Paper Tiger or Barrier to Proliferation? What Accessions Reveal about NPT Effectiveness. San Jose, CA Ithaca, NY 14853

Transcription

1 Paper Tiger or Barrier to Proliferation? What Accessions Reveal about NPT Effectiveness Karthika Sasikumar Christopher Way Assistant Professor Associate Professor Department of Political Science Department of Government San Jose State University Cornell University San Jose, CA Ithaca, NY Paper to be delivered at CISAC, Stanford University, April 15, 2010

2 Abstract Despite declarations by world leaders about the need to halt the proliferation of nuclear weapons, and the increased controversy about the effectiveness of the Nuclear Non- Proliferation Treaty (NPT) regime, surprisingly little systematic scholarship has investigated this treaty. Is the NPT a paper tiger, as critics claim, or the foundation of a successful anti-proliferation strategy? To answer this question, we need to understand why countries sign the NPT in the first place or hesitate to do so. If the NPT is a paper tiger that may even aid countries in advancing their nuclear technology, as critics allege, why did so many countries hesitate to sign it? On the other hand, if the treaty imposes real restrictions, as advocates claim, then why do countries that could potentially benefit from a nuclear deterrent nonetheless sign away their right to make the bomb? Understanding who signs the treaty, why, and when is an important step towards a better understanding of how the NPT functions. We develop a simple cost-benefit framework to understand the pros and cons of joining the NPT. We evaluate our framework with statistical models of accessions to the NPT, covering 187 countries from 1968 to Results suggest that eagerness to join the NPT is strongly associated with level of economic development, high energy needs, external threat environment, and lack of great power security guarantees. Our results strengthen the idea that the NPT is not a paper tiger but a barrier to proliferation: states certainly act as if the NPT poses a serious constraint.

3 Page 1 1. INTRODUCTION The Nuclear Nonproliferation Treaty (NPT) was concluded at the end of the 1960s, a decade which saw the drama of the Cuban Missile Crisis, the height of the nuclear arms race between the superpowers, and the entry of France and China into the club of countries that had tested nuclear devices. The basic bargain underlying the NPT allows countries to surrender their right to develop nuclear weapons in return for access to international assistance in civilian nuclear technology. Five countries (the United States, the Soviet Union, the United Kingdom, France, and China) that had tested nuclear devices before 1 January 1967, were conferred the status of Nuclear Weapon State (NWS). All other signatories (Non Nuclear Weapon States or NNWSs) pledged to abjure the development and diffusion of nuclear weapons technology. Events in recent years have heightened debate over the utility of the NPT as a means of stemming nuclear proliferation. In the Persian Gulf, European countries placed renewed faith in an Iranian re-commitment to the NPT at the center of their strategy for stopping Teheran s efforts to acquire nuclear arms. 1 The reinforcement of multilateral agreements such as the NPT is at the center of the European Union s strategy against weapons of mass destruction, unveiled in December Meanwhile, on the Korean peninsula, the relentless pursuit of nuclear arms by a former NPT signatory seems to underscore the futility of the regime. 2 North Korean actions point to a weakness of the regime long emphasized by critics: countries can make clear progress towards the nuclear threshold under NPT auspices and then simply withdraw shortly before assembling weapons. In July 2005, the US and India announced an agreement on nuclear cooperation. India, though not a party to the NPT, had violated international norms with its nuclear tests of May The agreement was therefore widely perceived as strengthening the

4 Page 2 legitimacy of India s nuclear capability and lowering that of the NPT. Has the NPT effectively slowed proliferation, or do recent events point to its futility? Assessing the NPT s effectiveness is difficult because almost every country has signed it, and very few countries have tried to acquire nuclear weapons. We attempt to get around this difficulty by asking another, closely related question: what explains the large variance in willingness to sign the NPT? Why did some states accede quickly whereas others dragged their feet for decades? What are the costs and benefits that states weigh when deciding whether or not to join the regime? In exploring the patterns of NPT accession, if we find that states decisions do follow the logic of the NPT s fundamental bargain, then we can infer that accessions to the treaty are meaningful and that it is perceived as a costly constraint by potential proliferators. That is, patterns of accession to the NPT might function as a screening mechanism that separates countries that are considering nuclear weapons from those that are not. On the other hand, if states view the NPT as irrelevant and/or ineffective, we would expect to see no clear pattern in accession. States that might desire nuclear weapons, in this second scenario, may be among the forefront in signing up, whereas states without pressing security concerns might lag behind. Understanding who signs the treaty, why, and when is an important step towards a better understanding of how the NPT functions and whether it works as designed. Systematic studies of decisions to join the NPT are lacking. Single country studies often chronicle the debates about the NPT in great detail. Yet there have been few, if any, comparative cross-country studies designed to provide generalizable answers to these questions. We seek to fill this void by presenting a simple cost-benefit framework for understanding decisions to join (or not to join) the NPT, and by testing our framework with statistical tests that span the entire life of the NPT and include all potential signatory countries. To this end, we estimate event history models of the decision to join the NPT, using data covering 185 potential signatories spanning the time period.

5 Page 3 Our results show that in choosing whether to accede to the NPT, states do follow the logic embodied in the treaty s design. Security and energy are the most important costs making countries hesitant to sign the NPT. States with acute security concerns are less likely to join the NPT quickly, states with high energy needs are more likely to join it, and states for whom developing nuclear weapons would be a simple and inexpensive task are more reluctant to join. Specifically, states facing a challenging security environment are only about half as likely to join the NPT in a given year, although a defense pact with a great power can largely cancel out this effect. States at a higher level of economic development (those for whom developing nuclear weapons is highly feasible) are less likely to join quickly compared to less developed states, whereas states with high electricity generation needs join more quickly. These patterns correspond quite well with the logic underlying the NPT. States paying higher opportunity costs in giving up the nuclear option -- that is, those facing severe security challenges and for whom developing and fielding weapons would be relatively easy -- are cautious about signing on. On the other hand, states giving up little by forgoing nuclear weapons -- those enjoying benign security environments or for whom developing weapons would be technologically too difficult or too expensive -- sign on relatively quickly. Although we cannot trace the causal mechanisms in these cases in such a large-n study, these results strengthen the idea that the NPT is not a paper tiger but a barrier to proliferation: states certainly act as if the NPT poses a serious constraint. The next section outlines our cost-benefit framework, describing the potential economic, security, and political costs and benefits of joining the NPT. After that we briefly describe the data, including the measures of costs and benefits, and our dependent variable: ratification of the treaty. We then describe the results of event history models of the decision to join the NPP, followed by an investigation of potentially influential observations and outliers. By way of a conclusion, we discuss the implications of our

6 Page 4 findings for understanding the mechanisms through which the NPT works, as well as and for crafting strategies to strengthen the treaty. 2. TO SIGN OR NOT TO SIGN: A COST-BENEFIT FRAMEWORK Overview of the NPT To investigate the spread of the NPT, we adopt a cost-benefit framework in which we imagine states weighing the pros and cons of accession. As a preliminary, let us briefly review the core of the NPT bargain. The impetus for the NPT grew out of dissatisfaction with the Atoms for Peace policies promoted by the Eisenhower administration during the 1950s. Under this program, the United States offered unrestricted access to nuclear fuel in exchange for the promise that it be used only for peaceful purposes. The Atoms for Peace framework established the core bargain that would underpin the NPT: countries that give up the military potential of the atom should be able to enjoy the full peaceful benefits of the atom. To enlist multilateral support for this initiative, President Eisenhower supported vesting oversight tasks in a United Nations agency; this led in 1957 to the creation of the International Atomic Energy Agency (IAEA). Part of its charge was to ensure the peaceful use of nuclear materials transferred under Atoms for Peace. However, shortcomings in the Atoms for Peace/IAEA framework soon became apparent. By focusing only on nuclear materials transferred explicitly under specific agreements, the arrangement neglected to regulate technology, material, and knowledge developed indigenously (or copied from transferred material). The emergence of France and then China as nuclear powers was an obvious signal of the need for a more robust and wide-ranging agreement. At the initiative of the United States, United Kingdom, and Soviet Union, a new framework designed to monitor and regulate all nuclear material was negotiated. 3 This became the Nuclear Non-Proliferation Treaty, which opened for

7 Page 5 signatures in 1968 and entered into force in Under the treaty, non-nuclear weapons states must accept safeguards -- intended to detect the transfer of nuclear materials from peaceful to military uses -- on all nuclear materials in their territory, no matter what its source. 4 The costs and benefits of joining the NPT laid out in the text of the treaty itself are quite straightforward. On the cost side, non-nuclear states give up the right to possess or create nuclear weapons. 5 In return, parties agree to the fullest possible exchange of knowledge, material, and equipment useful for peaceful purposes. In addition, some hope of taming vertical nuclear proliferation is offered by the NWS pledge in Article VI to negotiate towards ending the nuclear arms race and to pursue international nuclear disarmament. Given these costs and benefits, what types of countries should rush to sign up, and what type should resist? Here we make two observations. First, the weighing of costs and benefits is more complicated than suggested by this simple statement for several reasons. Beyond the formal stipulations of the treaty, powerful states create informal costs and benefits by their actions. They may, for example, link the NPT to other issues, allowing it to affect diplomacy more broadly. Moreover, the calculation of the security and economic costs is mediated by beliefs and domestic politics: the specific context of a country and its domestic politics are likely to shape its calculation of costs and benefits. Second, since the treaty offers a generic bargain to a heterogeneous group of states, its appeal will vary widely according to the security and economic situations of specific states. To develop a more fine-grained assessment of the relative attractiveness of the NPT, we turn to a more nuanced and comprehensive accounting of costs and benefits under three headings: security, economic, and political.

8 Page 6 Costs of Signing the NPT Security Costs Signing any international agreement imposes some costs on states because it implies a surrender of sovereignty. In the case of security regimes, the most significant cost is the loss of long-term strategic freedom. Nuclear weapons are often seen as a way for weaker powers to level the playing field with more powerful rivals or to provide an effective deterrent at relatively low expense {{99 Kapur,A. 2001;48 Potter,W.C. 1982; 50 Quester,G.H. 1973; 51 Quester,G.H. 1977; }}. However, this does not translate into a real burden for many NPT signatories, who never had any intention of developing nuclear weapons, or lacked the ability to do so at reasonable expense. For some states, the security environment is not sufficiently threatening to make costly acquisition of nuclear weapons worthwhile. For others, the economic burden of developing weapons is too great to contemplate; many small undeveloped countries are not giving up a realistic option in eschewing nuclear weapons. Why might these countries nonetheless hesitate to sign the NPT? The reason is that, for them, hesitating to sign the NPT may actually provide a way to extract benefits from the United States and other treaty promoters. They could use their accession as a bargaining chip, wresting concessions from great powers who have an interest in controlling proliferation in exchange for giving up their freedom of maneuver; thus they retain some incentive to delay signing the treaty. In sum, for states in challenging security environments who possess the wherewithal to develop nuclear weapons, the costs of acceding to the NPT are substantial. However, even states lacking the incentives and ability to develop the bomb may drag their feet a bit in an effort to extract concessions from treaty promoters (chiefly the United States) who are eager to see widespread adoption, although in general such countries should sign on much faster than states facing a genuine security threat.

9 Page 7 Yet even for states with serious security concerns, the costs of forswearing the nuclear option and joining the NPT may be reduced by a defense pact with a nuclear armed-great power. In this view, the acquisition of nuclear weapons and forging of alliances serve as substitutes in the quest for security {{8 Betts,R.K. 1993;83 Davis,Z.S. 1993; 65 Thayer,B.A. 1995; }}. With the backing of a nuclear-armed great power ally, states may perceive lower costs to joining the NPT. By committing themselves to the NPT, states also face the possibility of exit costs should they choose to leave the treaty in the future or punishment costs should they choose to violate it. Aside from the loss of reputation in the international arena, there are specific legal consequences attached to the violation of the NPT. Although the NPT itself has no provisions to enforce compliance, all members sign an enforcement agreement with the IAEA. In the extreme case, violations uncovered by the IAEA can serve as a legal justification for international military action against the transgressor state. When North Korea announced its intention to withdraw from the NPT in 1993, the IAEA referred the issue to the United Nations (UN) Security Council. The implication was that action could be taken under Chapter VII of the UN Charter, which permits military intervention by other states in the face of a threat to international peace and security. 6 After two months of negotiations, North Korea ultimately stayed in the NPT, only to declare its withdrawal again in In general, treaty violators can expect signatories to impose economic and diplomatic sanctions upon them. Economic Costs Signing the NPT entails no direct economic costs, with one possible exception: a state with a nuclear power industry counting on large exports may feel hamstrung by the requirement not to export nuclear equipment or material to non-nuclear weapons states except under IAEA safeguards. If a country wished to export nuclear technology and materials to states not submitting to safeguards, the costs of joining the NPT rise. France,

10 Page 8 for example, with the world s largest nuclear energy industry, declined to restrict nuclear exports to IAEA-safeguard countries until 1991 (and finally ratified the NPT in 1992). We also note an important indirect factor. More developed states -- those with greater industrial and scientific infrastructure -- face much lower economic opportunity costs in developing and deploying nuclear weapons. Thus, they are in effect paying a higher price in acceding to the NPT than are less developed states; only the former are giving up a relatively low-cost, eminently feasible option. Because they pay lower opportunity costs (in terms of foregone alternative uses of scarce economic resources) if they develop nuclear weapons, they pay a higher opportunity cost (in terms of the attractiveness and feasibility of the foregone nuclear option) in joining the NPT. Because the economic (opportunity) cost of nuclear weapons is lower for more developed states, 7 the cost of joining the NPT is higher. Political Costs In addition to security and economic costs, signing the NPT implies possible domestic political costs. A government that accedes to the NPT can face the wrath of domestic pro-nuclear constituencies. These could be influential religious leaders (as in Iran, where Ayatollah Ahmad Janati, head of the influential and conservative Guardians Council, has at times been a vigorous opponent of the NPT), the armed forces (as in Pakistan), opposition parties (as in Israel), or important sections of the bureaucracy (as in India). In India, for example, the doyen of Indian strategic analysts, K. Subrahmanyam, wrote in 1985 that the NPT was a means for the nuclear weapons states to legitimize their nuclear arsenals, license further vertical proliferation, and establish hegemony over the development of nuclear technology in the developing world {{147 Subrahmanyam, K. 1985; }}. In a strategic culture such as this, where Subrahmanyam s views are typical, any government that accedes to the NPT exposes itself to substantial symbolic political costs. Thus in April 1994, for instance, the ruling party in India was forced to clarify that it was

11 Page 9 not negotiating a possible signature of the treaty at secret talks with the United States in London. Similarly, as Brazilian and Argentinean policy-makers moved toward renunciation of nuclear weapons, they rejected IAEA inspections in favor of bilateral inspections because of the history of domestic political opposition to the NPT {{146 Reiss, Mitchell 1995; }}. Benefits of signing the NPT Security benefits The NPT does not feature an explicit security guarantee, and this has often been cited by critics as a fatal flaw in its design. Some NWS have, however, given an assurance that they will never use nuclear weapons against a non-nuclear signatory to the treaty. Russia, for example, pledged in 1993 not to use nuclear weapons against NPT signatory countries, and the United States made a similar pledge in In addition, the nuclear weapons states, through a 1968 Security Council Resolution, promised to come to the aid of a NNWS threatened with nuclear weapons. 9 Aside from formal institutional guarantees, the NPT may provide a tool for countries that wish to signal peaceful intentions, perhaps in trying to de-escalate an enduring rivalry. By formally eschewing the nuclear option, a state can credibly demonstrate a commitment to alleviating tensions, potentially easing the security dilemma with its neighbors and avoiding a costly arms race. For example, signing the NPT was seen as a firm signal of resolve to change policy priorities on the part of the new civilian governments in Brazil and Argentina, reducing tensions in their sometimes rocky relationship {{63 Solingen,E. 1998; }}. In a similar fashion, accessions by Germany and by Japan to the treaty were seen in the 1960s as a final marker of their pacification and renunciation of militarism.

12 Page 10 Economic benefits The NPT holds out the promise of access to nuclear energy technology, with all the concomitant economic benefits, to states that would otherwise be unable to develop a civilian nuclear infrastructure. Furthermore, signatories to the NPT, having signaled their peaceful intentions, are more likely to be treated favorably by export control regimes relating to dual-use technology, such as CoCom or the Nuclear Suppliers Group, further facilitating their development of nuclear power as an energy source. Less transparently, it has often been alleged that economic aid, both bilateral and multilateral, is used as a carrot to induce countries to join the NPT regime. When this is the case, states can receive direct economic benefits from accession to the treaty. For instance, the US threatened to block IMF loans to Pakistan until the latter made some movement towards signing the NPT. Countries may also receive concessions in other areas as a reward for signing the NPT. Many observers believe, for example, that the nuclear powers provide NPT signatories with favored access to conventional arms. In short, multilateral loans and aid, trade agreements, and conventional arms trade can all be linked to a county s stance towards the NPT, providing benefits for signatories and posing opportunity costs for hold-outs. Political Benefits On the domestic political front, governments may use the NPT to downgrade the status and power of their rivals, as Fernando Collor and Carlos Menem did in Latin America, and as the last apartheid government did in South Africa. In addition, for governments with an uncertain hold on power, signing an international treaty may provide an effective way to bind their successors. By signing the NPT, governments can raise the costs of pursuing nuclear weapons for future governments. Moreover, domestic antinuclear constituencies may be strengthened by accession, furthering the lock-in effect of making an international commitment. In short, political volatility may actually induce governments to join the NPT.

13 Page 11 Domestic political benefits for joining the NPT may also be greater for governments in democracies. The literatures on the democratic peace and legalization in international politics both suggest that states tend to externalize their domestic conflict resolution procedures. The essence of democracy is rule of law, and it is thus not surprising that domestic political constituencies in democracies would push for greater legalization of the international realm, delivering greater rewards to governments that join and promote legalized international institutions {{141 Kahler, Miles 2000; }}. Finally, for many countries signing the NPT provides status benefits that derive from a sense that they have joined the mainstream of international society. It has been suggested that a nuclear taboo begun to grow steadily in strength fairly soon after the invention of nuclear weapons {{148 Tannenwald, Nina 1999; }}. Especially after the end of the Cold War, attempts to develop or acquire weapons of mass destruction have been seen as a marker of rogueness, in contrast to earlier periods in which acquisition of nuclear arms was often seen as a marker of modernity {{53 Sagan,S.D. 2000; }}. Signing the NPT may thus provide diplomatic benefits in the form of coveted recognition as a modern state; as the treaty spreads across the globe, peer pressure to join mounts 3. DATA Dependent Variable: NPT Ratification Accession to the Nuclear Proliferation Treaty is our dependent variable. 10 We collected data on signing and ratification dates from the United Nations. Our analysis below focuses on ratification dates because the decision to ratify the NPT has often proved particularly contentious and subject to delay. 11 South Korea, for example, signed the treaty quickly in 1969, but deferred ratification to keep its nuclear options open, only ratifying the treaty under extreme pressure from the United States in Figure 1 about here ---

14 Page 12 As a preliminary, let us examine patterns over time and across regions in ratification of the NPT. The treaty opened for signature and ratification in 1968, and by 2001 (the end point of our statistical analysis) nearly all sovereign states had acceded. 12 Figure 1 plots the spread of the NPT across the globe from 1968 to The solid blue line plots the percentage of all possible members that had ratified the treaty by any given year against the left-hand axis. The spike-plot indicates the number of countries joining each year, with the scale shown on the right-hand axis. (A list of acceding countries by year is included in an appendix). As Figure 1 indicates, after an initial rush to ratify the treaty, the number of member states expands in a fairly even progression. The percentage of eligible states belonging to the NPT jumps up from zero to over 40 per cent in the first three years, and then follows a loosely linear expansion until At that time, the percentage briefly dips -- reflecting the increase in the number of eligible states following the collapse of the Soviet Union and the break-up of Yugoslavia -- only to continue its upward path as the successor states signed on quickly. By 2000, over 96 per cent of eligible states were parties to the NPT. The expansion process, however, is not perfectly linear, and some spikes do occur, most notably in 1975, 1992, and The 1975 spike follows the May, 1974 Indian nuclear test at Pokhran, an event which spurred a deepening of the nuclear non-proliferation regime, including the formation of the Nuclear Suppliers Group and the passage of the US Nuclear Non-proliferation Act. In 1992 another rush to join occurs, as Eastern bloc successor states acceded to the NPT in large numbers. 13 Finally, the 1995 spike features a large number of micro-states and coincides with efforts by US diplomats to achieve near universal participation in the treaty Figure 2 about here --- Although the world-wide story is one of steadily increasing coverage, regional differences in the rate of spread are apparent. Figure 2 provides a glimpse at this pattern. For each of five regions, it plots the percentage of eligible states within the region that had ratified for each year. Globally, just over 70 per cent of eligible states had signed the

15 Page 13 treaty by 1980, but this obscures some variation across regions. In Europe, accessions to the treaty spread rapidly, with 90 per cent of eligible states on board by In Africa, Asia, and the Americas accession rates are somewhat slower with about 65 per cent on board by 1980 and 80 percent by 1990 and similar across the three regions. Asia s relatively quick ascent may seem somewhat surprising given the number of well-known NPT hold-outs in the region -- India, China, North Korea, and Burma among them -- yet the majority of Asian countries joined the NPT relatively quickly. The Middle East provides the major outlier, with an accession rate notably below that of the rest of the world, at least up until the 1990s. Only about 55 per cent of eligible states in this strifetorn region had ratified the NPT by 1980, and membership rates only caught up to the other regions as the new millennium approached. Explanatory Variables: Economic, Security, and Political In accord with our cost/benefit framework, we group explanatory variables under the headings of economic, security, and political variables. In keeping with our emphasis on the core logic of the original NPT bargain, we give pride of place to variables emphasizing states security environment and needs for electricity generating capacity. Of course, states join the NPT for different reasons, some of which we cannot operationalize systematically, and the treaty does not offer equal benefits and costs to all states. Still, if there is merely random variation across countries, we would not be able to pick up any systematic effects. It is these potential systematic effects, and in particular those reflecting the core bargain of the NPT, that we hope to identify. -- Table 1 about here Security Variables Enduring Rivalry. Although subjective perceptions of security threats can vary substantially, participation in an enduring rivalry can be safely taken as an indicator of a

16 Page 14 significant security threat, especially since the vast majority of wars and militarized disputes occur in the context of enduring rivalries {{84 Diehl,P.F. 1998}}. Drawing on Bennett s coding for enduring rivalries and their dates {{6 Bennett,S.D. 1998}}, we create a dichotomous variable indicating whether or not a state was involved in one or more enduring rivalries in a given year. Giving up the nuclear option is costly for countries in an enduring rivalry, so we expect a negative coefficient. Frequency of Dispute Involvement. As an alternate measure of intensity of the security threat, we calculated the five-year moving average of the number of militarized interstate disputes per year in which a state is involved, drawing on Version 3.0 of the Militarized Interstate Dispute (MID) dataset {{240 Ghosn,Faten 2003}}. As with enduring rivalries, we expect a negative coefficient: states with a recent history of frequent MIDs pay a higher cost in giving up the atomic option. Security Guarantee. The allure of nuclear weapons as an avenue to security may be attenuated by security assurances in the form of a defense pact from a nuclear-armed great power. 15 A pact thus reduces the costs of acceding to the NPT. Drawing on Singer and Small s standard list of great powers {{126 Singer,J.D. 1982}}we count the United States, the Soviet Union/Russia, the United Kingdom (from 1952), France (from 1960), and China (from 1964) as nuclear-capable great power allies. Economic Variables Level of development. We use two variables to tap level of economic development: gross domestic product per capita and urbanization. Whereas aggregate economic size indicates total resources available, per capita GDP more accurately reflects level of economic development, which is closely linked to the sophisticated technical, engineering, and manufacturing knowledge necessary for the development and construction of nuclear arms. Purchasing power parity GDP data is taken from version 6.1

17 Page 15 of the Penn World Tables {{29 Heston,A. 2002; }}. 16 We supplement this with data on urban population as a share of total population, drawing on version 3.01 of the Correlates of War project s national material capabilities dataset {{149 Singer,J.D. 1987; }}. We expect a negative coefficient on the economic development variables because highly developed countries are giving up a potentially attractive policy option by signing the NPT. Aggregate Economic Size. Total GDP is taken from same sources as per capita GDP. In contrast to per capita GDP, aggregate GDP proxies for the total resources available to the state; if they have large enough populations, even states with very low levels of per capita GDP may be able to acquire nuclear weapons (China providing a prominent example). We expect a positive coefficient on total GDP. Electricity Demand/Needs. Countries with greater electricity generation needs should find the NPT bargain more attractive. To create a proxy for high electricity needs, we regressed electricity generation on variables for size of population and the economy. 17 The residuals from this regression can be interpreted as a proxy for the attractiveness of nuclear energy; countries that are under-producing electricity have more to gain from additional sources and technologies such as nuclear power. In addition, we include a dummy variable for countries that export more than 100,000 barrels of oil; these countries are expected to have a less pressing need for the potential benefits of nuclear power. 18 Research Reactor. We also include a dummy variable for operation of at least one research or commercial nuclear reactor. 19 This could cut either way: countries with nascent nuclear technology may find the assistance offered by the IAEA more attractive, but they may also be loathe to place their reactors under safeguards if the country harbors any weapons ambitions. Political Variables

18 Page 16 Democracy. We use the Polity IV data to see if democracies may be quicker to sign the NPT than authoritarian states {{30 Jaggers,K. 1995; }}. Our measure assesses democracy for each country-year: we create a derived measure of the level of democracy within each state by combining the two separate eleven-point scales for democracy and autocracy from Polity IV: dem i = democ i - autoc i. Involvement in international institutions. Countries that participate widely in international institutions and have absorbed systemic, multilateral norms should be more likely to sign on to the NPT quickly. As a proxy for multilateral proclivity and enthusiasm for systemic norms, we use the extent of participation in international organizations. This was constructed using version 2.1 of the International Governmental Organizations Dataset {{150 Pevehouse, Jon 2005; }}, which contains information on membership of any organization with at least three states as members from We excluded all organizations that were restricted to countries of particular regions, such as the European Union. We also excluded geographically-limited functional organizations, such as the association of tin-producing countries. Using the resulting set of truly systemic organizations -- organizations which any country in the world can join -- we calculated IO involvement for each country year as the number of all organizations joined as a ratio of all organizations to which it could possibly belong. The resulting indicator varies from zero (no IO memberships at all) to 100 (membership in every possible IO). NPT Membership Rate. If the NPT has proved successful at establishing nuclear abstinence as a marker of modernity, then we would expect peer pressure to join to mount as membership becomes a systemic norm. International institutions are sometimes thought to strengthen norms by socializing actors and by actively promoting norms through advocacy and epistemic communities {{151 Johnson, Alastair Iain 2001; }}. To tap this possibility, we include the percentage of eligible states which have acceded to the NPT for each year as a variable, expecting a positive coefficient. 20

19 Page QUANTITATIVE RESULTS Methods We employ event history models to evaluate the usefulness of our cost-benefit approach for understanding decisions to sign the NPT. Event history models -- also called survival, hazard, or duration models -- provide estimates of the probability of an event occurring at a particular time, given that it has not yet happened {{3 Allison,P.D. 1984;10 Box-Steffensmeier,J.M. 1997; }}. For our purposes, the relevant question is: how likely is a country to sign the NPT in a given year, given that it has not done so up until this point? In the language of event history analysis, this probability is given by the hazard rate, which tells us the risk that a country will sign the NPT in a given year. Event history models can be either parametric, requiring the specification of a particular distributional form (such as the Weibull, exponential, or Gompertz) for the baseline hazard function, or semi-parametric, allowing one to avoid making such assumptions when there is no strong a priori reason to favor one distributional form over another. Because we have no strong theoretical priors about the about the distributional form of the duration times, we estimate Cox proportional hazard models. The Cox model is attractive for our purposes since it allows us to estimate the relationship between the covariates and the hazard rate without having to make any assumptions about the shape of the baseline hazard rate {{142 Box- Steffensmeir, Janet M. 2004; }}. 21 The hazard rate is then given by: h i (t) = h 0 (t)exp ( x) where h o (t) is the baseline hazard function, t is time, and x are the regression parameters and covariates. 22 Since survival models are non-linear, interpretation of coefficients is not straightforward. Unlike those in standard OLS regression models, the beta coefficients do

20 Page 18 not represent the marginal effect on the dependent variable of a one-unit change in the independent variable. For the Cox model, a positive coefficient means that the hazard is increasing as a function of the covariate; a negative one means that it is decreasing. Exponentiating the Cox coefficients expresses them in terms of hazard ratios, which are more readily interpretable. In this form, the coefficient can be read as the number by which we would multiply the risk (hazard) of, for example, ratifying the NPT in a given year for a one-unit increase in the independent variable. To illustrate, suppose the exponentiated coefficient on the dichotomous enduring rivalry variable is.71. This would indicate that the risk of signing the NPT in a given year among countries in an enduring rivalry is only 71 percent of the risk for countries not involved in enduring rivalries (in other words, it is about 29 percent lower). As another example, suppose the coefficient on great power alliance is This implies a risk of signing the NPT for such countries that is 127 percent the risk for countries not having a great power ally (in other words, it is about 27 percent higher). We report exponentiated coefficients in the results below, along with the standard errors. 23 Results To explore decisions to sign the NPT, we estimated a series of models with the variables in our cost-benefit framework. In these event history models, durations consist of strings of country-years. When a country ratifies the NPT, it exits the risk pool and thus the analysis. In principle, a country can re-enter the risk pool if it withdraws from the NPT: at this point, it is again deemed at risk and can choose to re-enter the regime. In practice, however, only North Korea has withdrawn from the NPT, and it did so after the end-date of our analysis Table 2 about here --

21 Page 19 We first explored the effects of variables representing the three types of costs and benefits -- economic, security, and political -- separately, and then combined variables from all three approaches in a comprehensive model. Starting with model one in Table 2, two economic variables have strong and significant effects on the hazard rate. Countries with greater economic resources, as proxied by aggregate GDP, are more likely to ratify the NPT quickly. In contrast, more developed countries are slower to sign on, although neither the coefficient on urban population or GDP per capita reach statistical significance at standard levels. Reflecting the logic of the core NPT bargain, countries with high energy needs -- as measured by lower electricity generation than would be expected given their size and economic development -- are more likely to sign the treaty quickly (presumably because access to nuclear energy is more attractive to them). Neither status as an oil exporter nor prior operation of a research reactor is associated with propensity to join. Model Two focuses on the security environment. Here we expect that the opportunity costs of joining the NPT are higher for states facing a challenging security environment, inducing them to be NPT laggards. This logic gains considerable support from the hazard estimations: all three variables are both statistically and substantively significant. Participation in an enduring rivalry sharply lowers the likelihood a state will join the NPT in a given year; a state in an enduring rivalry is 38 percent less likely to join than one not participating in a rivalry. More frequent militarized dispute involvement in the recent past also reduces the likelihood of ratifying the NPT: increasing the five year moving average of dispute involvement by one reduces the hazard by 17 percent. However, Model 2 also suggests that great power security guarantees can reduce the opportunity costs of signing away the nuclear option: a defense pact with a nuclear-armed power increases the predicted probability of joining the NPT by 34 percent. The third model taps political variables. Democracy, as proxied by Polity IV data, appears unrelated to propensity to ratify the NPT. Countries that participate more broadly

22 Page 20 in international institutions are indeed more likely to join the NPT, as indicated by the hazard ratio greater than one, but the p-value is only.17. As NPT membership grows, however, countries remaining outside the treaty become increasingly likely to join. 25 The comprehensive model reported in column four reinforces most of these relationships. All three security variables retain their substantive and statistical significance. Countries with higher GDP per capita are slower to sign on, whereas those with greater electricity needs and larger aggregate GDPs are quicker to join. Finally, states with higher IO participation rates are more likely to join, and a higher membership rate induces countries to join the club. The Cox models used to produced the estimates reported in Table 2 make a rather strong assumption in requiring that parameters are stable over time. Specifically, the proportional hazard assumption entails that the effect of a change in an explanatory variable is to shift the hazard by a factor of proportionality that remains constant over time. One might suspect that this is not the case over 30-plus years of experience with the NPT. Effects may increase or decrease, or even reverse sign, perhaps as a result of changing norms about the non-proliferation regime or the acceptability of nuclear weapons. Fortunately, we can test the non-proportionality assumption and, if necessary, make corrections {{143 Box-Steffensmeir, Janet M. 2003; }}. Table 3 reports the results of both a global proportionality test and covariate specific tests; very small p-values indicate likely violations of the proportionality assumption. -- Table 3 about here The global test reported in the final row of Table 3 indicates that we cannot reject the null hypothesis of proportionality. However, the covariate specific tests do reject the hypothesis of proportional effects for one variable: participation rate in international organizations. To explore the consequences of this variable-specific violation of the proportionality assumption, we re-estimated the models allowing for non-proportionality of hazards in the participation in international organizations variable. 26 We interacted this

23 Page 21 variable with the log of time and included the interaction term in the model. 27 This allows the influence of the variable to increase or decrease over time. Table 4 repeats the earlier analysis allowing for non-proportional hazards in the participation in international organizations variable. Both this variable and its interaction are significant in both the reduced political model and in the comprehensive model. The coefficient on the IO participation variable means that countries with a high propensity to join IOs in general are more likely to sign on to the NPT, but the interaction effect indicates that this effect diminishes over time. Countries that are joiners in general were more likely to ratify the NPT quickly, but if they did not join the treaty in the first few years they become increasingly less likely to do so. This may be because joiners that (uncharacteristically) do not sign on quickly are countries with particularly good reasons to stay out of the NPT. The necessity of explaining resistance to the treaty may also, over time, endow opposition to the NPT with meaning and legitimacy which renders it politically difficult to join at a later date (as perhaps has happened in India). --- Table 4 about here --- In sum, however, our overall analysis of the NPT is very robust to consideration of non-proportional hazards. The effects of the covariates are very stable across time. Influence and Outlier Analysis Having explored the possibility of non-proportional hazards, we now turn to influence analysis and the exploration of outliers. With a heterogeneous set of countries, a handful of which have fairly extreme values on some covariates, it important to determine if any observations have a disproportionate influence on the estimated values of parameters. Dfbetas were calculated for each of the variables, and we then used both graphical and numerical analysis to look for countries that might be strongly influencing the parameter estimates. Overall, few possibilities for influential countries emerged. 28

24 Page 22 Employing the commonly used cut-point of 2/ {{144 Belsley, D.A 1980; }}, specific countries emerged as potentially influential for only two covariates: oil exporter and participation rate in international organizations. To assess their importance, we reestimated the models deleting these countries. For the oil exporter variable, only Nigeria breaches the cut-point, although Canada, Libya, and Indonesia also score fairly highly -- all four of these countries exert upward influence on the value of the parameter estimate. Omitting Nigeria reduces the coefficient (in hazard ratio terms) from 1.24 to 1.12, and leaving out all four countries reduces it to.99. With or without these potentially influence countries, the conclusion is thus the same: the effect of oil exporter status is indistinguishable from zero. The other possible case of influential countries arose with the international organization participation variable. Four countries -- France, Spain, St. Lucia, and the United Kingdom collectively exert downward influence on the main variable and upwards bias on the interaction with time. Excluding all four pushes up the coefficient on the IO variable and reduces that on the time-interaction variable, with both remaining statistically significant (p-values staying <.01). Overall, our analysis turns up very few potentially influential countries, and omitting those few does not alter the substance of the results for the variables potentially affected. We further explored the model s fit by calculating deviance residuals, which are fairly symmetrically distributed around zero with larger residuals indicating cases that are poorly predicted by the model (possible outliers). 29 Observations that might be considered outliers are scarce. No observations have deviance residuals exceeding three in absolute magnitude, a threshold sometimes used to identify outliers {{145 Thompson, Laura A. 2003; }}. Three countries score more than two standard deviations above the mean (that is, they joined the NPT very quickly compared to the model s prediction): Nigeria, Lithuania, and Iraq. Four countries feature residuals more than two standard deviations below the mean (they were NPT holdouts relative to the model s prediction): Cuba,

25 Page 23 Brazil, Chile, and Guyana. Leaving out either or both sets of countries does not alter the conclusions of our analysis. However, a closer look at countries with relatively large positive residuals suggests an intriguing possibility. Several of the countries who signed more quickly than anticipated were involved in enduring rivalries or otherwise difficult security situations, notably Iraq, Syria, Taiwan, and Jordan. 30 Syria, Jordan, and Iraq were all embroiled in an ongoing rivalry with Israel, which was widely rumored to be on the brink of acquiring nuclear weapons in the late 1960s. With little hope of matching Israel on the nuclear front, signing the NPT may have been a way to seize the moral high ground and hopefully forestall Israel from going nuclear. In many ways, Taiwan was in a similar position vis a vis China: unable to compete with China in a nuclear arms race, Taiwan could use the NPT as a strategic resource to perhaps help forestall a nuclear race and to seize the normative high ground. For these four early signers, it is plausible that nuances of their security situation not captured by our indicators provide a reason for their alacrity in acceding to the NPT. 5. CONCLUSIONS The nonproliferation regime is one of the most important efforts at international cooperation today, and the Nuclear Non-proliferation Treaty is the cornerstone of that regime. Understanding why and when countries ratify it is of utmost importance and provides valuable clues about the functioning and effectiveness of the treaty. The NPT commands near-universal adherence. With very few known exceptions, its signatories have not pursued nuclear weapons. It is therefore hard to assess whether signing the NPT itself makes any difference to countries decisions to pursue nuclear weapons. The results presented in this paper provide important hints towards the answer. We ask: what explains the large variance in willingness to join the NPT? If the behavior of potential members does not reflect the logic of the bargain offered by the NPT, we would begin to suspect

26 Page 24 that the NPT does not work in the way that it was intended to function, bolstering fears that it is an ineffective paper tiger with little real importance. In the worst case, countries desiring nuclear weapons might eagerly sign on as a short-cut to gaining access to much of the necessary technology and knowledge. In fact, we find that a cost-benefit framework reflecting the logic of the NPT bargain can explain much of the variance in joining the NPT. Specifically, states that have serious security concerns are less likely to sign the NPT, states with high energy needs are more likely to sign it, and states that could easily develop atomic weapons at low cost are more reluctant to join. Hesitation to sign the NPT is closely related to factors shaping a country s willingness and ability to acquire nuclear weapons. Thus, it seems that states believe that the NPT does have an effect. They do not treat it as a paper tiger; they act as if it poses real constraints. In effect, reactions to the NPT appear to sort states according to their interest in nuclear weapons; the NPT may function as a screening mechanism for inducing states to reveal their preferences over nuclear arms. While we do not claim that this is conclusive proof that the NPT stops countries from going nuclear, we do suggest that since countries that have good reasons to want nuclear weapons hesitate to accede to the treaty, we should give it the benefit of the doubt. Our model also enables us to identify surprisingly quick joiners and puzzling holdouts. The reluctance of some countries to join the NPT when our variables suggest that they should do so quickly (and those of countries that join more quickly than predicted) can tell us about things that the conventional wisdom may have omitted. We see that several countries have joined the treaty even when they were facing security threats. Why is this so? What motivates these surprisingly quick NPT joiners? We hypothesize that the NPT becomes a strategic resource for these countries. They signal their peaceful intentions and try to garner international support with their renunciatory act. States may use the NPT as a resource for strategic signaling of their preferences or attempting to gain