Managing foreign affairs is in no small measure about anticipating the actions (and non-actions) of others, and about taking steps to limit the unexpected—and the undesired. Law has long been recognized as important to these tasks.Footnote 1 For goals ranging from lowering trade barriers and erecting environmental and resource protections to avoiding armed conflict, governments and officials frequently look to international law as a medium for coordination. As a consequence, there is a substantial body of international relations (IR) scholarship that examines when and why states make international legal commitments and how those commitments affect the behavior of governments and others subject to international law. Nevertheless, much current conventional wisdom about international law's effects and efficacy is built on undertheorized foundations that overlook three critically important properties of law and legally mediated decision making.
The first of these properties is semantic indeterminacy. Legal rules are formulated for general use and thus require interpretation to determine their applicability and meaning in relation to specific facts. Consequently, it is possible for different actors to reasonably reach different conclusions about whether a given rule applies, and what it permits or requires.Footnote 2 Second, no international legal rule operates in isolation. By this I mean not only that few issues of importance to international affairs plausibly implicate only a single law, but also that law is a system of rules—including many that directly concern the operation of other rules. Whether other rules facilitate or complicate any given task of applied interpretation is thus a legal-empirical question. Third, laws do not invoke, interpret, or apply themselves; doing so requires agents and agency. Different agents opt to invoke law and legal obligations at different times and for different purposes. This is because agents may vary with regard to their resources, capabilities, time horizons, and degrees of good faith, as well as their underlying profiles of legal obligation and relative legal authority.
Not considering how and to what degree this cluster of properties may influence who gets what, when, and how in legally mediated interactions risks nontrivial errors of causal inference and ascription in research featuring law or legal commitments as important variables. For example, efforts to study the effects of legal rules in isolation can lead researchers to attribute influence where it is wholly or largely spurious. It also may obscure how other (related) legal rules reinforce, condition, or qualify the rules of primary interest, or offer alternative legal pathways to similar objectives.Footnote 3 More broadly, ignoring these (admittedly messy) properties of law in favor of more methodologically tractable assumptions—for instance, that all actors hold identical understandings about the scope and meaning of particular legal rules—essentially blinds IR scholars to a universe of deeply political interactions in which legal rules and legally mediated behavior feature centrally.Footnote 4
A growing body of research into international regimes and “regime complexes” offers a valuable—albeit partial—corrective to overly siloed approaches to explaining the incidence, design, and effects of international agreements.Footnote 5 The core idea is that whenever institutions in a particular domain begin to cluster and overlap, this generates (formally) nonhierarchical and potentially contradictory sets of demands on agreement members. This multiplicity of crosscutting demands creates both challenges and opportunities for international governance.Footnote 6 For example, regime complexity may undermine legal-political bargains by permitting strategic actors to “shift” activities to more preferred domains, or to “shop” for more favorable dispute-resolution forums.Footnote 7 Other research shows how complexity within and across regimes can enable productive forms of bargaining that facilitate institutional specialization and regulatory burden sharing.Footnote 8 Still far from widespread, however, are clearly articulated ideas about the properties and processes of law itself that allow motivated actors to use it strategically.Footnote 9
I describe and theorize two crosscutting interpretative tactics that derive from the properties of law I sketched earlier. The first, legal mingling, involves drawing upon two or more legal obligations to interpret the content of one in relation to the other(s). The interpreter's objective may be to promote a favored interpretation by mingling mutually reinforcing features. Conversely, mingling also may be used to qualify, or dilute, less favored interpretations—by highlighting discretion-enabling tensions between rules or agreements, by establishing conditional hierarchies among them, or by setting conditions that trigger particular priorities. For either type of goal, mingling may involve repurposing obligations to make legal claims in ways that may not have been foreseen when time commitments were made.
The second tactic is strategic augmentation. This entails modifying (or encouraging others to modify) existing configurations of legal commitments in anticipation of future instances of legal mingling. Strategic augmentation can involve enacting new bilateral or multilateral commitments. It also can involve modifying or removing existing obligations.Footnote 10 Notably, the objective(s) for which mingling and augmentation are undertaken need not be limited to the primary purpose of the agreements.
Semantic indeterminacy coupled with the sources of contextual variance I describe strongly suggests all legal rules can support some n > 1 range of applied interpretations, not all of which may be consistent. To be clear, I am not suggesting that individual laws can be interpreted to mean absolutely anything an interpreter desires. To the contrary, natural language constraints, formal and informal frameworks that guide the interpretation of legal texts,Footnote 11 and observed patterns of prior usage all limit when and who may invoke law, and the meanings it may support.Footnote 12 A further practical impediment to playing fast and loose with legal rules is anticipated reactions from wider audiences of legal experts (including, but not limited to, judges), government bureaucrats, and other engaged actors who together make up ever-shifting “interpretative communities” whose tacit or active cooperation may be needed.Footnote 13 As a consequence even very powerful actors generally cannot unilaterally reinterpret international legal rules to make them mean something altogether different from how they are customarily used.Footnote 14
One important implication of the argument thus far is that international legal rules and commitments may provide noisy signals concerning expected behavior. Nevertheless, it is common for rationalist IR treatments to assume that governments (or responsible agencies within governments) have fully formed, durable preferences about when, how, and for what ends their international legal obligations may be invoked, and what they may be held to require. In practice, however, such preferences may have latent aspects that emerge only in response to specific configurations of facts, or in interactions with particular categories of actors. Even highly capable users of international law thus may not be able to anticipate with any precision how others—left to their own devices—will assess the applicability, meaning, and weight of legal rules in every situation in which those rules may be invoked.
For this reason, where political stakes are high, concerned actors may not be content to leave interpretative options undefined. Instead they may try proactively to shape how prospective interlocutors interpret their own and others’ obligations in order to reveal, stabilize, or alter their preferences and intentions.Footnote 15
This account resonates strongly with Riker's concept of heresthetics—the “strategy value” of making particular arguments in particular circumstances—and extends it to the international legal arena.Footnote 16 Riker defines heresthetics as separate from, but related to logic, grammar, and rhetoric, in that each involves the “art” of “us[ing] language to accomplish some purpose: to arrive at truth, to communicate, to persuade, and to manipulate.” Importantly, decision makers need not be persuaded in any normative sense by the heresthetician's rhetoric; they need only feel “forced by circumstances” to act in a manner consistent with its preferences.Footnote 17 Framed in Riker's terms, this can entail aspects of both agenda control and “manipulating the dimensionality of issue space.”Footnote 18 Competent legal advisors are thus masters of heresthetics—selecting and interpreting legal rules (perhaps in conjunction with related cost-benefit considerations and normative arguments) to construct reliable pathways of (rational) choice leading to particular outcomes or actions. This task is hardly limited to skillful drafting of agreements or legislation. On the contrary, it is an ongoing endeavor—not least because users of law do not cease to be strategic once agreements are concluded or laws are enacted.Footnote 19
Anticipating when, how, and in whose hands, legal-interpretive wiggle room may be used (or created) and working to shape it can be difficult. Governments that try to use international law to formulate or further policy goals quickly discover that a single approach almost never fits all states, since no rule, or cluster of rules, can be expected to influence all actors identically. First, common baselines of legal obligation are rare, which makes the expected marginal impact of new commitments variable.Footnote 20 Second, governments (or their subcomponents) also may have different complementary or competing objectives, or preferences about how to achieve shared objectives. Eliciting and maintaining credible assurances from others about how they will configure their obligations thus demands investments of political and legal-interpretative resources.
Of course, not all states are equally capable of configuring legal environments and choice situations using mingling and augmentation.Footnote 21 Drezner observes that powerful states are often more capable of leveraging institutional complexity to create advantageous forms of legal and policy flexibility.Footnote 22 This requires convincing others to go along. The type of “power” that enables states to establish heresthetic dominance over high-priority issues is a combination of interpretative authority and legal-administrative capacity. The first of these entails an ability to craft broadly plausible accounts of legality and illegality tailored to the legal profiles of different interlocutors, drawing on available legal materials both within and across issue domains.Footnote 23 Often this requires substantial legal-administrative capacity in the form of dedicated institutions (like ministries of foreign affairs) staffed with experienced legal experts. Not surprisingly, administratively capable states are also relatively more able to design new international agreements to address identified legal lacunae, and, all else equal, to secure international political support for their enactment.
If efforts to leverage legal obligations alone fail to deliver desired objectives, states with high degrees of administrative capacity often also have other, extra-legal means to help tip the balance in favor of their preferences, such as threats to withdraw benefits or to impose sanctions.Footnote 24 These should be used sparingly, however, since reliance on such mechanisms can easily undercut the efficacy of legal-normative logics vis-à-vis both the objects of manipulation, and wider interpretative communities. In short, successful legal argumentation involves finding arguments that are both compelling to the targets of influence, and sufficiently plausible to outside audiences to avoid challenge.Footnote 25 Where these conditions are satisfied, deciding when, how, and in which forums to leverage legal obligations becomes a matter of political choice.
I now turn to demonstrating how opportunities for agreement mingling and augmentation function in complex legal environments by reexamining US efforts to insulate its citizens from unwanted exercises of jurisdiction by the International Criminal Court (ICC).Footnote 26 Conveniently, other scholars have previously used this example to test competing ideas about behavior related to treaty commitments and treaty design.Footnote 27 These studies thus provide ready alternative accounts.
From Early Sponsorship to Threats to “Invade the Hague”
For much of the 1990s the United States supported the creation of a permanent court for international criminal matters. Nevertheless, the agreement adopted by 120 states at a United Nations Conference of Plenipotentiaries in Rome on 17 July 1998 was, in many respects, a US defeat. At the core of US dissatisfaction was the independent authority granted to the chief prosecutor to open investigations and issue indictments without the approval of the indictee's state of citizenship. In addition, the Rome Statute obligates its members to cooperate in securing custody over all persons under ICC indictment. Thus, even if the United States did not join the ICC, US nationals, and those of other nonmember countries, could be (in theory) forced to stand trial in The Hague for international crimes if apprehended in the territory of an ICC member.Footnote 28
During pre-Rome Statute negotiations and in subsequent Preparatory Commission meetings,Footnote 29 the United States lobbied hard to secure a UN Security Council veto over the court's prosecutorial decisions, and a blanket exemption from ICC jurisdiction for US officials and military personnel.Footnote 30 A core US concern was that the ICC might be used to bring “political” charges against American officials and military personnel under controversial formulations of international crimes.Footnote 31 Having largely failed to achieve these objectives, in May 2002 the George W. Bush administration withdrew the Clinton administration's earlier signature of the Rome Statute.Footnote 32
Less than four years after the Rome Statute opened for signature, it had the sixty ratifications required to enter into force and on 1 July 2002, the ICC was established. Two weeks later the United States initiated a policy whereby it asked every state with which it had diplomatic relations to sign an agreement promising not to “surrender” any US citizen to the ICC without US consent.Footnote 33 These bilateral nonsurrender agreements exploited a safety clause in Article 98(2) of the Rome Statute that prohibits the court from issuing surrender requests if it would require the requested state to violate another legal obligation.Footnote 34 These agreements’ sole purpose was to create a conflicting obligation.Footnote 35 In response, several states insisted that these agreements violated the spirit, if not the letter, of Article 98, and thus questioned their legality.Footnote 36 The architects of this initiative insisted that its objective was not to undermine the ICC, but to protect US persons from any “politically motivated criminal accusations, investigations, and prosecutions” that might arise out of the United States’ global engagement in international security matters.Footnote 37 The US government thus sought to use bilateral means to get what it failed to achieve multilaterally, and to use international law and legal processes where it had fallen short with politics alone.
These efforts were reinforced by the US Congress with the 2002 American Servicemembers Protection Act (ASPA). This statute prohibits US officials from assisting the ICC, and infamously authorizes the president to use “all means necessary and appropriate” to secure release of US entities detained by, or on behalf of, the ICC—a provision that earned ASPA the epithet, the “Hague Invasion Act.”Footnote 38 More salient to many foreign governments, ASPA also stipulated cuts in military aid to ICC members that refused to sign an ICC nonsurrender agreement. These sanctions were slated to come into effect in August 2003. In December 2004, additional sanctions for development aid were added under the Nethercutt Amendment.Footnote 39
However, the ASPA exempts from sanction all North Atlantic Treaty Organization (NATO) members and other major non-NATO allies.Footnote 40 Several additional states were de facto exempt, since they requested little or no ASPA-targeted aid. This includes all non-NATO Western and Central European states, several oil exporters (e.g., Brunei, Qatar), and states that were already under US sanction for other reasons (e.g., Burma, Côte d'Ivoire, Syria). The act allows the president to waive sanctions on a case-by-case basis, though such waivers were used sparingly prior to 2006.Footnote 41 In 2003, twenty-five states had ASPA-related aid cut for refusing to adopt an ICC nonsurrender agreement. In 2004 and 2005, twenty-four states had aid cut, and in 2006, twenty-three states experienced cuts.
Several states (including many ICC ratifiers) adopted ICC nonsurrender agreements early in the initiative. Others did not, even after cuts in US aid. By 2005, however, evidence was mounting that ASPA sanctions were harming US efforts in its “wars” on drugs and terror.Footnote 42 In late 2006, President Bush waived ASPA sanctions against all remaining nonexempt states. The waivers were renewed biannually, adding any new ICC members. The Obama administration continued this policy.Footnote 43 Although the United States subsequently took a more relaxed stance toward the ICC, it has not renounced ICC nonsurrender agreements—nor have any of the 101 states that ultimately adopted one.
Minimizing US Legal Exposure to ICC Jurisdiction
Did the United States succeed in its efforts to create legal impediments for foreign states opting to surrender indicted US citizens to the ICC without its permission? Framing the effort narrowly in terms of extracting commitments to bilateral nonsurrender agreements might suggest that the US was at best only moderately successful. Indeed, nearly half the states in the international system, including many ICC members, refused to adopt nonsurrender agreements. Among the ICC members that refused were several of the US's closest economic and security partners.Footnote 44 However, looking only to willingness to adopt a nonsurrender agreement together with ICC membership omits a great deal of relevant information about the degree to which the US government was able to insulate US citizens from ICC jurisdiction. This is because US officials did not design and carry out the ICC nonsurrender agreement campaign in a legal vacuum.
Other international agreements—most notably, extradition treaties, status of forces agreements (SOFAs), and mutual legal assistance treaties (MLATs)—also had direct relevance to US vulnerability to unwanted exercises of ICC jurisdiction before, during, and after the ICC nonsurrender agreement initiative. It follows that limiting attention to foreign states’ decisions about whether to sign a single agreement (in this case a bilateral nonsurrender agreement) in a complex legal and policy environment yields an incomplete and potentially faulty understanding of US efforts, and of other governments’ responses.
Kelley describes refusals to sign ICC nonsurrender agreements with the United States among states that had joined or intended to join the ICC as “principled” behavior.Footnote 45 Although this is doubtlessly true to a degree, taking a principled stance had very different price tags for different states. This is not only because, as Kelley notes, economic and military dependency on the United States varies, but also because the wider regime of legal obligations around extradition in many cases offered the US other legal levers to pull if it ever needed to prevent the unwanted surrender of a US citizen to the ICC. This broader menu of options comes into focus only by stepping back from a siloed view of political and legal interactions around ICC nonsurrender agreements
Mingling and Augmentation of Extradition Obligations
ICC nonsurrender agreements are but one feature in a broader politics of US efforts to frustrate ICC authority over US citizens. Extradition treaties, SOFAs, and MLATs all contain provisions that the US government can draw upon and mingle to block unwanted surrenders.Footnote 46 When interpreted together, the resulting obligations embodied in these other agreements approximate the blocking ability of an ICC nonsurrender agreement.
• Extradition treaties obligate signatories to transfer indicted individuals upon request to the custody of another treaty partner to face criminal charges, provided the terms of the treaty are met. Older extradition treaties generally enumerate lists of serious crimes for which extradition may be requested. Contemporary extradition treaties instead require “dual criminality”—the idea that extraditable offenses must be a crime in both the requesting and the sending state. All extradition requests are subject to the “rule of specialty,” which requires states to charge extradited individuals with only crimes specified in the request.Footnote 47 In 2002, the United States had extradition arrangements with 101 states.
• Status of forces agreements allocate criminal and civil jurisdiction over military forces that are deployed abroad with the foreign government's consent. Because most international crimes within the ICC's mandate involve use of lethal force, such arrangements directly affect how vulnerable US military personnel are to unwanted surrender. Although the scope and details of US SOFAs vary, most grant the United States exclusive criminal jurisdiction over alleged crimes by its military personnel in the course of official duty. Several also grant exclusive jurisdiction over accompanying civilian personnel. Most SOFAs are bilateral. However, two multilateral SOFAs—one among NATO members and another for members of the “Partnership for Peace”—account for roughly half of the dyadic SOFA arrangements to which the United States is party.Footnote 48 The United States had SOFAs with 118 states in 2002.Footnote 49
• Mutual legal assistance treaties set terms for information sharing and cooperation in international law enforcement and judicial proceedings in criminal matters. They can be bilateral or multilateral. MLATs, like extradition treaties, typically contain limitations on the types of information that can be requested, and the purposes for which it can be used.Footnote 50 Where custody of a US citizen is obtained abroad using information provided in part by US officials, MLATs can complicate efforts to surrender that individual to a third party, including the ICC, and to provide testimonial or documentary evidence required for ICC proceedings.Footnote 51
Mingling SOFAs, extradition treaties, and MLATs creates a barrier to the surrender of indicted US citizens that is comparable to an ICC nonsurrender agreement.Footnote 52 However, not all options for placating the United States on ICC matters were available in equal measure to all states. In countries where the ICC was popular domestically, and the United States less so, ICC nonsurrender agreements were denounced as examples of American arrogance and hypocrisy, and signing one risked prompting political backlash.Footnote 53 Both the EU and MERCOSUR issued common positions, in 2002 and 2003 respectively, discouraging their members from adopting ICC nonsurrender agreements. Although formally nonbinding, these positions increased the stakes for acquiescing to US demands, particularly for weaker and aspiring member states.Footnote 54
In Eastern Europe during this period most states eagerly sought (independent of ICC dynamics) membership in NATO or the Partnership for Peace. Both include SOFAs that give the United States strong extradition priority. This gave these governments an alternative for providing the United States with assurances against unwanted surrender of US citizens to the ICC. In most Latin American countries, by contrast, both SOFAs and ICC nonsurrender agreements were politically contentious, particularly when they were tied to US basing agreements.Footnote 55 These governments thus had fewer options to generate nonsurrender-related legal assurances (if this was an objective).
Similarly, states with few economic or political ties to the United States also had a truncated menu of options for generating assurances because ICC-related concerns were not the only, nor arguably the most important, considerations in play with substitutes for ICC nonsurrender agreements. States lacking minimal domestic respect for rule of law could not hope to conclude new extradition or mutual legal assistance treaties with the United States. Likewise, where the US did not already have, or actively seek, cooperative defense interests with foreign governments, the option of concluding (or upgrading) a SOFA was remote.
Defining Pathways to US Assurance
I turn now to an empirical examination of the mingling and strategic augmentation of extradition-related arrangements between the United States and other countries. The dependent variable here is us legal assurance on the question of blocking the unauthorized surrender of US citizens to the ICC. To assess this cross-nationally, I construct a scale with weighted scores for different agreements according to the amount of US legal assurance each creates or eliminates. I then trace how each state's score changes over time and by what mechanisms. I distinguish between mingling based solely on agreements that were in place prior to August 2002 and mingling with elements of strategic augmentation. I focus on the period from 2002, when the Rome Statute entered into force, to 2006—the natural cut point for analyzing the effects of heavy-handed US approaches to securing formal assurances against unauthorized surrenders.Footnote 56
The components of the scale and the weights assigned to each agreement type are summarized in Table 1. These weights are based upon the degree of coverage each agreement type provides relative to the blanket prohibition on surrender of US citizens contained in bilateral ICC nonsurrender agreements, factoring in their scope and conditions of use. Although the elements and values in Table 1 are specific to this exercise, the general approach can be used to map virtually any legally mediated policy space.
Elements of scale for US legal exposure to unwanted ICC surrender
In a world with no ICC, US exposure to ICC extradition would be 0. Joining the ICC decreases US assurance by three points, and an ICC nonsurrender agreement increases it by an equal amount. An extradition treaty or an MLAT with the United States increases US assurance by one point, since each provides a mechanism for blocking an unwanted ICC surrender, but each also has shortcomings or contingencies that make it less than a full substitute.Footnote 57 The scale differentiates between “strong” and “weak” SOFAs. Strong SOFAs give the United States exclusive or primary criminal jurisdiction over US service personnel for all activities undertaken in an official capacity.Footnote 58 Weak SOFAs—of which the United States has relatively fewFootnote 59—provide only concurrent jurisdiction between the US and the “host” state for alleged crimes.Footnote 60 In recognition of this difference, a strong SOFA increases US assurance by two points on the scale, and a weak SOFA increases it by one point.Footnote 61 The scoring is conservative in that it is purely additive. But, as noted, when a foreign state has several reinforcing obligations, their mingling for the purpose of determining particular decisions or courses of action can strengthen a government's hand beyond the separate value of each agreement.
How might legal mingling influence efforts to convince, cajole, or coerce an unwilling government (so identified by their unwillingness to sign a bilateral nonsurrender agreement) to disregard a future ICC request? In line with the broader ideas here, US officials’ ability to draw upon several agreements between states that serve similar purposes (here states’ interests in charging or prosecuting their own citizens for serious criminal offenses), combined with interpretative leeway provided by indeterminacies of treaty language, and thin or nonexistent underlying practice, is a resource in diplomatic bargaining.
While these separate legal obligations, negotiated primarily for other reasons, may not provide US officials with airtight legal claims against the unwanted surrender of US citizens in every instance, they nevertheless provide a foundation for bargaining that also implicates other aspects of bilateral or multilateral relations. If a government were to come under direct US pressure not to follow through with an ICC surrender request, the ability to appeal to an overriding set of legal commitments could provide both international legal cover, and domestic political cover, for conceding to US demands.
Tracing Change and Stasis in US Legal Exposure
A critical test of my claims about the use and strategic utility of agreement mingling and augmentation involves assessing how much the reduction in US exposure to unwanted extraditions between 2002 and 2006 can be attributed to ICC nonsurrender agreements versus other strategies. As emphasized, ICC nonsurrender agreements were not the only means for states to cooperate with the United States in the post-Rome Statute period. The option of mingling existing agreements with functionally similar assurances, or augmenting agreements to provide such assurances, was an option in many instances. Non-ICC members who were unwilling, or politically unable, to take any of these paths also retained the option of not joining the ICC.Footnote 62
Figure 1 offers a comparative snapshot of US legal exposure in 2002 and 2006 as represented by the assurance scores of all 187 states with which the United States had official relations when the ICC nonsurrender agreement initiative was unveiled. Figure 1(a) shows states that ratified the Rome Statute prior to July 2002 (“Early Joiners”) together with states that had not joined the ICC at that point (“Other States”). Note that without a single ICC nonsurrender agreement having been signed, only six states were on the far left of the scale in July 2002, meaning that they joined the ICC with no extradition obligations toward the United States. The other seventy early ICC joiners had some prior relevant obligations in the form of an extradition treaty, SOFA, MLAT, or some combination. Of these, thirty-eight (or 50 percent) were at the 0 mark or above on the US assurance scale, meaning they were already heavily committed to respect US extradition requests through other extradition mechanisms.
Movement in US assurance against unwanted ICC surrenders between 2002 and 2006
As Kelley observes, states that ratified the Rome Statute after July 2002 had different information about the costs of joining the ICC in terms of inviting US pressure on nonsurrender. Among the 111 states that were not early ICC joiners, twenty-six (or 23 percent) were at 3 or higher on the assurance scale in July 2002. Thus, even if these states subsequently joined the ICC—which several did—the United States had several mechanisms in place to prevent (or seriously complicate) unwanted surrenders, even if they declined to adopt a nonsurrender agreement.Footnote 63 For the remaining eighty-five states, a post-July 2002 decision to join the ICC without any other changes to their extradition obligations toward the United States would flip them into a position where US citizens might be vulnerable to ICC surrender requests.Footnote 64 The US objective was thus to shift ICC members on the left (negative) side of the assurance scale to the right (positive) side, and to undercut the ability of non-ICC members to pose a future threat by joining the ICC.
Figure 1(b) reveals considerable movement toward US legal assurance during the period in which the US nonsurrender agreement initiative was active. Whereas in 2002 a sizeable proportion of the states that had joined the ICC fell below 0 on the assurance scale, by 2006 much of the gray-shaded area has shifted to the right, with a clear spike of ICC members at or exceeding 0 on the scale. Indeed, only fifteen of the thirty-eight early ICC joiners with negative assurance scores in July 2002 still had negative scores in 2006. Of these, two (Serbia, and Venezuela) increased their assurance by one point. Only three early joiners (Andorra, Namibia, and Niger) remained on the far left of the scale with no mingleable extradition obligations toward the United States.
Also worth noting is that among the 110 states that had not joined the ICC as of July 2002, thirty had no extradition obligations toward the United States (i.e., were at 0 on the scale in Figure 1a) making them likely sources of concern for US officials. By late 2006 only seven non-ICC members remained without any such obligations.Footnote 65 During this period, only seven states made the leap to negative assurance, meaning they joined the ICC after July 2002 without having, or enacting, a full suite of substituting agreements, or adopting a nonsurrender agreement.Footnote 66 Even so, each had some extradition-related obligations toward the United States.
Mingling and Augmentation in Action
Table 2 provides a descriptive breakdown of how states moved along the assurance scale. A state is tallied as having a set of “mingleable” commitments adequate to assure US officials if it was an ICC member with a score of 0 or above on the assurance scale in July 2002, or a non-ICC member at 3 or above. States that added non-ICC surrender agreement extradition obligations toward the United States after July 2002 are counted as having used strategic augmentation, whether or not the added obligations put the state over the conjectured threshold of “adequate” assurance.
Tally of assurance by mechanism between 2002 and 2006
The even-numbered rows in Table 2 indicate that a nontrivial amount of the US increase in legal assurance is attributable to states adopting ICC nonsurrender agreements. By the end of 2006, ninety-eight states had done so. Among them were forty-five ICC members, twenty-seven of which were “early joiners.” But this does not tell the entire story—about either ICC nonsurrender agreements or US effectiveness in achieving its broader policy objective. Obligation mingling and augmentation were also important to gathering assurance that US citizens would not be subjected to unauthorized surrender.
As Table 2 shows, already in July 2002, sixty-two states (including forty-five ICC members) had extradition obligations that could be mingled to approximate the assurance provided by an ICC nonsurrender agreement (rows 3, 4, 7, 8). Despite having this level of prior obligations, twenty-three of these states also signed a nonsurrender agreement (rows 4 and 8).
Rows 3 to 8 show how the US government could use agreement mingling to create critical masses of bilateral obligation on a situational basis among ICC members and non-ICC members alike toward, and sometimes considerably beyond, the conjectured minimum US threshold for assurance. Row 5 (which contains states that used alternative commitments concluded after July 2002 only) is particularly noteworthy, since these states, by all appearances, wanted to reassure the United States, but lacked the willingness, or the ability, to sign an ICC nonsurrender agreement.Footnote 67
Rows 5 to 8 show that between July 2002 and late 2006, thirty-two states entered into at least one new agreement that was not an Article 98 nonsurrender agreement, but nonetheless contained provisions governing the investigation, extradition, or transfer of US citizens.Footnote 68 The United States secured new or renegotiated SOFA arrangements with sixteen states.Footnote 69 Also in this period, the US concluded new extradition treaties with two ICC members—Lithuania and Peru. (If use of extradition treaties as an augmentation vehicle seems modest in the critical period, consider that between the adoption of the Rome Statute in July 1998 and August 2002, the United States negotiated, or renegotiated, extradition treaties with twenty-two states—a rate far above normal.)
When the Rome Statute was adopted, the United States had MLATs with just twenty-one states. Between July 1998 and August 2002, the United States formalized judicial assistance obligations with an additional twenty-seven states, including sixteen ICC members. During the period of the ICC nonsurrender agreement push, a further twenty-five states entered into MLAT arrangements involving the United States. Of these, eighteen were with ICC members. For eight of these states, the 2003 multilateral MLAT between the European Union and the United States created obligations where none existed previously.Footnote 70 In Latin America during this period, ten states (including five ICC members) likewise ratified the Inter-American Convention on Mutual Assistance in Criminal Matters, which the United States had joined in January 2001.Footnote 71
To clarify, the argument is not that insulating US citizens from the ICC's reach was the sole reason the United States enacted extradition treaties, SOFAs, and MLATs with other states during this period. The often reciprocal nature of extradition treaties and MLATs in particular ensures such agreements are not entered into lightly, but only after substantial vetting of a partner state's domestic institutions and laws.Footnote 72 Still, where such agreements were at least plausible for other reasons, they offered an alternative means for increasing US assurance about unwanted ICC surrenders.
Embedded ICC clauses and senate understandings. For states in a position to have, or to consider, extradition arrangements other than an ICC nonsurrender agreement, an additional option to provide reassurance involved embedding clauses in new or updated agreements to expressly prohibit extradition to the ICC (or in the case of MLATs, use of information acquired from US officials).Footnote 73 Closely related are so-called “understandings” precluding ICC extradition (or assistance) appended to the US Senate's ratification. Where tacitly accepted by multilateral treaty partners, understandings become an element of the treaty vis-à-vis the issuing state. In this way, they provided a back-door mechanism for foreign governments to assure the United States of how they would interpret, or allow others to interpret, their extradition obligations in the event of an ICC surrender request.
Table 3 summarizes how and when different states used these mechanisms. The left column lists states that embedded express assurances in other treaties prior to the launch of the ICC nonsurrender agreement initiative. The right column lists states that used these mechanisms after August 2002. States that had an assurance deficit, meaning a score of less than 0 for ICC members, or less than 3 for nonmembers, at the time an indicated agreement was concluded are in italics.
States with embedded ICC clauses or Senate understandings precluding ICC surrender or cooperation in extradition arrangements with the US as of December 2006
Notes: It was not possible for states to sign ICC nonsurrender agreements with the United States prior to August 2002. Therefore, checkmarks indicate action taken at a later date. ET = extradition treaty, MLAT = mutual legal assistance treaty, SOFA = status of forces agreement. Italics indicate an actual or potential assurance deficit at the time the agreement was concluded. Parentheses indicate country joined ICC after July 2002.
†Indicates obligation exclusively via the Inter-American Convention on Mutual Legal Assistance.
Note that eleven of the sixteen bilateral SOFAs concluded after the Rome Statute was enacted include clauses prohibiting extradition of US military personnel to the ICC or any “international court or tribunal.”Footnote 74 Although only one extradition treaty negotiated between August 2002 and December 2006 contained such provisions, all but one of the twenty-two negotiated between July 1998 and August 2002 contained an ICC clause or Senate understanding.Footnote 75 Six bilateral MLATS created after August 2002 had Senate understandings, as did twenty of the pre-2002 MLATS. The Inter-American Convention on Mutual Assistance in Criminal Matters also includes a Senate understanding prohibiting use of acquired information in ICC-related investigations applicable from 5 January 2001 when the United States ratified.Footnote 76 At the end of 2006, thirty-eight of seventy MLAT arrangements between the United States and other state dyads included either an ICC clause or a Senate understanding.Footnote 77
As Table 3 shows, many states that ultimately declined to sign ICC nonsurrender agreements were willing to provide the United States with specific assurance against ICC surrender prior to the launch of the bilateral nonsurrender agreement push. This includes several EU member states (France, Greece, Luxembourg, Spain, Portugal), as well as Eastern European states that were aspiring to membership in both NATO and the EU. A third well-represented group in this figure is Western Hemisphere states, which had fewer politically viable options for openly assuring the US government of their intentions. Eighteen states used these low-profile mechanisms more than once, including six that used them in both the pre- and post-August 2002 period.Footnote 78 One state (Cyprus) used them in all three agreement categories.Footnote 79
Of course, adding an ICC clause or a Senate understanding to any one of these agreements does not transform it into the full equivalent of an ICC nonsurrender agreement because of limitations in the scope of coverage I described earlier.Footnote 80 For this reason, I retain the scoring described in Table 1 in subsequent analyses to provide the toughest possible test of my argument.
Targeting US sanctions and exemptions. The ASPA framework imposed differential costs on states that declined to adopt an ICC nonsurrender agreement. These costs provide an opportunity to test claims about US strategy in managing the issue of exposure to ICC jurisdiction. In a resource-constrained world it would make little sense for the United States to push hard for ICC nonsurrender agreements where it already had solid legal claims to extradition priority—even if some Bush administration officials preferred to minimize US legal exposure wherever possible. If my argument about the potential for agreement mingling and augmentation using interpretative tools to provide assurance is correct, this should be reflected in how sanctions and exemptions were targeted. Specifically, ASPA exemptions should focus on countries with which the United States already had robust extradition arrangements, and sanctions threats should be aimed at ICC members with few substituting obligations. Further, US officials should have been less concerned about targeting states with no affinity toward the ICC, and those with few US economic or military ties, particularly if very small or geographically remote, since these factors would substantially reduce the possibility that US citizens under ICC indictment might be taken into custody in those states.
Kelley's examination of state responses to US requests for nonsurrender agreements does not encompass ASPA exemptions or sanctions threats, although she controls for sanctions imposed, and finds them negatively related to ICC nonsurrender agreement adoption.Footnote 81 Another empirical study of behavior around ICC nonsurrender agreements grapples with this surprising finding by exploring the determinants of ASPA exemptions. Noorudin and Payton argue that the contemporaneous US war on terror, coupled with the need to build a “coalition of the willing” in preparation for the 2003 invasion of Iraq, made US officials “loath to endanger relationships with close allies” by punishing ICC nonsurrender agreement resistance.Footnote 82 While plausible, their claim is not well supported by statistical or anecdotal evidence. To the contrary, of the thirty-four coalition members that contributed troops to Iraq operations, twenty-three were ICC members but only twelve were exempted under ASPA.Footnote 83 Tellingly, among this group, seven countries received no US aid, and one (Portugal) received a trivial amount. Only Poland and Hungary received enough ASPA-targeted aid to be used coercively.Footnote 84 Thus, if coalition building was the main driver of exemption allocation, it was a poorly executed strategy.
Noorudin and Payton also argue that between 2003 and 2007, the United States sanctioned only ICC nonsurrender agreement resisters over which it had little other leverage (“hard cases”), and that it chose not to sanction other “at risk” states.Footnote 85 They find further that democracies and states with high scores on the World Bank's Worldwide Governance Indicators (WGI) scale were more likely to receive US aid, but less likely to be sanctioned.Footnote 86 However, Noorudin and Payton's “at risk” pool includes several states that requested no US aid during the period of interest, or for a substantial number of years prior.Footnote 87 Excluding de facto exempt states reveals that the US government did sanction all ICC members that both requested ASPA-targeted aid and refused a nonsurrender agreement.Footnote 88 Consequently, explanations for allocation of exemptions and sanctions must lie elsewhere.
Looking to the full range of mingleable agreements around extradition provides a straightforward explanation both for ICC nonsurrender agreement exemptions and for the targeting of US sanction threats. First, all twenty-four ASPA-exempt states had extradition treaties with the United States in July 2002, and all but one also had a strong SOFA. Only New Zealand had no SOFA. This suggests the United States declined to press its close military allies to adopt ICC nonsurrender agreements not because it feared rupturing ties, but because ICC nonsurrender agreements were unnecessary. Of the thirty-one states that were de facto but not statutorily exempt in 2002 (and therefore “not targeted” by statutory design), six had both a strong SOFA and an extradition treaty with the United States, putting them at 0 on the assurance scale. By 2006, Andorra, Liechtenstein, San Marino, and the (then) new state of Montenegro were the only de facto exempt ICC members with negative assurance scores. Six other de facto exempt states remained outside the ICC.
Economic and Military Dependency on the United States
Another possible determinant of other states’ propensity to offer the United States assurance is dependency on trade, aid, and/or defense. The intuition is that states with high levels of dependency on the United States may be more willing to commit to not surrender US citizens to the ICC. Kelley finds that receiving trade benefits under the US Generalized System of Preferences (GSP) correlates positively with ICC nonsurrender agreement adoption.Footnote 89 Noorudin and Payton find that states with defensive pacts with the United States were slower than others to adopt nonsurrender agreements, even when they ultimately did so. They also find that as a country's percentage of trade with the US increases, its likelihood of adopting an ICC nonsurrender agreement increases by a small, but significant, amount.Footnote 90
To explore this relationship, I look to data on (a) the percentage of the foreign state's total export trade received by the United States lagged by one year; (b) the amount of US aid per capita (total combined defense and security assistance and economic development aid divided by population) lagged by one year; (c) whether the country receives GSP benefits; and (d) whether it has a defensive pact with the United States.Footnote 91 I consider diverse measures since countries with little US trade may still be highly dependent on US military or development aid, and states that receive no US aid may have trade or alliance dependencies.
Figure 2 encompasses four quartets of small scatterplots. Each quartet contains the universe of 188 states, divided into two-by-two subsets. In each scatterplot, the y axis is the assurance score as of December 2006. The x axis in the upper two quartets is the percentage of export trade with the United States. In the lower two quartets, the x axis is the natural log of US aid disbursements per capita. The two left quartets of the figure track whether the state has a defensive pact with the US. The quartets on the right track the presence or absence of GSP benefits. Within each, the plots in the left column contain states that joined the ICC prior to July 2002, and those in the right column contain states that joined after that date or not at all. Finally, each small plot shows a standard regression line and also a locally weighted (loess) regression line.Footnote 92
Weak relationship between trade, aid, and military dependency
This treatment of the data allows for visual assessment of several possible relationships simultaneously and highlights the general absence of any relationship, regardless of how one cuts the data. The largely flat standard regression lines in most of the scatterplots indicates no consistent relationship between key dependence indicators and a state's position on the assurance scale as of December 2006.Footnote 93 The sole exception is the top pair of scatterplots in the lower left quartet, which show a positive and significant relationship between receiving different amounts of US development aid and levels of legal assurance against unwanted ICC extradition among states with a defensive pact with the United States. Even here, however, the high dispersion of the data at low levels of aid suggests that little is explained by these variables. There is also great sensitivity to outliers receiving especially high amounts of US aid per capita.
Taken together, the scatterplots also suggest that early ICC joiners differed from late and nonjoiners in terms of their willingness—or ability—to provide assurance to the United States. This can be seen by comparing the slopes of the local regression curves on the left side of each quartet to those on the right and is especially visible with regard to exports. Why this is the case may not be immediately obvious, however.
Prior work views economic and security relations with the United States purely as pressure points for coercing unwilling states into signing ICC nonsurrender agreements. In contrast, my analysis sees configurations of trade and security relationships as providing possibilities for alternative pathways to assurance—whether through mingling existing legal obligations or by augmenting them with new commitments. In many instances, mechanisms that provided the United States assurance against unwanted surrender of US indictees pre-dated the need for assurance on this count. As such, they could not have been intended as purposeful “cooperation.” Nevertheless, through mingling, those commitments could be made to serve this function. Similarly, for non-Article 98 agreements relevant to ICC surrender concluded or updated after adoption of the Rome Statute, neither the United States nor its bilateral interlocutors likely viewed the ICC-relevant aspects of those arrangements as the sole reason for their adoption.
A key implication of this analysis is that opportunities for alternatives to ICC nonsurrender agreements were far from uniformly distributed. Not only are wealthier states more likely to have robust economic relations with the United States than poorer ones but they are also more likely to have (or are better positioned to seek) extradition treaties or MLATs with the United States because of these economic relationships. Conversely, being a recipient of US aid or GSP benefits—both of which indicate wealth asymmetries—is negatively associated with having an extradition treaty or an MLAT with the United States.Footnote 94 Countries in this group simply had, practically speaking, fewer options for offering legal assurance that did not involve an ICC nonsurrender agreement.
Security ties provided a narrower slate of opportunities for mingling and augmentation. One might expect that having a defensive pact with the United States would be strongly associated with having a SOFA (and thus a ready mechanism for providing assurance against unwanted ICC extraditions of US military personnel). Although chi-squared tests indicate this is true overall, the association disappears when NATO countries are excluded.Footnote 95 This makes sense if one considers that many states with which the United States has defensive alliances are in the Western Hemisphere, and are thus generally wary of having US military forces in their territories.Footnote 96 At the same time, several post-Cold-War-era SOFA arrangements with the United States were created with former Soviet Bloc states via the Partnership for Peace, which is not a defensive pact. One further source of new SOFA arrangements with the United States has been with states characterized by security unrest, which are poor candidates for defensive alliances.
It is thus not surprising that of the twenty-two states that ultimately remained in negative assurance territory at the end of 2006, the largest group (eight states) was in Latin America, where both ICC nonsurrender agreements and SOFAs were politically toxic. Other well-represented groups in the low-assurance category include five African states (Kenya, Mali, Namibia, Niger, and Tanzania), three Balkan countries (Serbia, Croatia, and Montenegro), and several small island and microstates, many of which were also de facto exempt from sanction. Among major US partners, only New Zealand retained an assurance deficit. Only three countries (Andorra, Namibia, and Niger) remained with absolutely no alternative extradition obligations toward the United States. Whether because of thin or highly asymmetric trade and security relations with the United States, or domestic political attitudes, most of these countries had fewer options for providing ICC-related assurance than did countries with deeper and more diversified relations.Footnote 97
Conclusion
Empirical scholars of international law frequently fail to consider whether and how prior legal commitments may affect incentives to join (or exit, or follow, or violate) the specific commitments they are analyzing.Footnote 98 As a result, how states and other actors use interpretative tools to prospectively configure the substantive, institutional, and procedural features of their own and others’ legal obligations to achieve—or avoid—particular outcomes has been largely overlooked. Correcting for this requires attention both to details of how law functions, and to how heresthetic power is produced and leveraged within, and across, legal systems.
I demonstrate here how there is often more than one pathway to cooperation, even for relatively esoteric types of commitments. The empirical approach—spatially mapping the direction and strength of multilateral and bilateral legal obligations in a relation to specific policy objectives—can be adapted to any issue area in which multiple reinforcing or cross-cutting legal obligations may affect how states conduct their international affairs in domains. Thus, for example, a state that refuses to sign the Nuclear Non-Proliferation Treaty may commit to other international arrangements to monitor or limit transfer of nuclear materials and components.Footnote 99 Similarly, a state that balks at adopting the International Convention on the Rights of the Child (CRC) on domestic constitutional grounds can still join other agreements geared toward countering transnational threats to the safety and well-being of children, or apply its own domestic laws extraterritorially against the sexual exploitation of minors to combat sex tourism.Footnote 100
With regard to US behavior under the ICC nonsurrender agreement initiative, the theory and empirical approach used here provides a more coherent, intuitive, and complete explanation than prior studies. Understanding whether and how international law factors into international affairs is not simply a matter of tracking discrete instances of formal rule making and patterns of commitment. Nor can its influence be captured by pointing to legal complexity as a background condition. It is equally a matter of theorizing agency and its roots in legal authority and political power, together with the legal and administrative capacity to identify where and to what degree extant legal rules can be persuasively mingled, or frameworks of legal obligation selectively augmented, to create stable pathways to desired ends.
Among governments that were unwilling to adopt ICC nonsurrender agreements on legal-normative grounds, or were unable to do so, whether out of deference to the directives of regional-organizations, or fears of domestic political backlash, several had extradition-related obligations toward the United States in place. Where these could be mingled to create a rough equivalent of a bilateral nonsurrender agreement, governments were asked to sign, but were not sanctioned for declining to do so. In cases where existing arrangements did not give the US government sufficient assurance, several states opted instead to engage in some form of strategic augmentation of their existing obligations toward the United States. However, this behavior is legible only if scholars look beyond membership in isolated agreements and look instead to the existence and operation of functionally similar agreements inside and outside established regime complexes.
To date none of these arrangements have been tested by an ICC request to surrender a US citizen. If that were to happen, it would ignite heated debate within relevant interpretative communities over whether and how the legal obligations of the state unfortunate enough to find itself in that position can, or should, be interpreted. However, to the extent this scenario remains largely beyond the pale of plausibility, this may be itself in part a function of US success in stacking the legal deck in its favor.
Efforts to design and tailor slates of mingleable legal obligations to influence others in advance of critical decision points is an underappreciated dimension of political strategy. Nevertheless, it is a critically important bridge between law creation and the emergence of formalized legal disputes—both of which have been intensively studied by IR scholars. When strategic actors succeed in preconfiguring how others understand and prioritize their legal obligations in applied terms it might prevent disputes from taking shape and even preclude situations where anticipated choices or tradeoffs may have to be made from arising at all.Footnote 101
One important bottom line is that when thinking through the consequences of widespread international commitment making (or “legalization”) we should not automatically equate complexity with uncertainty—especially without taking power differences into account. Instead scholars should focus on exploring the theoretical and empirical consequences of complexity.
Another takeaway is that theories of strategic behavior involving law that better reflect its properties as a medium for coordination and coercion complicate liberal-normative ideas about the international “rule of law.”Footnote 102 The case examined here involves using legal mingling and augmentation for the arguably illiberal purpose of insulating US citizens from ICC jurisdiction. Still, before jumping to normative judgment on this point, it is worth considering the extent to which the US intervention was both narrowly tailored and thoroughly “legalist.” Had the US government intended to seriously hobble or destroy the court, rather than simply walling itself off from its jurisdictional reach, US officials could have opted for other, much blunter political means. Although much about the ICC nonsurrender case is atypical—such as the universality, simultaneity, and nonforeseeability of the US “ask”—in this important respect it engages a common and critically important set of maneuvers by which states that find aspects of international agreements unworkable can try to address those issues without undermining cooperation altogether.
Other examples where governments mingle and augment legal arrangements in efforts to achieve important (to them) objectives that are not shared by all include preferential trade agreements (PTAs) and regional trade arrangements, which are sometimes criticized as undermining the “most favored nation” principle at the foundation of the current international trade regime.Footnote 103 Similarly, defensive alliances—commitments to come to the aid of specific states if they are attacked militarily—also create tiers of priority within broader international legal commitments to collective security.Footnote 104 In neither case are these more specialized instruments intended to undermine the general legal arrangement. To the contrary, both involve selective tailoring and deepening of cooperation beyond what may be immediately achievable on a large scale. Furthermore, this reformulated cooperation may depend heavily on (or even be motivated by) the existence of the prior, more general arrangement.
Still, pointing out the coexistence of legal arrangements alone may tell us little about whether and under what conditions those arrangements influence behavior. This results in part from the potential for different interpretations of the applicability and meaning of legal rules to confound even good-faith efforts at legal rule following—whether due to the sematic indeterminacy of individual rules, or to different approaches to prioritizing and configuring relevant rules and principles. Far from being a marginal or isolated activity, agreement mingling is a core and never-ending operation by which participants describe, and thereby discursively construct, how those obligations manifest in applied terms.
Supplementary Material
Supplementary material for this article is available at <https://doi.org/10.1017/S0020818319000377>.
Acknowledgments
I thank several institutions and individuals for feedback and suggestions. These include the International Relations Workshop at Yale University, the International Law and Politics Colloquium at Cornell Law School, the Georgetown University International Theory and Research Seminar, the International Relations Speaker Series at the Department of Politics at the University of Pennsylvania, Nisha Fazal, Julia Gray, Peter Katzenstein, Abe Newman, Erik Voeten, and my anonymous reviewers. Jeffrey Lax deserves special thanks both for substantive inputs and for extensive consultations on data presentation. I also thank Hyun Jae (Ryan) Bae, Neeti Deoras, and Cara Feldcamp for excellent research assistance. Finally, I thank Gabriela Jara, whose 2010 Columbia honors thesis prompted the initial foray into the empirical materials.
