Introduction
Punishment (or expectation of punishment) for norm violations in international relations is commonly thought to be important for norm maintenance and an indicator of norm strength.Footnote 1 In practice, economic sanctionsFootnote 2 are go-to tools for international norm enforcement. Severe human rights abuses, conflict, and nuclear weapons development are consistently cited as grounds for sanctions, suggesting some shared understanding about how to punish non-compliance with core national and human security norms. Since 2013, multilateral sanctions have been issued against state or non-state actors for norm violations in Burundi, Central African Republic, Egypt, Guinea-Bissau, Libya, Ukraine, Russia, South Sudan, Venezuela, and Yemen, to say nothing of a plethora of pre-existing regimes and national-level sanctions against others. Yet many more escaped such punishment despite similar behaviour. Clearly, not all norm violators get slapped with sanctions.
This article seeks to understand international metanorms – international norms about enforcing norms – and when ‘bad behaviour’ faces material punishment in world politics. Why do states administer costly punishment for norm violations in some cases but not others, and what do those practices reveal about international metanorms? For many experts, compliance requires the threat of material or social sanctions, especially with norms actors have not internalised or prefer to ignore.Footnote 3 Among the few tools of norm enforcement available in international politics, economic sanctions can impose direct material costs on norm violators (unlike ‘naming and shaming’ and other social sanctions) without resorting to military force. They can also constrain targets’ ability to engage in offending behaviours and publicly signal disapproval.Footnote 4 Yet sanctions imposition is often complicated by special relationships in international politics and special interests in domestic politics. Whether international metanorms create obligations for states to enforce norms when faced with social and material ties has important implications for norm development but is unknown.
With the exception of research on social tools like naming and shaming and on specific norms in isolation, patterns of international norm enforcement and metanorms have not been systematically investigated. Using the Threat and Imposition of Sanctions (TIES) and Arms Embargo Datasets, I examine multilateral and unilateral sanctions by 22 potential sending (that is, sanctions-imposing) states against 189 potential target (that is, sanctions-receiving) states. I compare six security norms – nuclear weapons development, internal and interstate conflict, human rights abuses, mass killing, and support for terrorism – and find that although violations are positively associated with sanctions, valuable relationships often discourage their imposition. These cross-norm trends point to international metanorms, especially among major sending states, as a likely cause, rather than norm-specific attributes. However, I also show that norm violations are themselves relatively rare, despite the even greater rarity of material punishment. This suggests that compliance has deeper roots and broader tools than strict enforcement approaches to compliance expect.
Overall, I therefore argue that international metanorms may limit costly material punishments by protecting interstate relationships, recognising that compliance may depend even more on social sanctions and internalisation for which fostering interstate ties is valuable. Finally, while I do not make claims about sanctions effectiveness, the findings have implications for debates about when and why sanctions may or may not ‘work’. I show that the ingredients believed necessary for meaningful economic sanctions – valued target-sender relationships – reduce the likelihood of their imposition. States most often selected for sanctions may be least likely to comply with them. Thus, metanorms could complicate sanctions effectiveness from the start. Nevertheless, if protecting relationships encourages compliance through tools such as socialisation, social pressures, and other forms of norm enforcement, the results may be less discouraging than they at first glance seem.
Norm enforcement in world politics
Norms are collective expectations for appropriate behaviour by actors of a given identity.Footnote 5 In international politics, norms guide states’ interactions with one another, play a constitutive role in identity formation, and can shape how they treat their own populations. By creating a sense of obligation to follow certain behavioural rules, norms can be a source of order and cooperation. Adherence to norms can also provide strategic and material benefits like foreign aid, and cognitive benefits like social approval and positive self-image. Certainly, no norm is followed or bought into by all actors at all times. States may violate norms when strong self-interests cannot be reconciled with social obligations, when social obligations are weak, or when they lack the capacity or incentives to comply.Footnote 6
Scholars from both rationalist and constructivist traditions have argued that the international community's response to norm violations can influence norm development, persistence and strength, and compliance. Enforcement can socialise actors into norms, provide social and material incentives to comply with them, and establish or reinforce their value in world politics.Footnote 7 The extent of those social and material costs – and who is willing to impose them and on whom – can both shape and communicate a norm's legitimacy and social value. In short, enforcement helps define what is accepted as ‘normal’ in international relations by conveying social information about what is appropriate and valued and by whom. Enforcement expectations may in turn shape actors’ decisions to violate norms and their perceptions of when it is acceptable to do so.Footnote 8 For strict enforcement approaches to compliance, material enforcement may be necessary to ensure states’ cost-benefit calculus in favour of compliance. Membership (or desired membership) in an in-group can also affect enforcement, as actors seek conformity with the norms of groups with which they identify.Footnote 9 Finally, norms whose violations are not punished may eventually disappear or be replaced.Footnote 10 Practices and institutionalised tools of norm enforcement are therefore often also used as indicators or proxies of norm strength.Footnote 11
Despite its importance, however, international norm enforcement is not well understood. With the exception of ‘naming and shaming’, existing studies tend to focus on the enforcement of individual norms rather than on broader practices. This single-norm approach tends to highlight either case-specific factors such as costs, capacity, and interests behind enforcement decisions,Footnote 12 or characteristics that make a norm more or less ‘enforceable’. For example, non-compliance is more difficult to determine for norms that are ambiguously defined,Footnote 13 and for norms whose violations cannot readily be observed or monitored.Footnote 14 Norms with well-resourced or powerful patron states may also be better enforced, thanks to those states’ interest in and capacity to respond to violations.Footnote 15 While these factors undoubtedly shape norm enforcement in important ways, broader theoretical and empirical discussions of international metanorms and enforcement practices across different norms are largely missing from the theoretical and empirical literature.Footnote 16
International metanorms
This article examines patterns of sanctions imposition to uncover what metanorms might exist (if any) for the enforcement of international security norms. Metanorms are norms that create expectations and obligations of norm enforcement and regulate how actors legitimately punish norm violations.Footnote 17 Like any norms, metanorms may of course be contested or lack buy-in by some members of the international community. Nevertheless, because international norm enforcement tends to be decentralised, metanorms may be essential for motivating responses to violations and identifying legitimate forms and procedures of enforcement.Footnote 18 Without some shared understanding of when or what types of punishments are appropriate, actors may be less willing and able to enforce norms, collectively or individually. In theory, metanorms also compensate or reward actors who bear the costs of punishing violators on behalf of a group (although beyond social compensation, this may be rare in international politics). They can also ensure that actors who do not punish norm violators are themselves punished.Footnote 19
Institutionalised metanorms can provide actors with formal prescriptions for responding to norm violations, legal and normative obligations to accept some enforcement costs, and access to selective incentives to mitigate those costs.Footnote 20 However, norm enforcement is generally not well institutionalised in international politics.Footnote 21 Even when norms are codified in formal agreements, those agreements often eschew specifying punishments and delegating responsibility for executing them in the event of violations. The Nuclear Non-Proliferation Treaty (NPT), for example, establishes a firm non-proliferation norm and mandates participation in safeguards, but contains no formal measures to deal with non-compliance.Footnote 22 Similarly, human rights agreements may establish investigative mechanisms and victims’ right to domestic legal recourse but often do not contain international-level measures to punish violations. Thus, although institutionalisation may assist with enforcement by articulating norm content and strengthening compliance obligations, metanorms may not require it.
Indeed, international metanorms may be even more important in the absence of formal enforcement mechanisms. They can establish shared expectations about how and when punishment for non-compliance is acceptable or expected – especially costly, material forms of punishment, which may convey stronger disapproval for and greater illegitimacy of an actor's behaviour but may also need to adhere to procedural and proportionality expectations. Audie Klotz, for example, argues that states voluntarily adopted sanctions against apartheid South Africa because the racial equality norm had made rhetorical condemnation alone insufficient to demonstrate rejection of apartheid.Footnote 23 At the same time, states may have more latitude to interpret non-institutionalised metanorms, making enforcement less consistent and more contested. Institutionalised or not, however, international metanorms can save actors extra political effort that may otherwise be required to decide whether to punish norm violations and with what tools. Otherwise, even when violations are clear, deciding whether and how to punish them could be difficult.
Tools of norm enforcement
International actors have three main tools of norm enforcement: social sanctions, economic sanctions, and military action.Footnote 24 In practice, only states and international organisations acting on behalf of states employ economic sanctions and military action; non-state actors engaged in norm enforcement are typically limited to social sanctions, like naming and shaming. All three call negative attention to an actor's bad behaviour and can impose social costs, including negative self-image, social opprobrium, and loss of legitimacy. The material costs to targets (that is, the norm violators/punishment receivers) and senders (that is, the norm enforcers/punishment executers) vary with the type of punishment and can have social implications. As material costs rise, punishments are collectively seen as more severe and may invite criticism for senders if metanorms suggest they are inappropriate or disproportionate for the norm violation. Relationships between targets and senders, and between senders, also affect punishment costs. Punishments that disrupt alliances, special relationships, or trade between states, for example, will entail higher costs to both targets and senders than those that do not. The use of costly material sanctions, in turn, can convey information about and shape the social value of the norm being enforced.
Social sanctions generally entail punishments that rely on social or moral leverage, the removal of social status, and targets’ embarrassment and concern for social standing to provoke behavioural change.Footnote 25 Various forms of rhetorical condemnation and diplomatic isolation, such as suspension of institutional membership rights, may have material costs but use social exclusion and embarrassment as their primary source of leverage.Footnote 26 Perhaps most commonly, naming and shaming draws public attention to norm violations in order to mobilise public opinion and pressure violators to change their behaviour.Footnote 27 It can include press releases and media reports, official citations, or rankings by governmental or non-governmental organisations. Naming and shaming has been well studied, especially in human rights, where it is ‘the principal weapon of choice among many international organizations and governments’.Footnote 28 Its low material costs to senders make it a popular form of enforcement, especially for those without significant material resources. Although direct material costs to targets are negligible, social costs and, sometimes, indirect material costs from social stigmatisation can incentivise behavioural change.Footnote 29
Associated with high material costs, military tools rare for two reasons. First, they introduce the greatest proportionality considerations, which make them an inappropriate punishment in most cases. Second, they impose greater costs on norm enforcers, related to organising and carrying out the punishment and the extreme disruption of target-sender relationships. Renee De Nevers, for example, argues that great powers’ use of force against norm-violating states depends on the social standing and relative material power of the violator; states that are weak and socially excluded are more likely to be targets of force than other forms of enforcement.Footnote 30 Of course, military tools need not be administered in isolation. As a measure of last resort, they are often preceded by lower-cost punishments that can raise the profile of a target's norm violations and increase the likelihood of military action.Footnote 31
The material costs of economic sanctions generally fall between social sanctions and military action.Footnote 32 Economic sanctions are ‘the deliberate, government-inspired withdrawal, or threat of withdrawal, of customary trade or financial relations’.Footnote 33 They may be imposed unilaterally by individual states or multilaterally, by groups of states through institutions. They may be comprehensive (prohibiting all economic activity with a target) or targeted or ‘smart’ (prohibiting economic activity in certain sectors or with certain actors in a target). Arms embargoes are targeted economic sanctions that seek to prevent legal/state-sanctioned arms sales to governments or other actors within a state. Sanctions use has grown since the end of the Cold War, as the UN Security Council deadlock broke and the EU acquired greater foreign policymaking powers. In the process, organisational norms about their use have also developed.Footnote 34 Over time, many have come to view comprehensive sanctions excessive and inhumane. Since 2000, targeted sanctions have therefore become more accepted forms of norm enforcement, a development that may increase perceived sanctioning opportunities by making them a more proportional option for a wider range of norm violations and by potentially lowering their aggregate cost to senders.
The costs sanctions impose on a target depend partly on its relationships with senders. Sanctions should provide stronger incentives for behavioural change when they disrupt close economic and political ties. Moreover, when sanctions impose significant costs on senders who share in those close relations, they signal stronger disapproval of norm-violating behaviour. When economic and political ties are minimal, sanctions impose negligible costs on targets and senders. Under these conditions, sanctions may be easier to enact but struggle to push targets to change or constrain their behaviour or signal collective condemnation.
In theory, higher punishment costs thus not only generate greater incentives for targets to comply with norms but also force senders to shoulder a heavier burden. States typically prefer multilateral sanctions in order to share costs among senders, more effectively cut off goods to targets, and signal stronger disapproval. The international community also tends to see UN sanctions particularly as more legitimate.Footnote 35 Yet high costs to senders (and relationships among senders) can render decisions to impose multilateral sanctions contentious, especially if metanorms do little to offset those costs. In cases like the Syrian conflict or 2009 Gaza war, the Security Council could not overcome divisions to impose arms embargoes. Often, it will not even discuss sanctions when a permanent-member veto appears inevitable. Other multilateral organisations or individual states may issue sanctions instead (risking criticism for going outside the UN), but norm violations often avoid sanctions entirely. Such variation in sanctions use introduces important questions about metanorm content and when and how material punishment occurs in world politics.
Hypotheses
Economic sanctions offer several advantages for exploring international metanorms. First, patterns of sanctions imposition can reveal potential metanorms by showing costly norm enforcement (or lack thereof) in action across a broad range of norm violations and punishment episodes. Second, sanctions can test the effects of target-sender relations and institutionalisation on norm enforcement. Finally, sanctions provide a wider range of punishment functions, in addition to symbolic condemnation and ‘sticks’ to coerce behavioural change. For example, arms embargoes can take away or make more costly to acquire the tools by which an actor perpetrates its violations.
Empirical research has primarily focused on the effects of sanctions, not their imposition. This limits scholars’ ability to understand metanorms as well as how decisions about imposition may influence sanctions’ ability to change targets’ behaviour. While sanctions are more commonly imposed against adversaries,Footnote 36 adversaries are less likely to concede to senders’ demands. Moreover, limiting analyses to ‘sanctions-relevant dyads’ – cases in which targets and senders have existing economic tiesFootnote 37 – may bias results by overlooking many sanctions episodes. States often impose sanctions on states without close economic or political ties to them, even if those sanctions may lack the leverage needed to coerce behavioural change. Thus, research on the norms of sanctions imposition may also reveal insights for debates about sanctions effectiveness.
I hypothesise that norm-violating behaviour, norm characteristics, and sender relationships with potential target states may shape sanctions imposition. At a most basic level, states that violate widely shared international norms invite punishment, and sanctions may be seen as legitimate tools with which to punish them. For example, severe human rights abuses, conflict, nuclear weapons programmes, and support for terrorismFootnote 38 are among the most commonly cited grounds for arms embargoes. Both North Korea and Iran have faced extensive sanctions packages linked to their nuclear weapons programmes and been targeted by the US as ‘state sponsors of terrorism’. Democracies especially may use sanctions to promote human rights and democracy,Footnote 39 even against their strategic partners.Footnote 40 With the growth of targeted sanctions after 2000, moreover, sanctions may become more common, as they become more proportional and less costly to senders. Thus, the absence of a consistent significant relationship across norm violations and sanctions would indicate that international metanorms may not extend to material punishments or even exist at all.
Hypothesis 1
States that violate international norms are more likely to become the targets of sanctions than those states that do not violate norms.
In addition, the literature suggests that three characteristics may make some norms more ‘enforceable’ than others. From this perspective, sanctions are more likely for some types of norm-violating behaviour than others. First, norms that are formally institutionalised may attract more consistent enforcement by establishing what constitutes compliance, whether or not non-compliance merits punishment, and, if so, how to punish it. Second, norms whose violations are more readily monitored and observable might be more enforceable simply by providing more punishment opportunities. Norms against nuclear proliferation, human rights abuses and mass killings, and interstate conflict are institutionalised in formal agreements with (relatively) clear-cut obligations but varying levels of violation visibility. In contrast, despite treaties condemning specific types of terrorist activities, terrorism lacks an internationally agreed-upon definition, making both it and state sponsorship of it potentially more difficult to identify and punish. Public attacks aside, it is also typically difficult to observe and lacks easy-to-employ monitoring mechanisms. Norms against internal conflict also tend to be less well institutionalised than interstate conflict, although both types of conflict are generally observable.
Hypothesis 2a
Violations of more institutionalised norms are more likely to be sanctioned than violations of less institutionalised norms.
Hypothesis 2b
Violations of more ‘observable’ or better-monitored norms are more likely to be sanctioned than violations of less ‘observable’ or poorly monitored norms.
Third, states – especially powerful ones – may impose sanctions to protect their own material self-interest,Footnote 41 prioritising the enforcement of norms with direct links to their perceived national security interests. Thus, nuclear proliferation, some interstate conflict, and potential terrorism may particularly attract national security concerns. Conversely, states may be less motivated to incur costs to punish violations they perceive as minimally linked to their own security, such as human rights or internal conflict.
Hypothesis 2c
Violations of norms directly affecting potential senders’ perceived national security interests are more likely to be sanctioned than violations of other norms.
Finally, interstate relations may influence decisions about costly norm enforcement. As Gary Clyde Hufbauer et al. note, ‘Cross-cutting economic and security interests with the target regime complicate the construction of a sanctions package’,Footnote 42 both between targets and senders and between senders themselves. Disrupting economic ties between states can entail domestic costs, such as higher prices, unemployment, and reduced business profits. States may also want to avoid sanctions on military partners. For example, NATO allies have often been hesitant to embargo arms sales to Turkey, overlooking human rights concerns in favour of alliance cohesiveness and security.Footnote 43 Close relationships may also affect whether states even perceive norm violations as punishable: whereas an out-group's ‘bad’ behaviour is often ascribed to faulty character and more likely to be punished, similar behaviour by an in-group member is often excused as situation-specific.Footnote 44 Moreover, these material and social incentives to avoid sanctions may reinforce one another as interstate relations evolve over time, deepening economic, military, and social ties.
Hypothesis 3
Close economic, military, or political relationships between states will have a negative effect on sanctions imposition.
Research design and data
The statistical analyses focus on the post-Cold War period, when sanctions and arms embargoes became more commonly used tools of international norm enforcement (Figure 1). Although they are a type of economic sanction, I provide separate analyses of arms embargoes. While multilateral arms embargoes have generally been well recorded, unilateral arms embargoes have been less fully tracked in sanctions databases, meaning that documented sender-target dyads targeted by sanctions in general and arms embargoes specifically are similar but not the same. Yet, arms embargoes are the most frequently used type of economic sanction and often one of the first employed against violations of security norms, both on their own and as part of broader sanctions packages.Footnote 45 They can impose direct costs to punish security norm violations, since arms availability can affect actors’ ability to engage in those violations. Importantly, because arms embargoes are a specific type of sanction (not an aggregate measure), the analyses can better assess whether and how interstate relationships may affect embargoing decisions.
Total exporter-importer dyads with arms embargoes and sanctions.Footnote 46
Aggregate sanctions measures entail many different costs to many different actors. The arms embargo analyses, in contrast, can isolate whether costs to specific domestic interests influence norm enforcement. Arms transfers are commonly seen as crucial for the health of national defence industries, which tend to have close relations with their governments. In turn, governments may be reluctant to jeopardise their hard-won arms transfer relationships, which themselves are often the result of close political and security commitments. Nevertheless, in targeting an actor's military capabilities, arms embargoes can reduce norm violators’ ability to wage war, build weapons programmes, or repress their populations.Footnote 47 Arms embargoes are also less likely to be inhibited by humanitarian controversy than sanctions packages that target an economy more broadly.
The analyses use state-level dyadic data covering top arms-exporting states and potential customers, in order to examine relationships between potential norm-enforcing states and potential norm-violating states. States are a more relevant unit of analysis than international institutions. Although unilateral sanctions may be seen as less legitimate and less effective than multilateral sanctions, they are not uncommon. Even when norms are embedded in international institutions, member states must agree on and implement sanctions, either through those institutions or on their own. Dyadic data, in turn, can account for material costs to senders, depending on their chosen targets, capturing whether and how target behaviour and relationships affect norm enforcement. Dyadic data also reflect the reality that targets can face sanctions from multiple senders, or from some states and not others.
Dependent variables: Arms embargoes and economic sanctions
I conduct analyses of sanctions practices with the Threat and Imposition of Sanctions (TIES) data.Footnote 48 TIES provides broad coverage of sanctions through 2005, including tariffs, import or travel bans, export controls, embargoes, asset freezes, foreign aid cuts, and blockades. As an aggregate measure of sanctions, TIES includes sanctions that may less directly address a target's norm violations, and costs to senders and targets may be more diffuse and difficult to track than specific sanctions types, like arms embargoes. I convert TIES into a dichotomous variable, covering unilateral and multilateral sanctions from 1990–2005 by the 22 potential sending states in the Arms Embargo Dataset described next. If no sanctions are recorded between a potential sender and potential target in a specific year (that is, dyad-year), that year is coded 0.
For the arms embargo analyses, I update the Arms Embargo Dataset,Footnote 49 which covers 22 top small and major conventional arms exporting states (potential senders) and 189 potential importing states (potential targets) from 1981–2004, to 2010.Footnote 50 I create a single dichotomous variable indicating whether or not there is a record of a unilateral or multilateral arms embargo in each dyad-year. The embargo variable is coded 0 when there is no record of an arms embargo in a dyad-year. When an embargo is identified, I code the dyad-years for which it is in place the full year 1. Between 1990 and 2005, there are 3,229 dyad-years for which arms embargoes are documented in the Arms Embargo Dataset but no sanctions at all are documented in TIES. This is likely because unilateral arms embargoes in particular can require research in specialist defence industry publications that are not readily available in digital form or would not otherwise be examined for a general sanctions database.
Independent variables: Accounting for variation in norm enforcement
The primary independent variables of interest are six norm-violating behaviours of potential target states. I conduct separate analyses for each, in order to limit the control variables to those that may affect both the likelihood of that behaviour and the likelihood of sanctions,Footnote 51 rather than all variables that may influence sanctions imposition. This strategy avoids potential distorting effects such as collinearity and non-linearity that complicate or hinder understanding the specific relationships of interest for this study.Footnote 52 Appendix Table A1 summarises all variables. Because norm violations create enforcement opportunities (hypothesis 1), the independent variables capture security norms whose violations are frequently cited as justifications for economic sanctions. They also represent a range of characteristics, including international institutionalisation, compliance observability, links to potential sender interests, and target-sender relations (hypotheses 2a–c, 3).
I account for potential target states’ human rights records using the Political Terror Scale (PTS), which measures physical integrity rights.Footnote 53 Human rights are well institutionalised and monitored norms whose implementation powerful states do not typically view as directly important to their national security. Based on annual US State Department (DOS) and Amnesty International reports, PTS data are coded from 1 (very good human rights) to 5 (very bad human rights).Footnote 54 Because the PTS scale is ordinal rather than continuous, I create a dummy variable for ‘bad’ human rights with the scores most likely to attract sanctions (scores 4 and 5). Since ‘better’ human rights scores are less likely to incur sanctions, they serve as the reference category. I rely primarily on PTS scores coded from DOS reports in order to gain insight into government perceptions of human rights violations and their relation to norm enforcement. I also include a log of total human rights organisations with reported members in a potential target, which can make sanctions more likely due to greater monitoring and advocacy.Footnote 55
For mass-killing events, I code and update a dichotomous variable from the Political Instability Task Force (PITF) list. These norms are also well institutionalised and observable but in most cases do not directly affect powerful states’ security interests. PITF defines mass killing ‘as any event in which the actions of state agents result in the intentional death of at least 1,000 non-combatants from a discrete group in a period of sustained violence’.Footnote 56 It accounts for intentional direct and indirect mass civilian death caused by state officials against residents of their own countries. The variable is coded 1 for years in which mass killing events occur in a state and 0 otherwise.
Dummy variables for internal conflict and interstate conflict from the Uppsala Armed Conflict DatasetFootnote 57 cover potential target states’ conflict participation, based on annual battle-deaths of 25 or higher. Conflict participation tends to be well monitored internationally. Despite political and normative gray areas for when interstate war is permissible, it tends to be better regulated than internal conflict and may more directly threaten other states’ security interests.
For nuclear weapons development, I update Sonali Singh and Christopher Way's data.Footnote 58 Non-nuclear states that actively pursue or explore nuclear weapons development in a year are coded 1. Non-nuclear states with no demonstrated interest in nuclear weapons development are 0. Nuclear weapons states are excluded, since the 0 category (‘no interest’ in nuclear weapons) is obviously different than the possession of nuclear weapons. Non-proliferation norms are institutionalised in the NPT, monitored by the International Atomic Energy Agency and others, and linked to powerful states’ security interests.
Finally, I examine state support for terrorism, which involves ‘intentional assistance to a terrorist group to help it use violence, bolster its political activities, or sustain the organization’.Footnote 59 Although most states condemn terrorism and many see it as a threat, there is no agreed-upon definition in international law, making both it and state support for it contentious and poorly defined in practice. State support for terrorism can include financial support, weapons provision, training, or safe haven. Concrete information on terrorist groups and their sponsors is vague at best, leaving many scholars to rely on the DOS annual list.Footnote 60 However, the DOS list is highly political and automatically triggers US arms embargoes. I therefore create three rough dichotomous measures: The DOS list from Patterns of Global Terrorism 1981–2003 and Country Reports on Terrorism 2004–2010; an expanded version of the DOS list with additional cases based on state support for terrorist organisations identified in the Terrorist Organization Profiles (TOP) database;Footnote 61 and membership in counterterrorism conventions.Footnote 62 I rely on the expanded DOS list and report significant differences in results from the others in the footnotes.
The remaining control variables account for potential target states’ characteristics and relationships with potential senders that may affect both their norm-violating behaviour and senders’ willingness to punish that behaviour. I include logged GDP per capita as a measure of potential targets’ wealth and level of economic development,Footnote 63 which may affect their value as trading partners. Wealthier states are also less likely to engage in human rights violations, political violence, and interstate conflict,Footnote 64 but are more likely to develop nuclear weapons.Footnote 65 Select models include two additional economic variables. First, a state's oil production is commonly associated with human rights abuses and interstate conflict. It also indicates its value as a potential trading partner and source of privileged treatment on the global arms market.Footnote 66 I use an updated measure of millions of barrels of oil produced per day per capita.Footnote 67 Second, I use logged annual bilateral trade flowsFootnote 68 in the interstate conflict model, where higher trade may depress the likelihood of conflictFootnote 69 and sanctioning, and in the nuclear weapons model, where lower trade may make nuclear weapons pursuit more likely.Footnote 70 Since senders may treat export and import relationships differently, flow1 covers potential targets’ exports to potential senders in current US millions of dollars and flow2 covers potential targets’ imports from potential senders.
Several variables address interstate relationships. Since arms transfers can facilitate norm violations, indicate positive security and economic relations between states, and raise the costs of arms embargoes for some senders, I include a dichotomous variable for conventional arms transfers between states.Footnote 71 States in formal alliances also have security relationships, which may influence sanctions decisionsFootnote 72 as well as a target's sense of security.Footnote 73 I use an updated version of the dichotomous Correlates of War alliance variableFootnote 74 in the nuclear weapons and interstate conflict models, along with militarised interstate dispute counts to assess a state's propensity to engage in those bad behaviours as a result of its external threat environment.
Democracy may make states more sensitive to international rulesFootnote 75 and is associated with better human rights, reduced mass killings, and lower likelihood of support for terrorism. I therefore use Polity IV's polity2, classifying potential targets’ from full autocracy (-10) to full democracy (10)Footnote 76 in those three models. In the terrorism model, I also use the Index of Agreement with the United States,Footnote 77 since state support for terrorism in this period may be associated with a wish to counter US hegemony or a negative relationship with the United States.Footnote 78 1 indicates that a state always votes with the US in the UN General Assembly and 0 indicates that it never does. Finally, I include two norm-violating behaviours as control variables in other models. For internal conflict, I control for bad human rights, since repressive practices are associated with the onset of both sanctions and internal conflict. For mass killing, I use the dichotomous internal conflict variable, as internal conflict is associated both with sanctions onset and mass killing.Footnote 79
Models and results
Importantly, the vast majority of dyad-years do not exhibit relevant norm-violating behaviour; most states appear to be relatively compliant with security norms (Figure 2). Among the six ‘bad behaviours’, severe human rights violations are the most common at 19 per cent of the total dyad-years between 1990 and 2010. Interstate conflict is least common, with 2 per cent of total dyad-years in this period.
Percentage of dyad-years engaging in norm violations.
Across all norm-violating behaviours, sanctions are even more rare, however well institutionalised or monitored the norm (hypotheses 2a–b).Footnote 80 Cross-tabulations illustrate whether dyad-years in which violations are recorded have been punished with arms embargoes or economic sanctions (Figure 3). Ninety-two per cent of dyad-years with a state engaged in interstate conflict and 81 per cent engaged in internal conflict are not subject to arms embargoes; that figure drops to 72 per cent and 76 per cent respectively with general economic sanctions. Seventy-five per cent of dyad-years with major human rights violations and 76 per cent with mass killing events also lack embargoes (sanctions figures are similar). Frequency is highest for states pursuing nuclear weapons (38 per cent receive embargoes and sanctions) or supporting terrorism (32 per cent for embargoes, 40 per cent for sanctions), perhaps because prominent states see them as posing security threats (hypothesis 2c). Yet even here, sanctions are rare relative to norm violations. Most states need not clean up their acts in order to avoid sanctions.
Percentages of norm-violating dyad-years sanctioned.
Regression analyses permit a closer examination of norm enforcement by introducing control variables that might help to account for the low incidence of material punishment. For each ‘bad behaviour’, I use a logit model with a dichotomous arms embargo dependent variable (embargo/no embargo) or a dichotomous sanctions dependent variable (sanctions/no sanctions).Footnote 81 Control variables are limited to those with a potential confounding relationship between the dependent variables and the independent variables of interest to limit problems with non-linearity and collinearity. I use panel-corrected standard errors and year dummies to deal with contemporaneous autocorrelation in the error terms. All independent variables are lagged one year to give time for information about states’ bad behaviour to travel to decision-makers in potential sending states and for them to decide whether and how to deal with it.Footnote 82 Tables 1 and 2 report the coefficients for arms embargoes for all years (1990–2010) and sanctions for all years (1990–2005) respectively. Each column reports the results for each norm-violation model. I also calculate the differences in predicted probabilities for each of the ‘bad behaviour’ variables, as a source of insight into the magnitude and significance of their effects on sanctions imposition.Footnote 83 In addition to the predicted probabilities, which indicate how well the ‘bad behaviour’ variables perform, the chi-squared figures allow me to reject the null hypothesis that all of the model's coefficients except the constant are equal to zero and have no effect on the dependent variables.
Logit analysis of arms embargo targets.
Note: *significant at the .05 level; **significant at the .01 level; year-dummy coefficients omitted for space.
Logit analysis of sanctions targets.
Note: *significant at the .05 level; **significant at the .01 level; year-dummy coefficients omitted for space.
The results reveal that targets’ norm-violating behaviours are positively and significantly correlated with sanctions and arms embargoes (hypothesis 1), regardless of norm characteristics (hypotheses 2a–c). States that had poor human rights records,Footnote 84 mass killing events, interstate conflict, nuclear weapons pursuit, and terrorism supportFootnote 85 were more likely to be punished with sanctions and arms embargoes than states that did not. Only internal conflict is insignificant in the arms embargo models.Footnote 86 Even so, the differences in predicted probabilities expose a more sobering story about the magnitude of effects. Holding all other variables constant, a change from not engaging in each norm violation to engaging in that norm violation has no significant effect, regardless of institutionalisation, observability, or links to national security. While norm violators are likely to be sanctioned, the probability is nevertheless low.
Most importantly for understanding the role of interstate relations, arms transfers between senders and targets significantly decrease the chances of an arms embargo across all six norm violations. Senders are less likely to embargo arms to states with which they have an arms transfer relationship (hypothesis 3). However, for the aggregate sanctions measure, arms-transfer relationships are insignificant, with the exception of mass killing and interstate conflict, where they increase and decrease the chances of sanctions respectively. Alliances are significantly negatively correlated with arms embargoes in the interstate conflict and nuclear weapons models but positively correlated with economic sanctions.
Most other control variables perform as expected, although there are some differences between arms embargoes and economic sanctions. Interestingly, the number of human rights organisations decreases the chances of arms embargoes but is insignificant for economic sanctions. Oil production has no significant effect for economic sanctions, but, for arms embargoes, it is significantly negative with human rights and significantly positive with interstate conflict.Footnote 87 Agreement with the US is significantly negatively correlated with arms embargoes for terrorism support but significantly positively correlated for economic sanctions. States thus appear much more careful to protect relationships linked to arms embargoes specifically, rather than with economic sanctions in general.
I next conduct ‘moving windows’ or ‘moving regression’ analyses for each model, in order to examine how sanctioning practices have changed over time (if at all). These analyses can indicate whether metanorms might evolve as sanctions become more common tools of norm enforcement and push states to overcome some of their reluctance to employ sanctions in valuable relationships (hypothesis 1). In particular, as targeted sanctions become the state of the art after 2000, states may find sanctions more ‘useable’ for reasons of both proportionality and avoiding excessive material costs to themselves. The moving regressions analyse five years at a time (five-year windows), shifting the window by one year at a time, in order to provide a more fine-grained look at sanctioning practices over time. Thus, the coefficients for 1990 are based on logit regressions for the years 1986–1990, the coefficients for 1991 are based on 1987–1991, and so on.
Figures 4 and 5 illustrate the results for arms embargoes and sanctions respectively. These results provide insight into the likelihood of sanctions given a potential target state's engagement in that behaviour, but not the magnitude of the effect. The figures show the coefficient and the upper and lower 95 per cent confidence intervals. The years for which the full confidence interval is above or below the zero line are the years for which the coefficient is significantly positive (above the line) or significantly negative (below the line).
Norm enforcement by arms embargo (1990–2010).
Norm enforcement by economic sanctions (1990–2005).
Although the results show that norm-violating behaviour tends to make sanctions more likely, there is some variation across different types of norm violations. Some types even appear to have weaker effects in recent years, rather than increase the likelihood of imposition as targeted sanctions become more common. Norm-violating behaviour has the most significant effects on the likelihood of arms embargoes from the late 1990s and early to mid-2000s (Figure 4). Overall, the mid-2000s seem a high point for arms embargoes. Only nuclear weapons development is significant across all years, although human rights comes close.Footnote 88 Internal conflict and mass killing gain momentum after the early 1990s. Support for terrorism and internal conflict both weaken in the most recent years. Interstate conflict briefly shows positively significant coefficients in the early 1990s and 2000–04, but it is otherwise insignificant (with the exception of 2005, a significantly negative year).
The picture for economic sanctions as a whole (Figure 5) is much more consistently significant. Over all years, human rights, nuclear weapons development, and support for terrorism significantly increase the chances of sanctions, as does mass killing after 1992. After 1995–6, both internal and interstate conflict also appear to attract sanctions. Although it appears as though sanctions may occur more consistently over time, these analyses lack data for post-2005, the period when the arms embargo data began to reveal some weakening trends. Interestingly, coefficients for conflict and nuclear weapons development appear to be moving closer to zero in the most recent years, in contrast expectations in light of growing prominence for targeted sanctions after 2000.
Discussion
The results broadly suggest that metanorms – and not simply the relative ‘enforceability’ of different norms or state interests – may guide sanctioning behaviour. Sanctions are commonly invoked for violations of international and human security norms, both before and after the rise of targeted sanctions. All six norm-violating behaviours are linked to increased chances of costly punishment in world politics (hypothesis 1), not just those that have more ‘enforceable’ qualities (hypotheses 2a–c). Violations of norms that are better institutionalised, more observable, or directly affect national security interests do not systematically produce stronger effects on sanctions imposition.
These findings may seem straightforward, but they nevertheless establish an important baseline that links sanctions to norm violations in general and not simply to specific norm characteristics or state interests. Indeed, the overall consistency of sanctions practices points to a role for international metanorms in costly norm enforcement, which may promote certain acceptable sanctioning behaviour and provide general standards for sanctions imposition. In doing so, metanorms may also lay out when sanctions are inappropriate. For example, they may condemn sanctions on compliant behaviour – such as the US had planned against Iran despite the Joint Comprehensive Plan of Action – or secondary sanctions against other states for not joining those sanctions.
Even so, the majority of violations in the analyses go without material punishment. As much as experts worry that sanctions are overused,Footnote 89 the frequency of the behaviour sanctions seek to punish is greater than the frequency of the punishment itself. In particular, interstate relations emerge as a feature, rather than a footnote, of international norm enforcement (hypothesis 3). Most notably, arms-transfer relationships exercise a consistent dampening effect on arms embargoes, but not economic sanctions (with the exception of interstate conflict). Senders may strategically select not just sanctions targets, but also sanctions types, to avoid directly upsetting important relationships. During the Arab Spring, for example, arms-trade relationships between potential targets and major arms-exporting states prevented most UN-level conversations about arms embargoes. Rather than reducing weapons flows, European arms transfers to the region doubled from 2007 to 2012.Footnote 90 While met with some domestic criticism, the absence of norm enforcement roused little concern from other states. This was not an isolated incident. Indeed, the findings show that sanctions are systematically limited by interstate relationships, which may have both social and material value to states that shape their sanctioning practices.
Such widespread behavioural patterns may indicate that protecting relationships and associated interests is part of the metanorm, not unacceptable behaviour itself subject to reprimand. Clearly, with the exception of some top security concerns (hypothesis 2c), senders are frequently unwilling to sanction valuable partners (hypothesis 3). It may be that at times they turn a blind eye to norm violations in line with psychological explanations, or they may simply prefer to protect their interests and avoid the costs of punishing their partners, choosing less costly social sanctions or eschewing any punishment at all. Such behaviour may even be built into acceptable norm enforcement practices, helping actors avoid the expectation of making costly punishments in the first place. If so, metanorms may recommend social forms of enforcement not examined here, which lean on relationships and social in-groups for influence. Yet inconsistent application could also open norm enforcers to charges of hypocrisy and reinforce social divisions in the international community that could make global security governance more difficult. Thus while selective imposition may protect interstate relations (and with them, potentially increase the value of social sanctions), it may also risk norm erosion and send mixed messages about the social value of norms and sanctions as a legitimate tool of norm enforcement.
Conclusions and implications
Overall, the empirical analyses show that sanctions imposition is positively and significantly related to violations of all six security norms, despite their varied ‘enforceability’. This suggests that there is some shared understanding in international politics – particularly among major producing states – that economic sanctions are useful and appropriate tools of norm enforcement. In other words, there is at least a basic metanorm that leads states (even if not universally) to accept sanctions as a legitimate form of norm enforcement. Since the end of the Cold War, using sanctions to achieve coercive goals other than norm enforcement may consequently even be considered inappropriate. However, the story does not end there. The analyses also show that sanctions imposition is negatively and significantly related to valuable interstate relationships and that sanctions are rare relative to norm-violating behaviour. To the extent that international metanorms exist, they may accommodate and not condemn selective punishment. Yet even as norm violations often escape costly sanctions, they remain uncommon, highlighting the potential value of other tools of enforcement and socialisation.
By revealing cross-norm patterns of sanctions imposition, this article provides insights into international metanorms and the practices that sustain them. It also opens up additional questions and implications for research. First, not all sanctions are created equal, and metanorms may evolve. Multilateral sanctions are seen as more useful than unilateral sanctions, and UN sanctions as more legitimate than non-UN sanctions. Similarly, targeted sanctions have become more acceptable than comprehensive sanctions. Aggregate sanctions measures may therefore provide less insight into sanctions imposition and norm enforcement than specific types like arms embargoes, making it important to study target selection in combination with the types of sanctions chosen.
Second, the infrequency of sanctions relative to norm-violating behaviour may suggest that international metanorms are ‘weak’. However, metanorms may simply function differently in international politics. States’ relationships and associated interests may win out with little cost to the potential enforcer when they come into conflict with sanctions. Social psychologists argue that metanorms are stronger where communities are highly interdependent, those carrying out punishments receive collective support, and the norms being enforced offer greater benefits to the community.Footnote 91 This is a high bar for a decentralised system of sovereign states, even with the help of international institutions. International metanorms may therefore work differently. Rather than compensation for high-cost norm enforcement, states may get wide berth in choosing to enact or avoid costly punishments without being criticised themselves. In doing so, international metanorms could allow states to separate norm violations in one area from beneficial relationships in others to keep international politics functioning. Those important social, political, and economic relationships may, in turn, better promote norm-compliant behaviour by other means, such as engagement, internalisation, and social sanctioning. Thus, interstate relationships may not rule out enforcement so much as complicate the selection and implementation of enforcement tools. In response, scholars may find it useful – if not necessary – to consider material and social variables and rationalist and non-rationalist arguments alike in their research.
Third, the results encourage scholars to explicitly consider senders’ relationships with and selection of targets as potential components of understanding sanctions effectiveness.Footnote 92 Despite (or perhaps because of) the perception of sanctions as a popular norm enforcement tool, their effectiveness has frequently been questioned.Footnote 93 Sceptics argue that sanctions cannot create sufficient incentives for targets to follow the will of the senders.Footnote 94 Proponents argue that sanctions can help motivate behavioural change when properly designed.Footnote 95 If sanctions on friends and allies provide greater incentives to resolve disputes, change policies, and comply with norms, then sanctions that avoid these targets may struggle to change behaviour. Senders may have the most leverage in cases they are the least willing to exercise it. Research should therefore account for the complexities of interstate relations in how states choose to employ sanctions (or select other tools), in order to avoid the selection bias of assessing the outcomes of only those sanctions imposed.
Finally, it is important to note that norm violations are rare in world politics – but not as rare as costly material punishments of them. If the promise of material punishment for norm violations were the primary source of norm compliance, the findings here suggest that violations should be more common. States could make a reasonably safe bet that, as long as they are somewhat well connected, violations might avoid material sanctions. Yet the data indicate that violations of these six norms are relatively uncommon. For most states, compliance may not require credible threats of economic sanctions. Other sources of norm compliance, such as state interest, norm internalisation, and social sanctions, may be more important. Scholars and policymakers have long debated the merits of engagement versus punishment strategies for changing international actors’ behaviour. Based on these findings, it is worth considering whether, in preserving relationships, international metanorms strengthen these other forms of enforcement, if even perhaps at the expense of material sanctions.
Even so, if egregious norm violations are inconsistently punished or regularly not met with any form of punishment – social or material – at all, it may lead to norm erosion. States that do weigh the costs and benefits of non-compliance may reconsider their calculations; others may begin to see enforcement efforts as less legitimate. Reports from Bahrain and Egypt, for example, suggest that leaders have taken a heavier hand against opponents since US President Donald Trump took office, expecting support and access to arms rather than US sanctions and condemnation.Footnote 96 As once-accepted international norms and institutions are called into question, metanorms may be even more important to communicate their social value and the consequences of their violation. Eventually, inconsistent or frequent non-enforcement may undermine sender credibility, weaken norms, and signal that the risks of engaging in norm violations are low. Over the short term, if norms are sufficiently well entrenched, this may not affect compliance. However, over the long term, or if norms are new or contested, states may reassess whether norm compliance is in their interest at the expense of one important source of order in international affairs.
Acknowledgements
I thank the editors and two anonymous reviewers for their insights and suggestions. I am also grateful to Michael Brzoska, Jennifer Dixon, Jerry Easter, Bob Faulkner, Michael Herron, Isaac Kamola, Kathrin Kranz, Jennifer Lind, Sara Moorman, Daryl Press, Benjamin Valentino, Chris Way, William Wohlforth, and Amy Yuen for their comments and assistance. Thanks also to Caroline Tilden, Elizabeth Wall, and Sean Sudol for their research assistance.
Appendix
Summary of variables.
