The Institutional Design of Arms Control Agreements

The Institutional Design of Arms Control Agreements Abstract The question of international cooperation on nuclear security presents states with a conundrum. While states are wary of the potential for cheating and opportunism that would favor more legalized agreements, highly legalized agreements also create the risk of being constrained in an agreement with which others might not comply, leaving one state vulnerable. How do states balance these competing incentives? Through the study of all arms control agreements concluded, negotiated, and seriously considered since 1945, this analysis finds that at least some aspects of the legalization parameters that would make commitments more credible and sustainable through the future—in particular obligation and to a lesser extent delegation—also make bargaining over the terms of cooperation more difficult, sometimes delaying if not altogether subverting that process. These findings have implications not just for how theories of institutional design and cooperation apply to nuclear issues but also for policies related to the conduct of arms control negotiations. Introduction In 2014, Iran began working with major international powers—including the permanent five nuclear-weapon states and Germany (P5 + 1)—on a nuclear deal. The negotiations illustrated the tensions inherent in striking an arms control agreement. Some states sought an agreement where the enforcement mechanism and punishment of Iran’s noncompliance would be automatic rather than requiring an international resolution that other countries could veto. Yet, these same states aimed for an agreement that would be sufficiently binding that it would not have to be frequently renegotiated. Several months into the negotiations, one diplomat reported that “We haven’t found a mechanism that works for everyone yet” (Charbonneau 2015). The negotiations illustrate a fundamental problem for nuclear cooperation. How do states negotiate an agreement that is worth the paper it is written on but also grants the ability to hedge against future uncertainty? How does the degree of institutional legalization affect when or whether enough states participate in arms control agreements for those agreements to be concluded and entered into force? This article answers those questions with an analysis of all arms control agreements proposed, negotiated, and concluded between 1945 and the present. It first engages the institutional design literature by way of evaluating the theoretical trade-offs of legalization in the context of arms control agreements. Second, it introduces a new data set of all arms control agreements designed to ban, regulate, reduce, and limit nuclear weapons since 1945.1 Third, it tests the aspects of legalization that affect the propensity for international cooperation on arms control agreements compared to other factors. Fourth, it illustrates the key legalization measurements with reference to the Iran deal concluded in 2015. Finally, it concludes with implications, especially in reference to ongoing debates about the design of international arms control agreements. A number of scholars have investigated aspects of these questions, for example why particular arms control agreements have been designed with particular sets of legalization features (Koremenos 2001), the correlates of state participation in specific arms control agreements (Sasikumar and Way 2009), and the effect of institutional design on cooperation in other issue areas (see, inter alia, Hafner-Burton, Mansfield, and Pevehouse 2015; Von Stein 2008). However, this is the first systematic study of all agreements in the area of arms control. Balancing state sovereignty with the advantages of making credible commitments is a conundrum that scholars have appreciated in other issue areas (Simmons and Danner 2010), but nowhere are the stakes as high as in the nuclear arena, given the consequences of defection. Yet, this cost of defection could theoretically cut two ways. It could make states more eager to negotiate more credible commitments through agreements that are more highly legalized, but the prospect of becoming enmeshed in a durable cooperation arrangement can also make states wary of concluding an agreement at all (Fearon 1998, 270; Milner and Rosendorff 2001, 836), given that the agreement would have enduring consequences even in the face of changing distributions of international power, future conflicts, or shifting domestic political winds. Through the study of arms control agreements, this analysis finds that at least some aspects of the legalization parameters that would make commitments more credible and sustainable through the future—in particular obligation, and to a lesser extent delegation—also make bargaining over the terms of cooperation more difficult, sometimes delaying if not altogether terminating that process. These findings have implications not just for how theories of institutional design and cooperation apply to nuclear issues but also for policies related to the conduct of arms control negotiations. Institutional Design and Arms Control During the 1990s, international relations scholars debated at length the question of whether institutions “matter” (Mearsheimer 1994/95; Keohane and Martin 1995). Alexander Wendt (2001, 1019) observed that the focus on whether institutions matter had “left IR (international relations) with less to say” about the types of institutions that governments create and why. More than a decade later, the literature has filled that void, with scholars identifying the dimensions along which institutions vary in their degree of legalization—operationalized along the dimensions of obligation, precision, and delegation (Abbott et al. 2000; Koremenos, Lipson, and Snidal 2001)—but also raising new sources of debate about how these dimensions affect the likelihood that states will reach a cooperative outcome. Briefly, obligation refers to the degree to which an arrangement is legally binding, for example whether there are escape, withdrawal, or sunset clauses that provide mechanisms for withdrawal should circumstances change in the future (Milner and Rosendorff 2001), or reservations, declarations, and interpretive statements that exempt states from particular aspects of an agreement (Helfer 2012, 179). Precision refers to the specificity or ambiguity of rules outlined in an agreement. Delegation is the degree to which states grant authority to an agent that can help enforce an agreement, whether through diplomacy or an entity such as a court (Hawkins et al. 2006, 7; Abbott et al. 2000, 18). High or hard legalization offers less latitude in terms of how states interpret and implement international legal commitments (with low legalization or soft law as the opposite). Legalization introduces inherent trade-offs between the advantages of a firm, credible commitment and the disadvantages of being locked into an agreement in the event that circumstances change in the future (Goldstein et al. 2001). On the one hand, legalization can be attractive for coping with future uncertainty (Helfer 2012, 177; Koremenos 2001, 289). Lower-obligation, non-binding agreements give states an off-ramp if circumstances change in the future (Abbott and Snidal 2000, 446). States might also be drawn to lower degrees of precision to avoid “undesirable rigidity” in terms of how they interpret an agreement, or to agreements that do not delegate responsibility for enforcement lest doing so introduce higher sovereignty costs (Abbott and Snidal 2000, 433, 435). Moreover, lower degrees of legalization need not come at the expense of institutional effectiveness. As Lipson suggests (1991, 500), insofar as it can “commit states to characteristic forms of discourse and procedure, soft law provides a way of achieving compromise over time.” If low legalization makes it easier for states to come to an agreement, why do states ever conclude more highly legalized agreements? The reason is that they also value the credibility of a commitment. Martin (2000, 3) argues that “the forms of international cooperation that offer states the highest benefits require them to make credible commitments to one another.” Several different attributes of an agreement can make those commitments more credible. Harder law in the form of binding commitments creates higher reputational costs for reneging (Guzman and Meyer 2010; Shaffer and Pollack 2011, 1162). Precision also helps limit opportunism since it limits “self-serving auto-interpretation” of an agreement (Abbott and Snidal 2000, 427) and, as Simmons (2009, 121) puts it, “reduces the scope for plausible deniability of violation.” Delegation helps identify and punish violators in ways that guard against defection and bring state behavior back into compliance (Koremenos 2007). Assessing Legalization in Arms Control Agreements While a number of scholars have examined legalization in the context of the environment (Von Stein 2008) and trade (Milner and Rosendorff 2001; Kucik and Reinhardt 2008, 478), the nature of coordination trade-offs in those issue areas may be qualitatively different than in a security issue such as arms control, which presents states with complicated payoff structures that require balancing the rewards of mutual restraint with the potential risks if one side cooperates and the other defects (Axelrod and Keohane 1985, 231–32). On the one hand, the logic of the spiral model should “provide the intellectual basis for arms control” (Kydd 2000, 228), as it implies that the accrual of nuclear weapons makes states worse off by inducing fear that can occasion either preemptive war or a nuclear accident (Jervis 1978). By implication, effective arms control institutions would entail credible commitments, given the risks of unilateral disarmament. Reagan had said that “he would gladly accept nuclear disarmament if he was sure he could get the genuine article” (quoted in Jervis 1990, 3). One way to do that is to make those commitments binding, with states seeking a more formal commitment to try to militate against opportunistic behavior and reneging. The more binding the contract (higher obligation), the higher the reputational costs for reneging and the more credible the commitment (Mitchell and Hensel 2007, 724). Jervis suggests (1978, 174) that in the absence of such binding obligations, defection will become a “self-fulfilling prophecy in which each person rushes to defect because he fears others are going to.” Similarly, credibility concerns could explain the observation that “arms control agreements display increasing precision” (Goldstein et al. 2001, 2), with states trying to combat cheating or opportunism that could arise from ambiguity in the language of the agreement. As Oye (1985, 17) observes, “the very act of clarifying standards of conduct, of defining cooperative and uncooperative behavior, can permit more effective resort to strategies of reciprocity” that are central to principles of arms control. Many agreements, especially the bilateral agreements between the United States and Russia cite specific numbers of weapon cuts to an arsenal, and others, such as the 1963 Treaty Banning Nuclear Weapons Tests in the Atmosphere, in Outer Space, and Underwater, state an unequivocal, sweeping prohibition. Such concerns about credible commitments might also mean states favor a robust verification and monitoring mechanism as a way to identify and punish noncompliance. As Jervis (1990, 59) suggested, “the state not only wants to know what its adversary is doing, but under many conditions also wants its adversary to know what it is doing.” Similarly, Barrett (1998, 322) argues that “countries will only agree to do something collectively if that something can be monitored,” especially under conditions of uncertainty where states value additional checks on others’ commitments (Koremenos 2007, 193–94). This could mean providing legal authority to conduct on-site inspections or use national technical means to identify whether state parties to an agreement are in compliance. If these arguments about credible commitments are operative in the context of nuclear weapons, the observable implication is that less highly legalized arrangements would be unusual or even nonexistent in the arms control arena (Raustiala 2005, 592–93). Legalization as an Obstacle to Cooperation Another interpretation of the paradox described above—in which states might value some form of arms control but be concerned with whether others’ commitments are credible and reciprocal—is that states might be wary of committing to highly legalized commitments on arms control in order to hedge against an uncertain future. As Fearon (1998, 270) suggests, the prospects of cooperation over the long term can work at cross-purposes with actually making that agreement happen because it can “give states an incentive to bargain harder, delaying agreement in hopes of getting a better deal.” Based on this formulation, cooperation is a two-phase process in which the longer-term enforcement of the agreement affects states’ initial approach to negotiation. Whereas the implications for Fearon are that an agreement is delayed or even altogether scuttled, another possibility links the prospects for long-term cooperation with the shorter-term prospects of reaching an agreement. States might ease some of the legalization features that foster longer-term cooperation but obstruct the shorter-term agreement in order to accommodate the source of state concerns about future uncertainty. As Milner and Rosendorff (2001, 833) suggest in the context of trade, “uncertainty about the state of the world” can give states incentives to craft agreements that allow them to hedge, for example by including escape clauses. Some of the uncertainties that arise in the context of arms control are not dissimilar to those that would arise in other issue areas. As Nelson (2013) notes, changes in domestic leadership, concerns about the potential for future conflict, and changes to the distribution of power can all create uncertainties and hesitation about binding states to future commitments. States might deem it appropriate to limit nuclear weapons at one moment, but not want to tie their hands for a time in the future when circumstances have changed (Koremenos 2001). Beyond the question of uncertainty that could affect how states regard highly legalized agreements across issue areas, there are other arms control–specific factors that link the enforcement phase with how states approach the initial decision about how to craft an agreement. First, as Fearon (1998, 282) suggests, “when states care a lot about future payoffs, the expected long-run benefits of getting the better deal are very large.” Nuclear strategy is an area where states care about future payoffs insofar as defection involves “potentially grave losses to a player who attempts to cooperate without reciprocation” (Lipson 1984, 14). States care deeply about minimizing the costs of defection, where in the absence of a sovereign, one state could defect while another disarms (Oye 1985, 13), leaving the disarmed state vulnerable. Second, the attempts to create something resembling a sovereign through robust verification and enforcement mechanisms create their own dilemmas. Under the NPT, non-nuclear-weapon states that are party to the treaty are permitted to produce fissile material for peaceful purposes (Article IV), but many of the processes used in the service of civilian nuclear power can be channeled toward military ends, which makes it difficult to verify whether facilities are being employed in the service of peaceful rather than military purposes (Fuhrmann 2012). On-site inspections can help verify, but the IAEA is limited in its capacity to monitor each nuclear site. Even for the sites with NPT safeguards, such as Iran, the IAEA had reported that “it continues to have serious concerns regarding possible military dimensions to Iran’s nuclear program” (Dahl 2012). As the Congressional Research Service concludes, “a determined cheater could probably find a way to conceal some types of violations” (Woolf, Nikitin, and Kerr 2013, 9).2 Third, while states might prefer to have more information about the nuclear activities of other states (Jervis 1990, 59), delegation comes at a cost in terms of sovereignty. Indeed, what makes an inspection regime capable of uncovering violations are the intrusive reporting and inspection standards that are “clearly negatives from the point of view” of the target state (Braun and Chyba 2004, 31). Historical evidence suggests that on-site inspections were a sticking point in arms control negotiations such as the LTBT in 1963, with the United States favoring inspections as a way to corroborate Soviet compliance and the Soviets viewing inspections as a violation of sovereignty (MacDonald 2015).3 Thus, not only is delegation costly in terms of sovereignty, the advantages may be limited, since states may be able to elude detection. Even the most intrusive verification regime, consisting of on-site inspections, has been seen as “an encroachment on national sovereignty” (Blix 2001), while still not necessarily producing incontrovertible evidence about proliferation. As the preceding discussion suggests, states face a dilemma in fashioning international agreements. While states are wary of the potential for cheating and opportunism, which would favor more legalized agreements, highly legalized agreements also create the risk of constraining states in an agreement with which others might not comply, leaving one state vulnerable. Taken together, this suggests that higher legalization might not necessarily be a prerequisite of cooperation but actually make it less likely, with certain aspects of legalization more vexing than others. When it comes to the uncertainties associated with committing to a long-term agreement and the potential consequences of other states reneging, I argue that states are therefore likely to favor arms control agreements that incorporate flexibility mechanisms. The most straightforward manifestation is through lower obligation agreements, for example those that are non-binding or allow for declarations and reservations, duration provisions, amendment procedures, and withdrawal and termination clauses (Helfer 2012; Milner and Rosendorff 2001). This discussion about uncertainty in the context of arms control produces the first hypothesis: H1: High-obligation agreements should make cooperation on arms control less likely. Although scholars tend to treat flexibility mechanisms primarily as ways to sidestep, withdraw, or otherwise give “wiggle room” to states as they comply with an agreement over time (Pelc 2013), degrees of precision may be a conceptual cousin of these tools. If so, aspirational language in the form of a less precise agreement could provide an analogous way for states to hedge against future uncertainty, as it can give states latitude to interpret behavior in ways that are consistent with an agreement even if the compliant nature of those actions becomes a stretch over time (Economist 2014). If this is the case, then we would see support for the second hypothesis: H2: High-precision agreements should make cooperation on arms control less likely. Finally, states may also find the degree of delegation reassuring in one sense but problematic in another, since the types of delegation that could help verify states’ compliance with an agreement also introduce costs to states’ sovereignty while not being failsafe. Thus, we might also expect to see evidence for the third hypothesis: H3: High-delegation agreements should make cooperation on arms control less likely. The next section turns to the empirical analysis of how the degree and attributes of legalization influence the likelihood of cooperation and under what circumstances. Research Design To address this question of how different forms of legalization affect international cooperation on arms control agreements, I compiled a data set of all nuclear arms control measures negotiated between 1945 and 2010.4 The data set is drawn from sources such as the Nuclear Threat Initiative, the Federation of American Scientists, and Goldblat’s (1994). Since the main focus is on how the degrees of legalization affects the propensity for cooperation, the data set includes both formal treaties and informal agreements. In addition, it includes both successful agreements and proposed agreements that are never concluded or enter into force, so that the sample does not select for a subset of agreements that are better suited to a cooperative outcome. Indeed, one-fifth of the proposed agreements have never been concluded or entered into force, suggesting that the sample is not systematically excluding untenable arms control ideas. While it might be possible, in principle, to go even further toward addressing the potential for selection bias by trying to include arms control norms or ideas proposed behind closed doors that do not ever reach a formal proposal, this approach would raise different concerns, which is whether these ideas come to light in a systematic, unbiased way.5 Thus, conditional on proposing an agreement, the analysis evaluates why some are successful—and sometimes quickly—and others are not.6 Based on these parameters, a formal agreement enters the data set when a state representative, as defined by the 1969 Vienna Convention on the Law of the Treaties (Article 7), publicly proposes a particular arms control problem be addressed.7 Often, such heads of state or ministers of foreign affairs have heralded proposals in international forums such as the UN General Assembly, where President Clinton, for example, called for fissile material cutoff negotiations in a September 1993 address. In that same session of the UNGA, the UNGA passed Resolution 48/75L endorsing a “prohibition of the production of fissile material for nuclear weapons or other nuclear explosive devices” (Kimball 2013). For informal agreements that tend not to arise through formal UNGA resolutions, I turn to high-level officials’ publicly endorsing a particular set of arms control principles alongside bilateral or multilateral negotiations. For example, while the United States was initially concerned with the transfer of missiles in the 1970s, concrete efforts to curb missile proliferation were restricted to internal US debates about how to control the transfer of missile technology. In 1982, the United States issued a new policy called “Nuclear Capable Missile Technology Transfer Policy” that outlined the objectives of limiting missile exports and concurrently began negotiations on the Missile Technology Control Regime (MTCR) with West Germany, Italy, and France that same year (Mistry 2005, 46). Thus, the MTCR enters the data set with this proposal and the beginning of negotiations in 1982. Dependent Variable The outcome of interest for this analysis, international cooperation in arms control, is captured in two ways: first, in terms of the duration required to achieve cooperation, and second, more simply by whether states were able to achieve cooperation at all, each discussed in turn. Theoretical and policy reasons suggest some virtue in a relatively expedient negotiation process. Indeed, international organizations often act as though expediency is important. They frequently pass resolutions or statements urging rapid progress on an agreement—for example a 2013 UNGA resolution urging the “speedy establishment of a nuclear-weapon-free zone in the Middle East—intimating a belief that the successful conclusion and entry into force of an agreement have an important effect on state behavior.8 Moreover, scholars have shown that delay in reaching an agreement is costly. As Fearon (1998, 277) concludes, “delay means more time spent without the benefits an agreement would bring; second, as time passes there may be some growing risk that one side will break off negotiations entirely.” Within the context of duration, I evaluate two different outcomes. One is from the time of proposal to the successful conclusion of an agreement. According to Article 9 of the 1969 Vienna Convention, this formally occurs when participating states adopt the text of the treaty, an important mark of diplomatic progress that would theoretically create reputational costs for states that renege (Guzman 2002). Another way to measure duration is from proposal to entry into force (EIF). The Vienna Convention on the Law of Treaties specifies that the agreement enters into force whether as a function of the negotiating states’ consent to comply with the agreement or by a date specified by the treaty. To the extent that reputational loss is higher once the agreement has entered into force (EIF), it is a more consequential marker of cooperation (Schneider and Urpelainen 2013, 14; Haftel 2010). To summarize, in terms of using duration to international cooperation as the dependent variable, I consider both conclusion of an agreement and EIF. For the evaluation of duration as the dependent variable, I use survival analysis to assess the likelihood of state participation with the agreement-year as the unit of analysis. Whereas epidemiologists use survival analysis to assess the efficacy of a treatment (e.g., the recurrence of a disease), political scientists have applied it for evaluating the effect of one variable on the duration of a particular outcome, for example whether outside mediation brings a more rapid end to interstate disputes (e.g., Regan and Stam 2000). Here, I use survival analysis, specifically a Cox proportional hazard model,9 in the context of nuclear agreements to estimate the way legalization impacts the duration required to reach an agreement and, where applicable, to go into effect. This measures the “risk” or likelihood of an event taking place over time, which helps explain variation in the rate at which a particular outcome—the conclusion or EIF of an agreement—takes place (Haftel and Thompson 2013b, 370). A second dependent variable that captures cooperation on arms control agreements is whether the agreement is concluded (and ratified) or not, a dichotomous measure. Perhaps international organizations are misguided in pushing for speedy conclusion of agreements and more effective, durable agreements take longer because countries have to settle their differences in an effective manner (Haftel 2010, 351–52), in which case duration may not be as important as whether the agreement is successfully reached and enters into effect. I therefore employ an alternative measure of international cooperation, which is whether the agreement is concluded and enters into force, regardless of duration, and then use probit analysis to evaluate the reasons why agreements succeed or fail, with the agreement-year as the unit of analysis. In sum, I include two separate measures for the dependent variable. The first measures international cooperation as a function of state participation over time—with conclusion of an agreement and EIF as two different ways of capturing this outcome. The second is whether an agreement is concluded and enters into force successfully at all. Both sets of dependent variables speak to the ways that state participation culminates in an international agreement. Some observers might respond that certain agreements are crafted precisely to screen out states that are not relevant to or expected to comply with the agreement. For example, in most arms control negotiations between the United States and Russia, other countries were explicitly excluded from the negotiations by design and objective. The dependent variable focusing on international cooperation based on reaching a threshold of state participation is not meant to imply maximization but rather achieving the particular levels required by the agreement. Independent Variable: Legalization I measure the main independent variable, degree of legalization, based on the obligation, precision, and delegation typology discussed earlier. As Abbott et al. (2000, 17) suggest, each of these dimensions is a continuum, which the authors illustrate in reference to a variety of agreements. How these dimensions are operationalized in the context of arms control has not been outlined in the existing literature, but is important for the current study that investigates how the design of institutions affects the speed and likelihood of international cooperation in arms control agreements. As a first step, then, Table 1 outlines the categories of legalization and indicators in the specific context of arms control; while the obligation dimension is generalizable beyond arms control, the other two require more arms control–specific indicators cited in the table and elaborated upon below.10 Table 1. Indicators of legalization along the dimensions of obligation, precision, and delegation Legalization dimension  Indicator of legalization  Obligation  Political treaty rather than agreement, unconditional obligation and intent to be legally bound  No reservations, understandings, declarations permitted  No temporal modifications (e.g., renegotiations, escape, withdrawal, or sunset clauses)  Precision  Specific types of nonproliferation behavior or progress toward nonproliferation  Specific arsenal numbers, thresholds, delivery vehicles, military capabilities cited  Delegation  Reporting of compliance through regular meetings or information exchanges  Monitoring through states’ National Technical Means  Verification through on-site inspections  Legalization dimension  Indicator of legalization  Obligation  Political treaty rather than agreement, unconditional obligation and intent to be legally bound  No reservations, understandings, declarations permitted  No temporal modifications (e.g., renegotiations, escape, withdrawal, or sunset clauses)  Precision  Specific types of nonproliferation behavior or progress toward nonproliferation  Specific arsenal numbers, thresholds, delivery vehicles, military capabilities cited  Delegation  Reporting of compliance through regular meetings or information exchanges  Monitoring through states’ National Technical Means  Verification through on-site inspections  View Large In the context of a nonproliferation agreement, obligation deals first with whether it is legally binding. For example, the Nuclear Suppliers Group (NSG) was assembled to regulate the transfer of nuclear-related materials with an eye to stemming proliferation. As Belcher (2014) notes, the NSG was based on a “gentleman’s agreement,” with members taking a pledge rather than making a formal contract. By contrast, the Comprehensive Test Ban Treaty (CTBT) and the Negative Security Assurances agreement were both—were they to have entered into force—legally binding. A second indicator of obligation is whether the agreement allows for reservations, understandings, declarations. The CTBT explicitly does not allow for such treaty actions, whereas the Treaty for the Prohibition of Nuclear Weapons in Latin America and the Caribbean permitted countries such as the United States to issue specific understandings and declarations that “adherence to Protocol II will not prevent collaboration by the US with Contracting Parties for the purpose of carrying out explosions of nuclear devices.”11 A last indicator of obligation is whether there is temporal conditionality that allows states to revise their commitments in the future. As discussed earlier, escape or renegotiation clauses give states latitude over time in terms of their commitment to a treaty. In the case of the nonproliferation treaty (NPT), negotiating states included a built-in expiration date after which the treaty would be reopened for discussion, a flexible trial period that allowed states to overcome uncertainty about the future and come to an agreement (Koremenos 2001). An agreement is coded as high precision if the language is categorical and unambiguous, giving little room for interpretation. The most straightforward indicator is an arms reduction treaty that specifies the number or type of weapons allowed in a country’s arsenal. Whereas the New Start treaty specified the number of deployed missiles, warheads, and launchers that the United States and Russia could possess,12 the Prevention of Dangerous Military Activities Agreement proscribes damage to another party, while being vague about the technical thresholds or definitions of such damage (ILM 1989). Some agreements, such as the NPT, vary in the degree of precision within the agreement (Abbott and Snidal 2000, 450). Article I prohibits nuclear-weapon states from the “transfer to any recipient whatsoever of nuclear weapons or other nuclear explosive devices,” which is quite precise and categorical, whereas Article VI requires that nuclear-weapon states “pursue negotiations in good faith on effective measures relating to cessation of the nuclear arms race at an early date.” While there may be heterogeneity within an agreement, the coding of precision used here relies on the highest level of precision within an agreement. Another indicator of precision relates to compliance behavior, specifically whether it spells out timelines for compliance or types of proliferation behavior. For example, the START I treaty notes the specific time frame for accomplishing each phase of the treaty (e.g., no later than 36 months after EIF for the first phase).13 A contrasting example is the Accident Measures Agreement, whose Article 1 requires that “Each Party undertakes to maintain and to improve, as it deems necessary, its existing organizational and technical arrangements to guard against the accidental or unauthorized use of nuclear weapons under its control.” Finally, whether an agreement is legalized along the dimension of delegation requires some explanation before turning to the indicators. High delegation in the context of a multilateral treaty would mean “vesting decision-making authority in a third party institution with strong adjudicative capacity or independent administrative power” (Abbott and Snidal 2000, 424). Such a definition may be appropriate in the context of institutions such as the World Trade Organization, but would not capture what it means to delegate authority in the context of bilateral arms control agreements where creating a third-party enforcement and monitoring organization is unusual. Russia and the United States often negotiated questions of verification and monitoring on a bilateral basis, with responsibility for adjudication vested in the other country, for example through national technical means or inspections (Woolf 2011, 8–10). In this context, delegation would mean “the creation of monitoring and enforcement mechanisms” that increase the sovereignty costs of the agreement by granting authority for implementation to third parties or other states in the agreement (Hafner-Burton, Mansfield, and Pevehouse 2015, 7). In the context of arms control, this could first refer to whether the agreement stipulates non-voluntary reporting on armaments (and locations) related to the agreement, whether through regular meetings of state parties or information exchanges (Vaynman 2014), such as the extensive data exchanges stipulated by Article IX of the INF treaty. A second indicator of delegation deals with monitoring—for example through technology—that is carried out by state parties to the agreement. START I has a verification regime that includes satellites to collect data and detect violations (Woolf, Nikitin, and Kerr 2013, 9). The third includes verification through on-site inspections (Brown 2010, 152). The LTBT sidestepped such inspections of facilities for the reasons mentioned earlier, whereas START I had an intrusive regime of on-site inspections.14 As these examples suggest, the same agreement can exhibit strong legalization on some dimensions and not others. Rather than aggregate into one index of legalization, since each dimension has different set of theoretical expectations than others, I consider each separately. Indeed, this also helps produce more useful policy implications, speaking to design features that increase the likelihood of conclusion and EIF. Each is associated with a legalization score calculated by summing the number of indicators present for each dimension; for example, an agreement that exhibits all indicators of obligation would be scored a 3 for that dimension. The Appendix shows the coding of each institutional design variable. Beyond the core legalization measures, I also control for the group size or degree of multilateralism involved in the coordination. Larger group sizes could create a collective action problem that complicates coordination by creating uncertainty about cheating and enforcement (Axelrod and Keohane 1985; Oye 1985). Controlling for group size also addresses concerns about the comparability between bilateral and multilateral treaties, where a bilateral treaty enters into force as soon as the two countries have ratified while a multilateral treaty has different thresholds for EIF, potentially creating higher hurdles for the latter. Keohane (1990, 731) observes that “multilateralism can be defined as the practice of coordinating national policies in groups of three or more states, through ad hoc arrangements or by means of institutions.” Multilateralism is therefore a dummy variable that takes on a 1 for arrangements such as a Nuclear Weapon Free Zone or the NPT, and a 0 for bilateral agreements, such as the Non-Attack Agreement between India and Pakistan. I account for whether great powers—specifically the five nuclear weapon states designated by the NPT—propose or draft an arms control proposal, as their involvement might be more likely to yield a successful outcome (Kroenig 2014). For example, the United States used bargaining leverage in the form of loans and guarantees to pressure South Korea to ratify the NPT in 1974 (Tzeng 2014, 476–77). The approach of the NPT, which was drafted partly by the United States, contrasts with that of the Nuclear Weapons Convention, which has been shunned by the permanent five other than China. Although the agreements all address arms control, I control for the possibility that heterogeneous goals could affect the speed or success with which an agreement is concluded or goes into effect. As Downs, Rocke, and Barsoom point out (1996, 389), many arms control agreements were designed to “maintain the status quo trajectory rather than to alter it significantly,” and did not require that participating states deviate from their current behavior, thereby making it an easier path toward conclusion, EIF, and ultimately compliance. I include a dummy variable depth for cooperation that requires deviation from the status quo, compared to agreements that did not alter a state’s behavior. For example, agreements such as the Hot Line agreement did “not regulate an arms output such as the number and/or location of a weapons system” (Downs, Rocke, and Barsoom 1996, 388–89), whereas agreements such as the intermediate-range nuclear forces (INF) and bilateral treaties such as START required the elimination or reduction of existing forces (390). I also control for what Copelovitch and Putnam (2014, 472) refer to as “institutional context,” which refers to “the presence or absence of prior agreements between prospective partners in ‘new’ cooperation.” Context for nuclear agreements means that some states will have successfully engaged in prior negotiations, which should create a longer shadow of the future and make subsequent cooperation more likely. I code prior design successes as a count variable based on the number of prior agreements involving the same parties. For example, the Threshold Test Ban Treaty is coded as a 3 because it built on the institutional context of the Baruch Plan, the formation of the IAEA, and the Limited Test Ban Treaty. I also include a dummy variable, successor, for agreements that were negotiated as follow-on agreements to the earlier agreement; for example, SALT II and START II were successors to the earlier, related agreement, which could have made them more likely because they had clear diplomatic antecedents.15 Finally, I account for circumstances that could affect both interest in pursuing a particular arms control objective in the first place and the determination to conclude and facilitate EIF quickly. As Kelley and Pevehouse (2015) note, the consent process underlying treaty ratification requires time and political capital. It may be that leaders’ willingness to expend time and capital to generate a fruitful outcome to negotiations and EIF depends in part on the security context. Some accounts suggest that the introduction of thermonuclear weapons provided the impetus for proposing the LTBT (Bohlen 2009), and that the close call of the Cuban Missile Crisis further helped parties reach an agreement. To address the potential effect of the Cold War on incentives to diffuse tensions through arms control, I include a Cold War dummy that takes on a 1 until 1991, the year of Soviet dissolution, including its associated nuclear weapons. Results Figure 1 tracks the number of agreements based on conclusion and EIF over time. As the figure shows, events characterized by whether an agreement was concluded are concentrated more up front whereas the frequency with which those agreements come into force are scattered across more decades. As the descriptive statistics show,16 the mean time to conclude an agreement is 9.02 years, compared to 10.85 for EIF. Figure 1. View largeDownload slide Frequency of nonproliferation agreements concluded and entered into force as a function of time (years) Figure 1. View largeDownload slide Frequency of nonproliferation agreements concluded and entered into force as a function of time (years) In terms of the overall success of proposed agreements, of the 48 arms control agreements that have been proposed, a total of nine were either never concluded or did not enter into force (see Table 2). Six, such as the Baruch Plan and Antisatellite Agreement, were neither concluded nor ratified. Three, including SALT II and the Comprehensive Test Ban Treaty, never entered into force. The fact that negotiations do fail at fairly frequent rates suggests that the sample is not biased in favor only of arms control objectives that states expected to succeed. Table 2. Cross-tabulations of proposals that were concluded and entered into force Agreement concluded  Entered into force  Yes  No  Total  Yes  39  3  42  No  N/A  6  6  Total  39  9  48  Agreement concluded  Entered into force  Yes  No  Total  Yes  39  3  42  No  N/A  6  6  Total  39  9  48  View Large Having offered a broad overview of arms control agreement behavior, the analysis first turns to a discussion of the dependent variable that captured duration. Table 3 shows the results from two models: Models 1 and 2 estimate how the degree of legalization influences the likelihood that an agreement will be concluded (signed), without and with controls, respectively, while Models 3 and 4 consider outcomes based on the agreement entering into force (without and with controls). The table reports hazard ratios, which refer to the risk at which an agreement will be concluded or enter into force during the time period. It is interpreted relative to 1, such that ratios below 1 mean a lower likelihood of an agreement occurring in a given time period (and longer time required) and ratios greater than 1 mean a higher likelihood of an agreement occurring under those conditions (decreasing the amount of time required). Table 3. Cox proportional hazard models for treaties concluded and entering into force   DV: Concluded  DV: EIF    Model 1 (no controls)  Model 2 (with controls)  Model 3 (no controls)  Model 4 (with controls)  Legalization           Obligation  .08 (.03)***  .08 (.03)***  .13 (.04)***  .13 (.04)***   Precision  1.15 (.34)  1.07 (.36)  1.14 (.40)  1.2 (.46)   Delegation  .98 (.18)  .79 (.17)  .17 (.06)  .47 (.12)**  Multilateralism    2.35 (.98)**    1.19 (.82)**  Cold War    .47 (.23)    .37 (.18)**  Great Power    2.11 (1.47)    3.22 (2.17)  Depth    1.59 (.82)    1.6 (.80)  Context    .93 (.04)    .88 (.04)**  Successor    23.99 (15.80)***    2.92 (3.29)  N  389  389  482  482  Time “at risk”  35-0  3509  5232  5232  Log-likelihood  –114.28  –104.08  –118.55  –113.03  p > x2  0.00  0.00  0.00  0.00    DV: Concluded  DV: EIF    Model 1 (no controls)  Model 2 (with controls)  Model 3 (no controls)  Model 4 (with controls)  Legalization           Obligation  .08 (.03)***  .08 (.03)***  .13 (.04)***  .13 (.04)***   Precision  1.15 (.34)  1.07 (.36)  1.14 (.40)  1.2 (.46)   Delegation  .98 (.18)  .79 (.17)  .17 (.06)  .47 (.12)**  Multilateralism    2.35 (.98)**    1.19 (.82)**  Cold War    .47 (.23)    .37 (.18)**  Great Power    2.11 (1.47)    3.22 (2.17)  Depth    1.59 (.82)    1.6 (.80)  Context    .93 (.04)    .88 (.04)**  Successor    23.99 (15.80)***    2.92 (3.29)  N  389  389  482  482  Time “at risk”  35-0  3509  5232  5232  Log-likelihood  –114.28  –104.08  –118.55  –113.03  p > x2  0.00  0.00  0.00  0.00  Note: Dependent variable is whether an agreement was reached (models 1, 2) or entered into force (model 3, 4). ***Significant at the 1 percent level. **Significant at the 5 percent level. *Significant at the 10 percent level. Note: Hazard ratios > 1 are interpreted as lower risk of achieving agreement, hazard ratios < 1 means higher risk of achieving agreement. View Large As Table 3 suggests, not all features of legalization are created equal in terms of affecting whether the agreement will be concluded and entered into force in a timely way. For example, agreements that are marked by the highest obligation decrease the likelihood of reaching an agreement or that agreement going into effect in a particular year by 92 and 87 percent, respectively, offering strong support for H1. The analysis provides more qualified support for H3. High-delegation agreements also impacted the prospect of cooperation, though in terms of delaying the agreement going into effect rather than reaching an agreement altogether, with agreements having the highest degree of delegation delaying the prospect of EIF in a given year by 53 percent. The hypothesis related to precision (H2), on the other hand, found no support, as it has no meaningful impact on the prospects for cooperation. As suggested above, some qualitative evidence points to cases where less precision in an agreement has contributed to its successful conclusion of an agreement, but preferences regarding precision do not appear to be a factor that systematically affects cooperation outcomes. Beyond the main legalization variables, the analysis shows that several of the control variables affected outcomes. Several factors made it more likely that an agreement would be reached or entered into force during a particular period, including whether the agreement was multilateral. This finding is surprising given the theoretical expectations that coordination among more parties should raise more potential differences to reconcile. Yet, in the case of arms control agreements, the reason that multilateralism facilitates the speed of conclusion and EIF may be that many multilateral agreements are actually non-binding agreements like the Proliferation Security Initiative that are negotiated quickly. Indeed, interacting obligation and multilateralism produces a hazard ratio of 5.27 that is significant at the 5 percent level, suggesting that multilateralism in its association with informal agreements may be driving the finding rather than multilateralism per se. Whether the agreement is a successor to an earlier set of negotiations made it more likely that the agreement would be concluded, which is not surprising since the previous agreement likely established the groundwork for subsequent negotiations. For example, START I took the longest to negotiate compared to its successors. Conversely, institutional context slowed the prospect of EIF, perhaps because whereas the successor agreement is more likely to capture differences in degree, context is more likely to capture differences in kind, where prior negotiations mean that some of the easier cooperation problems were already solved and that the more difficult ones were shelved until a subsequent time. The Cold War also made EIF during a given period less likely, which is consistent with a number of agreements, such as the Conventional Forces in Europe agreement, whose EIF stalled until after the Cold War. Turning to the factors that affect whether agreements are successful, as opposed to the timing of cooperation, the results look quite compatible with the duration analysis. As Table 4 shows, obligation is the most significant attribute and the coefficient is negative; the more binding the agreement, the less likely it is to be concluded or entered into force. Thus, irrespective of how the dependent variable is measured—by duration or by whether the agreement is ultimately successful in its conclusion and entry into force—obligation is revealed to be highly significant. Similarly, as with the duration models, delegation is significant for EIF but not whether it is concluded, suggesting that higher-delegation agreements can ultimately be a barrier to success. Table 4. Probit models for treaties concluded and entering into force   DV: Concluded  DV: EIF    Model 1 (no controls)  Model 2 (with controls)  Model 3 (no controls)  Model 4 (with controls)  Legalization           Obligation  –-.81 (.16)***  –.87 (.19)***  –.60 (.13)***  –.57 (.15)***   Precision  .03 (.87)  –.04 (.19)  .02 (.18)  .13 (.21)   Delegation  .00 (.10)  –.04 (.12)  –.19 (.10)*  –.29 (.13)**  Multilateralism    .32 (.25)    .35 (.26)  Cold War    –.68 (.29)**    –.72 (.26)**  Great Power    .67 (.43)    .78 (.39)**  Depth    –.03 (.29)    .02 (.26)  Context    –.05 (.03)*    –.07 (.03)**  Successor    .92 (.41)**    –.09 (.51)  N  389  389  482  482  Log-likelihood  –104.25  –98.30  –109.41  –104.12  Constant  –.43 (.20)  –.39 (.39)  –.41 (.20)  .35 (.38)    DV: Concluded  DV: EIF    Model 1 (no controls)  Model 2 (with controls)  Model 3 (no controls)  Model 4 (with controls)  Legalization           Obligation  –-.81 (.16)***  –.87 (.19)***  –.60 (.13)***  –.57 (.15)***   Precision  .03 (.87)  –.04 (.19)  .02 (.18)  .13 (.21)   Delegation  .00 (.10)  –.04 (.12)  –.19 (.10)*  –.29 (.13)**  Multilateralism    .32 (.25)    .35 (.26)  Cold War    –.68 (.29)**    –.72 (.26)**  Great Power    .67 (.43)    .78 (.39)**  Depth    –.03 (.29)    .02 (.26)  Context    –.05 (.03)*    –.07 (.03)**  Successor    .92 (.41)**    –.09 (.51)  N  389  389  482  482  Log-likelihood  –104.25  –98.30  –109.41  –104.12  Constant  –.43 (.20)  –.39 (.39)  –.41 (.20)  .35 (.38)  Note: Dependent variable is whether an agreement was reached (models 1, 2) or entered into force (models 2, 3). ***Significant at the 1 percent level. **Significant at the 5 percent level. *Significant at the 10 percent level. View Large Precision is again not significant in either model, raising the question of why. As Wallace (2014, 114) notes, “precision is frequently relegated to the least consequential component of legalization.” Scholars such as Abbott et al. (2000, 22) code an agreement that is high precision but low obligation and delegation agreement as soft legalization, and vice versa, with a high obligation and delegation agreement that is low precision as harder legalization. One reason for privileging obligation and delegation is that these both impose higher costs. Obligation creates potential costs in the form of reducing decision-making autonomy should circumstances change, and delegation imposes sovereignty costs for the most intrusive inspections. Precision is a matter of interpretation and is not as directly associated with a set of costs as the other two dimensions. In short, the findings discussed here lend further support to the rationale for relegating precision to the least consequential component of legalization. Beyond the key legalization variables, it is noteworthy that again, the Cold War negatively impacts whether an agreement was concluded or entered into force, which is broadly consistent with the earlier finding. Similarly, as institutional context increased the speed with which an agreement enters into force, it also made the likelihood of reaching an agreement and that agreement entering into force less likely. As in the duration analysis, the presence of successor agreement made conclusion of a subsequent agreement more likely, even if not affecting EIF. Finally, the presence of a great power sponsoring the agreement made ultimate EIF more likely, suggesting that even if great power sponsorship does not facilitate other aspects of arms control agreements, on the most basic measure of whether it eventually facilitates EIF, irrespective of time, it has a positive impact. To investigate the robustness of these results, I carried out a number of checks. To cope with the potential subjectivity associated with hand-coding the arms control agreements based on legalization (Hopkins and King 2010), two other sets of coders were also deployed to code the agreements along the three dimensions of obligation, precision, and delegation, using the same criteria outlined in Table 1 above. I also ran a number of secondary robustness checks, including creating ordinal variables for several of the variables that were dummies in the preliminary analysis—including multilateralism and depth—and removed the Cold War dummy. The results remained the same, suggesting that they do not vary depending on particular coding decisions; these results are reported in the Appendix. Moreover, the results presented without control variables increase the confidence that the findings are not driven by the choice of controls. The Iran Agreement as an Illustrative Case As these findings suggest, the most important legalization variables are obligation and to a lesser extent delegation, with precision being inconsequential for reasons posited earlier. While a full case study illustrating these findings is beyond the scope of this article, the recent Iran nuclear deal (called the Joint Comprehensive Plan of Action) helps offer a concrete illustration of the concept, measurement, and consequences of obligation, delegation, and precision in the arms control context. Consistent with the findings outlined here, the two most significant sticking points—affecting both the duration and ultimate success of the negotiations—fell under the heading of obligation and delegation. Reaching an agreement required softening legalization along each of these dimensions, especially obligation. By contrast, high degrees of precision were not an obstacle, corroborating the earlier finding that hard legalization along the dimension of precision does not tend to obstruct cooperative outcomes. In terms of obligation, the agreement essentially has an elaborate set of sunset provisions. As a diplomat who helped negotiate the deal observed (Nephew 2015), “Iran has made it clear that any future nuclear restrictions it agrees to will have to be limited and time-bound,” indicating that “sunset provisions are a common feature of international arms control.” Thus, whereas the United States and its allies might have favored a set of restrictions that continued indefinitely, Iran insisted on being able to hedge against medium- to longer-term uncertainties about its security situation rather than agreeing to stringent nuclear restrictions in perpetuity. Based on the agreement, Iran will enrich uranium only up to 3.67 percent, can begin testing up to 30 IR-6 and IR-8 centrifuges after 8.5 years, will keep its enrichment capacity at Natanz up to a total of 5060 IR-1 centrifuges for 10 years, and will not have heavy water production, spent fuel reprocessing, or uranium stockpile development for 15 years. After each of these periods, Iran is entitled to continue each of these activities. In short, the agreement does not dismantle the program indefinitely, a softening of legalization along the dimension of obligation that made an agreement possible. Consistent with the findings of this research, hard legalization along the dimension of delegation proved to be less of an obstacle, although the negotiating parties did include provisions that softened aspects of the inspection regime’s intrusiveness. The agreement provides for “unprecedented verification measures” that can “establish with near-perfect confidence that the agreement is being respected” (Parmentola and Postol 2015). As part of these provisions, the IAEA has access to declared nuclear sites such as Fordow and Natanz and the number of inspectors increases from the current number of 50 up to 150. For undeclared sites, however, inspectors will come from countries with whom Iran has diplomatic relations. Iran is able to restrict IAEA inspectors at the Parchin military site, where it has previously tested explosives, and Iranians will be able to play a role in inspections that do occur. Against the backdrop of a quite rigorous inspection regime, some sovereignty-preserving concessions appeared to be necessary to reach an agreement, which is also consistent with the more qualified support for the delegation variable outlined earlier. The dimension along which negotiators did not back away at all was precision, and indeed, the specificity of the agreement is extremely high. Instead of stating qualitatively that Iran had to keep its uranium enrichment significantly below the enrichment level needed to create a bomb, the agreement states a specific number. Similarly, the agreement acknowledges that Iran needs tens of thousands of centrifuges to create highly enriched uranium for a bomb and currently has about 20,000 centrifuges between their Natanz and Fordow facilities. The agreement requires Iran to reduce its centrifuges not to a level below that needed to make a bomb but precisely to 6,10417 (White House 2015). Iran must reduce its stockpile of uranium by 98 percent and keep its level of uranium enrichment at 3.67 percent. These examples suggest that high precision was not an impediment to reaching a cooperation agreement compared to obligation, where the negotiating parties crafted a set of sunset clauses, and to a lesser extent delegation, where the parties reduced some of the sovereignty costs associated with the inspection regime. In sum, after 22 months of arduous negotiations, the P5 + 1 concluded an agreement but did so by virtue of softening two aspects of legalization in ways consistent with the statistical findings presented earlier. In this case, it was a matter of stepping back from an indefinite commitment in terms of time (high obligation) and to a lesser extent from airtight inspection regime (high delegation), not by having to pursue vaguer, less precise terms of the agreement (Jervis 2015). Taken together, the sketch helps situate the analysis and findings of this research in the context of the recent Iran nuclear deal, offering some concrete implications for foreign policy. Conclusion While scholars have written at length about institutional design in the areas of trade and the environment (Milner and Rosendorff 2001; Von Stein 2008) or about particular arms control agreements (Koremenos 2001), this is the first study to empirically test the logic of legalization in the context of all arms control agreements. As the discussion suggests, cooperation on arms control presents states with a conundrum. States have little interest in initiating negotiations on a problem of international cooperation if the regime is not effective, defined as being able to “embody, and affect actors’ expectations…and make governments concerned about precedents” (Axelrod and Keohane 1985, 234). To shape expectations, the constraints and commitments must be credible, but the prospect of becoming involved in a highly legalized agreement that would make commitments credible might also make states leery of concluding an agreement at all, or at least bargaining harder and longer to produce an agreement most advantageous to its own interests, which can mean delaying the cooperation objectives that led states to propose an agreement in the first place (Fearon 1998, 270; Milner and Rosendorff 2001, 836). With this conundrum in mind, this research asked how the dimensions of legalization that would help foster longer-term cooperation affect the shorter-term prospects of concluding an agreement and that agreement going into force. Based on this study of all arms control agreements proposed and negotiated between 1945 and the present, I find that particular legalization variables make bargaining over the terms of cooperation more difficult. Obligation consistently obstructs the conclusion of an agreement and entry into force, which is consistent with previous research showing that flexibility in terms of obligation can alleviate the impediments to reaching an agreement (Milner and Rosendorff 2001) since mechanisms such as escape and withdrawal clauses may be particularly well positioned to help cope with future uncertainty (Kucik and Reinhardt 2008), giving states an out when circumstances change in the future. In much more limited circumstances, namely the time for entry into force, delegation also discourages states from participating in an arms control agreement. Delegation in the context of arms control may generate more information, which could make it attractive in circumstances where states face uncertainty and risk (Koremenos 2007), but does so at a cost, namely in terms of sovereignty. Moreover, determined cheaters will be able to evade verification regimes, reducing the advantages in terms of whether delegation can be effective in assuaging uncertainty. Precision, relegated in other accounts of institutional design (Abbott et al. 2000, 22), also takes a backseat to the other variables in this analysis. In short, states may be concerned about opportunism in terms of enforcement, but are more concerned about tying their hands in the face of future uncertainty through highly formalized forms of arms control cooperation, which is why they value features of an agreement that give them ways to opt out if new security threats present themselves in the future. This should not suggest that legalization is less helpful in enforcing that cooperation once the terms of cooperation have been successfully negotiated, but that they can impede the initial stage of cooperation, delaying and in some cases preventing altogether the successful negotiation of an agreement. By clarifying which institutional design features are relevant in the context of negotiating arms control agreements successfully, this research has important implications for policy, such as for crafting the types of agreements that are likely to affect vertical or horizontal proliferation. The findings suggest that international negotiators might be better served drafting less highly legalized agreements that offer latitude in states’ commitment to the agreement, since the prospect of tying their hands will discourage states from engaging in higher-obligation commitments. Backing away from aggressive delegation measures might also help states buy into an agreement, a dynamic illustrated by experiences of Cold War arms control negotiations in which the Soviets in particular were leery of the intrusiveness of on-site inspections (Miller 1984, 720). Focusing on precision is likely to offer less purchase in terms of reaching an international agreement and that agreement entering into force. On the contrary, as the Iran example suggests, negotiators can include quite high precision and still have confidence that the agreement will be successful. When it comes to questions of policy and drafting specific agreements such as the Fissile Material Cutoff Treaty, a proposed binding treaty that would ban fissile material but has languished, this analysis also offers some guidance. In particular, it suggests that easing the degree of obligation, such as by offering escape clauses or even making the agreement more voluntary for the negotiating phase, is more likely to produce a successful outcome in terms of concluding the agreement, which can then foster the conditions under which states accomplish more highly legalized agreements and achieve their arms control objectives over the longer term (Gold 1983). While a first step in systematically evaluating legalization in the context of arms control agreements, it is not the last word. The focus on the attributes of an agreement that facilitate cooperation in terms of concluding the agreement and that agreement entering into force raises questions about cooperation in terms of compliance. Thus, an important next step would be to assess whether state behavior is consistent with arms control agreements that have entered into force, and whether certain attributes of legalization are more closely correlated with compliance. Similarly, future research should evaluate how particular attributes of legalization interact with different states, and whether particular types of states are more receptive to participating in and complying with some types of agreements than others. This analysis focused on the agreements themselves, but if prior research is any guide, factors such as regime type (Hafner-Burton, Mansfield, and Pevehouse 2015; Haftel and Thompson 2013a) and nuclear status (Fuhrmann and Li 2008; Way and Sasikumar 2009) may affect international cooperation on arms control agreements. I leave such questions for subsequent research. Footnotes 1 According to Goldblat (1994, 3), limiting proliferation and reducing arsenals are two types of arms control objectives, which is why I sometimes refer to agreements in terms of proliferation and reductions, but the overall category for this analysis is nuclear arms control. 2 See Brown 2011 for a discussion of IAEA verification problems in the context of Iraq in 1991 and Iran today. This concern, however, applies largely to nuclear weapons proliferation rather than nuclear testing, where national technical means can be valuable. 3 The US State Department Office of the Historian highlights the importance of inspections from the standpoint of the United States and United Kingdom; see https://history.state.gov/milestones/1961-1968/limited-ban 4 The Appendix lists all arms control agreements in the data set. 5 Author interviews with James Goodby, who negotiated the Limited Test Ban Treaty for both Eisenhower and Kennedy, revealed that agreements are proposed based not on likelihood of success but on strategic need. Negotiators then determine appropriate legal contours, implying that the idea precedes the legal arrangement, not vice versa. 6 While it would be interesting to consider various drafts of the same agreement, this approach could be problematic since negotiations take place under a cloak of secrecy and we could not be certain that drafts emerge in a systematically unbiased way. Future work might investigate the iterations of negotiations either within one agreement or across agreements, including the associated time-varying measures. 7 The 1969 Vienna Convention on Law of the Treaties (Article 7) indicates that a state representative can perform acts related to the conclusion of a treaty; these members include heads of state, ministers of foreign affairs, heads of diplomatic missions, and “representatives accredited by States to an international conference…for the purpose of adopting the text of a treaty.” 8 “Increasing Pressure on the Nine CTBT Hold-Outs at the UN,” http://www.ctbto.org/press-centre/highlights/2008/increasing-pressure-on-the-nine-ctbt-hold-outs-at-the-united-nations/. 9 To ascertain that the proportional hazards assumption is appropriate, I ran a global test based on Schoenfeld residuals (following Box-Steffensmeier, Reiter, and Zorn 2003, 36, and Fortna 2004, 282) and determined that the proportional hazard assumption is not violated by the models and variables. 10 For an analogous approach, see Hafner-Burton, Mansfield, and Pevehouse (2015), who code human-rights institutions based on obligation, precision, and delegation. 11 Treaty of Tlatelolco available at http://www.state.gov/p/wha/rls/70658.htm. 12 New START, http://www.state.gov/t/avc/newstart/index.htm. 13 Start I treaty, http://fas.org/nuke/control/start1/text/. 14 Treaty Banning Nuclear Weapon Tests in the Atmosphere, in Outer Space and Under Water, available at http://www.state.gov/t/isn/4797.htm. 15 Institutional context and successor might seem to be related, but the correlation between the two is .31. 16 See Appendix for descriptive statistics. 17 See White House Iran Deal Facts, available at https://www.whitehouse.gov/issues/foreign-policy/iran-deal. References Abbott Kenneth, Snidal Duncan. 2000. “ Hard and Soft Law in International Governance.” International Organization  54: 421– 56. Google Scholar CrossRef Search ADS   Abbott Kenneth, Keohane Robert, Moravcsik Andrew, Slaughter Anne-Marie, Snidal Duncan. 2000. “ The Concept of Legalization.” International Organization  54: 17– 35. Axelrod Robert, Keohane Robert. 1985. “ Cooperation under Anarchy: Strategies and Institutions.” World Politics  38: 226– 54. Google Scholar CrossRef Search ADS   Barrett Scott. 1998. “ On the Theory and Diplomacy of Environmental Treaty-Making.” Environmental and Resource Economics  13: 317– 33. Google Scholar CrossRef Search ADS   Belcher Emma. 2014. Origins of the Nuclear Suppliers Group as a Gentlemen’s Agreement. Unpublished manuscript. Bohlen Avis. 2009. “ Arms Control in the Cold War.” Foreign Policy Research Institute , http://www.fpri.org/articles/2009/05/arms-control-cold-war. Box-Steffensmeier Janet, Reiter Dan, Zorn Christopher. 2003. “ Nonproportional Hazards and Event History Analysis in International Relations.” Journal of Conflict Resolution  47: 33– 53. Google Scholar CrossRef Search ADS   Braun Chaim, Chyba Christopher. 2004. “ Proliferation Rings: New Challenges to the Nuclear Nonproliferation Regime.” International Security  29: 5– 49. Google Scholar CrossRef Search ADS   Brown Robert. 2010. “ Measuring Delegation.” Review of International Organizations  5: 141– 75. Google Scholar CrossRef Search ADS   Brown Robert. 2011. “ International Nonproliferation: Why Delegate to the International Atomic Energy Agency.” In Causes and Consequences of Nuclear Proliferation , edited by Rauchhaus Robert, Kroenig Matthew, Gartzke Erik. New York: Routledge. Copelovitch Mark, Putnam Tonya. 2014. “ Design in Context: Existing International Agreements and New Cooperation.” International Organization  68: 471– 93. Google Scholar CrossRef Search ADS   Dahl Fredrik. 2012. “ Iran Has Expanded Sensitive Nuclear Work: UN Agency.” Reuters . Accessed April 30, 2016. http://www.reuters.com/article/us-nuclear-iran-iaea-idUSTRE81N0HS20120224. Downs Anthony, Rocke David, Barsoom Peter. 1996. “ Is the Good News about Compliance Good News about Cooperation?” International Organization  50: 379– 406. Google Scholar CrossRef Search ADS   Fearon James. 1998. “ Bargaining, Enforcement, and International Cooperation.” International Cooperation  52: 269– 305. Fortna Page. 2004. Peace Time: Cease-Fire Agreements and the Durability of Peace . Princeton, NJ: Princeton University Press. Fuhrmann Matthew. 2012. Atomic Assistance: How “Atoms for Peace” Programs Cause Nuclear Insecurity . Ithaca, NY: Cornell University Press. Fuhrmann Matthew, Li Xiaojun. 2008. “ Rethinking Ratification: Legalization and Nuclear Weapon Free Zone Treaties.” Working Paper. Gold Joseph. 1983. “ Strengthening the Soft International Law of Exchange Arrangements.” American Journal of International Law  77: 443– 89. Google Scholar CrossRef Search ADS   Goldblat Josef. 1994. Arms Control: A Guide to Negotiations and Agreements . Oslo, Norway: Sage. Goldstein Judith, Kahler Miles, Keohane Robert, Slaughter Anne-Marie. 2001. Legalization and World Politics . Cambridge, MA: MIT Press. Guzman Andrew. 2002. “ A Compliance-Based Theory of International Law.” California Law Review  90: 1826– 1888. Google Scholar CrossRef Search ADS   Guzman Andrew, Meyer Timothy. 2010. “ International Soft Law.” Journal of Legal Analysis  2: 171– 225. Google Scholar CrossRef Search ADS   Hafner-Burton Emilie, Mansfield Edward, Pevehouse Jon. 2015. “ Human Rights Institutions, Sovereignty Costs, and Democratization.” British Journal of Political Science  45: 1– 27. Google Scholar CrossRef Search ADS   Haftel Yoram. 2010. “ Ratification Counts: US Investment Treaties and FDI Flows into Developing Countries.” Review of International Political Economy  17: 348– 77. Google Scholar CrossRef Search ADS   Haftel Yoram, Thompson Alexander. 2013a. “ Delayed Ratification: The Domestic Fate of Bilateral Investment Treaties.” International Organization  67: 355– 87. Google Scholar CrossRef Search ADS   Haftel Yoram, Thompson Alexander. 2013b. “ When Do States Renegotiate International Agreements: The Case of Bilateral Investment Treaties.” Hawkins Darren, Lake David, Nielson Daniel, Tierney Michael. 2006. Delegation and Agency in International Organizations . New York: Cambridge University Press. Google Scholar CrossRef Search ADS   Helfer Laurence. 2012. “ Flexibility in International Agreements.” In Interdiscilpinary Perspectives on International Law and International Relations , edited by Dunoff Juffrey and Pollack Mark, 175– 96. New York, NY: Cambridge University Press. Google Scholar CrossRef Search ADS   Hopkins Dan, King Gary. 2010. “ A Method of Automated Nonparametric Content Analysis for Social Science.” American Journal of Political Science  54: 229– 47. Google Scholar CrossRef Search ADS   ILM (International Legal Materials). 1989. “ Union of Soviet Socialist Republics-United States: Agreement on the Prevention of Dangerous Military Activities.” International Legal Materials  28: 877– 95. Jervis Robert. 1978. “ Cooperation under the Security Dilemma.” World Politics  30: 167– 214. Google Scholar CrossRef Search ADS   Jervis Robert. 1990. The Meaning of the Nuclear Revolution: Statecraft and the Prospect of Armageddon . Ithaca, NY: Cornell University Press. Jervis Robert. 2015. “ Turn Down for What? The Iran Deal and What Will Follow.” ForeignAffairs.com. Kelley Judith, Pevehouse Jon. 2015. “ An Opportunity Cost Theory of US Treaty Behavior.” International Studies Quarterly  59: 531– 43. Google Scholar CrossRef Search ADS   Keohane Robert. 1990. “ Multilateralism: An Agenda for Research.” International Journal  44: 731– 64. Google Scholar CrossRef Search ADS   Keohane Robert, Martin Lisa. 1995. “ The Promise of Institutionalist Theory.” International Security  20: 39– 51. Google Scholar CrossRef Search ADS   Kimball Daryl. 2013. “ Fissile Material Cut-Off Treaty at a Glance.” Arms Control Association , August. Koremenos Barbara. 2001. “ Loosening the Ties That Bind: A Learning Model of Agreement Flexibility.” International Organization  55: 289– 325. Google Scholar CrossRef Search ADS   Koremenos Barbara. 2007. “ If Only Half of International Agreements Have Dispute Resolution Provisions, Which Half Needs Explaining?” Journal of Legal Studies  36: 189– 212. Google Scholar CrossRef Search ADS   Koremenos Barbara, Lipson Charles, Snidal Duncan. 2001. “ The Rational Design of International Institutions.” International Organization  55: 761– 99. Google Scholar CrossRef Search ADS   Kroenig Matthew. 2014. “ Force of Friendship? Explaining Great Power Nonproliferation Policy.” Security Studies  23: 1– 32. Google Scholar CrossRef Search ADS   Kucik Jeffrey, Reinhardt Eric. 2008. “ Does Flexibility Promote Cooperation? An Application to the Global Trade Regime.” International Organization  62: 477– 505. Google Scholar CrossRef Search ADS   Kydd Andrew. 2000. “ Arms Races and Arms Control: Modeling the Hawk Perspective.” American Journal of Political Science  44: 228– 44. Google Scholar CrossRef Search ADS   Lipson Charles. 1984. “ International Cooperation in Economic and Security Affairs.” World Politics  37: 1– 23. Google Scholar CrossRef Search ADS   Lipson Charles. 1991. “ Why Are Some International Agreements Informal?” International Organization  45: 495– 538. Google Scholar CrossRef Search ADS   MacDonald Julia. 2015. “ Eisenhower’s Scientists: Policy Entrepreneurs and the Test-Ban Debate 1954–1958.” Foreign Policy Analysis  11: 1– 15. Google Scholar CrossRef Search ADS   Martin Lisa. 2000. Democratic Commitments: Legislatures and International Cooperation . Princeton, NJ: Princeton University Press. Google Scholar CrossRef Search ADS   Mearsheimer John. 1994/95. “ The False Promise of International Institutions.” International Security  19: 5– 49. Google Scholar CrossRef Search ADS   Miller Steven. 1984. “Politics over Promise: Domestic Impediments to Arms Control.” International Security 8: 67–90. Milner Helen V, Rosendorff B. Peter. 2001. “ Democratic Politics and International Trade Negotiations.” The Journal of Conflict Resolution  47: 117– 46. Mistry Dinshaw. 2005. Containing Missile Proliferation . Seattle: University of Washington Press. Mitchell Sara McLaughlin, Hensel Paul. 2007. “ International Institutions and Compliance with Agreements.” American Journal of Political Science  51: 721– 37. Google Scholar CrossRef Search ADS   Nelson Amy. 2013. “Arms Control and Uncertainty: A New Dataset and Its Implications for Arms Control Agreements.” Unpublished manuscript. Nephew Richard. 2015. “ False Flag: The Bogus Uproar over Iran’s Nuclear Sunset.” National Interest , March 6. Oye Kenneth. 1985. “ Explaining Cooperation under Anarchy: Hypotheses and Strategies.” World Politics  38: 1– 24. Google Scholar CrossRef Search ADS   Parmentola John, Postol Theodore. 2015. “ How to Make Iran Inspections Work.” Politico , August 31. Pelc Krzysztof. 2013. “ The Cost of Wiggle-Room: Looking at the Welfare Effects of Flexibility in Tariff Rates at the WTO.” International Studies Quarterly  53: 349– 68. Google Scholar CrossRef Search ADS   Raustiala Kal. 2005. “ Form and Substance in International Agreements.” American Journal of International Law  99: 581– 614. Google Scholar CrossRef Search ADS   Regan Patrick, Stam Allan. 2000. “ In the Nick of Time: Conflict Management, Mediation Timing, and the Duration of Interstate Disputes.” International Studies Quarterly  44: 239– 60. Google Scholar CrossRef Search ADS   Schneider Christina, Urpelainen Johannes. 2013. “ Distributional Conflict Between Powerful States and International Treaty Ratification.” International Studies Quarterly  57: 13– 27. Google Scholar CrossRef Search ADS   Shaffer Gregory, Pollack Mark A.. 2011. “ Hard Law, Soft Law, and International Security: The Cases of Nuclear Weapons and the Responsibility to Protect.” Boston College Law Review  52: 1147– 1241. Simmons Beth. 2009. Mobilizing for Human Rights: International Law in Domestic Politics . New York: Cambridge University Press. Google Scholar CrossRef Search ADS   Simmons Beth, Danner Allison. 2010. “ Credible Commitments and the International Criminal Court.” International Organization  64: 225– 56. Google Scholar CrossRef Search ADS   The Economist, “ An Uneven Deal,” November 12, 2014. Tzeng Peter. 2014. “ Nuclear Leverage.” Nonproliferation Review  20: 473– 92. Google Scholar CrossRef Search ADS   Vaynman Jane. 2014. “ Enemies in Agreement: Domestic Politics, Uncertainty, and Cooperation between Adversaries.” Unpublished manuscript. Von Stein Jana. 2008. “ The International Law and Politics of Climate Change.” Journal of Conflict Resolution  52: 243– 68. Google Scholar CrossRef Search ADS   Wallace Geoffrey. 2014. “ International Law and Public Attitudes Toward Torture: An Experimental Study.” International Organization  67: 105- 40. Google Scholar CrossRef Search ADS   Way Christopher, Sasikumar Karthika. 2009. “ Testing Theories of Proliferation: The Case of Nuclear South Asia.” In Inside Nuclear South Asia , edited by Sagan Scott D.. Stanford, CA: Stanford University Press. Wendt Alexander. 2001. “ Driving with the Rearview Mirror: On the Rational Science of Institutional Design.” International Organization  55: 1019– 1049. Google Scholar CrossRef Search ADS   White House. 2015. “ The Historic Deal that Will Prevent Iran from Acquiring a Nuclear Weapon.” Accessed April 30, 2016. https://www.whitehouse.gov/issues/foreign-policy/iran-deal. Woolf Amy. 2011. Monitoring and Verification in Arms Control . Washington, DC: Congressional Research Service. Woolf Amy, Nikitin Mary Beth, Kerr Paul. 2013. Arms Control and Nonproliferation: A Catalog of Treaties and Agreements . Washington, DC: Congressional Research Service. © The Author (2016). Published by Oxford University Press on behalf of the International Studies Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Foreign Policy Analysis Oxford University Press

The Institutional Design of Arms Control Agreements

Loading next page...
 
/lp/ou_press/the-institutional-design-of-arms-control-agreements-JeBE0jjJYz
Publisher
Wiley-Blackwell
Copyright
© The Author (2016). Published by Oxford University Press on behalf of the International Studies Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
1743-8586
eISSN
1743-8594
D.O.I.
10.1093/fpa/orw045
Publisher site
See Article on Publisher Site

Abstract

Abstract The question of international cooperation on nuclear security presents states with a conundrum. While states are wary of the potential for cheating and opportunism that would favor more legalized agreements, highly legalized agreements also create the risk of being constrained in an agreement with which others might not comply, leaving one state vulnerable. How do states balance these competing incentives? Through the study of all arms control agreements concluded, negotiated, and seriously considered since 1945, this analysis finds that at least some aspects of the legalization parameters that would make commitments more credible and sustainable through the future—in particular obligation and to a lesser extent delegation—also make bargaining over the terms of cooperation more difficult, sometimes delaying if not altogether subverting that process. These findings have implications not just for how theories of institutional design and cooperation apply to nuclear issues but also for policies related to the conduct of arms control negotiations. Introduction In 2014, Iran began working with major international powers—including the permanent five nuclear-weapon states and Germany (P5 + 1)—on a nuclear deal. The negotiations illustrated the tensions inherent in striking an arms control agreement. Some states sought an agreement where the enforcement mechanism and punishment of Iran’s noncompliance would be automatic rather than requiring an international resolution that other countries could veto. Yet, these same states aimed for an agreement that would be sufficiently binding that it would not have to be frequently renegotiated. Several months into the negotiations, one diplomat reported that “We haven’t found a mechanism that works for everyone yet” (Charbonneau 2015). The negotiations illustrate a fundamental problem for nuclear cooperation. How do states negotiate an agreement that is worth the paper it is written on but also grants the ability to hedge against future uncertainty? How does the degree of institutional legalization affect when or whether enough states participate in arms control agreements for those agreements to be concluded and entered into force? This article answers those questions with an analysis of all arms control agreements proposed, negotiated, and concluded between 1945 and the present. It first engages the institutional design literature by way of evaluating the theoretical trade-offs of legalization in the context of arms control agreements. Second, it introduces a new data set of all arms control agreements designed to ban, regulate, reduce, and limit nuclear weapons since 1945.1 Third, it tests the aspects of legalization that affect the propensity for international cooperation on arms control agreements compared to other factors. Fourth, it illustrates the key legalization measurements with reference to the Iran deal concluded in 2015. Finally, it concludes with implications, especially in reference to ongoing debates about the design of international arms control agreements. A number of scholars have investigated aspects of these questions, for example why particular arms control agreements have been designed with particular sets of legalization features (Koremenos 2001), the correlates of state participation in specific arms control agreements (Sasikumar and Way 2009), and the effect of institutional design on cooperation in other issue areas (see, inter alia, Hafner-Burton, Mansfield, and Pevehouse 2015; Von Stein 2008). However, this is the first systematic study of all agreements in the area of arms control. Balancing state sovereignty with the advantages of making credible commitments is a conundrum that scholars have appreciated in other issue areas (Simmons and Danner 2010), but nowhere are the stakes as high as in the nuclear arena, given the consequences of defection. Yet, this cost of defection could theoretically cut two ways. It could make states more eager to negotiate more credible commitments through agreements that are more highly legalized, but the prospect of becoming enmeshed in a durable cooperation arrangement can also make states wary of concluding an agreement at all (Fearon 1998, 270; Milner and Rosendorff 2001, 836), given that the agreement would have enduring consequences even in the face of changing distributions of international power, future conflicts, or shifting domestic political winds. Through the study of arms control agreements, this analysis finds that at least some aspects of the legalization parameters that would make commitments more credible and sustainable through the future—in particular obligation, and to a lesser extent delegation—also make bargaining over the terms of cooperation more difficult, sometimes delaying if not altogether terminating that process. These findings have implications not just for how theories of institutional design and cooperation apply to nuclear issues but also for policies related to the conduct of arms control negotiations. Institutional Design and Arms Control During the 1990s, international relations scholars debated at length the question of whether institutions “matter” (Mearsheimer 1994/95; Keohane and Martin 1995). Alexander Wendt (2001, 1019) observed that the focus on whether institutions matter had “left IR (international relations) with less to say” about the types of institutions that governments create and why. More than a decade later, the literature has filled that void, with scholars identifying the dimensions along which institutions vary in their degree of legalization—operationalized along the dimensions of obligation, precision, and delegation (Abbott et al. 2000; Koremenos, Lipson, and Snidal 2001)—but also raising new sources of debate about how these dimensions affect the likelihood that states will reach a cooperative outcome. Briefly, obligation refers to the degree to which an arrangement is legally binding, for example whether there are escape, withdrawal, or sunset clauses that provide mechanisms for withdrawal should circumstances change in the future (Milner and Rosendorff 2001), or reservations, declarations, and interpretive statements that exempt states from particular aspects of an agreement (Helfer 2012, 179). Precision refers to the specificity or ambiguity of rules outlined in an agreement. Delegation is the degree to which states grant authority to an agent that can help enforce an agreement, whether through diplomacy or an entity such as a court (Hawkins et al. 2006, 7; Abbott et al. 2000, 18). High or hard legalization offers less latitude in terms of how states interpret and implement international legal commitments (with low legalization or soft law as the opposite). Legalization introduces inherent trade-offs between the advantages of a firm, credible commitment and the disadvantages of being locked into an agreement in the event that circumstances change in the future (Goldstein et al. 2001). On the one hand, legalization can be attractive for coping with future uncertainty (Helfer 2012, 177; Koremenos 2001, 289). Lower-obligation, non-binding agreements give states an off-ramp if circumstances change in the future (Abbott and Snidal 2000, 446). States might also be drawn to lower degrees of precision to avoid “undesirable rigidity” in terms of how they interpret an agreement, or to agreements that do not delegate responsibility for enforcement lest doing so introduce higher sovereignty costs (Abbott and Snidal 2000, 433, 435). Moreover, lower degrees of legalization need not come at the expense of institutional effectiveness. As Lipson suggests (1991, 500), insofar as it can “commit states to characteristic forms of discourse and procedure, soft law provides a way of achieving compromise over time.” If low legalization makes it easier for states to come to an agreement, why do states ever conclude more highly legalized agreements? The reason is that they also value the credibility of a commitment. Martin (2000, 3) argues that “the forms of international cooperation that offer states the highest benefits require them to make credible commitments to one another.” Several different attributes of an agreement can make those commitments more credible. Harder law in the form of binding commitments creates higher reputational costs for reneging (Guzman and Meyer 2010; Shaffer and Pollack 2011, 1162). Precision also helps limit opportunism since it limits “self-serving auto-interpretation” of an agreement (Abbott and Snidal 2000, 427) and, as Simmons (2009, 121) puts it, “reduces the scope for plausible deniability of violation.” Delegation helps identify and punish violators in ways that guard against defection and bring state behavior back into compliance (Koremenos 2007). Assessing Legalization in Arms Control Agreements While a number of scholars have examined legalization in the context of the environment (Von Stein 2008) and trade (Milner and Rosendorff 2001; Kucik and Reinhardt 2008, 478), the nature of coordination trade-offs in those issue areas may be qualitatively different than in a security issue such as arms control, which presents states with complicated payoff structures that require balancing the rewards of mutual restraint with the potential risks if one side cooperates and the other defects (Axelrod and Keohane 1985, 231–32). On the one hand, the logic of the spiral model should “provide the intellectual basis for arms control” (Kydd 2000, 228), as it implies that the accrual of nuclear weapons makes states worse off by inducing fear that can occasion either preemptive war or a nuclear accident (Jervis 1978). By implication, effective arms control institutions would entail credible commitments, given the risks of unilateral disarmament. Reagan had said that “he would gladly accept nuclear disarmament if he was sure he could get the genuine article” (quoted in Jervis 1990, 3). One way to do that is to make those commitments binding, with states seeking a more formal commitment to try to militate against opportunistic behavior and reneging. The more binding the contract (higher obligation), the higher the reputational costs for reneging and the more credible the commitment (Mitchell and Hensel 2007, 724). Jervis suggests (1978, 174) that in the absence of such binding obligations, defection will become a “self-fulfilling prophecy in which each person rushes to defect because he fears others are going to.” Similarly, credibility concerns could explain the observation that “arms control agreements display increasing precision” (Goldstein et al. 2001, 2), with states trying to combat cheating or opportunism that could arise from ambiguity in the language of the agreement. As Oye (1985, 17) observes, “the very act of clarifying standards of conduct, of defining cooperative and uncooperative behavior, can permit more effective resort to strategies of reciprocity” that are central to principles of arms control. Many agreements, especially the bilateral agreements between the United States and Russia cite specific numbers of weapon cuts to an arsenal, and others, such as the 1963 Treaty Banning Nuclear Weapons Tests in the Atmosphere, in Outer Space, and Underwater, state an unequivocal, sweeping prohibition. Such concerns about credible commitments might also mean states favor a robust verification and monitoring mechanism as a way to identify and punish noncompliance. As Jervis (1990, 59) suggested, “the state not only wants to know what its adversary is doing, but under many conditions also wants its adversary to know what it is doing.” Similarly, Barrett (1998, 322) argues that “countries will only agree to do something collectively if that something can be monitored,” especially under conditions of uncertainty where states value additional checks on others’ commitments (Koremenos 2007, 193–94). This could mean providing legal authority to conduct on-site inspections or use national technical means to identify whether state parties to an agreement are in compliance. If these arguments about credible commitments are operative in the context of nuclear weapons, the observable implication is that less highly legalized arrangements would be unusual or even nonexistent in the arms control arena (Raustiala 2005, 592–93). Legalization as an Obstacle to Cooperation Another interpretation of the paradox described above—in which states might value some form of arms control but be concerned with whether others’ commitments are credible and reciprocal—is that states might be wary of committing to highly legalized commitments on arms control in order to hedge against an uncertain future. As Fearon (1998, 270) suggests, the prospects of cooperation over the long term can work at cross-purposes with actually making that agreement happen because it can “give states an incentive to bargain harder, delaying agreement in hopes of getting a better deal.” Based on this formulation, cooperation is a two-phase process in which the longer-term enforcement of the agreement affects states’ initial approach to negotiation. Whereas the implications for Fearon are that an agreement is delayed or even altogether scuttled, another possibility links the prospects for long-term cooperation with the shorter-term prospects of reaching an agreement. States might ease some of the legalization features that foster longer-term cooperation but obstruct the shorter-term agreement in order to accommodate the source of state concerns about future uncertainty. As Milner and Rosendorff (2001, 833) suggest in the context of trade, “uncertainty about the state of the world” can give states incentives to craft agreements that allow them to hedge, for example by including escape clauses. Some of the uncertainties that arise in the context of arms control are not dissimilar to those that would arise in other issue areas. As Nelson (2013) notes, changes in domestic leadership, concerns about the potential for future conflict, and changes to the distribution of power can all create uncertainties and hesitation about binding states to future commitments. States might deem it appropriate to limit nuclear weapons at one moment, but not want to tie their hands for a time in the future when circumstances have changed (Koremenos 2001). Beyond the question of uncertainty that could affect how states regard highly legalized agreements across issue areas, there are other arms control–specific factors that link the enforcement phase with how states approach the initial decision about how to craft an agreement. First, as Fearon (1998, 282) suggests, “when states care a lot about future payoffs, the expected long-run benefits of getting the better deal are very large.” Nuclear strategy is an area where states care about future payoffs insofar as defection involves “potentially grave losses to a player who attempts to cooperate without reciprocation” (Lipson 1984, 14). States care deeply about minimizing the costs of defection, where in the absence of a sovereign, one state could defect while another disarms (Oye 1985, 13), leaving the disarmed state vulnerable. Second, the attempts to create something resembling a sovereign through robust verification and enforcement mechanisms create their own dilemmas. Under the NPT, non-nuclear-weapon states that are party to the treaty are permitted to produce fissile material for peaceful purposes (Article IV), but many of the processes used in the service of civilian nuclear power can be channeled toward military ends, which makes it difficult to verify whether facilities are being employed in the service of peaceful rather than military purposes (Fuhrmann 2012). On-site inspections can help verify, but the IAEA is limited in its capacity to monitor each nuclear site. Even for the sites with NPT safeguards, such as Iran, the IAEA had reported that “it continues to have serious concerns regarding possible military dimensions to Iran’s nuclear program” (Dahl 2012). As the Congressional Research Service concludes, “a determined cheater could probably find a way to conceal some types of violations” (Woolf, Nikitin, and Kerr 2013, 9).2 Third, while states might prefer to have more information about the nuclear activities of other states (Jervis 1990, 59), delegation comes at a cost in terms of sovereignty. Indeed, what makes an inspection regime capable of uncovering violations are the intrusive reporting and inspection standards that are “clearly negatives from the point of view” of the target state (Braun and Chyba 2004, 31). Historical evidence suggests that on-site inspections were a sticking point in arms control negotiations such as the LTBT in 1963, with the United States favoring inspections as a way to corroborate Soviet compliance and the Soviets viewing inspections as a violation of sovereignty (MacDonald 2015).3 Thus, not only is delegation costly in terms of sovereignty, the advantages may be limited, since states may be able to elude detection. Even the most intrusive verification regime, consisting of on-site inspections, has been seen as “an encroachment on national sovereignty” (Blix 2001), while still not necessarily producing incontrovertible evidence about proliferation. As the preceding discussion suggests, states face a dilemma in fashioning international agreements. While states are wary of the potential for cheating and opportunism, which would favor more legalized agreements, highly legalized agreements also create the risk of constraining states in an agreement with which others might not comply, leaving one state vulnerable. Taken together, this suggests that higher legalization might not necessarily be a prerequisite of cooperation but actually make it less likely, with certain aspects of legalization more vexing than others. When it comes to the uncertainties associated with committing to a long-term agreement and the potential consequences of other states reneging, I argue that states are therefore likely to favor arms control agreements that incorporate flexibility mechanisms. The most straightforward manifestation is through lower obligation agreements, for example those that are non-binding or allow for declarations and reservations, duration provisions, amendment procedures, and withdrawal and termination clauses (Helfer 2012; Milner and Rosendorff 2001). This discussion about uncertainty in the context of arms control produces the first hypothesis: H1: High-obligation agreements should make cooperation on arms control less likely. Although scholars tend to treat flexibility mechanisms primarily as ways to sidestep, withdraw, or otherwise give “wiggle room” to states as they comply with an agreement over time (Pelc 2013), degrees of precision may be a conceptual cousin of these tools. If so, aspirational language in the form of a less precise agreement could provide an analogous way for states to hedge against future uncertainty, as it can give states latitude to interpret behavior in ways that are consistent with an agreement even if the compliant nature of those actions becomes a stretch over time (Economist 2014). If this is the case, then we would see support for the second hypothesis: H2: High-precision agreements should make cooperation on arms control less likely. Finally, states may also find the degree of delegation reassuring in one sense but problematic in another, since the types of delegation that could help verify states’ compliance with an agreement also introduce costs to states’ sovereignty while not being failsafe. Thus, we might also expect to see evidence for the third hypothesis: H3: High-delegation agreements should make cooperation on arms control less likely. The next section turns to the empirical analysis of how the degree and attributes of legalization influence the likelihood of cooperation and under what circumstances. Research Design To address this question of how different forms of legalization affect international cooperation on arms control agreements, I compiled a data set of all nuclear arms control measures negotiated between 1945 and 2010.4 The data set is drawn from sources such as the Nuclear Threat Initiative, the Federation of American Scientists, and Goldblat’s (1994). Since the main focus is on how the degrees of legalization affects the propensity for cooperation, the data set includes both formal treaties and informal agreements. In addition, it includes both successful agreements and proposed agreements that are never concluded or enter into force, so that the sample does not select for a subset of agreements that are better suited to a cooperative outcome. Indeed, one-fifth of the proposed agreements have never been concluded or entered into force, suggesting that the sample is not systematically excluding untenable arms control ideas. While it might be possible, in principle, to go even further toward addressing the potential for selection bias by trying to include arms control norms or ideas proposed behind closed doors that do not ever reach a formal proposal, this approach would raise different concerns, which is whether these ideas come to light in a systematic, unbiased way.5 Thus, conditional on proposing an agreement, the analysis evaluates why some are successful—and sometimes quickly—and others are not.6 Based on these parameters, a formal agreement enters the data set when a state representative, as defined by the 1969 Vienna Convention on the Law of the Treaties (Article 7), publicly proposes a particular arms control problem be addressed.7 Often, such heads of state or ministers of foreign affairs have heralded proposals in international forums such as the UN General Assembly, where President Clinton, for example, called for fissile material cutoff negotiations in a September 1993 address. In that same session of the UNGA, the UNGA passed Resolution 48/75L endorsing a “prohibition of the production of fissile material for nuclear weapons or other nuclear explosive devices” (Kimball 2013). For informal agreements that tend not to arise through formal UNGA resolutions, I turn to high-level officials’ publicly endorsing a particular set of arms control principles alongside bilateral or multilateral negotiations. For example, while the United States was initially concerned with the transfer of missiles in the 1970s, concrete efforts to curb missile proliferation were restricted to internal US debates about how to control the transfer of missile technology. In 1982, the United States issued a new policy called “Nuclear Capable Missile Technology Transfer Policy” that outlined the objectives of limiting missile exports and concurrently began negotiations on the Missile Technology Control Regime (MTCR) with West Germany, Italy, and France that same year (Mistry 2005, 46). Thus, the MTCR enters the data set with this proposal and the beginning of negotiations in 1982. Dependent Variable The outcome of interest for this analysis, international cooperation in arms control, is captured in two ways: first, in terms of the duration required to achieve cooperation, and second, more simply by whether states were able to achieve cooperation at all, each discussed in turn. Theoretical and policy reasons suggest some virtue in a relatively expedient negotiation process. Indeed, international organizations often act as though expediency is important. They frequently pass resolutions or statements urging rapid progress on an agreement—for example a 2013 UNGA resolution urging the “speedy establishment of a nuclear-weapon-free zone in the Middle East—intimating a belief that the successful conclusion and entry into force of an agreement have an important effect on state behavior.8 Moreover, scholars have shown that delay in reaching an agreement is costly. As Fearon (1998, 277) concludes, “delay means more time spent without the benefits an agreement would bring; second, as time passes there may be some growing risk that one side will break off negotiations entirely.” Within the context of duration, I evaluate two different outcomes. One is from the time of proposal to the successful conclusion of an agreement. According to Article 9 of the 1969 Vienna Convention, this formally occurs when participating states adopt the text of the treaty, an important mark of diplomatic progress that would theoretically create reputational costs for states that renege (Guzman 2002). Another way to measure duration is from proposal to entry into force (EIF). The Vienna Convention on the Law of Treaties specifies that the agreement enters into force whether as a function of the negotiating states’ consent to comply with the agreement or by a date specified by the treaty. To the extent that reputational loss is higher once the agreement has entered into force (EIF), it is a more consequential marker of cooperation (Schneider and Urpelainen 2013, 14; Haftel 2010). To summarize, in terms of using duration to international cooperation as the dependent variable, I consider both conclusion of an agreement and EIF. For the evaluation of duration as the dependent variable, I use survival analysis to assess the likelihood of state participation with the agreement-year as the unit of analysis. Whereas epidemiologists use survival analysis to assess the efficacy of a treatment (e.g., the recurrence of a disease), political scientists have applied it for evaluating the effect of one variable on the duration of a particular outcome, for example whether outside mediation brings a more rapid end to interstate disputes (e.g., Regan and Stam 2000). Here, I use survival analysis, specifically a Cox proportional hazard model,9 in the context of nuclear agreements to estimate the way legalization impacts the duration required to reach an agreement and, where applicable, to go into effect. This measures the “risk” or likelihood of an event taking place over time, which helps explain variation in the rate at which a particular outcome—the conclusion or EIF of an agreement—takes place (Haftel and Thompson 2013b, 370). A second dependent variable that captures cooperation on arms control agreements is whether the agreement is concluded (and ratified) or not, a dichotomous measure. Perhaps international organizations are misguided in pushing for speedy conclusion of agreements and more effective, durable agreements take longer because countries have to settle their differences in an effective manner (Haftel 2010, 351–52), in which case duration may not be as important as whether the agreement is successfully reached and enters into effect. I therefore employ an alternative measure of international cooperation, which is whether the agreement is concluded and enters into force, regardless of duration, and then use probit analysis to evaluate the reasons why agreements succeed or fail, with the agreement-year as the unit of analysis. In sum, I include two separate measures for the dependent variable. The first measures international cooperation as a function of state participation over time—with conclusion of an agreement and EIF as two different ways of capturing this outcome. The second is whether an agreement is concluded and enters into force successfully at all. Both sets of dependent variables speak to the ways that state participation culminates in an international agreement. Some observers might respond that certain agreements are crafted precisely to screen out states that are not relevant to or expected to comply with the agreement. For example, in most arms control negotiations between the United States and Russia, other countries were explicitly excluded from the negotiations by design and objective. The dependent variable focusing on international cooperation based on reaching a threshold of state participation is not meant to imply maximization but rather achieving the particular levels required by the agreement. Independent Variable: Legalization I measure the main independent variable, degree of legalization, based on the obligation, precision, and delegation typology discussed earlier. As Abbott et al. (2000, 17) suggest, each of these dimensions is a continuum, which the authors illustrate in reference to a variety of agreements. How these dimensions are operationalized in the context of arms control has not been outlined in the existing literature, but is important for the current study that investigates how the design of institutions affects the speed and likelihood of international cooperation in arms control agreements. As a first step, then, Table 1 outlines the categories of legalization and indicators in the specific context of arms control; while the obligation dimension is generalizable beyond arms control, the other two require more arms control–specific indicators cited in the table and elaborated upon below.10 Table 1. Indicators of legalization along the dimensions of obligation, precision, and delegation Legalization dimension  Indicator of legalization  Obligation  Political treaty rather than agreement, unconditional obligation and intent to be legally bound  No reservations, understandings, declarations permitted  No temporal modifications (e.g., renegotiations, escape, withdrawal, or sunset clauses)  Precision  Specific types of nonproliferation behavior or progress toward nonproliferation  Specific arsenal numbers, thresholds, delivery vehicles, military capabilities cited  Delegation  Reporting of compliance through regular meetings or information exchanges  Monitoring through states’ National Technical Means  Verification through on-site inspections  Legalization dimension  Indicator of legalization  Obligation  Political treaty rather than agreement, unconditional obligation and intent to be legally bound  No reservations, understandings, declarations permitted  No temporal modifications (e.g., renegotiations, escape, withdrawal, or sunset clauses)  Precision  Specific types of nonproliferation behavior or progress toward nonproliferation  Specific arsenal numbers, thresholds, delivery vehicles, military capabilities cited  Delegation  Reporting of compliance through regular meetings or information exchanges  Monitoring through states’ National Technical Means  Verification through on-site inspections  View Large In the context of a nonproliferation agreement, obligation deals first with whether it is legally binding. For example, the Nuclear Suppliers Group (NSG) was assembled to regulate the transfer of nuclear-related materials with an eye to stemming proliferation. As Belcher (2014) notes, the NSG was based on a “gentleman’s agreement,” with members taking a pledge rather than making a formal contract. By contrast, the Comprehensive Test Ban Treaty (CTBT) and the Negative Security Assurances agreement were both—were they to have entered into force—legally binding. A second indicator of obligation is whether the agreement allows for reservations, understandings, declarations. The CTBT explicitly does not allow for such treaty actions, whereas the Treaty for the Prohibition of Nuclear Weapons in Latin America and the Caribbean permitted countries such as the United States to issue specific understandings and declarations that “adherence to Protocol II will not prevent collaboration by the US with Contracting Parties for the purpose of carrying out explosions of nuclear devices.”11 A last indicator of obligation is whether there is temporal conditionality that allows states to revise their commitments in the future. As discussed earlier, escape or renegotiation clauses give states latitude over time in terms of their commitment to a treaty. In the case of the nonproliferation treaty (NPT), negotiating states included a built-in expiration date after which the treaty would be reopened for discussion, a flexible trial period that allowed states to overcome uncertainty about the future and come to an agreement (Koremenos 2001). An agreement is coded as high precision if the language is categorical and unambiguous, giving little room for interpretation. The most straightforward indicator is an arms reduction treaty that specifies the number or type of weapons allowed in a country’s arsenal. Whereas the New Start treaty specified the number of deployed missiles, warheads, and launchers that the United States and Russia could possess,12 the Prevention of Dangerous Military Activities Agreement proscribes damage to another party, while being vague about the technical thresholds or definitions of such damage (ILM 1989). Some agreements, such as the NPT, vary in the degree of precision within the agreement (Abbott and Snidal 2000, 450). Article I prohibits nuclear-weapon states from the “transfer to any recipient whatsoever of nuclear weapons or other nuclear explosive devices,” which is quite precise and categorical, whereas Article VI requires that nuclear-weapon states “pursue negotiations in good faith on effective measures relating to cessation of the nuclear arms race at an early date.” While there may be heterogeneity within an agreement, the coding of precision used here relies on the highest level of precision within an agreement. Another indicator of precision relates to compliance behavior, specifically whether it spells out timelines for compliance or types of proliferation behavior. For example, the START I treaty notes the specific time frame for accomplishing each phase of the treaty (e.g., no later than 36 months after EIF for the first phase).13 A contrasting example is the Accident Measures Agreement, whose Article 1 requires that “Each Party undertakes to maintain and to improve, as it deems necessary, its existing organizational and technical arrangements to guard against the accidental or unauthorized use of nuclear weapons under its control.” Finally, whether an agreement is legalized along the dimension of delegation requires some explanation before turning to the indicators. High delegation in the context of a multilateral treaty would mean “vesting decision-making authority in a third party institution with strong adjudicative capacity or independent administrative power” (Abbott and Snidal 2000, 424). Such a definition may be appropriate in the context of institutions such as the World Trade Organization, but would not capture what it means to delegate authority in the context of bilateral arms control agreements where creating a third-party enforcement and monitoring organization is unusual. Russia and the United States often negotiated questions of verification and monitoring on a bilateral basis, with responsibility for adjudication vested in the other country, for example through national technical means or inspections (Woolf 2011, 8–10). In this context, delegation would mean “the creation of monitoring and enforcement mechanisms” that increase the sovereignty costs of the agreement by granting authority for implementation to third parties or other states in the agreement (Hafner-Burton, Mansfield, and Pevehouse 2015, 7). In the context of arms control, this could first refer to whether the agreement stipulates non-voluntary reporting on armaments (and locations) related to the agreement, whether through regular meetings of state parties or information exchanges (Vaynman 2014), such as the extensive data exchanges stipulated by Article IX of the INF treaty. A second indicator of delegation deals with monitoring—for example through technology—that is carried out by state parties to the agreement. START I has a verification regime that includes satellites to collect data and detect violations (Woolf, Nikitin, and Kerr 2013, 9). The third includes verification through on-site inspections (Brown 2010, 152). The LTBT sidestepped such inspections of facilities for the reasons mentioned earlier, whereas START I had an intrusive regime of on-site inspections.14 As these examples suggest, the same agreement can exhibit strong legalization on some dimensions and not others. Rather than aggregate into one index of legalization, since each dimension has different set of theoretical expectations than others, I consider each separately. Indeed, this also helps produce more useful policy implications, speaking to design features that increase the likelihood of conclusion and EIF. Each is associated with a legalization score calculated by summing the number of indicators present for each dimension; for example, an agreement that exhibits all indicators of obligation would be scored a 3 for that dimension. The Appendix shows the coding of each institutional design variable. Beyond the core legalization measures, I also control for the group size or degree of multilateralism involved in the coordination. Larger group sizes could create a collective action problem that complicates coordination by creating uncertainty about cheating and enforcement (Axelrod and Keohane 1985; Oye 1985). Controlling for group size also addresses concerns about the comparability between bilateral and multilateral treaties, where a bilateral treaty enters into force as soon as the two countries have ratified while a multilateral treaty has different thresholds for EIF, potentially creating higher hurdles for the latter. Keohane (1990, 731) observes that “multilateralism can be defined as the practice of coordinating national policies in groups of three or more states, through ad hoc arrangements or by means of institutions.” Multilateralism is therefore a dummy variable that takes on a 1 for arrangements such as a Nuclear Weapon Free Zone or the NPT, and a 0 for bilateral agreements, such as the Non-Attack Agreement between India and Pakistan. I account for whether great powers—specifically the five nuclear weapon states designated by the NPT—propose or draft an arms control proposal, as their involvement might be more likely to yield a successful outcome (Kroenig 2014). For example, the United States used bargaining leverage in the form of loans and guarantees to pressure South Korea to ratify the NPT in 1974 (Tzeng 2014, 476–77). The approach of the NPT, which was drafted partly by the United States, contrasts with that of the Nuclear Weapons Convention, which has been shunned by the permanent five other than China. Although the agreements all address arms control, I control for the possibility that heterogeneous goals could affect the speed or success with which an agreement is concluded or goes into effect. As Downs, Rocke, and Barsoom point out (1996, 389), many arms control agreements were designed to “maintain the status quo trajectory rather than to alter it significantly,” and did not require that participating states deviate from their current behavior, thereby making it an easier path toward conclusion, EIF, and ultimately compliance. I include a dummy variable depth for cooperation that requires deviation from the status quo, compared to agreements that did not alter a state’s behavior. For example, agreements such as the Hot Line agreement did “not regulate an arms output such as the number and/or location of a weapons system” (Downs, Rocke, and Barsoom 1996, 388–89), whereas agreements such as the intermediate-range nuclear forces (INF) and bilateral treaties such as START required the elimination or reduction of existing forces (390). I also control for what Copelovitch and Putnam (2014, 472) refer to as “institutional context,” which refers to “the presence or absence of prior agreements between prospective partners in ‘new’ cooperation.” Context for nuclear agreements means that some states will have successfully engaged in prior negotiations, which should create a longer shadow of the future and make subsequent cooperation more likely. I code prior design successes as a count variable based on the number of prior agreements involving the same parties. For example, the Threshold Test Ban Treaty is coded as a 3 because it built on the institutional context of the Baruch Plan, the formation of the IAEA, and the Limited Test Ban Treaty. I also include a dummy variable, successor, for agreements that were negotiated as follow-on agreements to the earlier agreement; for example, SALT II and START II were successors to the earlier, related agreement, which could have made them more likely because they had clear diplomatic antecedents.15 Finally, I account for circumstances that could affect both interest in pursuing a particular arms control objective in the first place and the determination to conclude and facilitate EIF quickly. As Kelley and Pevehouse (2015) note, the consent process underlying treaty ratification requires time and political capital. It may be that leaders’ willingness to expend time and capital to generate a fruitful outcome to negotiations and EIF depends in part on the security context. Some accounts suggest that the introduction of thermonuclear weapons provided the impetus for proposing the LTBT (Bohlen 2009), and that the close call of the Cuban Missile Crisis further helped parties reach an agreement. To address the potential effect of the Cold War on incentives to diffuse tensions through arms control, I include a Cold War dummy that takes on a 1 until 1991, the year of Soviet dissolution, including its associated nuclear weapons. Results Figure 1 tracks the number of agreements based on conclusion and EIF over time. As the figure shows, events characterized by whether an agreement was concluded are concentrated more up front whereas the frequency with which those agreements come into force are scattered across more decades. As the descriptive statistics show,16 the mean time to conclude an agreement is 9.02 years, compared to 10.85 for EIF. Figure 1. View largeDownload slide Frequency of nonproliferation agreements concluded and entered into force as a function of time (years) Figure 1. View largeDownload slide Frequency of nonproliferation agreements concluded and entered into force as a function of time (years) In terms of the overall success of proposed agreements, of the 48 arms control agreements that have been proposed, a total of nine were either never concluded or did not enter into force (see Table 2). Six, such as the Baruch Plan and Antisatellite Agreement, were neither concluded nor ratified. Three, including SALT II and the Comprehensive Test Ban Treaty, never entered into force. The fact that negotiations do fail at fairly frequent rates suggests that the sample is not biased in favor only of arms control objectives that states expected to succeed. Table 2. Cross-tabulations of proposals that were concluded and entered into force Agreement concluded  Entered into force  Yes  No  Total  Yes  39  3  42  No  N/A  6  6  Total  39  9  48  Agreement concluded  Entered into force  Yes  No  Total  Yes  39  3  42  No  N/A  6  6  Total  39  9  48  View Large Having offered a broad overview of arms control agreement behavior, the analysis first turns to a discussion of the dependent variable that captured duration. Table 3 shows the results from two models: Models 1 and 2 estimate how the degree of legalization influences the likelihood that an agreement will be concluded (signed), without and with controls, respectively, while Models 3 and 4 consider outcomes based on the agreement entering into force (without and with controls). The table reports hazard ratios, which refer to the risk at which an agreement will be concluded or enter into force during the time period. It is interpreted relative to 1, such that ratios below 1 mean a lower likelihood of an agreement occurring in a given time period (and longer time required) and ratios greater than 1 mean a higher likelihood of an agreement occurring under those conditions (decreasing the amount of time required). Table 3. Cox proportional hazard models for treaties concluded and entering into force   DV: Concluded  DV: EIF    Model 1 (no controls)  Model 2 (with controls)  Model 3 (no controls)  Model 4 (with controls)  Legalization           Obligation  .08 (.03)***  .08 (.03)***  .13 (.04)***  .13 (.04)***   Precision  1.15 (.34)  1.07 (.36)  1.14 (.40)  1.2 (.46)   Delegation  .98 (.18)  .79 (.17)  .17 (.06)  .47 (.12)**  Multilateralism    2.35 (.98)**    1.19 (.82)**  Cold War    .47 (.23)    .37 (.18)**  Great Power    2.11 (1.47)    3.22 (2.17)  Depth    1.59 (.82)    1.6 (.80)  Context    .93 (.04)    .88 (.04)**  Successor    23.99 (15.80)***    2.92 (3.29)  N  389  389  482  482  Time “at risk”  35-0  3509  5232  5232  Log-likelihood  –114.28  –104.08  –118.55  –113.03  p > x2  0.00  0.00  0.00  0.00    DV: Concluded  DV: EIF    Model 1 (no controls)  Model 2 (with controls)  Model 3 (no controls)  Model 4 (with controls)  Legalization           Obligation  .08 (.03)***  .08 (.03)***  .13 (.04)***  .13 (.04)***   Precision  1.15 (.34)  1.07 (.36)  1.14 (.40)  1.2 (.46)   Delegation  .98 (.18)  .79 (.17)  .17 (.06)  .47 (.12)**  Multilateralism    2.35 (.98)**    1.19 (.82)**  Cold War    .47 (.23)    .37 (.18)**  Great Power    2.11 (1.47)    3.22 (2.17)  Depth    1.59 (.82)    1.6 (.80)  Context    .93 (.04)    .88 (.04)**  Successor    23.99 (15.80)***    2.92 (3.29)  N  389  389  482  482  Time “at risk”  35-0  3509  5232  5232  Log-likelihood  –114.28  –104.08  –118.55  –113.03  p > x2  0.00  0.00  0.00  0.00  Note: Dependent variable is whether an agreement was reached (models 1, 2) or entered into force (model 3, 4). ***Significant at the 1 percent level. **Significant at the 5 percent level. *Significant at the 10 percent level. Note: Hazard ratios > 1 are interpreted as lower risk of achieving agreement, hazard ratios < 1 means higher risk of achieving agreement. View Large As Table 3 suggests, not all features of legalization are created equal in terms of affecting whether the agreement will be concluded and entered into force in a timely way. For example, agreements that are marked by the highest obligation decrease the likelihood of reaching an agreement or that agreement going into effect in a particular year by 92 and 87 percent, respectively, offering strong support for H1. The analysis provides more qualified support for H3. High-delegation agreements also impacted the prospect of cooperation, though in terms of delaying the agreement going into effect rather than reaching an agreement altogether, with agreements having the highest degree of delegation delaying the prospect of EIF in a given year by 53 percent. The hypothesis related to precision (H2), on the other hand, found no support, as it has no meaningful impact on the prospects for cooperation. As suggested above, some qualitative evidence points to cases where less precision in an agreement has contributed to its successful conclusion of an agreement, but preferences regarding precision do not appear to be a factor that systematically affects cooperation outcomes. Beyond the main legalization variables, the analysis shows that several of the control variables affected outcomes. Several factors made it more likely that an agreement would be reached or entered into force during a particular period, including whether the agreement was multilateral. This finding is surprising given the theoretical expectations that coordination among more parties should raise more potential differences to reconcile. Yet, in the case of arms control agreements, the reason that multilateralism facilitates the speed of conclusion and EIF may be that many multilateral agreements are actually non-binding agreements like the Proliferation Security Initiative that are negotiated quickly. Indeed, interacting obligation and multilateralism produces a hazard ratio of 5.27 that is significant at the 5 percent level, suggesting that multilateralism in its association with informal agreements may be driving the finding rather than multilateralism per se. Whether the agreement is a successor to an earlier set of negotiations made it more likely that the agreement would be concluded, which is not surprising since the previous agreement likely established the groundwork for subsequent negotiations. For example, START I took the longest to negotiate compared to its successors. Conversely, institutional context slowed the prospect of EIF, perhaps because whereas the successor agreement is more likely to capture differences in degree, context is more likely to capture differences in kind, where prior negotiations mean that some of the easier cooperation problems were already solved and that the more difficult ones were shelved until a subsequent time. The Cold War also made EIF during a given period less likely, which is consistent with a number of agreements, such as the Conventional Forces in Europe agreement, whose EIF stalled until after the Cold War. Turning to the factors that affect whether agreements are successful, as opposed to the timing of cooperation, the results look quite compatible with the duration analysis. As Table 4 shows, obligation is the most significant attribute and the coefficient is negative; the more binding the agreement, the less likely it is to be concluded or entered into force. Thus, irrespective of how the dependent variable is measured—by duration or by whether the agreement is ultimately successful in its conclusion and entry into force—obligation is revealed to be highly significant. Similarly, as with the duration models, delegation is significant for EIF but not whether it is concluded, suggesting that higher-delegation agreements can ultimately be a barrier to success. Table 4. Probit models for treaties concluded and entering into force   DV: Concluded  DV: EIF    Model 1 (no controls)  Model 2 (with controls)  Model 3 (no controls)  Model 4 (with controls)  Legalization           Obligation  –-.81 (.16)***  –.87 (.19)***  –.60 (.13)***  –.57 (.15)***   Precision  .03 (.87)  –.04 (.19)  .02 (.18)  .13 (.21)   Delegation  .00 (.10)  –.04 (.12)  –.19 (.10)*  –.29 (.13)**  Multilateralism    .32 (.25)    .35 (.26)  Cold War    –.68 (.29)**    –.72 (.26)**  Great Power    .67 (.43)    .78 (.39)**  Depth    –.03 (.29)    .02 (.26)  Context    –.05 (.03)*    –.07 (.03)**  Successor    .92 (.41)**    –.09 (.51)  N  389  389  482  482  Log-likelihood  –104.25  –98.30  –109.41  –104.12  Constant  –.43 (.20)  –.39 (.39)  –.41 (.20)  .35 (.38)    DV: Concluded  DV: EIF    Model 1 (no controls)  Model 2 (with controls)  Model 3 (no controls)  Model 4 (with controls)  Legalization           Obligation  –-.81 (.16)***  –.87 (.19)***  –.60 (.13)***  –.57 (.15)***   Precision  .03 (.87)  –.04 (.19)  .02 (.18)  .13 (.21)   Delegation  .00 (.10)  –.04 (.12)  –.19 (.10)*  –.29 (.13)**  Multilateralism    .32 (.25)    .35 (.26)  Cold War    –.68 (.29)**    –.72 (.26)**  Great Power    .67 (.43)    .78 (.39)**  Depth    –.03 (.29)    .02 (.26)  Context    –.05 (.03)*    –.07 (.03)**  Successor    .92 (.41)**    –.09 (.51)  N  389  389  482  482  Log-likelihood  –104.25  –98.30  –109.41  –104.12  Constant  –.43 (.20)  –.39 (.39)  –.41 (.20)  .35 (.38)  Note: Dependent variable is whether an agreement was reached (models 1, 2) or entered into force (models 2, 3). ***Significant at the 1 percent level. **Significant at the 5 percent level. *Significant at the 10 percent level. View Large Precision is again not significant in either model, raising the question of why. As Wallace (2014, 114) notes, “precision is frequently relegated to the least consequential component of legalization.” Scholars such as Abbott et al. (2000, 22) code an agreement that is high precision but low obligation and delegation agreement as soft legalization, and vice versa, with a high obligation and delegation agreement that is low precision as harder legalization. One reason for privileging obligation and delegation is that these both impose higher costs. Obligation creates potential costs in the form of reducing decision-making autonomy should circumstances change, and delegation imposes sovereignty costs for the most intrusive inspections. Precision is a matter of interpretation and is not as directly associated with a set of costs as the other two dimensions. In short, the findings discussed here lend further support to the rationale for relegating precision to the least consequential component of legalization. Beyond the key legalization variables, it is noteworthy that again, the Cold War negatively impacts whether an agreement was concluded or entered into force, which is broadly consistent with the earlier finding. Similarly, as institutional context increased the speed with which an agreement enters into force, it also made the likelihood of reaching an agreement and that agreement entering into force less likely. As in the duration analysis, the presence of successor agreement made conclusion of a subsequent agreement more likely, even if not affecting EIF. Finally, the presence of a great power sponsoring the agreement made ultimate EIF more likely, suggesting that even if great power sponsorship does not facilitate other aspects of arms control agreements, on the most basic measure of whether it eventually facilitates EIF, irrespective of time, it has a positive impact. To investigate the robustness of these results, I carried out a number of checks. To cope with the potential subjectivity associated with hand-coding the arms control agreements based on legalization (Hopkins and King 2010), two other sets of coders were also deployed to code the agreements along the three dimensions of obligation, precision, and delegation, using the same criteria outlined in Table 1 above. I also ran a number of secondary robustness checks, including creating ordinal variables for several of the variables that were dummies in the preliminary analysis—including multilateralism and depth—and removed the Cold War dummy. The results remained the same, suggesting that they do not vary depending on particular coding decisions; these results are reported in the Appendix. Moreover, the results presented without control variables increase the confidence that the findings are not driven by the choice of controls. The Iran Agreement as an Illustrative Case As these findings suggest, the most important legalization variables are obligation and to a lesser extent delegation, with precision being inconsequential for reasons posited earlier. While a full case study illustrating these findings is beyond the scope of this article, the recent Iran nuclear deal (called the Joint Comprehensive Plan of Action) helps offer a concrete illustration of the concept, measurement, and consequences of obligation, delegation, and precision in the arms control context. Consistent with the findings outlined here, the two most significant sticking points—affecting both the duration and ultimate success of the negotiations—fell under the heading of obligation and delegation. Reaching an agreement required softening legalization along each of these dimensions, especially obligation. By contrast, high degrees of precision were not an obstacle, corroborating the earlier finding that hard legalization along the dimension of precision does not tend to obstruct cooperative outcomes. In terms of obligation, the agreement essentially has an elaborate set of sunset provisions. As a diplomat who helped negotiate the deal observed (Nephew 2015), “Iran has made it clear that any future nuclear restrictions it agrees to will have to be limited and time-bound,” indicating that “sunset provisions are a common feature of international arms control.” Thus, whereas the United States and its allies might have favored a set of restrictions that continued indefinitely, Iran insisted on being able to hedge against medium- to longer-term uncertainties about its security situation rather than agreeing to stringent nuclear restrictions in perpetuity. Based on the agreement, Iran will enrich uranium only up to 3.67 percent, can begin testing up to 30 IR-6 and IR-8 centrifuges after 8.5 years, will keep its enrichment capacity at Natanz up to a total of 5060 IR-1 centrifuges for 10 years, and will not have heavy water production, spent fuel reprocessing, or uranium stockpile development for 15 years. After each of these periods, Iran is entitled to continue each of these activities. In short, the agreement does not dismantle the program indefinitely, a softening of legalization along the dimension of obligation that made an agreement possible. Consistent with the findings of this research, hard legalization along the dimension of delegation proved to be less of an obstacle, although the negotiating parties did include provisions that softened aspects of the inspection regime’s intrusiveness. The agreement provides for “unprecedented verification measures” that can “establish with near-perfect confidence that the agreement is being respected” (Parmentola and Postol 2015). As part of these provisions, the IAEA has access to declared nuclear sites such as Fordow and Natanz and the number of inspectors increases from the current number of 50 up to 150. For undeclared sites, however, inspectors will come from countries with whom Iran has diplomatic relations. Iran is able to restrict IAEA inspectors at the Parchin military site, where it has previously tested explosives, and Iranians will be able to play a role in inspections that do occur. Against the backdrop of a quite rigorous inspection regime, some sovereignty-preserving concessions appeared to be necessary to reach an agreement, which is also consistent with the more qualified support for the delegation variable outlined earlier. The dimension along which negotiators did not back away at all was precision, and indeed, the specificity of the agreement is extremely high. Instead of stating qualitatively that Iran had to keep its uranium enrichment significantly below the enrichment level needed to create a bomb, the agreement states a specific number. Similarly, the agreement acknowledges that Iran needs tens of thousands of centrifuges to create highly enriched uranium for a bomb and currently has about 20,000 centrifuges between their Natanz and Fordow facilities. The agreement requires Iran to reduce its centrifuges not to a level below that needed to make a bomb but precisely to 6,10417 (White House 2015). Iran must reduce its stockpile of uranium by 98 percent and keep its level of uranium enrichment at 3.67 percent. These examples suggest that high precision was not an impediment to reaching a cooperation agreement compared to obligation, where the negotiating parties crafted a set of sunset clauses, and to a lesser extent delegation, where the parties reduced some of the sovereignty costs associated with the inspection regime. In sum, after 22 months of arduous negotiations, the P5 + 1 concluded an agreement but did so by virtue of softening two aspects of legalization in ways consistent with the statistical findings presented earlier. In this case, it was a matter of stepping back from an indefinite commitment in terms of time (high obligation) and to a lesser extent from airtight inspection regime (high delegation), not by having to pursue vaguer, less precise terms of the agreement (Jervis 2015). Taken together, the sketch helps situate the analysis and findings of this research in the context of the recent Iran nuclear deal, offering some concrete implications for foreign policy. Conclusion While scholars have written at length about institutional design in the areas of trade and the environment (Milner and Rosendorff 2001; Von Stein 2008) or about particular arms control agreements (Koremenos 2001), this is the first study to empirically test the logic of legalization in the context of all arms control agreements. As the discussion suggests, cooperation on arms control presents states with a conundrum. States have little interest in initiating negotiations on a problem of international cooperation if the regime is not effective, defined as being able to “embody, and affect actors’ expectations…and make governments concerned about precedents” (Axelrod and Keohane 1985, 234). To shape expectations, the constraints and commitments must be credible, but the prospect of becoming involved in a highly legalized agreement that would make commitments credible might also make states leery of concluding an agreement at all, or at least bargaining harder and longer to produce an agreement most advantageous to its own interests, which can mean delaying the cooperation objectives that led states to propose an agreement in the first place (Fearon 1998, 270; Milner and Rosendorff 2001, 836). With this conundrum in mind, this research asked how the dimensions of legalization that would help foster longer-term cooperation affect the shorter-term prospects of concluding an agreement and that agreement going into force. Based on this study of all arms control agreements proposed and negotiated between 1945 and the present, I find that particular legalization variables make bargaining over the terms of cooperation more difficult. Obligation consistently obstructs the conclusion of an agreement and entry into force, which is consistent with previous research showing that flexibility in terms of obligation can alleviate the impediments to reaching an agreement (Milner and Rosendorff 2001) since mechanisms such as escape and withdrawal clauses may be particularly well positioned to help cope with future uncertainty (Kucik and Reinhardt 2008), giving states an out when circumstances change in the future. In much more limited circumstances, namely the time for entry into force, delegation also discourages states from participating in an arms control agreement. Delegation in the context of arms control may generate more information, which could make it attractive in circumstances where states face uncertainty and risk (Koremenos 2007), but does so at a cost, namely in terms of sovereignty. Moreover, determined cheaters will be able to evade verification regimes, reducing the advantages in terms of whether delegation can be effective in assuaging uncertainty. Precision, relegated in other accounts of institutional design (Abbott et al. 2000, 22), also takes a backseat to the other variables in this analysis. In short, states may be concerned about opportunism in terms of enforcement, but are more concerned about tying their hands in the face of future uncertainty through highly formalized forms of arms control cooperation, which is why they value features of an agreement that give them ways to opt out if new security threats present themselves in the future. This should not suggest that legalization is less helpful in enforcing that cooperation once the terms of cooperation have been successfully negotiated, but that they can impede the initial stage of cooperation, delaying and in some cases preventing altogether the successful negotiation of an agreement. By clarifying which institutional design features are relevant in the context of negotiating arms control agreements successfully, this research has important implications for policy, such as for crafting the types of agreements that are likely to affect vertical or horizontal proliferation. The findings suggest that international negotiators might be better served drafting less highly legalized agreements that offer latitude in states’ commitment to the agreement, since the prospect of tying their hands will discourage states from engaging in higher-obligation commitments. Backing away from aggressive delegation measures might also help states buy into an agreement, a dynamic illustrated by experiences of Cold War arms control negotiations in which the Soviets in particular were leery of the intrusiveness of on-site inspections (Miller 1984, 720). Focusing on precision is likely to offer less purchase in terms of reaching an international agreement and that agreement entering into force. On the contrary, as the Iran example suggests, negotiators can include quite high precision and still have confidence that the agreement will be successful. When it comes to questions of policy and drafting specific agreements such as the Fissile Material Cutoff Treaty, a proposed binding treaty that would ban fissile material but has languished, this analysis also offers some guidance. In particular, it suggests that easing the degree of obligation, such as by offering escape clauses or even making the agreement more voluntary for the negotiating phase, is more likely to produce a successful outcome in terms of concluding the agreement, which can then foster the conditions under which states accomplish more highly legalized agreements and achieve their arms control objectives over the longer term (Gold 1983). While a first step in systematically evaluating legalization in the context of arms control agreements, it is not the last word. The focus on the attributes of an agreement that facilitate cooperation in terms of concluding the agreement and that agreement entering into force raises questions about cooperation in terms of compliance. Thus, an important next step would be to assess whether state behavior is consistent with arms control agreements that have entered into force, and whether certain attributes of legalization are more closely correlated with compliance. Similarly, future research should evaluate how particular attributes of legalization interact with different states, and whether particular types of states are more receptive to participating in and complying with some types of agreements than others. This analysis focused on the agreements themselves, but if prior research is any guide, factors such as regime type (Hafner-Burton, Mansfield, and Pevehouse 2015; Haftel and Thompson 2013a) and nuclear status (Fuhrmann and Li 2008; Way and Sasikumar 2009) may affect international cooperation on arms control agreements. I leave such questions for subsequent research. Footnotes 1 According to Goldblat (1994, 3), limiting proliferation and reducing arsenals are two types of arms control objectives, which is why I sometimes refer to agreements in terms of proliferation and reductions, but the overall category for this analysis is nuclear arms control. 2 See Brown 2011 for a discussion of IAEA verification problems in the context of Iraq in 1991 and Iran today. This concern, however, applies largely to nuclear weapons proliferation rather than nuclear testing, where national technical means can be valuable. 3 The US State Department Office of the Historian highlights the importance of inspections from the standpoint of the United States and United Kingdom; see https://history.state.gov/milestones/1961-1968/limited-ban 4 The Appendix lists all arms control agreements in the data set. 5 Author interviews with James Goodby, who negotiated the Limited Test Ban Treaty for both Eisenhower and Kennedy, revealed that agreements are proposed based not on likelihood of success but on strategic need. Negotiators then determine appropriate legal contours, implying that the idea precedes the legal arrangement, not vice versa. 6 While it would be interesting to consider various drafts of the same agreement, this approach could be problematic since negotiations take place under a cloak of secrecy and we could not be certain that drafts emerge in a systematically unbiased way. Future work might investigate the iterations of negotiations either within one agreement or across agreements, including the associated time-varying measures. 7 The 1969 Vienna Convention on Law of the Treaties (Article 7) indicates that a state representative can perform acts related to the conclusion of a treaty; these members include heads of state, ministers of foreign affairs, heads of diplomatic missions, and “representatives accredited by States to an international conference…for the purpose of adopting the text of a treaty.” 8 “Increasing Pressure on the Nine CTBT Hold-Outs at the UN,” http://www.ctbto.org/press-centre/highlights/2008/increasing-pressure-on-the-nine-ctbt-hold-outs-at-the-united-nations/. 9 To ascertain that the proportional hazards assumption is appropriate, I ran a global test based on Schoenfeld residuals (following Box-Steffensmeier, Reiter, and Zorn 2003, 36, and Fortna 2004, 282) and determined that the proportional hazard assumption is not violated by the models and variables. 10 For an analogous approach, see Hafner-Burton, Mansfield, and Pevehouse (2015), who code human-rights institutions based on obligation, precision, and delegation. 11 Treaty of Tlatelolco available at http://www.state.gov/p/wha/rls/70658.htm. 12 New START, http://www.state.gov/t/avc/newstart/index.htm. 13 Start I treaty, http://fas.org/nuke/control/start1/text/. 14 Treaty Banning Nuclear Weapon Tests in the Atmosphere, in Outer Space and Under Water, available at http://www.state.gov/t/isn/4797.htm. 15 Institutional context and successor might seem to be related, but the correlation between the two is .31. 16 See Appendix for descriptive statistics. 17 See White House Iran Deal Facts, available at https://www.whitehouse.gov/issues/foreign-policy/iran-deal. References Abbott Kenneth, Snidal Duncan. 2000. “ Hard and Soft Law in International Governance.” International Organization  54: 421– 56. Google Scholar CrossRef Search ADS   Abbott Kenneth, Keohane Robert, Moravcsik Andrew, Slaughter Anne-Marie, Snidal Duncan. 2000. “ The Concept of Legalization.” International Organization  54: 17– 35. Axelrod Robert, Keohane Robert. 1985. “ Cooperation under Anarchy: Strategies and Institutions.” World Politics  38: 226– 54. Google Scholar CrossRef Search ADS   Barrett Scott. 1998. “ On the Theory and Diplomacy of Environmental Treaty-Making.” Environmental and Resource Economics  13: 317– 33. Google Scholar CrossRef Search ADS   Belcher Emma. 2014. Origins of the Nuclear Suppliers Group as a Gentlemen’s Agreement. Unpublished manuscript. Bohlen Avis. 2009. “ Arms Control in the Cold War.” Foreign Policy Research Institute , http://www.fpri.org/articles/2009/05/arms-control-cold-war. Box-Steffensmeier Janet, Reiter Dan, Zorn Christopher. 2003. “ Nonproportional Hazards and Event History Analysis in International Relations.” Journal of Conflict Resolution  47: 33– 53. Google Scholar CrossRef Search ADS   Braun Chaim, Chyba Christopher. 2004. “ Proliferation Rings: New Challenges to the Nuclear Nonproliferation Regime.” International Security  29: 5– 49. Google Scholar CrossRef Search ADS   Brown Robert. 2010. “ Measuring Delegation.” Review of International Organizations  5: 141– 75. Google Scholar CrossRef Search ADS   Brown Robert. 2011. “ International Nonproliferation: Why Delegate to the International Atomic Energy Agency.” In Causes and Consequences of Nuclear Proliferation , edited by Rauchhaus Robert, Kroenig Matthew, Gartzke Erik. New York: Routledge. Copelovitch Mark, Putnam Tonya. 2014. “ Design in Context: Existing International Agreements and New Cooperation.” International Organization  68: 471– 93. Google Scholar CrossRef Search ADS   Dahl Fredrik. 2012. “ Iran Has Expanded Sensitive Nuclear Work: UN Agency.” Reuters . Accessed April 30, 2016. http://www.reuters.com/article/us-nuclear-iran-iaea-idUSTRE81N0HS20120224. Downs Anthony, Rocke David, Barsoom Peter. 1996. “ Is the Good News about Compliance Good News about Cooperation?” International Organization  50: 379– 406. Google Scholar CrossRef Search ADS   Fearon James. 1998. “ Bargaining, Enforcement, and International Cooperation.” International Cooperation  52: 269– 305. Fortna Page. 2004. Peace Time: Cease-Fire Agreements and the Durability of Peace . Princeton, NJ: Princeton University Press. Fuhrmann Matthew. 2012. Atomic Assistance: How “Atoms for Peace” Programs Cause Nuclear Insecurity . Ithaca, NY: Cornell University Press. Fuhrmann Matthew, Li Xiaojun. 2008. “ Rethinking Ratification: Legalization and Nuclear Weapon Free Zone Treaties.” Working Paper. Gold Joseph. 1983. “ Strengthening the Soft International Law of Exchange Arrangements.” American Journal of International Law  77: 443– 89. Google Scholar CrossRef Search ADS   Goldblat Josef. 1994. Arms Control: A Guide to Negotiations and Agreements . Oslo, Norway: Sage. Goldstein Judith, Kahler Miles, Keohane Robert, Slaughter Anne-Marie. 2001. Legalization and World Politics . Cambridge, MA: MIT Press. Guzman Andrew. 2002. “ A Compliance-Based Theory of International Law.” California Law Review  90: 1826– 1888. Google Scholar CrossRef Search ADS   Guzman Andrew, Meyer Timothy. 2010. “ International Soft Law.” Journal of Legal Analysis  2: 171– 225. Google Scholar CrossRef Search ADS   Hafner-Burton Emilie, Mansfield Edward, Pevehouse Jon. 2015. “ Human Rights Institutions, Sovereignty Costs, and Democratization.” British Journal of Political Science  45: 1– 27. Google Scholar CrossRef Search ADS   Haftel Yoram. 2010. “ Ratification Counts: US Investment Treaties and FDI Flows into Developing Countries.” Review of International Political Economy  17: 348– 77. Google Scholar CrossRef Search ADS   Haftel Yoram, Thompson Alexander. 2013a. “ Delayed Ratification: The Domestic Fate of Bilateral Investment Treaties.” International Organization  67: 355– 87. Google Scholar CrossRef Search ADS   Haftel Yoram, Thompson Alexander. 2013b. “ When Do States Renegotiate International Agreements: The Case of Bilateral Investment Treaties.” Hawkins Darren, Lake David, Nielson Daniel, Tierney Michael. 2006. Delegation and Agency in International Organizations . New York: Cambridge University Press. Google Scholar CrossRef Search ADS   Helfer Laurence. 2012. “ Flexibility in International Agreements.” In Interdiscilpinary Perspectives on International Law and International Relations , edited by Dunoff Juffrey and Pollack Mark, 175– 96. New York, NY: Cambridge University Press. Google Scholar CrossRef Search ADS   Hopkins Dan, King Gary. 2010. “ A Method of Automated Nonparametric Content Analysis for Social Science.” American Journal of Political Science  54: 229– 47. Google Scholar CrossRef Search ADS   ILM (International Legal Materials). 1989. “ Union of Soviet Socialist Republics-United States: Agreement on the Prevention of Dangerous Military Activities.” International Legal Materials  28: 877– 95. Jervis Robert. 1978. “ Cooperation under the Security Dilemma.” World Politics  30: 167– 214. Google Scholar CrossRef Search ADS   Jervis Robert. 1990. The Meaning of the Nuclear Revolution: Statecraft and the Prospect of Armageddon . Ithaca, NY: Cornell University Press. Jervis Robert. 2015. “ Turn Down for What? The Iran Deal and What Will Follow.” ForeignAffairs.com. Kelley Judith, Pevehouse Jon. 2015. “ An Opportunity Cost Theory of US Treaty Behavior.” International Studies Quarterly  59: 531– 43. Google Scholar CrossRef Search ADS   Keohane Robert. 1990. “ Multilateralism: An Agenda for Research.” International Journal  44: 731– 64. Google Scholar CrossRef Search ADS   Keohane Robert, Martin Lisa. 1995. “ The Promise of Institutionalist Theory.” International Security  20: 39– 51. Google Scholar CrossRef Search ADS   Kimball Daryl. 2013. “ Fissile Material Cut-Off Treaty at a Glance.” Arms Control Association , August. Koremenos Barbara. 2001. “ Loosening the Ties That Bind: A Learning Model of Agreement Flexibility.” International Organization  55: 289– 325. Google Scholar CrossRef Search ADS   Koremenos Barbara. 2007. “ If Only Half of International Agreements Have Dispute Resolution Provisions, Which Half Needs Explaining?” Journal of Legal Studies  36: 189– 212. Google Scholar CrossRef Search ADS   Koremenos Barbara, Lipson Charles, Snidal Duncan. 2001. “ The Rational Design of International Institutions.” International Organization  55: 761– 99. Google Scholar CrossRef Search ADS   Kroenig Matthew. 2014. “ Force of Friendship? Explaining Great Power Nonproliferation Policy.” Security Studies  23: 1– 32. Google Scholar CrossRef Search ADS   Kucik Jeffrey, Reinhardt Eric. 2008. “ Does Flexibility Promote Cooperation? An Application to the Global Trade Regime.” International Organization  62: 477– 505. Google Scholar CrossRef Search ADS   Kydd Andrew. 2000. “ Arms Races and Arms Control: Modeling the Hawk Perspective.” American Journal of Political Science  44: 228– 44. Google Scholar CrossRef Search ADS   Lipson Charles. 1984. “ International Cooperation in Economic and Security Affairs.” World Politics  37: 1– 23. Google Scholar CrossRef Search ADS   Lipson Charles. 1991. “ Why Are Some International Agreements Informal?” International Organization  45: 495– 538. Google Scholar CrossRef Search ADS   MacDonald Julia. 2015. “ Eisenhower’s Scientists: Policy Entrepreneurs and the Test-Ban Debate 1954–1958.” Foreign Policy Analysis  11: 1– 15. Google Scholar CrossRef Search ADS   Martin Lisa. 2000. Democratic Commitments: Legislatures and International Cooperation . Princeton, NJ: Princeton University Press. Google Scholar CrossRef Search ADS   Mearsheimer John. 1994/95. “ The False Promise of International Institutions.” International Security  19: 5– 49. Google Scholar CrossRef Search ADS   Miller Steven. 1984. “Politics over Promise: Domestic Impediments to Arms Control.” International Security 8: 67–90. Milner Helen V, Rosendorff B. Peter. 2001. “ Democratic Politics and International Trade Negotiations.” The Journal of Conflict Resolution  47: 117– 46. Mistry Dinshaw. 2005. Containing Missile Proliferation . Seattle: University of Washington Press. Mitchell Sara McLaughlin, Hensel Paul. 2007. “ International Institutions and Compliance with Agreements.” American Journal of Political Science  51: 721– 37. Google Scholar CrossRef Search ADS   Nelson Amy. 2013. “Arms Control and Uncertainty: A New Dataset and Its Implications for Arms Control Agreements.” Unpublished manuscript. Nephew Richard. 2015. “ False Flag: The Bogus Uproar over Iran’s Nuclear Sunset.” National Interest , March 6. Oye Kenneth. 1985. “ Explaining Cooperation under Anarchy: Hypotheses and Strategies.” World Politics  38: 1– 24. Google Scholar CrossRef Search ADS   Parmentola John, Postol Theodore. 2015. “ How to Make Iran Inspections Work.” Politico , August 31. Pelc Krzysztof. 2013. “ The Cost of Wiggle-Room: Looking at the Welfare Effects of Flexibility in Tariff Rates at the WTO.” International Studies Quarterly  53: 349– 68. Google Scholar CrossRef Search ADS   Raustiala Kal. 2005. “ Form and Substance in International Agreements.” American Journal of International Law  99: 581– 614. Google Scholar CrossRef Search ADS   Regan Patrick, Stam Allan. 2000. “ In the Nick of Time: Conflict Management, Mediation Timing, and the Duration of Interstate Disputes.” International Studies Quarterly  44: 239– 60. Google Scholar CrossRef Search ADS   Schneider Christina, Urpelainen Johannes. 2013. “ Distributional Conflict Between Powerful States and International Treaty Ratification.” International Studies Quarterly  57: 13– 27. Google Scholar CrossRef Search ADS   Shaffer Gregory, Pollack Mark A.. 2011. “ Hard Law, Soft Law, and International Security: The Cases of Nuclear Weapons and the Responsibility to Protect.” Boston College Law Review  52: 1147– 1241. Simmons Beth. 2009. Mobilizing for Human Rights: International Law in Domestic Politics . New York: Cambridge University Press. Google Scholar CrossRef Search ADS   Simmons Beth, Danner Allison. 2010. “ Credible Commitments and the International Criminal Court.” International Organization  64: 225– 56. Google Scholar CrossRef Search ADS   The Economist, “ An Uneven Deal,” November 12, 2014. Tzeng Peter. 2014. “ Nuclear Leverage.” Nonproliferation Review  20: 473– 92. Google Scholar CrossRef Search ADS   Vaynman Jane. 2014. “ Enemies in Agreement: Domestic Politics, Uncertainty, and Cooperation between Adversaries.” Unpublished manuscript. Von Stein Jana. 2008. “ The International Law and Politics of Climate Change.” Journal of Conflict Resolution  52: 243– 68. Google Scholar CrossRef Search ADS   Wallace Geoffrey. 2014. “ International Law and Public Attitudes Toward Torture: An Experimental Study.” International Organization  67: 105- 40. Google Scholar CrossRef Search ADS   Way Christopher, Sasikumar Karthika. 2009. “ Testing Theories of Proliferation: The Case of Nuclear South Asia.” In Inside Nuclear South Asia , edited by Sagan Scott D.. Stanford, CA: Stanford University Press. Wendt Alexander. 2001. “ Driving with the Rearview Mirror: On the Rational Science of Institutional Design.” International Organization  55: 1019– 1049. Google Scholar CrossRef Search ADS   White House. 2015. “ The Historic Deal that Will Prevent Iran from Acquiring a Nuclear Weapon.” Accessed April 30, 2016. https://www.whitehouse.gov/issues/foreign-policy/iran-deal. Woolf Amy. 2011. Monitoring and Verification in Arms Control . Washington, DC: Congressional Research Service. Woolf Amy, Nikitin Mary Beth, Kerr Paul. 2013. Arms Control and Nonproliferation: A Catalog of Treaties and Agreements . Washington, DC: Congressional Research Service. © The Author (2016). Published by Oxford University Press on behalf of the International Studies Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

Journal

Foreign Policy AnalysisOxford University Press

Published: Jan 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve Freelancer

DeepDyve Pro

Price
FREE
$49/month

$360/year
Save searches from
Google Scholar,
PubMed
Create lists to
organize your research
Export lists, citations
Read DeepDyve articles
Abstract access only
Unlimited access to over
18 million full-text articles
Print
20 pages/month
PDF Discount
20% off