds little leverage. Krustev (2010) uses the example of US sanctions against Taiwan over its trade of endangered animal parts, citing the lack of similar sanctions – or threats thereof – against China, ostensibly a worse offender. Krustev notes that the lack of demands against China were probably strategic; China was too powerful economically to
10
coerce.8 I also assume that leaders willing to change policy in order to prevent being sanctioned prefer to do so before being threatened with sanctions. This assumption follows because backing down to an explicit sanction threat signals weakness (e.g., Lacy and Niou, 2004). Leaders who appear weak risk losing the support of their winning coalition and, therefore, risk losing power (e.g., Fearon, 1994; Weeks, 2008). Leaders facing sanction threats are victim to a catch-22; they face costs whether they change policy or endure sanctions. However, even if acquiescence is preferable to enduring sanctions, leaders would be better off changing policy without paying the additional audience cost associated with backing down to a sanction threat. Rational leaders seeking to maintain economic ties with known senders should be aware of this potential dilemma. Like senders, (potential) targets should behave strategically, considering potential sanctions outcomes before these episodes even begin. If a leader perceives a sufficient probability that sanctions could follow from its abuse of human rights, and this leader prefers maintaining trade to enduring sanctions, it is cost effective not to wait for a sanction threat before changing policy. The question then becomes: where can leaders obtain information with which to calculate the likelihood that they will be threatened with sanctions? Below, I discuss the potential of third-party sanctions to inform these leaders. 3.1
The Relevance of Third-party Human Rights Sanctions
A reputation effect of third-party sanctions strong enough to motivate proactive change in observing states is far from absolute. Lack of confidence in this phenomenon is due, at least in part, to an understanding that third-party sanctions likely hold little relevance for many (perhaps most) state leaders. Importantly, leaders engaging in fewer abuses than sanction targets may assume that their own practices are not sufficiently severe to provoke senders to take action. However, being at least as abusive as sanction targets is a minimum 8 However,
it does appear in the years since US sanctions against Taiwan that China has enacted new laws against the trade in these endangered animal parts, and has attempted to educate its citizens regarding their lack of medicinal value. However, poaching continues to escalate due to demand among the populace.
11
bar; there are a number of factors that could affect whether leaders in abusive states witnessing third-party human rights sanctions will decide to improve their own practices. For this reputation effect to occur, political leaders in abusive states must engage in vicarious learning, perceiving their state as sufficiently similar to the sanction target that they expect the sender’s willingness to impose sanctions on the target reflects a similar willingness to impose sanctions on them (Crescenzi, 2007). There are a number of characteristics (in addition to actual abuse level) that leaders could use to assess similarity. Crescenzi notes that similarity of military capabilities is an important dimension when examining third-party reputation effects on militarized conflict. The author notes that, for example, relatively weak states will learn more about a potential adversary’s likely action against them from its action against weak third parties. With regard to sanctions, economic development may be the more pertinent measure of power, as developed states can more easily endure lost commerce. The sanctions literature shows that, all else equal, senders hold more leverage over less wealthy states (e.g., Hufbauer et al., 2007). Yet, if a sender imposes sanctions against a relatively wealthy state, other wealthy states could be more likely to infer that the sender is willing to impose sanctions on them despite the fact that actual leverage is lower. Crescenzi (2007) also incorporates foreign policy similarity into his measure of reputation information. This factor could also influence the reputation effect of sanctions given that, when witnessing imposition against a third party, a state is likely to pay attention to how the target’s preexisting affinity with the sender compares to its own. For example, sender allies may largely ignore sanctions that are imposed against sender adversaries, and yet pay close attention if the sender imposes sanctions against an ally. States could also compare their own regime characteristics to those of the target; leaders in democracies may pay less attention to sanctions against autocracies, and vice-versa.9 The 9 Although
democracy correlates with higher respect for human rights, there are a number of democracies that engage in abusive practices, including India and Brazil (e.g., Cingranelli and Richards, 2010). I test for the robustness of all my results excluding these possible outliers. All results are robust.
12
sanctions literature has shown that democracies typically recover from sanctions more quickly (Lektzian and Souva, 2001). Accordingly, senders likely have less leverage over democracies on average. Yet, as with sanctions against wealthy states, if a sender imposes sanctions against a democracy, then other democracies may note its willingness to sanction even when its leverage is lower. Finally, keeping in mind the factors of development, political affinity, and regime type, states may be more sensitive to sanctions against others in their home region. This expectation is based on the premise that regime type, development, trade, and militarized conflicts, as well as cultural factors that are less easily measured, tend to cluster geographically (Gleditsch, 2002b; Brinks and Coppedge, 2006; Buhaug and Gleditsch, 2008; Weinhold, 2002). Beyond these dimensions of relevance, there are several other factors that likely condition the reputation effect of sanctions. First, I focus specifically on the imposition of sanctions against third parties, as opposed to threat behavior, as a critical cue of expected sender behavior. Krustev (2010) notes that, when sender credibility is questionable, little short of imposed sanctions will convince targets to change their policy. It is reasonable to assume that the credibility of a potential sender is inherently in doubt when that sender merely threatens a third party. Developed, human rights-respecting states often send mixed messages on the issue of spreading respect for human rights internationally. Although leaders in common senders like the US tend to give lip service to this notion, these leaders have been known to tolerate abuse in pursuit of strategic interests. Furthermore, foreign policymakers are not always in agreement on the question of supporting human rights improvements abroad. For example, George Kennan, a primary architect of US foreign policy during the Cold War, criticized US advocacy of human rights as detrimental to security interests. Indeed, sanctions policy is political for senders, dependent on the relative strength of interest groups favoring or opposing restricted trade with potential targets (e.g., Kaempfer and Lowenberg, 1988, 1989). Accordingly, a mere threat against a third party may not convince observing
13
states that a sender is serious about restricting trade.10 In light of this issue, I contend that imposed sanctions send the clearest signals to third parties. I also focus on the act of imposition rather than the presence of imposed sanctions as critical cues to third parties. The persistence of a previously imposed human rights sanction against a third party could be indicative of mere status quo bias rather than explicit choice by the sender not to lift sanctions.11 Similarly, states witnessing third parties acquiesce to the threat of human rights sanctions might be unsure whether this action resulted because the sender was very likely to impose sanctions, or if the third party acquiesced to a bluff. I contend that the act of imposition signals the sender’s willingness at a given time to endure lost economic ties.12 Thus, it is this act that best serves as a cue to third parties. Finally, it is important to consider how human rights policy changes occur over time. While minor abusive policies might be improved relatively quickly, larger structural changes to state institutions could take more time to implement.13 Ultimately, I contend that leaders witnessing a relevant third-party sanction over the issue of human rights will attempt to accelerate any changes that would lead to improved human rights practices. At a minimum, these leaders will attempt to slow the implementation of any policies that could worsen abuse. The argument above leads to the following hypotheses: 10 An
imposed sanction could signify only that domestic interests (for example, import competitors) are powerful enough to influence sender policy. Even if sanctions stem from the preferences of domestic interest rather than sender norms on human rights, these sanctions nonetheless inform third parties on the sender’s likelihood of restricting trade. Ultimately, a state’s assessment of interest group competition in the sender could lead it to improve human rights practices even if it thinks sender norms on the matter are weak. Accordingly, while the terms resolve and threat credibility are used often when discussing reputation effects of sanctions, perhaps a more useful concept is overall sender willingness to impose sanctions (Most and Starr, 1989), stemming from its domestic preferences and constraints. 11 For example, the persistence of US sanctions against Cuba arguably follow as much from status quo bias (and specifically from the effective veto power of anti-Castro interests in Florida) as from explicit policy choice. 12 One might also expect that sender capitulation to targets resisting its threats could lead third parties to infer that abuse will not be punished, such that they actually worsen abuses. Previous research suggests that future targets, inferring that a sender is prone to bluffing, are less likely to acquiesce to threats from senders that recently capitulated (Peterson, Forthcoming). However, I contend that even if the sender sends this signal of weakness to the international community, abusive regimes will, at most, remain confident that they need not improve; it is unlikely that states would suddenly decide to be more abusive. I test this contention empirically in robustness tests available by request. These models show that cases of the US backing down against resistant third-party threat targets do not influence respect for human rights. 13 An action more costly than restricted economic ties might produce a larger proactive policy change. For example, one might argue that the US invasion of Iraq served as a message to other dictators with a history of US antagonism. Notably, Muammar Gaddafi subsequently reversed decades of policy and began cooperating with the US. Future research may benefit from addressing broader reputation effects.
14
Hypothesis 1 The duration until improvement of government respect for physical integrity rights will be shorter in the presence of recently imposed human rights sanctions against a sufficiently similar third party. Hypothesis 2 The duration until worsening of government respect for physical integrity rights will be longer in the presence of recently imposed human rights sanctions against a sufficiently similar third party.
4
Research Design To test my hypotheses, I use country-year level data spanning 1977 to 2001. I specify
Cox proportional hazards regression models examining the duration until a state changes its respect for physical integrity rights. Given that I have expectations regarding the relative time until change in human rights practice, but have no expectations on the specific pattern of change, Cox models are especially useful because they do not impose a specific shape on the baseline hazard (Box-Steffensmeier and Jones, 2004). I utilize a latent survivor time approach to examine the competing risks of improvement and worsening human rights practices. Given that I have multiple (i.e., yearly) observations for each state, I include timevarying covariates in each Cox model. States re-enter the analysis immediately after a failure; accordingly, my specifications allow for multiple failures (i.e., multiple instances of improved or worsened human rights practices) per country. This specification has several advantages. First, it allows me to distinguish whether relevant sanctions encourage improvements in human rights, or merely discourage worsened abuse. It also allows me to examine whether any improvements are long-lived. For example, a slight increase in the hazard of improvement would be more promising if there is also a steep decline in the hazard of worsened abuse, but less-so if the hazard of worsened abuse also remains high.14 14 There
are a number of coding decisions that must be made with regard to flagging changes in human rights. For example, if human rights practices become more abusive in a given year, I code this event as a worsening of human rights. However, if human rights practices return to the original level in the following year, there could be disagreement regarding whether this event should be coded as improvement. I contend that such a change could indicate a delayed reputation effect of third party sanctions. The initial decline could have been set in motion before the impact of the third party sanction is absorbed. The observation of a relevant third party sanction then convinced leaders to reverse this policy of heightened abuse. Accordingly, I code this event as an improvement. However, all results are robust in models where improvement or worsening is coded with respect to human rights practices at entry (or re-entry in case of subsequent failures). Additionally, in the current analysis, an improvement in human rights practices is recorded as survival in the model examining worsening practices, and vice-versa. However, results are also consistent if I censor such observations.
15
In order to mitigate omitted variable bias by sender, I restrict my analyses to the examination of human rights sanctions imposed by the United States.15 Each model is stratified by existing human rights practices in order to allow all states with a given pre-existing level of respect for human rights (or lack thereof) to have a group-specific baseline hazard of change.16 Given that I have multiple observations and, in many cases, multiple failures for each state, I cluster standard errors on the state to account for correlated survival times. Finally, I lag explanatory variables 1 year to mitigate the potential for simultaneity bias.17 4.1
The Dependent Variables: Duration until Improved/Worsened Respect for Human Rights
I use two dependent variables to capture improvement and worsening of human rights practices, both of which are coded from the Political Terror Scale (PTS) human rights data (Wood and Gibney, 2010). Specifically, I take the Amnesty International indicator, which varies between 1 (least abusive) to 5 (most abusive). I code improvement in human rights as a decrease in the PTS indicator; accordingly, my first dependent variable is years until a state’s PTS indicator decreases by at least one point. I code worsening human rights as an increase in the PTS indicator. Thus, my second dependent variable is the time in years until a state’s physical integrity index increases by at least one point.18 In robustness check models presented in the supplemental appendix, I replicated results using variables coded from the Cingranelli and Richards (2010) (CIRI) physical integrity 15 This
decision is also useful because it allows me easily to code variables for each state’s preexisting economic and political ties to the US. If I were to look at all potential senders, I would have to use the dyad as the unit of analysis in order to capture these preexisting ties, problematic given that my dependent variable is at the country-year level. Another factor informing this choice is that US leadership in the international system could imply that its behavior is scrutinized more, leading states to have better information regarding likely US preferences and resolve on human rights at any given time than they would have for other senders. Nonetheless, future research could benefit from examining these causal mechanisms more broadly. 16 I do not simply include pre-existing human rights practices as a covariate in my models because doing so causes multiple violations of the proportional hazards assumption. This coding decision also allows me to include states with most abusive and most respectful human rights scores. For these strata, the hazard of worsening and improvement, respectively, is always equal to 0. 17 Accordingly, failure is recorded if my variable for human rights practices changes between year t-1 and year t, while my primary explanatory variables are coded for year t-1. 18 In robustness check models, I coded improvement and worsening as occurring if there is at least a two-point change in the PTS indicator. These alternate specifications are intended to account for the possibility that 1-point changes are due as much to random variation as to leader intention to change practices. Estimates for all explanatory variables are quite similar. The primary difference in these models is that the baseline hazard is lower with respect to both dependent variables.
16
index, finding very similar results. I also reran all models disaggregating the four components of the CIRI physical integrity index: political imprisonment, torture, extrajudicial killing, and disappearance. Results of these analyses support my hypotheses. However, for the purposes of this paper, I contend that an aggregate measure of human rights practices is useful to examine general patterns of abuse. Previous studies aggregating physical integrity rights abuses do so in order to capture a general willingness of leaders to use repression as a means to achieve their ends (e.g., Cingranelli and Richards, 2010). While studies have advocated the utility of examining abuses independently in order to understand the policy choices of abusive regimes (e.g., McCormick and Mitchell, 1997), I contend that the use of an aggregate indicator is particularly justified in an examination of the reputation effect associated with US sanctions because US policy-makers tend not to discriminate by abuse when discussing human rights sanctions.19 This is not to say that human rights practices are necessarily complements rather than substitutes. For example, Cingranelli and Richards (2010) note that incidents of torture have decreased since 1980, while incidents of political imprisonment have increased. A recent paper finds that attention to human rights abuse by naming and shaming organizations leads some states to substitute alternate abuse types (Conrad and DeMeritt, n.d.). The authors recommend not examining one abuse in isolation, as factors affecting one abuse type could also affect others. However, by aggregating abuses, my results should not suffer from this possible bias. 19 Rather,
US leaders tend to speak in quite broad terms about the promotion of “human rights,” without necessarily even defining their specific meaning thereof, providing third parties with information only at this aggregate level. Hufbauer et al. (2007) records several such statements made by US officials. For example, regarding systemic abuses in Myanmar, Representative Stephen J. Solarz (D-NY) notes that sanctions should “encourage political reform and discourage abuses of human rights” (US House of Representatives, 1989; from Hufbauer et al. 2007). After learning of Indonesia’s violation of human rights in East Timor, Representative Ron Machtley (R-RI) authored an amendment to bar military aid for Indonesia, noting that “[w]e are sending a message which says we certainly will not reward behavior which displays such a flagrant disregard for human rights” (US Information Service, 1992; from Hufbauer et al. 2007). Response from the executive is similar. For example, President George H. W. Bush responded to Chinese abuses in Tianamen square by taking a moderate stance, stating that “... I don’t want to see a total break in this relationship .... I want to see us stay involved and continue to work for restraint and for human rights and for democracy” (Washington Post, 1989; from Hufbauer et al. 2007).
17
4.2
Primary Explanatory Variables: Relevant Third Party Sanctions
My primary explanatory variable is a dichotomous, yearly indicator equal to 1 if a given state has human rights practices at least as abusive as those of a state against which the United States imposed sanctions. Sanctions are included only if human rights is an issue at stake.20 This variable is coded using the Threat and Imposition of Sanctions (TIES) data version 3.5 (Morgan, Bapat and Krustev, 2009) imposition year.21 However, I create four variations of relevant third party sanctions over human rights, each of which specifies a specific dimension that a state might consider when comparing its similarity to sanction targets.22 The first of these is an aggregate relevant sanction variable. This dichotomous variable is equal to 1 for states in years where the United States has imposed human rights sanctions against a third party state that scores the same or lower (i.e., less abusive) on the Political Terror Scale.23 Second, I code a region-specific variant of the relevant third-party sanction variable. This variable is equal to 1 if the criteria above apply and the state is in the same region as the target state. I identify regions using the Correlates of War region codes (Correlates of War Project, 2011). As noted above, states witnessing US sanctions in their region may pay more attention to this signal, given likely similarities associated with geographic clustering of human rights abuse and other state characteristics (e.g., Gleditsch, 2002a). Third, I code a state-specific variant of the relevant third-party sanction variable intended 20 Specifically,
one of the issue1, issue2 or issue3 variables must be equal to “8.” These sanctions are designed to “end repressive laws, policies, or actions” (Morgan, Bapat and Krustev, 2009). 21 Two TIES cases record imposition, but do not record a year thereof. I supplement these missing values with data from other sources, primarily Hufbauer et al. (2007). These missing values could result if imposition is announced, but then the target acquiesces before actual implementation begins. Indeed, this behavior is expected in cases where the target preferred acquiescence to imposed sanctions but judged the sender’s threat as not credible (Hovi, Huseby and Sprinz, 2005). I include these cases in my analysis because the announcement of imposition demonstrates US commitment to sanctions over human rights, a commitment that is observed by non-sanctioned states. 22 In the supplemental appendix, I create a continuous indicator of reputation that incorporates a count of the number of dimensions on which the observer state is similar to the sanction target. All results are robust. 23 There are three years where the US sanctions two states over human rights, and one year (1977) where the US sanctions eight states. In these years, I use the average of the political terror scale scores for comparison purposes. In the supplemental appendix, I create an alternate, weighted measure that adds the reputation score of each sanction imposed in a given year. Furthermore, in the supplemental appendix, I create a relative abuse weight scheme, where information from a third-party sanction is weighted down if the observer is more respectful than the target, and weighted up if the observer is more abusive. However, the creation of this variable requires arbitrary weights to be applied. Accordingly, I prefer the simpler, dichotomous measure used here.
18
to capture salient regime characteristics. This variable is equal to 1 if the observer state is at least as abusive as the sanction target and this state is similar in terms of democracy and low-income status. Specifically, I code democracies as states scoring at least a 6 on the Polity combined score (Marshall and Jaggers, 2010) and I code low-income states as those where income per capita is less than 1,025 dollars, adjusted for inflation, as defined by the World Bank. To illustrate: if the US sanctions a low-income, non-democracy in a given year, then the target-specific third-party sanction variable will equal 1 only for abusive states that are also low-income, non-democracies. As noted above, regime type and wealth are both correlated with state vulnerability to sanctions – in other words, the relative “strength” of a state vis-`a-vis a potential sender. Accordingly, as Crescenzi (2007) notes, states should pay more attention to the actions of a potential adversary (in this case, a potential sender) when witnessing that state’s behavior towards third parties of similar strength. Finally, I code an affinity-specific variant of the relevant third-party sanction variable. This variable is equal to 1 if the observer state is at least as abusive as the sanction target and this state has the same alliance status with the US as the sanction target. For example, if the US sanctions a non-ally in a given year, only other non-allies would be coded as 1 in that year. This variable is intended to capture potentially discriminatory US policy against abusers depending on its security interests. For example, if Israel were to witness sanctions against Syria over Syrian human rights abuses, it would not necessarily expect similar treatment even if it engages in human rights abuses in the Palestinian territory because it is a key US ally in a turbulent region.24 24 I
choose alliance status rather than a continuous measure of affinity such as the “S” score from Signorino and Ritter (1999) to avoid arbitrary coding. I would have to create arbitrary thresholds for “similar” foreign policy preferences if I were using a continuous indicator. Furthermore, some studies suggest that S scores, while having advantages, are also limited in ability to capture foreign policy preferences (e.g., Sweeney and Keshk, 2005). This problem would likely be exacerbated by the use of arbitrary thresholds. Finally, given that I am most interested in US security interests rather than overall preference similarity with other states, alliance status is a better indicator.
19
4.3
Control Variables
To isolate the third-party reputation effect of US human rights sanctions, I include control variables to account for other factors that could correlate with the observation of a relevant sanction and also influence human rights practices directly. First, I code three variables capturing direct sanctioning behavior. Specifically, I include dichotomous indicators equal to 1 if the US has: (1) imposed sanctions against a state in a given year, (2) backed down from imposing sanctions against a state in a given year, and (3) initiated a threat against a state in a given year. I code all of these variables using the TIES data. I take data on the imposition year using the episode imposition date in TIES; I take the backed-down year from the episode end date; and I code threat onset year using the episode start date. These variables are critical to include given that information communicated through thirdparty sanctions is probably less important for states already facing direct coercion attempts. Furthermore, previous research has demonstrated that states respond to sanctions typically by worsening abuses (Wood, 2008; Peksen, 2009). Direct threats could have a positive or negative influence on practices depending on leaders’ assessments of the costs and benefits of human rights abuse and enduring sanctions. I also include variables for several key indicators of respect for physical integrity rights that also affect vulnerability to sanctions – and therefore affect attention to cues suggesting that sanctions could be forthcoming. I include the 21-point combined Polity score to capture institutional democracy (Marshall and Jaggers, 2010). As noted above, democracies also tend to recover from sanctions more quickly than non-democracies (Lektzian and Souva, 2001). As a result, democracies could be less sensitive to third-party cues. However, more democratic states also tend to abuse their citizens less than authoritarian states, although there is variation in this relationship.25 In my data, there are few democracies with abusive records and many democracies with respectful records. Accordingly, given that I stratify 25 For
example India and Brazil tend to score highly on this measure of institutional democracy, but low on respect for physical integrity rights. The correlation between polity and the PTS score is modest: 0.27. All results are robust if I exclude the polity combined score.
20
the models by PTS score, I code democracy as change in Polity score to ensure that my explanatory variables have an equivalent effect on the baseline hazard for each stratum.26 Similarly, I include a variable for (logged) income per capita (in 2005 dollars), given that wealthier states tend to be less abusive. However, wealthier states can also endure sanctions more easily, and therefore could be less sensitive to third-party cues. Again, I code this variable as a change to ensure that its effect is equivalent across the five strata of human rights records, given that wealthier states tend to be less abusive. I code this variable using data from Gleditsch (2002b). To capture US economic leverage over a state, and therefore the state’s potential sensitivity to third-party cues regarding the likelihood of sanctions, I include a variable for the share of its exports to the US as a percentage of its total exports. This export share variable captures the cost to a state of terminating trade with the US, accounting for its alternate markets for exports (see, e.g., Hirschman, 1945; Galtung, 1967). Accordingly, a lower export share with the US suggests that it has more alternative markets that it can use to redirect trade if sanctioned by the US.27 Furthermore, prior research also finds that exports can sometimes lead to improved respect for physical integrity rights (e.g., Hafner-Burton, 2005; Cao, Greenhill and Prakash, 2012). It is important to note that not all sanctions are export restrictions; for example, some affect investment in the target. However, export share remains a useful proxy for US leverage in this case given that direct investment often is an indicator of the globalization of production, leading eventually to exports back to the US. I code this variable using trade data from Gleditsch (2002b).28 Again, I code this variable as a change in order to ensure that its effect is equivalent across the five strata of human rights records.29 26 Notably,
however, my primary results are consistent with those presented if I use level indicators instead of change variables for all continuous IVs. 27 However, this variable only captures economic ties to the US vs. third-parties. In robustness check models presented in the supplemental appendix, I add a variable counting the number of non-US allies a state maintains. These allies could provide political insulation to a sanctioned state, reducing its need to comply with US directives. 28 Aid sanctions are also relatively common. Results are consistent when I include an indicator of preexisting US aid to the potential target. However, this variable correlates highly with trade, so I exclude it from my primary models. 29 Dependence on the US is lower for more respectful states.
21
Finally, I include a dichotomous variable equal to 1 for states maintaining an alliance with the United States, using data from the Alliance Treaty Obligations and Provisions (ATOP) project (Leeds et al., 2002). Friendship with the US could, on one hand, have a direct effect on human rights by helping respectful human rights practices to diffuse. On the other hand, US allies might feel confident that they can maintain abusive practices without invoking US coercion. Therefore, alliance could render states less likely to improve practices, and generally less sensitive to third-party cues as well.30
5
Analysis I find strong support for the expectation that relevant US human rights sanctions are
associated with an increased hazard that a non-sanctioned state will improve its human rights practices, as well as a decreased hazard that it will worsen its human rights abuses. Table 1 presents Cox regression coefficients and robust standard errors for models examining duration until improvement of human rights as measured by the Political Terror Scale. For each set of models, I present results for the (1) aggregate, (2) region-specific context, (3) state-specific context, and (4) affinity-specific context variations of the relevant third-party sanction variable. I present minimum specifications including only my primary explanatory variable along with full specifications including all control variables, for a total of eight models in Table 1. [Table 1 about here] In all eight models, the coefficient for relevant third-party sanction is positive and significant, indicating that the observation of a relevant human rights sanction against a third party is associated with an increased hazard that a state improves its own human rights record. Coefficients for my primary explanatory variable look quite similar in the minimal 30 I
use this alliance variable as control variables in addition to incorporating it into a relevant sanction variable. I do so because I expect that the presence of an alliance with the US could have an effect independent of similarity to a third party with respect to alliance status.
22
and full specifications. Results also look consistent across the four variations of relevant sanction variable. Accordingly, I find support for hypothesis 1. Figure 1 presents the reputation effect of relevant human rights sanctions graphically (from Model 2). Given that the Cox models are stratified by abuse level, there is a distinct baseline hazard of improvement for each of the five possible values of the PTS. Accordingly, I present the survival until improvement in human rights practices for the twenty-fifth and seventy-fifth percentile of the PTS score.31 The left-hand graph shows the cumulative survival probability of human rights practices over time for relatively abusive states (where PTS is equal to 4). The right-hand graph presents the survival for relatively respectful states (where PTS is equal to 2). [Figure 1 about here] In both graphs, the proportion of states with existing human rights practices surviving (that is, not improving) is considerably higher in the absence of a relevant human rights sanction against a third party. The left-hand graph in Figure 1 shows that only ten percent of relatively abusive states (PTS equal to 4) will maintain this abuse level for five years after witnessing a relevant sanction. Conversely, over eighty percent of these abusive records survive for five years if leaders have not witnessed a relevant human rights sanction. The right-hand graph shows that survival until improvement of human rights practices is generally shorter for more respectful states (Where PTS is equal to 2). Essentially all such states will experience an improvement within five years if they have witnessed a relevant third-party sanction. Conversely, forty percent of these states will maintain their current practices (which are already fairly, although not fully, respectful) for five years if they do not witness a relevant sanction. [Table 2 about here] 31 I
present survival of existing human rights to demonstrate the short- and long-term influence of relevant third-party sanctions. Generally, the hazard of human rights improvement looks like an upside-down “U.” In the presence of a relevant sanction (from Model 2), the hazard increases sharply over the first two years, remaining steady for approximately 10 years, then declining back towards zero. This hazard function likely reflects the fact that changing human rights policy takes time to implement.
23
The survival until improvement in human rights explains only part of the reputation effect of US human rights sanctions against third parties. Table 2 presents coefficients and robust standard errors for Cox models examining the hazard of worsening human rights. The findings of these models are mirrors of those from Table 1: the coefficient for relative third-party sanction is negative and significant in all eight models. Again, the coefficients look relatively similar across all four models. Accordingly, this result supports hypothesis 2 that the hazard of worsening human rights records is lower when a state witnesses a similar third party endure sanctions over human rights abuse. Yet the magnitude of the relationships estimated in Table 2 adds important information regarding the difference in impact of relevant sanctions depending on pre-existing abuse levels. I demonstrate the survival until worsened human rights in Figure 2 (from Model 10). [Figure 2 about here] Figure 2 shows that the observation of a relevant third-party sanction has a strong impact on the duration until worsened human rights. Again, the left-hand graph presents survival until worsened rights for relatively abusive states (PTS equal to 4) and the right-hand graph shows this survival for relatively respectful states (PTS equal to 2). The figure shows that, in the absence of a third-party sanction, nearly 100 percent of relatively abusive states (PTS equal to 4) experience failure within six years, worsening their practices even further. However, in the presence of a relevant third-party sanction, nearly 70 percent of these states will survive six years without worsening abuse levels. Combined with the results from Figure 1, this result suggests that many of the states improving human rights after witnessing a human rights sanction will not return to previous abuse levels quickly. The results of the right-hand graph are even more striking. More than 90 percent of relatively respectful states (PTS equal to 2) witnessing a relevant sanction will survive ten years without worsening their practices. Combined with the results from Figure 1, this result suggests that there is a long-lasting reputation effect of US human rights sanctions for states that are already 24
relatively respectful of physical integrity rights. Control variables generally appear as expected. Interestingly, I find some support that direct US capitulation is associated with a reduced hazard of human rights improvement, although it does not appear to encourage worse abuse. I also find strong evidence that direct threat onset is associated with a greater hazard of human rights improvement. It is possible that, by controlling for imposition and US capitulation, this threat onset variable is capturing cases where a target acquiesces to US demands. One notable null result is that direct sanction imposition appears not to affect the hazard of improvement or decline when controlling for other sanctioning behavior.32 I find that an increasing export share, suggesting increasing US leverage over a state, is associated with a greater hazard of human rights improvement, yet has no association with the hazard of human rights decline. Increasing wealth and growing democracy appear to encourage improvement in these practices, while simultaneously discouraging worsening abuse. Finally, alliance with the US is associated with a lower hazard of human rights improvement. This result is unsurprising given that leaders could be insulated from pressure to improve respect for human rights if the US maintains a security interest in the state. Notably, alliance does not hasten worsening practices; US allies may forestall progress, but they appear not to intensify repression.
6
Conclusion Reputation effects are commonly attributed to costly – usually military – actions. Sanc-
tions are a more subtle tool of coercion, although they too can inflict considerable suffering on target populations. In this paper, I demonstrate that the influence of human rights sanctions transcends sanctioned states, influencing those witnessing this economic coercion. Indeed, human rights sanctions appear to have a third-party reputation effect strong enough to mo32 I
find that imposed sanctions are strongly associated with worsening abuse in OLS models examining 1-year (but not 2-year) changes in human rights practices. This difference in results by model type deserves more attention in future studies.
25
tivate proactive policy change in states that perceive themselves to be similar to sanction targets. My results have important implications for US foreign policy makers to consider when deciding whether or not to impose sanctions, particularly when they must contend with domestic interests opposing these restrictions. For example, several firms are wary of restrictions on “conflict minerals” from the Democratic Republic of the Congo (DRC). Yet, by putting pressure on the DRC, the US could motivate other states in the region to reduce abuses committed in the pursuit of mineral wealth. In short, sanctions could have a wider effect throughout the system; this subtle yet widespread effect of sanctions provides us with reason to reevaluate the definition of sanctions “effectiveness.” While sanctions could fail to change target behavior, and even lead to worse abuses (Wood, 2008; Peksen, 2009), resistant targets could serve as martyrs for the cause of improved human rights elsewhere, convincing third parties to comply with US demands to improve these practices. Furthermore, although I reevaluate the conclusion reached by Nossal (1989), I concur that improving the frequency with which the US threatens and imposes sanctions against abusive regimes could motivate wider perception that continued abuse is costly. While I find that the seemingly haphazard use of sanctions still does affect behavior throughout the international system, this behavior has the potential to have greater influence still. Increasing the consistency of sanctions policy could motivate wider and more sustained policy change. This paper opens considerable avenues for future research. First, the question I address here with regard to human rights can be extended to other issues, including nuclear proliferation and the use of militarized aggression. Research should also address whether senders other than the US can likewise communicate preferences and resolve widely though the use of sanctions. Indeed, the same principles applied here to sanctions could be extended to other foreign policy tools, including the use of military force. Full-scale war and occupation introduces an additional dynamic into the process explored here. While the US could conceivably sanction multiple states simultaneously, its use of military force is more constrained. For 26
example, the US occupation of Iraq could have sent a message to other dictators maintaining poor relations with the US, yet the occupation also strained considerably the resources available to use against its “next” target. States therefore must balance desire to communicate resolve with the necessity to maintain sufficient military capabilities to use in future coercion attempts. Future studies can examine how this more complex dynamic influences reputation effects. Finally, future research may benefit from examining more closely the micro-foundations on which I base my study. It is difficult to demonstrate vicarious learning directly; ultimately, my findings suggest only that patterns of human rights policy change are consistent with what we would expect if such learning did occur. Case studies or field experiments may improve our confidence that learning does occur. Importantly, my argument suggests that these studies should search for evidence of learning outside of sanction threat episodes, as states may incorporate what they learn from third-party sanctions into their behavior before these episodes begin.
27
References Allen, Susan Hannah. 2008. “The Domestic Political Costs of Economic Sanctions.” Journal of Conflict Resolution 52(6):916–944. Baldwin, David A. 1985. Economic Statecraft. Princeton: Princeton University Press. Barber, James. 1979. “Economic Sanctions As a Policy Instrument.” International Affairs 55(3):367–384. Box-Steffensmeier, Janet M. and Bradford S. Jones. 2004. Event History Modeling: A Guide for Social Scientists. Cambridge: Cambridge University Press. Brinks, Daniel and Michael Coppedge. 2006. “Diffusion Is No Illusion : Neighbor Emulation in the Third Wave of Democracy.” Comparative Political Studies 39(4):463–489. Bueno de Mesquita, Bruce, Alastair Smith, Randolph M. Siverson and James D. Morrow. 2003. The Logic of Political Survival. Cambridge: The MIT Press. Buhaug, Halvard and Kristian Skrede Gleditsch. 2008. “Contagion or Confusion? Why Conflicts Cluster in Space.” International Studies Quarterly 52:215–233. Cao, Xun, Brian Greenhill and Aseem Prakash. 2012. “Where Is the Tipping Point? Bilateral Trade and the Diffusion of Human Rights.” British Journal of Political Science . Cingranelli, David L. and David L. Richards. 2010. “The Cingranelli and Richards (CIRI) Human Rights Data Project.” Human Rights Quarterly 32:401–424. Conrad, Courtenay R. and Jacqueline H.R. DeMeritt. n.d. “Human Rights Advocacy and State Repression Substitutability.” Unpublished Manuscript. Correlates of War Project. 2011. http://correlatesofwar.org.
“State System Membership List, v2011.” Online
Crescenzi, Mark J. C. 2003. “Economic Exit, Interdependence, and Conflict.” Journal of Politics 65(3):809–832. Crescenzi, Mark J. C. 2007. “Reputation and Interstate Conflict.” American Journal of Political Science 51(2):382–396. Drezner, Daniel W. 2003. “The Hidden Hand of Economic Coercion.” International Organization 57(3):643–659. Drury, A. Cooper. 1998. “Revisiting Economic Sanctions Reconsidered.” Journal of Peace Research 35(4):497–509. Eaton, Jonathan and Maxim Engers. 1999. “Sanctions: Some Simple Analytics.” The American Economic Review 89(2):409–414. Escriba-Folch, Abel and Joseph Wright. 2010. “Dealing with Tyranny: International Sanctions and the Survival of Authoritarian Rulers.” International Studies Quarterly 54:335– 359. 28
Fearon, James D. 1994. “Signaling versus the Balance of Power and Interest: An Empirical Test of a Crisis Bargaining Model.” The Journal of Conflict Resolution 38(2):236–269. Finnemore, Martha and Kathryn Sikkink. 1998. “International Norm Dynamics and Political Change.” International Organization 52(4):887–917. Galtung, Johan. 1967. “On the Effects of International Economic Sanctions: With Examples from the Case of Rhodesia.” World Politics 19(3):378–416. Gleditsch, Kristian Skrede. 2002a. All International Politics is Local: The Diffusion of Conflict, Integration, and Democratization. Ann Arbor, MI: University of Michigan Press. Gleditsch, Kristian Skrede. 2002b. “Expanded Trade and GDP Data.” The Journal of Conflict Resolution 46:712–724. Hafner-Burton, Emilie M. 2005. “Trading Human Rights: How Preferential Trade Agreements Influence Government Repression.” International Organization 59(3):593–629. Hirschman, Albert O. 1945. National Power and the Structure of Foreign Trade. Berkeley: University of California Press. Hovi, Jon, Robert Huseby and Detlef F. Sprinz. 2005. “When Do (Imposed) Economic Sanctions Work?” World Politics 57(4):479–499. Hufbauer, Gary, Jeffrey Schott, Kimberly Ann Elliott and Barbara Oegg. 2007. Economic Sanctions Reconsidered: History and Current Policy. 3 ed. Washington, DC: Peterson Institute for International Economics. Jervis, Robert. 1976. Perception and Misperception in International Politics. Princeton: Princeton University Press. Kaempfer, William H. and Anton D. Lowenberg. 1988. “The Theory of International Economic Sanctions: A Public Choice Approach.” The American Economic Review 78:786– 793. Kaempfer, William H. and Anton D. Lowenberg. 1989. “The Theory of International Economic Sanctions: A Public Choice Approach: Reply.” The American Economic Review 79(5):1304–1306. Krustev, Valentin L. 2010. “Strategic Demands, Credible Threats, and Economic Coercion Outcomes.” International Studies Quarterly 54:147–174. Lacy, Dean and Emerson M. S. Niou. 2004. “A Theory of Economic Sanctions and Issue Linkage: The Roles of Preferences, Information, and Threats.” Journal of Politics 66(1):25–42. Leeds, Brett Ashley, Jeffrey M. Ritter, Sara McLaughlin Mitchell and Andrew G. Long. 2002. “Alliance Treaty Obligations and Provisions, 1815-1944.” International Interactions 28:237–260.
29
Lektzian, David and Mark Souva. 2001. “Institutions and International Cooperation: An Event History Analysis of the Effects of Economic Sanctions.” Journal of Conflict Resolution 45(1):61–79. Levy, Jack S. 1994. “Learning and Foreign Policy: Sweeping a Conceptual Minefield.” International Organization 48(2):279–312. Licht, Amanda A. n.d. “Falling Out of Favor: Economic Sanctions and the Tenure of Leaders.” Unpublished Manuscript. Lindsay, James M. 1986. “Trade Sanctions as Policy Instruments: A Re-examination.” International Studies Quarterly 30(2):153–173. Marinov, Nikolay. 2005. “Do Economic Sanctions Destabilize Country Leaders?” American Journal of Political Science 49(3):564–576. Marshall, Monty G. and Keith Jaggers. 2010. “Polity IV Project: Political Regime Characteristics and Transitions, 1800-2009.”. McCormick, James M. and Neil J. Mitchell. 1997. “Human Rights Violations, Umbrella Concepts, and Empirical Analyses.” World Politics 49(4):510–525. Morgan, T. Clifton, Navin Bapat and Valentin L. Krustev. 2009. “The Threat and Imposition of Economic Sanctions, 1971-2000.” Conflict Management and Peace Science 26(1):92–110. Morgan, T. Clifton and Valerie L. Schwebach. 1997. “Fools Suffer Gladly: The Use of Economic Sanctions in International Crises.” International Studies Quarterly 41(1):27–50. Morrow, James D. 1989. “Capabilities, Uncertainty, and Resolve: A Limited Information Model of Crisis Bargaining.” American Journal of Political Science 33(4):941–972. Most, Benjamin and Harvey Starr. 1989. Inquiry, Logic, and International Politics. Columbia: University of South Carolina Press. Nooruddin, Irfan. 2002. “Modeling Selection Bias in Studies of Sanctions Efficacy.” International Interactions 28(1):59–75. Nossal, Kim Richard. 1989. “International Sanctions as International Punishment.” International Organization 43(2):301–322. Olson, Richard Stuart. 1979. “Economic Coercion in World Politics: With a Focus on North-South Relations.” World Politics 41(4):471–494. Pape, Robert A. 1997. “Why Economic Sanctions Do Not Work.” International Security 22(2):90–136. Pape, Robert A. 1998. “Why Economic Sanctions Still Do Not Work.” International Security 23(1):66–77. Peksen, Dursun. 2009. “Better or Worse? The Effect of Economic Sanctions on Human Rights.” Journal of Peace Research 46(1):59–77. 30
Peterson, Timothy M. Forthcoming. “Sending a Message: The Reputation Effect of US Sanction Threat Behavior.” International Studies Quarterly . Poe, Steven C., C. Neal Tate and Linda Camp Keith. 1999. “Repression of the Human Right to Personal Integrity Revisited: A Global Cross-National Study Covering the Years 1976-1993.” International Studies Quarterly 43(2):291–313. Press, Daryl G. 2005. Calculating Credibility: How Leaders Assess Military Threats. Ithaca: Cornell University Press. Schelling, Thomas C. 1967. Arms and Influence. New Haven: Yale University Press. Signorino, Curtis S. and Jeffrey M. Ritter. 1999. “Tau-b or Not Tau-b: Measuring the Similarity of Foreign Policy Positions.” International Studies Quarterly 43:115–144. Sweeney, Kevin and Omar M. G. Keshk. 2005. “The Similarity of States: Using S to Compute Dyadic Interest Similarity.” Conflict Management and Peace Science 22:165–187. Weeks, Jessica L. 2008. “Autocratic Audience Costs: Regime Type and Signaling Resolve.” International Organization 62(1):35–64. Weinhold, Diana. 2002. “The Importance of Trade and Geography in the Pattern of Spatial Dependence of Growth Rates.” Review of Development Economics 6(3):369–382. Wood, Reed M. 2008. ““A Hand upon the Throat of the Nation”: Economic Sanctions and State Repression, 1976–2001.” International Studies Quarterly 52:489–513. Wood, Reed M. and Mark Gibney. 2010. “The Political Terror Scale (PTS): A Reintroduction and a Comparison to CIRI.” Human Rights Quarterly 32:367–400.
31
Table 1: Cox model coefficients and robust standard errors for third-party US sanctions and the hazard of improving human rights Aggregate Model Region Context State Context Affinity Context 1 2 3 4 5 6 7 8 Relevant 3rd party sanction Direct US imposition Direct US capitulation Direct US threat ∆ Export share with US 32 Alliance with US ∆ ln GDP per capita ∆ Polity combined score
2.907*** (0.170)
2.934*** (0.219) 0.276 (0.688) -0.706* (0.362) 1.143*** (0.367) 0.245** (0.105) -0.300*** (0.114) 1.337* (0.758) 0.054*** (0.020)
1.376*** (0.196)
1.198*** (0.227) 0.837 (0.650) -0.008 (0.424) 0.544 (0.443) 0.249** (0.113) -0.322*** (0.121) 1.110 (0.761) 0.064*** (0.018)
1.523*** (0.160)
1.405*** (0.160) 0.098 (0.576) -0.795** (0.391) 1.227*** (0.425) 0.223* (0.126) -0.360*** (0.125) 1.181 (0.744) 0.063*** (0.018)
1.788*** (0.204)
1.616*** (0.223) 0.235 (0.654) -0.549 (0.443) 1.103** (0.466) 0.257** (0.103) -0.189 (0.128) 1.279* (0.722) 0.063*** (0.018)
Observations 2997 2369 2997 2369 2997 2369 2997 2369 Log likelihood -1664.296 -1291.892 -1756.258 -1364.050 -1746.999 -1355.997 -1736.294 -1349.507 χ2 290.775*** 208.425*** 49.187*** 90.545*** 90.057*** 145.760*** 76.514*** 113.401*** *** p ≤ 0.01, ** p ≤ 0.05, * p ≤ 0.1, two-tailed tests Notes: model stratified by political terror scale value. Breslow method employed to handle ties. Standard errors clustered on the state. All explanatory variables are lagged one year.
Table 2: Cox model coefficients and robust standard errors for third-party US sanctions and the hazard of worsening human rights Aggregate Model Region Context State Context Affinity Context 9 10 11 12 13 14 15 16 Relevant 3rd party sanction Direct US imposition Direct US capitulation Direct US threat ∆ Export share with US 33 Alliance with US ∆ In GDP per capita ∆ Polity combined score
-2.653*** (0.217)
-3.189*** (0.295) -1.436 (1.177) 0.112 (1.111) -0.245 (1.012) -0.171 (0.191) 0.134 (0.090) -1.415*** (0.475) -0.024 (0.016)
-1.166*** (0.333)
-1.285*** (0.324) -1.140 (1.141) -0.651 (1.199) 0.134 (1.007) -0.183 (0.207) 0.068 (0.100) -1.958*** (0.536) -0.035** (0.016)
-1.250*** (0.256)
-1.239*** (0.269) -1.152 (1.222) -0.320 (1.228) -0.161 (1.030) -0.177 (0.203) 0.038 (0.100) -2.018*** (0.509) -0.028* (0.017)
-1.330*** (0.182)
-1.572*** (0.205) -1.226 (1.204) -0.265 (1.159) -0.135 (1.038) -0.226 (0.210) -0.066 (0.098) -1.909*** (0.506) -0.040** (0.016)
Observations 2997 2369 2997 2369 2997 2369 2997 2369 Log likelihood -1511.275 -1219.938 -1574.573 -1280.374 -1570.298 -1278.198 -1563.427 -1268.127 χ2 148.829*** 169.996*** 12.267*** 40.793*** 23.837*** 49.327*** 53.719*** 93.142*** *** p ≤ 0.01, ** p ≤ 0.05, * p ≤ 0.1, two-tailed tests Notes: model stratified by political terror scale value. Breslow method employed to handle ties. Standard errors clustered on the state. All explanatory variables are lagged one year.
Figure 1: Survival of human rights practices: failure = improvement (from Model 2)
.8 .6 0
.2
.4
.6 .4 0
.2
Proportion surviving
.8
1
PTS=2
1
PTS=4
0
5
10 15 Analysis time
20
25
0
5
10 15 Analysis time
20
No 3rd−party sanction
No 3rd−party sanction
3rd−party sanction observed
3rd−party sanction observed
Control variables held at means (if continuous) or modes (if dichotomous)
34
25
Figure 2: Survival of human rights practices: failure = worsening (from Model 10)
.8 .6 0
.2
.4
.6 .4 0
.2
Proportion surviving
.8
1
PTS=2
1
PTS=4
0
5
10 15 Analysis time
20
25
0
5
10 15 Analysis time
20
No 3rd−party sanction
No 3rd−party sanction
3rd−party sanction observed
3rd−party sanction observed
Control variables held at means (if continuous) or modes (if dichotomous)
35
25
