The Causes of Nuclear Weapons Proliferation · PL14CH11-Sagan ARI 18 February 2011 17:14 R E V I E W S I N A D V A N C E The Causes of Nuclear Weapons Proliferation Scott D. Sagan

PL14CH11-Sagan ARI 18 February 2011 17:14

RE V I E W

S

IN

AD V A

NC

E

The Causes of NuclearWeapons ProliferationScott D. SaganDepartment of Political Science, Stanford University, Stanford, California 94305;email: [email protected]

Annu. Rev. Polit. Sci. 2011. 14:225–44

The Annual Review of Public Health is online atpublhealth.annualreviews.org

This article’s doi:10.1146/annurev-polisci-052209–131042

Copyright c© 2011 by Annual Reviews.All rights reserved

1094-2939/11/0615-0225$20.00

Keywords

International Atomic Energy Agency (IAEA), nuclear latency,Non-Proliferation Treaty (NPT), plutonium reprocessing, uraniumenrichment

Abstract

This critical review of the new political science literature on the causesof nuclear weapons proliferation consists of four parts. The first sectionbriefly presents what we know about which states developed nuclearweapons and which states started but abandoned weapons developmentprograms. I highlight the problems that result from uncertainty aboutthe accuracy and completeness of the data. The second and third sec-tions review the literature on the spread of the technical capability todevelop nuclear weapons. We still lack robust knowledge about the re-lationship between the development of civilian nuclear power programsand nuclear weapons acquisition. The next two sections review the liter-ature on the demand for nuclear weapons. Comparative case studies andstatistical studies have improved our understanding of the diversity ofmotives for weapons development and restraints, but serious gaps in ourknowledge remain. The sixth section outlines alternative theories aboutthe potential impact of the nuclear Non-Proliferation Treaty (NPT) onnuclear weapons programs decisions. Finally, I lay out a future researchagenda to address the weaknesses in our current understanding of thecauses of nuclear proliferation.

225

Review in Advance first posted online on March 21, 2011. (Changes may still occur before final publication online and in print.)

Changes may still occur before final publication online and in print

Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


NPT: Non-Proliferation Treaty

INTRODUCTION

Concerns about the potential spread of nuclearweapons have increased during the past decade,as North Korea tested two nuclear weapons,Iran was placed under United Nations Secu-rity Council sanctions for refusing to stop itsuranium enrichment program, and more na-tions sought civilian nuclear power plants toproduce electricity and combat global warm-ing. The connections between the spread ofcivilian nuclear power and the proliferation ofnuclear weapons, however, are not clear-cut.The physical connection is simple: The fueland by-product of light-water nuclear powerreactors—enriched uranium and plutonium—can also be used to produce nuclear weapons.From a technical perspective, as NobelLaureate Hanes Alfven noted, “Atoms for peaceand atoms for war are Siamese twins” (as quotedin Miller 1979, p. 19). The political connec-tions between the growth nuclear power capa-bilities and nuclear proliferation risks, however,are more complex. Governments that developspecific technologies related to civilian nuclearpower facilities may be more capable of devel-oping nuclear weapons, but are they thereforemore likely to do so?

Although it is premature to claim, as somescholars and nuclear industry officials havedone, that a “nuclear renaissance” has startedwith respect to the spread of nuclear powerplants to new countries around the globe(Miller & Sagan 2009, Lester & Rosner 2009,Lavergeon 2009, Socolow & Glaser 2009),there has been a renaissance of scholarly inter-est among political scientists in explaining thecauses of nuclear weapons proliferation. Thenew political science literature on nuclear pro-liferation and nonproliferation should help usunderstand the crucial questions about states’capabilities and motives for building the bomb,but it has thus far been of limited utility. In theliterature, the technical capability to developnuclear weapons is called the “supply-side” as-pect of proliferation, and a government’s mo-tivation to develop nuclear weapons is calledthe “demand-side” aspect of proliferation. In

this review, I argue that the supply-side litera-ture has mistakenly focused on inaccurate mea-sures of nuclear weapons capabilities, while thedemand-side literature has paid inadequate at-tention to how the Non-Proliferation Treaty(NPT) and related institutions of the broadernonproliferation regime influence incentivesfor and against nuclear weapons acquisition. Inaddition, this traditional division of the subfieldinto supply and demand problems has in itselffocused scholarly attention away from under-standing the relationship between supply anddemand, between a state’s increased nuclear“latency” and its incentives for and against de-veloping nuclear weapons.

There have been many positive develop-ments in the past decade, as new theories havebeen developed and scholars have increasinglyused multiple methods to test their theories.Despite this progress, however, the new quan-titative and qualitative research projects on thecauses of proliferation have been limited bythe information available about the very smallnumber of cases of nuclear weapons prolifera-tion. Indeed, there is a deep and abiding uncer-tainty about the basic data on which states havesought nuclear weapons and when they startedand ended their programs. We must thereforeimproveon the historical proliferation data sets,and yet, given the secrecy surrounding nuclearprograms, some uncertainties are likely to re-main. To move forward beyond that, we willalso need more multidisciplinary research, withpolitical scientists developing better under-standings of the technology of nuclear powerand nuclear weapons and also the effectivenessof international law and export control regimes.

This critical review of the political scienceliterature on the causes of nuclear weapons pro-liferation consists of four parts. The first sec-tion below briefly presents what we know aboutwhich states developed nuclear weapons, andwhich states started weapons development pro-grams but then abandoned them. I highlightthe problems that result from remaining un-certainty about the accuracy and completenessof the existing data. The second and third sec-tions review the supply-side literature. Here I

226 Sagan


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


argue that political scientists have largely beengoing down a dead-end street that has led to lit-tle robust knowledge about the relationship be-tween the development of nuclear power pro-grams and nuclear weapons acquisition. Thefourth and fifth sections review the literatureon the demand for nuclear weapons. Here Iconclude that a mixture of comparative casestudies and statistical studies has improved ourunderstanding of the diversity of motives forweapons development and restraints, but thatserious gaps in our knowledge remain. Thesixth section outlines alternative perspectiveson the role of the NPT in nuclear decisionmaking and presents evidence on differencesbetween democratic and nondemocratic states’compliance with the treaty. The final sectionlays out a future research agenda to address theweaknesses in our current understanding of thecauses of nuclear proliferation.

NUCLEAR WEAPONS ANDNUCLEAR WEAPONS PROGRAMS

Figure 1 (see color insert) presents a famil-iar graph of when different states acquired nu-clear weapons, from the first U.S. nuclear test in1945 through the North Korean test in 2006.It shows a steady but slow increase. One newnuclear weapons state has emerged about everyfive years, and—with the exceptions of SouthAfrica and the three former Soviet Union statesthat inherited the weapons (but never had op-erational control over them)—each state haskept its nuclear weapons once it has them.There is virtually no disagreement about whatstates have acquired the bomb, although thereare still uncertainties about the precise yearin which “second-generation” nuclear weaponsstates—Israel, India, Pakistan, and NorthKorea—first acquired an operational nuclearweapon.

We are not only interested in which stateseventually acquired nuclear weapons, however,but also need to study what governments havetaken preliminary steps to develop nuclearweapons and then stopped such efforts, andwhat governments had serious and sustained

nuclear programs but nevertheless failed todevelop the bomb. Scholars have thus usefullydivided nuclear proliferation into phases(explore, pursue, and acquire), even thoughin practice it can be difficult to determinewhen a state actually falls into these categories.Figure 2 (see color insert) displays my esti-mates of when states started exploring nuclearweapons programs, and when they abandonedtheir nuclear weapons development efforts(with the exceptions of Iran and Syria, which arecoded as continuing a covert nuclear weaponsprogram). There is no single explanationfor why states give up their bomb programs(Campbell et al. 2004, Levite 2003, Paul2000, Reiss 1995, Rublee 2009). Some ofthese “nuclear reversal” cases were U.S. allieswho received strong diplomatic pressure fromWashington to stop suspected covert programs(Taiwan, South Korea); another resulted from“disarmament by force” (Iraq). Some govern-ments dismantled their bomb programs afteror in anticipation of domestic regime change(Romania, Brazil, South Africa); others endedtheir pursuit of nuclear weapons when an allyterminated joint nuclear weapons developmentagreements (Italy, Germany). But many acts ofboth nuclear reversal and abstinence appear tobe more voluntary in nature and, as discussedbelow, the emergence of the NPT in 1968and the later addition of related institutions,such as the Nuclear Suppliers Group (NSG),appear to have influenced many decisions torefrain from developing military applicationsof nuclear power programs.

It is important to note, however, that sig-nificant uncertainty exists about whether andwhen to code particular states as exploring orpursuing the bomb. This uncertainty exists inpart because many governments pursuing nu-clear weapons place veils of secrecy around theirprograms; in other cases, however, uncertaintyabout a government’s nuclear weapons pro-gram exists because the government itself neverdetermined whether early nuclear-related re-search was to be used only for future powerprogram development or for future nuclearweapons development.

www.annualreviews.org • Causes of Nuclear Weapons Proliferation 227


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


The small-numbers problem and the lin-gering uncertainty about the accuracy of thedata combine to create enormous difficultiesfor both the quantitative and case-study liter-atures on nuclear proliferation. In the quanti-tative literature, the relatively small number of“confirmed” states that have tried to develop thebomb has meant that the discovery of additionalstates, or the removal of states from the dataset because of new “disconfirming” evidence,can significantly reduce the robustness of manyauthors’ findings about the causes of prolifera-tion (Montgomery & Sagan 2009, 2011). Forexample, Montgomery & Sagan 2011) showthat adding new confirmed cases of states thatsought nuclear weapons to the data set devel-oped by Jo & Gartzke (2007a) significantly re-duces the magnitude of, and in some cases evenreverses, their statistical findings about correla-tions between a state’s membership in the NPT,the nuclear threats of its adversary, the diffu-sion of nuclear technology, and the status ofthat state’s nuclear weapons program.1

For the qualitative literature, the deep se-crecy surrounding many governments’ nuclearweapons decisions means that reasonable schol-ars can still defend conflicting interpretations ofwhen and why particular governments chose tomove forward or contain their nuclear weaponsprograms. For example, scholars disagree aboutwhether Argentina’s nuclear program was de-signed to fuel naval reactors for a submarinefleet or for a nuclear weapon (Hymans 2001,Muller & Schmidt 2010), while other observerscontinue to debate whether the Japanese gov-ernment has ever actively pursued a nuclearweapons option (Campbell & Sunohara 2004,Japan Times 2010). In short, the small num-ber of cases and remaining uncertainties aboutkey historical decisions present a significant andcontinuing challenge to understanding whenand why proliferation occurs.

1The cases added were Italy and West Germany 1957–1958(Nuti 1993, Muller 2003), Egypt 1960–1967 (Rublee 2009),Australia 1956–1972 (Walsh 1997), and Libya 1970–2003(Bowen 2006, Braut-Hegghammer 2009).

THE SUPPLY-SIDE LITERATURE

Ever since John F. Kennedy predicted in 1963that there might be as many as “fifteen, twenty,or twenty-five nuclear weapons powers” in the1970s, statesmen and scholars alike have beensurprised that the pace of nuclear prolifera-tion has been so much slower than predicted(Potter & Mukhatzhanova 2010). Indeed,Hymans (2010, p. 13) argues that “the basicpuzzle facing the study of nuclear prolifera-tion is why there is such a large and persistentgap between the technically nuclear weapons–capable states—at present there are perhapsas many as fifty—and the small number ofactual nuclear weapons states.” Much of thenew supply-side literature has been obsessedwith this puzzle. The underlying assumptionhere, that more and more states have become“nuclear weapons capable,” however, dependscrucially on how one defines and measures nu-clear weapons capability.

What does it mean for a state to be capa-ble of building nuclear weapons? The centralweakness of the coding rules by which politicalscientists have labeled states “nuclear capable”is clear once one delves into the methodologyused in their studies (Sagan 2010). The currentgeneration of proliferation specialists have builton the methodology and coding rules used inMeyer’s (1984) pioneering book The Dynamicsof Nuclear Proliferation and Stoll’s (1996) up-date of the Meyer data through 1992. Meyer(1984) carefully measured a set of ten tech-nical and economic indicators—national min-ing activity, indigenous uranium deposits, met-allurgists, steel production, construction workforce, chemical engineers, nitric acid produc-tion, electrical production capacity, nuclearengineers, physicists, chemists, and explosivesand electronics specialists—to produce what hecalled “a list of nations with latent capabilitiesto manufacture nuclear weapons” (p. 41). Be-cause he could not measure directly whetherthe quantity or quality of a state’s nuclear engi-neers and its explosives and electronics special-ists were sufficient to build a nuclear weapon,Meyer used two proxy indicators: whether the

228 Sagan


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


state had been operating a research reactor forthree years (the proxy for nuclear engineer-ing expertise) and whether the state manufac-tured automobiles, or assembled automobilesand manufactured radios and television sets (theproxy for expertise in explosives and electron-ics). Based on this particular set of indicators,Meyer found that 34 states held the latent ca-pability to build nuclear weapons in 1982.

Stoll (1996) updated the Meyer data set, witha hidden yet significant change in coding rules,in the mid-1990s: Whereas Meyer had mea-sured indigenous uranium sources, Stoll sim-ply assumed that all states had access to nu-clear materials, claiming that they were nowfreely available in the global marketplace. Stollthus assumed away the crucial technical bottle-neck of whether a state has access to uraniumthat, once enriched, could be used in a nuclearweapons program. Based on the resulting dataset, Stoll argued that 48 states had a latent nu-clear weapons capability in 1992 (Figure 3; seecolor insert).

Hymans (2004, p. 4; 2010) accepted thelogic of that argument, updated Stoll’s pro-liferation estimates, and claimed that the datademonstrate a “yawning gap between technicalpotential and military reality,” expressed by thecontrast between the number of states that havethe capability to produce nuclear weapons andthe number that have actually done so. A statecannot make a nuclear weapon, however, unlessit has highly enriched uranium or plutoniumfrom a large reactor, and Stoll’s hidden assump-tion that any state could acquire uranium on theopen market, coupled with his use of researchreactor experience as the measure of requirednuclear engineering expertise, essentiallyremoved those two technical constraints.

Jo & Gartzke (2007b) alter the Stoll cod-ing scheme by dropping three of the Stoll andMeyer indicators (construction workforce, steelproduction, and previous national mining ac-tivity) on the grounds that they are “too eas-ily available to be thresholds” and modifyingthe coding for necessary “uranium deposits”(which, as we have seen, were assumed by Stollto be available for all states) to include either a

IAEA: InternationalAtomic EnergyAgency

state with uranium deposits on its territory or astate that has acquired “produced uranium” fora research or power reactor. (See Table 1 for acomparison of the various coding schemes.)

Yet, although Jo & Gartzke (2007a, p. 169correctly note that “states that lack the basicmaterial capabilities will be excluded from thegroup of potential proliferators,” their codingrules do not in fact treat the possession of fissilematerials (enriched uranium or plutonium) as anecessary but not sufficient condition for build-ing nuclear weapons. Instead, Jo & Gartzkeimplicitly assume that “where there is a will,there is a way” because their coding rules leadto states being labeled as “nuclear weapons ca-pable” without having the necessary fissile ma-terials to make even one bomb. The evidencethat Jo & Gartzke’s coding rules do not ade-quately capture the necessary conditions for nu-clear weapons development is apparent in thefact that their data set coded North Korea asnot having full capability to develop nuclearweapons in 2001 (it still lacked sufficient chem-ical engineers, nitric acid production capability,and explosives specialists), even though NorthKorea was a major exporter of long-range mis-siles at the time and the North Koreans wereknown to have separated plutonium from thefuel rods of the Yongbyon reactor (May 2001).The Jo & Gartzke coding rules also led to theodd conclusion that South Africa lacked thecapability to build nuclear weapons in 2001—even though it had built six nuclear weapons inthe 1980s and, after dismantling the weaponsin the 1990s, still maintained 450–600 kg ofhighly enriched uranium according to Interna-tional Atomic Energy Agency (IAEA) safeguardinspections (Purkitt et al. 2002). (Within Jo &Gartzke’s coding rules, South Africa did nothave the “necessary” chemical engineers and ni-tric acid production capacity.)

By not focusing attention on enriched ura-nium and plutonium, the weak proxy measuresfor nuclear weapons capability used by Jo &Gartzke clearly lead to bizarre results. Trinidadand Tobago (which “only” lacks uranium de-posits, “produced uranium,” and any researchreactor experience) is assessed to have had a



Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


Table 1 Comparison of Meyer/Stoll (Meyer 1984, Stoll 1996) and Jo & Gartzke (2007b) nuclear capability indicators

Definition of indicator

Indicator Meyer/Stoll Jo & GartzkeMining activity Some fraction of labor force in mining activity ———Uranium deposits Known uranium deposits (Meyer)

Assumed market access (Stoll)Known uranium deposits or produceduranium already

Metallurgists Production of crude steel Production of crude steel or aluminumSteel Production of crude steel ———Construction workforce Production of steel and cement ———Chemical engineers Production of nitric acid or sulfuric acid Production of nitric acid or sulfuric acidNitric acid production capacity Nitric acid production or sulfuric acid

production and nonorganic nitrogenousfertilizer production

Nitric acid production or sulfuric acidproduction and nonorganic nitrogenousfertilizer production

Electricity production capacity Installed electrical capacity of 200 MWe Installed capacity of 200 MWe orproduces equivalent of 50,000 metrictons of oil

Nuclear engineers, physicists, chemists Three research-reactor years Three research-reactor yearsSpecialists in electronics and explosives Manufacture of motor vehicles, or assembly

of motor vehicles and manufacture of radiosor televisions

Manufacture or assembly of motorvehicles and manufacture of radios ortelevisions

higher degree of nuclear weapons latency in2001 than is North Korea, which was only fiveyears away from detonating its first nuclearweapon. Egypt, which had only two researchreactors in 2001, is assessed to have a higher de-gree of nuclear weapons latency than is SouthAfrica ( Jo & Gartzke 2007b).

These problems are examples of a commonweakness in the quantitative political science lit-erature on proliferation (Montgomery & Sagan2009, 2011). All too often the authors look forthe keys under the lamppost: They use “proxyvariables” that are easily available, rather thancollecting the data that reflect the substantivevariables of real interest. For example, scholarsinterested in measuring nuclear “weaponizationcapability” could have built original data sets onrelated arms-manufacturing experience and ex-pertise, but instead they used preexisting dataon automobile and television manufacturing.Furthermore, there are often hidden, but cru-cial, assumptions in this literature, such as “theopen market enables all states to get nuclearmaterials,” that have a strong but unacknowl-edged impact on the findings. Finally, the newquantitative proliferation literature in political

science frequently conflates two analytically dif-ferent phenomena under the same labels of “nu-clear weapons capability” or “nuclear latency”:first, what should properly be called “nuclearself-sufficiency”—a measure of how indepen-dent or self-sufficient a potential long-term nu-clear weapons program could be; and second,“nuclear latency”—a measure of how quickly astate could develop a nuclear weapon from itscurrent state of technological development if itchose to do so (Sagan 2010).

In short, by focusing our attention awayfrom the acquisition of the necessary fissile ma-terials, the supply-side proliferation literaturehas led to a grossly exaggerated estimate of howmany states currently have the technical capa-bility to build nuclear weapons. Indeed, the ap-propriate proliferation puzzle may not be whyso many “nuclear weapons capable” states haverefrained from building nuclear weapons, butwhy so few states have acquired the facilities andtechnology needed to enrich uranium or repro-cess plutonium. In 2010, ten states are knownto have uranium enrichment facilities (WorldNuclear Association 2010), and eight states,only one of which is a non–nuclear weapons

230 Sagan


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


state (NNWS), are believed to have nationalplutonium reprocessing facilities (InternationalPanel on Fissile Materials 2010). Why has therebeen so little proliferation of facilities for theproduction of fissile materials? Answering thisquestion would require deeper investigationsinto the economics of nuclear power and moreresearch on the development of nuclear exportcontrol regimes than political scientists haveconducted to date.

NEW SUPPLY-SIDE RESEARCH:BRINGING BACK THE FUEL

Two new contributions have helped improvethe literature by focusing precisely on thetechnologies needed to link nuclear powerto nuclear proliferation: nuclear exports andtechnology agreements regarding nuclear fuelproduction, reprocessing technology, nuclearengineering training, and power and researchreactors. First, Kroenig (2009a,b, 2010) arguesthat international sensitive nuclear assistancecontributes to the spread of nuclear weapons.Kroenig defines sensitive nuclear assistance asgiving information on weapons design or con-struction, providing large amounts of weapons-grade fissile material, or assisting in the con-struction of facilities for uranium enrichmentor plutonium reprocessing that could be usedto produce weapons-grade fissile material. Hetests his theory both statistically and throughthree case studies of countries that acquirednuclear weapons after having received sensitivenuclear assistance: Israel, China, and Pakistan.

Kroenig (2009a, 2010) finds that sensitivenuclear assistance is positively correlated to ac-quisition of a nuclear bomb, as is gross domesticproduct, the state’s overall industrial capacity,rivalry with nuclear weapons states, and regimetype. He carefully traces the importance of sen-sitive nuclear assistance in the history of histhree case studies and also examines other casesof states acquiring such assistance and exploringweapons programs. His central conclusion—“states that have the ability to produce nu-clear weapons, either through international as-sistance or domestic capacity, are much more

NNWS: non–nuclearweapons state

likely to do so” (2009a, p. 176)—is certainlytrue but hardly counterintuitive. Indeed, 5 ofKroenig’s 14 cases include not just assistancein building enrichment or reprocessing facili-ties but help in the design of nuclear weaponsthemselves. The fact that the states that weregiven bomb designs were more likely than otherstates to later acquire nuclear weapons is hardlysurprising.

Kroenig’s important contribution to ourunderstanding of proliferation, however, ison the causes of sensitive nuclear assistance(2009b, 2010). Why do states sell or givesensitive nuclear assistance to others, given thelikelihood that this might encourage weaponsproliferation? Most scholars and policy makershave assumed that governments do this foreconomic reasons or because they fail to predictthe nuclear proliferation consequences of theirnuclear technology export policy. Kroenig,however, finds that this is rarely the case.Instead, governments often provide sensitiveassistance for political and strategic reasons,and this is true because of, rather than despite,the anticipated proliferation consequences.Governments are more likely to transfer sen-sitive technologies into regions in which theycannot project conventional military powerand to states that are rivals of their enemies.“In other words, states may calculate that theenemy of their enemy is their best customer,”Kroenig (2010, p. 38) tersely argues. This is animportant (and disturbing) finding.2

The second new contribution to the supply-side proliferation literature is the work ofFuhrmann (2009a,b). In contrast to Kroenig,Fuhrmann (2009a, p. 8) claims that the spreadof all types of peaceful nuclear technology, notjust “sensitive” nuclear technology, increasesthe likelihood of proliferation: “The conven-tional wisdom is wrong—and dangerous. Alltypes of civilian nuclear assistance raise the risks

2It is also a controversial finding, for it in part depends on theaccuracy of Kroenig’s argument that the activities of the A.Q.Khan network were “state-sponsored by any reasonable def-inition of the term” and that the common enemy of Pakistanand its nuclear customers was the United States (Kroenig2010, pp. 135, 140).



Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


of proliferation. Peaceful nuclear cooperationand proliferation are causally connected be-cause of the dual-use nature of nuclear technol-ogy and know-how.” Fuhrmann acknowledgesthat the vast majority of states that have re-ceived civil nuclear assistance agreements havenot acquired weapons (in 99.77% of country-year observations, states receiving civilian nu-clear assistance did not acquire the bomb), buthe also insists that there is a strong statisti-cal and causal link between the number of nu-clear cooperation agreements (NCAs) and thelikelihood that a country will initiate a nuclearweapons program and eventually acquire thebomb. Fuhrmann asserts that “nuclear coopera-tion strongly influences whether a country goesdown the nuclear [weapons] path. Participationin at least one nuclear cooperation agreementincreases the likelihood of beginning a bombprogram by about 500%” (2009a, p. 32).

Fuhrmann’s central insight—that a statemay acquire dual-use technology with onlypeaceful intent, but then succumb to thetemptation to initiate weapons research wheninternational threats emerge—is an importantone. However, there are three problemswith his analysis. First, his estimate aboutthe substantive effects of NCAs on initiationof weapons programs is overstated. In hiscrosstabular analysis, 3 states are coded ashaving pursued nuclear weapons withoutNCAs (0.07% of observations) and 12 statespursued weapons after receiving NCAs (0.42%of observations). Although technically thisdoes mean that states with NCAs are >500%more likely to start nuclear weapons programsthan are states that did not have NCAs, thelikelihood of any state, even after signing anNCA, pursuing nuclear weapons is still lessthan half a percent. Moreover, the lag timeFuhrmann includes in his pursuit variable dropsRussia and the United States from this analysisentirely, further overemphasizing this effect.

Second, there is an endogeneity prob-lem: Fuhrmann did not adequately control forwhether the countries in question were alreadyexploring the nuclear weapons option whenthey signed NCAs. It seems reasonable that

a country’s interest in nuclear weapons tech-nology would make participation in an NCAmore likely, and thus it could be that NCAs arenot causing later weapons programs, but ratherthat plans for nuclear weapons programs aremotivating NCAs. The Singh & Way (2004)data set, which Fuhrmann uses for coding pur-suit and acquisition, also includes a measurefor when states first explored nuclear weaponsthat he does not utilize. Of Fuhrmann’s fivecase studies, three countries—Israel, Pakistan,and India—received substantial nuclear assis-tance after they were already coded by Singh& Way as exploring, but not pursuing, nuclearweapons. North Korea received aid from theSoviet Union at about the same time it beganexploring. Only South Africa received all of itsaid before exploring in 1969, and it is notablethat this assistance was provided in the 1950s,before the establishment of the NPT regime.

Third, Fuhrmann uses aggregate NCAs asthe crucial independent variable, which is prob-lematic. This variable places a great deal ofweight on countries, like India, that participatedin a large number of NCAs prior to acquisitionof nuclear weapons. Fuhrmann also includes anumber of cases in which NCAs were offered toa state that was known to be exploring a nuclearweapons option in an attempt to provide eco-nomic carrots to keep the state from pursuingnuclear weapons to completion. And Fuhrmannincludes NCAs that were negotiated and signedbut then canceled before coming into fruition(Kroenig in Bluth et al. 2010).

THE DEMAND-SIDELITERATURE

Different studies of why states are interested inacquiring nuclear weapons examine differentincentives and often focus on different actorswho influence government decisions. Earlywork on the demand for nuclear weaponsfocused on the utility of using internationalrelations theories to derive alternative testablehypotheses. Solingen’s 1994 article “The Polit-ical Economy of Nuclear Restraint” contrastedneorealist predictions about the importance

232 Sagan


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


of threats to her own prediction that “rulingcoalitions pursuing economic liberalizationare more likely to embrace regional nuclearregimes than their inward-looking, nation-alist, and radical-confessional counter-parts”(p. 136). Sagan’s 1996 article “Three Modelsin Search of a Bomb” compared the influenceof security threats, domestic political interests,and international norms on nuclear prolifer-ation decisions. The first of the three modelsfocused on military security motivations; thesecond emphasized domestic political actors orbureaucratic coalitions that form to support oroppose nuclear weapons for parochial reasons;and the third focused on norms and prestigeconsiderations that can encourage or discour-age acquisition of nuclear weapons (Sagan1996/1997, 1999). Sagan found strongestsupport for the security model, but he arguedthat domestic interests and normative concernsabout prestige were “sufficient, but not neces-sary” conditions for proliferation in a limitednumber of cases.

Scholarly debate has continued for twodecades about how to weigh these differentmotivations for proliferation. Some scholarsargue for the prime importance of securitymotivations (Mearsheimer 1990, Frankel 1993,Frankel & Davis 1993, Thayer 1995, Paul2000, Hecker 2010), others for the importanceof the constraints or incentives stemming fromdomestic economic interests (Lavoy 1993,Solingen 1998, Liberman 2001), and someconstructivist scholars focus on normativeconstraints and changing identities of indi-vidual leaders or governments (Chafetz et al.1996, Katzenstein 1996, Grillot & Long 2000,Tannenwald 2007, Rublee 2009).

The new quantitative studies of prolifer-ation have assessed states’ motivations usingdifferent data sets, different coding rules forkey variables, and different statistical methods,all of which could add to the robustness of theirresults if there were common findings. Unfor-tunately, that is not the case. Singh & Way’s(2004) hazard model finds strong support forboth enduring rivalries and militarized disputesas causes of proliferation, while Jo & Gartzke’s

(2007a) probit model finds strong supportfor status-driven motives and somewhat lesssupport for domestic political and internationalsecurity motives, with an especially interestingcounterintuitive finding that rivalry with anuclear weapons state is negatively correlatedwith both program and acquisition. Bleek’s(2010) hazard model, using an updated data setof states that explored, pursued, and acquiredthe bomb, finds in contrast that there is nocorrelation between states’ decisions to pursuenuclear weapons or eventually acquire themand one or more of that state’s enduring rivalshaving pursued or acquired nuclear weapons.Singh & Way (2004), Jo & Gartzke (2007a),and Bleek (2010) all do find that securityguarantees from a nuclear power have weakor no effect on a state’s decision to explore anuclear weapons option, yet this may be dueto selection effects: It seems reasonable toassume that a nuclear weapons state will bemore likely to offer a security guarantee toa country either facing a significant nuclearthreat or exploring nuclear weapons options.Thus, security guarantees may have a strongercausal effect in the real world than is capturedby the correlations in the observed data.

NEW DEMAND-SIDE RESEARCH:BRINGING BACK LEADERSHIP

Three recent case-study works by Abraham,Hymans, and Solingen have added significantlyto our understanding of specific national andregional cases of proliferation and nonprolifer-ation by developing new theoretical constructs.Each of these scholars focuses attention on theinterests and strategic preferences of govern-ment leaders and tests his or her theories againsta limited set of cases. It will be especially valu-able, therefore, for future researchers to testthese theories more broadly.

Abraham has not only written one of thebest studies of the Indian nuclear power andweapons program but has also introduced theimportant concept of “nuclear ambivalence”to the subfield (Abraham 1998, 2006, 2009).Abraham argues that the field’s obsession with



Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


nuclear weapons proliferation leads to a deepmisunderstanding of the relationship betweencivilian nuclear power and nuclear weapons.Most scholars (like most U.S. intelligence an-alysts) have examined the kinds of nuclear fa-cilities a country developed and have tried todiscern whether the government is really build-ing only a civilian energy capability or is se-cretly seeking to develop a latent or breakoutnuclear weapons capability. Thus, scholars of-ten write about “nuclear ambiguity” (is the gov-ernment seeking weapons or not?) or “nuclearopacity” (is the government hiding nuclear fa-cilities to cover its intentions?) when assess-ing the meaning of specific programs (Frankel1991, Frankel & Davis 1993). Abraham argues,in contrast, that government leaders may nothave strong and well-developed intentions inthe nuclear field at all. They may suffer frombasic ambivalence; they could be persuaded topush for weapons or for civilian power, depend-ing on a wide range of political and social factorsthat emerge long after they have constructedthe physical and organizational foundations fortheir nuclear program. Abraham tests his the-ory largely against evidence of the diversity andfluidity of motives seen inside the Indian nu-clear program between 1945 and the Pokhrannuclear weapons tests in 1998. He places primeimportance on the domestic power and politi-cal autonomy of what he calls the “strategic en-clave” of probomb scientists and governmentbureaucrats inside the Indian nuclear establish-ment (Abraham 1998; also see Bajpai 2009). Stillhe argues that the phenomenon of “nuclear am-bivalence” is common inside other capitals andnuclear bureaucracies, especially in the less de-veloped world. He further warns that nonpro-liferation policies to dissuade ambivalent actorsnot to do what they have no intention to do canbackfire, setting off national or bureaucratic re-sistance to external pressures.

Hymans (2006) makes a similarly creativetheoretical contribution, and his detailed casehistories of rarely studied incidents of prolif-eration and restraint add significantly to ourknowledge. Realism assumes that most lead-ers would want nuclear weapons for security

if they had the opportunity to get them; Hy-mans’ central argument turns this assumptionon its head. Very few leaders want nuclearweapons, he conjectures, because building thebomb is “a revolutionary decision” and “a leapinto the dark.” The decision makers cannotknow whether the state will be able to buildan arsenal, nor whether the strategic effect willenhance or damage national security. Nuclearproliferation is therefore rare, not because ofinstitutional constraints or alliance pressures,but rather “because few state leaders may in factneed to be constrained from seeking the bomb”(p. 12). Hymans develops a psychological the-ory, placing causal importance on what he callsthe “oppositional nationalist” identity of someleaders, who see their nation as naturally in ahostile relationship with other states and seethemselves as inherently superior to (or at leastthe equal of) their enemies. For such leaders,getting the bomb is not just a matter of a ra-tional cost–benefit calculation of state or evenparty self-interest: “Driven by fear and pride,oppositional nationalists develop a desire fornuclear weapons that goes beyond calculation,to self-expression” (Hymans 2006, p. 2).

Hymans’ case studies are particularly rich.He uncovers, for example, important new his-torical evidence suggesting that Argentina’suranium enrichment program, contrary to thecommon coding in the literature, may havebeen designed to produce nuclear fuel for a sub-marine fleet instead of being part of a covertnuclear weapons program. This is a fine ex-ample of how a case study can contribute tobuilding and revising the data sets needed forbroader tests of theories of proliferation. Andto his credit, Hymans does not rely on his his-torical case studies of nuclear decision makingin India, Argentina, France, and Australia tomeasure his key independent variable, whichwould be especially problematic for such an in-ductively derived theory. Instead, he developsa complex measure of “national identity con-ception” and uses a quantitative content anal-ysis of political leaders’ major public speeches(such as a state-of-the-union address) to assesswhether the leader can be accurately coded as an

234 Sagan


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


oppositional nationalist. This provides a checkto reduce the danger that Hymans might usehis knowledge of the final outcome to influencehis coding of the leaders’ national identity.

Although both Abraham and Hymans havedeveloped valuable new theories and haveconducted strong, albeit limited, tests of theirideas, their work shares a weakness: Not onlydo they themselves not test their theories ona wider set of cases, but it is also not clearhow they or other scholars could do so giventhe deliberately covert nature of many nuclearweapons programs. Abraham may be rightabout some governments or leaders holdingambivalent positions on nuclear matters, buthow could outside scholars measure the degreeto which this is the case when decisions arebeing discussed and made behind a veil ofsecrecy? Moreover, scholars have not yet de-veloped compelling measures of bureaucraticpower and autonomy that could help quantita-tive researchers determine whether the Indian“strategic enclave” phenomenon is an outlieror a more common occurrence (Sagan 2009a,Sasikumar & Way 2009). Similarly, Hymansmay be right about the influence of oppositionalnationalism in the cases he has studied, buthow can an outside analyst accurately measuresuch psychological characteristics for specificleaders making nuclear decisions who do notmake many public announcements, or whosepositions are strategically misrepresented toforeign or domestic audiences for bargainingpurposes? In short, Abraham and Hymans raiseimportant theoretical questions and providevaluable empirical understanding of specificcases, but it will be a major challenge for futurescholars to assess the broader strength of theirtheories about nonproliferation.

Finally, Solingen’s Nuclear Logics: Con-trasting Paths in East Asia and the MiddleEast (2007) compares the nuclear behaviorof states in those two regions, which hadsimilar qualities in the early 1970s but havesince experienced different nuclear trajecto-ries. Solingen’s focused regional comparisonmethodology uses detailed case studies inorder to test her preferred theory—a “global

integration” model that focuses on the degreeto which the government bases its domesticpolitical survival on economic integrationinto the global economy—against alternativerealist (security factors), neo-institutionalist(the NPT), and norms-and-constructivist (thenuclear taboo and global citizenship) theories.Her findings strongly support her global eco-nomic integration theory: States whose leadersor ruling coalitions advocate integration intothe global economy (mostly in East Asia) rejectnuclear weapons development, whereas leadersor coalitions that favor inward-looking relianceon domestic markets and nationalist values(mostly in the Middle East) are more prone topursue nuclear weapons.

The strength of case studies is the detailedtracing of the causal influences of economic in-terests, but the weakness of this methodologyis its inability to assess broader generalizabil-ity of the theory. Other scholars’ quantitativetests of Solingen’s global economic integrationmodel have produced mixed findings. Singh &Way’s (2004) hazard model tested the effects oftwo independent variables regarding a state’sintegration into the global economy—a state’s“interdependence” as measured by trade ratios(exports and imports as a share of GDP) anda state’s “liberalization” as measured by move-ment toward or away from higher trade inter-dependence. Singh & Way found that highertrade ratios were negatively and significantlycorrelated to all three measures of prolifera-tion (exploration, program, and acquisition),but that economic liberalization was insignif-icant at all three stages. In fact, in the explore-and-acquire phase, liberalization had a weakpositive correlation. Singh & Way’s quantita-tive analysis thus offers only tepid support forSolingen’s theory. Bleek (2010) uses a slightlydifferent data set on states that sought and ac-quired the bomb, but the same proxy measurefor economic interdependence and liberaliza-tion as Singh & Way, and also reports mixedfindings. However, they are not the same mixedfindings: Economic interdependence is nega-tively correlated with the exploration of nu-clear weapons, is not significantly correlated to



Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


pursuit, and is positively correlated to nuclearweapons acquisition. Like Singh & Way, how-ever, Bleek finds that economic liberalization isnot significantly correlated with proliferation atany stage.

Solingen’s case studies are impressive andhave produced counterintuitive findings. Herdetailed case study of Japanese nuclear decisionmaking in the 1970s, for example, reverses theusual causal arrow regarding the relationshipbetween security guarantees and nonprolif-eration commitments. In “a fine example ofintellectual judo” (Hymans 2010), Solingenpresents evidence that Japanese leaders, whowere already opposed to Japanese nuclearweapons on economic grounds, successfullysought tighter U.S. security assurances in orderto justify their antinuclear positions in domesticdebates in Tokyo. In addition, Solingen’s focuson what she calls changing “world time” inproliferation—the 1968 NPT is seen as a water-shed that alters the incentives and disincentivesfor states to seek nuclear weapons—is animportant warning to quantitative researchersnot to expect identical patterns of proliferationand restraint over the entire nuclear age.

This focus on the post-1968 NPT worldtime, however, makes it more puzzling thatSolingen denigrates the role of the treaty anddoes not examine whether the NPT was nec-essary for “liberalizing” governments to beconcerned that movement toward a nuclearweapons program would lead to internationalsanctions or other restrictions on the poten-tial benefits from integrating into the globaleconomy. In addition, Solingen does not fo-cus sufficient attention on the nuclear indus-try in NNWSs, which can be major engines ofglobal economic integration, if they feel thattheir access to foreign technology and fuel willbe restricted if there is any suspected weapons-related research in their nations. As noted be-low, the position of the nuclear industry maybe one of the most important determinants ofwhether a government eventually moves towarddeveloping nuclear weapons options or towardbecoming an even stronger supporter of nuclearnonproliferation.

THE NPT, REGIME TYPE, ANDNUCLEAR PROLIFERATION

Despite this lack of consensus about the centralcauses of nuclear proliferation, there are twospecific agreements in both the case-study andthe quantitative political science literature:(a) the NPT does not have significant effects onthe likelihood of proliferation and (b) regimetype has only a minimal effect on proliferation.Because of the (false) consensus that thecapability to build nuclear weapons is alreadywidely available, most political scientistsstudying proliferation are highly skeptical ofthe potential effect of efforts to control thesupply of nuclear technology. Hymans (2006,p. 220) argues that “given the widespreaddiffusion of nuclear capacities, supply-sidecontrol measures against potential proliferantstates are clearly of declining utility.” Jo &Gartzke (2007a, pp. 185–86) maintain that “theinhibiting effect of the NPT is overcome bythe stronger technological diffusion effect” andconclude that “enthusiasm for the NPT amongproliferation opponents thus appears to be mis-placed.” Solingen (2007, p. 31) also expressesskepticism about the effectiveness of the treaty:“Would more states have opted for nuclearweapons had the NPT never been concluded?Not necessarily . . . ” Betts (2000, p. 69) makesthe most extreme argument, from a realistperspective, about the irrelevance of the NPT:

As useful as treaties are, it is a misconcep-tion to see them as a solution. They areeffects of nonproliferation, not causes of it.The NPT and CTBT (Comprehensive TestBan Treaty) reflect the intent of their ad-herents to abjure nuclear weapons. To date,the countries considered problematic—thosethat might acquire nuclear weapons—simplydid not join the NPT (South Africa stayedout while it had a nuclear weapons pro-gram and joined when it decided to getrid of it). Or else they joined and cheated(Iraq and North Korea). . . . If the NPTor CTBT themselves prevent proliferation,one should be able to name at least onespecific country that would have sought

236 Sagan


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


nuclear weapons or tested them, but refrainedfrom doing so, or was stopped, because of ei-ther treaty. None comes to mind.

Betts’ statement is anachronistic. He ignoreshow many states not considered “problematic”today were in fact feared potential proliferatorsin the past. In 1963, for example, National Intel-ligence Estimate 4–63 and a Robert McNamaraDepartment of Defense report predicted thateight countries could develop nuclear weaponsby 1973: China, India, Sweden, Australia, Israel,South Africa, Japan, and West Germany (Lavoy2004). Egypt was considered to have “moder-ate to high” motivation and a capability to ac-quire nuclear weapons later in the 1970s; andArgentina, Brazil, Romania, Bulgaria, Hun-gary, and Yugoslavia were all feared to be ableto develop the bomb by the 1980s (Lavoy 2004).To understand the role of the NPT, one shouldnot select on the dependent variable and con-centrate only on recent “problematic” cases ofstates that acquired or try to acquire the bomb.Instead, one should also study the broader setof cases of “nuclear abstinence” (states that re-frained from starting a weapons program) andalso examine the history of the many states thatgave up their programs before they joined thetreaty and those that cheated on their NPTcommitments, were caught, and were encour-aged or forced to abide by their Article II com-mitment not to seek nuclear weapons.

Moreover, the NPT included requirementsfor NNWS members to accept safeguards—inspections run by the IAEA—on their nu-clear power facilities as a precondition for re-ceiving peaceful nuclear assistance (Scheinman1987, Schiff 1984). This IAEA inspection sys-tem has caught a number of states (includingIran, North Korea, South Korea, and Egypt)either cheating on their NPT commitments orat least engaging in ambiguous but suspiciousweapons-related activities (Acton 2009). Thus,the NPT should not be seen as irrelevant be-cause some states have not complied with theircommitments, for it is the treaty which sets upthe inspections that permit us to know whencheating occurs.

This relates, however, to the second area ofagreement in the literature, regarding regimetype and proliferation. Democracies and autoc-racies are seen to be similar in their proliferationbehavior; if anything, democracies are found tobe slightly more likely to go nuclear than non-democracies. Jo & Gartzke (2007a) find thatregime type makes no difference in whethera government initiates a nuclear weapons pro-gram, but that democracies are more likely thannondemocracies to acquire nuclear weaponsonce they have a program. The authors spec-ulate that this is due to one of two causes:“democracies are more vulnerable to national-ist pressure” ( Jo & Gartzke 2007a, p. 179) or“partial democratic states use nuclear weaponsproliferation as a diversion for domestic politi-cal reasons” (p. 184). Singh & Way (2004) alsofind that democracies are more likely to acquirenuclear weapons. The results of their hazardmodel show that democracy makes states 25%more likely to explore, 95% more likely to ac-quire, and slightly more likely to pursue nuclearweapons. Their multinomial logit model offerssimilar results, with the exception that it sug-gests that democracies are slightly less likely topursue nuclear weapons. Singh & Way (2004,p. 864) explain their findings on democracy byarguing that “democratic governments may betempted to pander to nationalist populations inan effort to boost their popularity and retainpower.” Kroenig (2009b, p. 172) similarly notesthat democracy measures are positively corre-lated with nuclear weapons acquisition, arguingthat his model “provides support for the ideathat democratic states may be more prone tonuclear proliferation because they may be sub-ject to pressure from domestic constituenciesthat favor nuclear development.” Fuhrmann(2009a, p. 36) finds the coefficient for democ-racy to be positive, but not significantly so.

These findings oversimplify the role ofregime type in determining nuclear prolif-eration behavior. Democratic countries havecertainly both pursued and acquired nuclearweapons, but the new quantitative literaturehas ignored the important observation thatno NNWS democracy has cheated on its



Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


NWS: nuclearweapons state

commitments under the NPT. In all pastcases, democracies that started and either aban-doned or completed nuclear weapons pro-grams either did so before the NPT cameinto force or did not join the NPT at all(Muller & Schmidt 2010, Miller & Sagan 2009).Figure 4 (see color insert) reproduces the datafrom Figure 2 on when different states startednuclear exploration and nuclear programs,adding their NPT ratification dates and a mea-sure of their regime type based on the PolityIV data set (http://www.systemicpeace.org/polity/polity4.htm). This figure clearly showsthat democracies have behaved differentlywith respect to nuclear weapons prolifera-tion. Democracies have both sucessfully de-veloped nuclear weapons and started but thenabandoned nuclear programs. But no demo-cratic NNWS has ever started a covert nuclearweapons program after the government ratifiedthe NPT.

NEW PROLIFERATION PUZZLESAND RESEARCH AGENDAS

Social science research should help us betterpredict the likely nuclear future as well as ex-plain the nuclear past. But to do so, scholars willneed to use multi-methods research designs totake advantage of the strengths of both quanti-tative and case-study methods, and also developmore accurate understandings of the technicaland institutional conditions under which nu-clear proliferation and restraint decisions aremade. In particular, scholars will need to focusmore attention on four related puzzles.

How the Non-ProliferationTreaty Works

We have a strong literature on institutionsdealing with whether international politicaleconomy and human rights treaties constraintheir members, but much less on internationalsecurity treaties, especially nuclear weaponsproliferation (Chayes & Chayes 1995; Moravc-sik 2000; Simmons 2000, 2010; Dai 2007).Indeed, it is revealing that Simmons’ (2010)Annual Review of Political Science article, “Treaty

Compliance and Violation,” discusses tradetreaties, human rights treaties, and alliances,but says nothing about the NPT.

How exactly does the NPT work? Neolib-eral institutional theory sees the NPT as a so-lution to two collective action problems. TheNWSs find it useful not to provide nuclearweapons to their allies, provided the otherNWSs likewise refrain, and Article I of thetreaty enshrines this joint commitment. TheNNWSs find it in their interest to constrainthemselves from developing nuclear weaponsonly if they can be reassured that their NNWSneighbors are practicing similar constraints,and Article II and the IAEA safeguards inspec-tors provide precisely that reassurance (Sagan1996/1997, Dai 2007). An alternative “domes-tic politics model” sees the NPT as promoting“responsible” use of civilian nuclear power, andthe actors who are empowered by the treatyand manage nuclear power plants in their statehave parochial reasons to adhere to their treatycommitments (Sagan 1996/1997). The IAEAcreates an inspection mechanism that makes iteasier for governments to monitor their ownscientists’ behavior, and the growth of nuclearpower in the state increases the potential eco-nomic costs that powerful actors would face ifsanctions were enacted because their govern-ment was accused of having a covert weaponsprogram. In addition, the NPT can tip the do-mestic balance of power in NNWSs as newbureaucratic actors favoring compliance withtreaty obligations are created and strengthenedthrough the NPT review conferences and re-lated meetings. Ratifying the treaty can havespecial strength in democracies if fear of payingdomestic audience costs encourages leaders tomaintain treaty commitments.

A third, more “realist” approach sees theNPT largely as a way to manage hypocrisy:The NWSs claim to be committed to nucleardisarmament under Article VI in order toencourage the NWWSs to accept a “second-class” nuclear status. In particular, the NPTunder this theory was designed in the 1960sprimarily to constrain Germany and India;opening the treaty for universal signature was

238 Sagan


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.

http://www.systemicpeace.org/polity/polity4.htm

http://www.systemicpeace.org/polity/polity4.htm


merely a face-saving device to make the con-straints more politically acceptable in Bonn andNew Delhi (Swango 2009). This view sees thereal strengths of the NPT in the additional in-ternational institutions that have been createdto manage export controls, such as the NuclearSuppliers Group (NSG), which coordinatesdecisions about what states can receive whatkinds of nuclear technologies (Davis 1993).

It will be important, however, not to seek amaster theory for how the treaty works, for I seeno reason to assume that the NPT has identicalpathways of influence for different states. In-deed, like many treaties, the NPT is a complexinstitution representing a set of compromisesbetween the governments that negotiated itscontent. We know very little about why dif-ferent governments joined the NPT and howtheir interests and interpretations have shapedthe patterns of their compliance behavior (see,however, Erickson & Way 2011 and Potter2010). There are likely to be some states whojoined the NPT to cement a nonproliferationbargain with regional rivals; some cheater statesthat joined while already having or planningcovert nuclear weapons programs; and manymore states that joined hoping membershipwould increase civilian nuclear technologytransfer to them. There are likely to be evenmore states that joined the NPT as a part ofjoining the world order and have no intentionof ever developing a civilian nuclear energyinfrastructure, much less nuclear weapons,such as Bhutan, Papua New Guinea, and Fiji.Indeed, even the Holy See, hardly a candidatefor nuclear power nor a nuclear weaponsconcern, is a member of the NPT.

Given this complex and diverse set of mo-tives for joining the treaty, we should not ex-pect to find that a state’s NPT status is re-lated to its nuclear proliferation behavior in anysimple manner. In addition, we should not as-sume that states’ interests in the NPT are setin stone. Legal research on the NPT and otherinternational treaties has suggested that devel-oping world actors have become less willing inthe postcolonial period to accept unequal treat-ment in treaty agreements and implementation

(Albin 2006). If this is true in the NPT, it wouldhelp explain the renewed focus at NPT ReviewConferences on the issue of whether the NWSshave been meeting their Article VI treaty com-mitments to work in good faith toward eventualnuclear weapons disarmament (Sagan 2009b).

As part of this effort to evaluate the NPT andthe broader nuclear nonproliferation regime inthe future, it will also be necessary to focus be-yond the treaty to understand the origins andeffectiveness of related nuclear technology ex-port control institutions. We have strong stud-ies of the origins of the NPT itself (Shaker1980, Bunn 1992, Swango 2009). But there areno equivalent studies of the origins or effec-tiveness of the NSG, or UN Security CouncilResolution 1540, or the Proliferation SecurityInitiative. These newer institutions are crucialelements of the nonproliferation regime andshould not be ignored.

How Regime Type InfluencesNuclear Trajectory

We need more work on how regime type influ-ences decisions about both nuclear power andnuclear proliferation. I have presented the datashowing that democracies do not cheat on theirNPT commitments, but we do not know whythat is the case. Are democracies more con-strained than nondemocracies by their treatycommitments, or are they simply more selec-tive in joining only treaties they intend to com-ply with fully (Simmons & Hopkins 2005, vonStein 2005)? If democratic leaders do feel moreconstrained than nondemocratic leaders, whyis this the case? Do they fear “audience costs”(punishment by voters) if they are caught cheat-ing on agreements, or do they believe the like-lihood of being caught cheating is higher be-cause of increased transparency and protectionof whistleblowers in a democracy (Tomz 2007)?With respect to the NPT, the simple divi-sion between democracies and nondemocraciesshould be supplemented with studies on the re-lationship between the political leadership andthe scientific community and military. Hymans(2008), for example, usefully theorizes that



Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


“neopatrimonial” or “sultanistic” regimes—governments characterized by extreme person-alized rule, use of state resources to buy offclients, and an absence of checks and balances—will take longer to develop advanced levelsof nuclear technology and will fail more of-ten in attempts to move from one techno-logical threshold to the next. Weeks’s (2008)work on “autocratic audience costs” in interna-tional crises suggests another research questionregarding proliferation: Do different types ofnondemocratic regimes show different propen-sities to signal their intent and ability to abideby their NPT commitments? Last, it is possi-ble that democracies may be less likely to movesuccessfully toward nuclear disarmament thannondemocracies because of the increased effectof “nuclear nationalism” among the electorate(Perkovich 1999).

How Nuclear Supply and NuclearDemand Influence Each Other

We need more interdisciplinary work on thespread of nuclear power and how it influences“nuclear weapons latency” and the “time lines”needed for states to move from different statesof civilian power status to nuclear weaponscapabilities (Wohlstetter et al. 1977, Zentneret al. 2005, Harney et al. 2006, Sagan 2010).To conduct this analysis accurately, scholarswill need to understand the history of differentstates’ open efforts to buy and sell sensitive nu-clear technology, and also to evaluate the effectsof the development of illicit nuclear technol-ogy networks such as the A.Q. Khan network(Braun & Chyba 2004, Montgomery 2005,Albright 2010).

Future efforts to understand nuclear latencyand proliferation will be most useful if theystudy the temporal connection between supplyand demand for nuclear weapons instead oftreating these two “sides” of nuclear prolifera-tion as separate issues. Three potential connec-tions are obvious. First, how hard a governmentworks on a nuclear weapons program—the re-sources it commits to the program and whetherit is engaged in a “crash” effort or normal con-struction effort—is likely to be strongly affected

by the severity of its demand for a weapon.Second, a high degree of nuclear capability orlatency could influence demand by enablingactors who favor a nuclear weapon to argue thatacquiring a weapon would be a relatively easyundertaking. Here it is worth differentiatingbetween “latency,” a measure of how long itwould take for a state to acquire a bomb, and“hedging,” which is a deliberate action by agovernment to make a nuclear weapons pro-gram easier in the future. Third, a high degreeof latency could make it easier for a pro–nuclearweapons party or individual leaders to imple-ment a decision to acquire nuclear weapons ifthey are in power for only a brief period of time.

How the Success or Failureof Civilian Nuclear PowerInfluences Proliferation

Does the civilian nuclear power industry con-strain states or does it make nuclear weaponsproliferation easier? Do expensive failures toproduce efficient nuclear power increase thelikelihood of proliferation? An important deter-minant of the nuclear future will be the degreeto which the spread of nuclear power producesnew actors in different states that want to main-tain peaceful programs and oppose turningcivilian energy programs into nuclear weaponsprograms. Indeed, how best to ensure thatcivilian nuclear power bureaucracies maintaina strong interest in opposing nuclear weaponsproliferation may be the single most criticalquestion to answer for reducing the potentiallydangerous effect of the global spread of nuclearpower on the likelihood of nuclear weaponsproliferation. This is ironic, for although somenonproliferation specialists may not want morecountries to start nuclear power programs,once a state starts a nuclear power programits nonproliferation behavior may be stronglyinfluenced by the degree to which its civiliannuclear industry is a successful contributor tonational energy production. The leaders andbureaucratic organizations that run successfulnuclear power enterprises may have increasedincentives to maintain strong ties to theglobal nuclear power industry, to international

240 Sagan


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


capital and technology markets, and to globalregulatory agencies—and hence may be morelikely to cooperate with the nuclear nonpro-liferation regime. Leaders of less successful orstruggling nuclear power enterprises, in con-trast, may be more likely to support clandestineor breakout nuclear weapons developmentprograms to justify their existence, prestige,and high budgets within their state.

CONCLUSIONS

A renaissance in nuclear proliferation stud-ies is under way. Yet, we need more

theoretically informed case studies of prolif-eration and nonproliferation decisions, bothto improve our data sets and to test whetherthe causal mechanisms posited in our theo-ries are actually present. We need better quan-titative studies to test the generalizability ofour theories. We need more interdisciplinaryresearch on the technical bottlenecks and le-gal restrictions on the spread of nuclear facil-ities. These complex research agendas shouldbe linked together to produce more robustfindings about the nuclear past and promotebetter predictions about the global nuclearfuture.

DISCLOSURE STATEMENT

The author is not aware of any affiliations, memberships, funding, or financial holdings that mightbe perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTS

I thank Philipp Bleek, James Fearon, Matthew Fuhrmann, Jacques Hymans, Gaurav Kampani,Matthew Kroenig, and Etel Solingen for their comments on this piece. Additionally, I am gratefulto Erin Dexter, Samantha Lang-Eppsteiner, Jane Esberg, Astasia Myers, Jaclyn Tandler, andJeremy Voss for research assistance.

LITERATURE CITED

Abraham I. 1998. The Making of the Indian Atomic Bomb: Science, Secrecy, and the Postcolonial State. London:Zed Books

Abraham I. 2006. The ambivalence of nuclear histories. Osiris 21:49–46Abraham I. 2009. Contra-proliferation: the Indian bomb and nuclear developmentalism. In Inside Nuclear

South Asia, ed. SD Sagan, pp. 106–36. Stanford, CA: Stanford Univ. PressActon JM. 2009. Nuclear power, disarmament, and technological restraint. Survival 51(4):101–126Albin C. 2006. Justice and Fairness in International Negotiation. Cambridge, UK: Cambridge Univ. PressAlbright D. 2010. Peddling Peril: How the Secret Nuclear Trade Arms America’s Enemies. New York: Free PressBajpai K. 2009. The BJP and the Bomb. In Inside Nuclear South Asia, ed. SD Sagan, pp. 25–67. Stanford, CA:

Stanford Univ. PressBetts RK. 2000. Universal deterrence or conceptual collapse? Liberal pessimism and utopian realism. In The

Coming Crisis: Nuclear Proliferation, US Interests and World Order, ed. VA Utgoff, pp. 51–85. Cambridge,MA: MIT Press

Bleek PC. 2010. Why do states proliferate? Quantitative analysis of the exploration, pursuit, and acquisitionof nuclear weapons. See Potter & Mukhatzhanova 2010, pp. 159–76

Bluth C, Kroenig M, Lee R, Sailor W, Fuhrmann M. 2010. Correspondence: civilian nuclear cooperation andthe proliferation of nuclear weapons. Int. Secur. 35(1):184–200

Bowen WQ. 2006. Libya and nuclear proliferation: stepping back from the brink. Adelphi Pap. 46(380)Braun C, Chyba C. 2004. Proliferation rings: new challenges to the nuclear nonproliferation regime.

Int. Secur. 29(2):5–49



Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


Braut-Hegghammer M. 2009. Nuclear entrepreneurs: drivers of nuclear proliferation. PhD thesis, Dep. Int. Relat.,London School Econ. Polit. Sci.

Bunn G. 1992. Arms Control by Committee: Managing Negotiations with the Russians. Stanford, CA: StanfordUniv. Press

Campbell KM, Einhorn RJ, Reiss MB, eds. 2004. The Nuclear Tipping Point: Why States Reconsider Their NuclearChoices. Washington, DC: Brookings Inst.

Campbell KM, Sunohara T. 2004. Japan: thinking the unthinkable. See Campbell et al. 2004, pp. 218–53Chafetz G, Abramson H, Grillot S. 2007. Role theory and foreign policy: Belarussian and Ukrainian compliance

with the nuclear nonproliferation regime. Polit. Psychol. 17:727–57Chayes A, Chayes AH. 1995. The New Sovereignty: Compliance with International Regulatory Agreements. Cam-

bridge, MA: Harvard Univ. PressCirincione J, Wolfstahl J, Rajkumar M. 2002. Deadly Arsenals: Tracking Weapons of Mass Destruction. Wash-

ington, DC: Carnegie Endow. Int. PeaceCohen A. 1999. Israel and the Bomb. New York: Columbia Univ. PressDai X. 2007. International Institutions and National Policies. New York: Cambridge Univ. PressDavis ZS. 1993. The realist nuclear perspective. Secur. Stud. 2(3/4):79–99Erickson JL, Way C. 2011. Membership has its priveleges: conventional arms and influence in the NPT.

In Causes and Consequences of Nuclear Proliferation, ed. R Rauchhaus, M Kroenig, E Gartzke. New York:Routledge. In press

Forland A. 1997. Norway’s nuclear odyssey: from optimistic proponent to nonproliferator. NonproliferationRev. 4(2):1–16

Frankel B. 1991. Opaque Nuclear Proliferation: Methodological and Policy Implications. London: Frank CassFrankel B. 1993. The brooding shadow: systemic incentives and nuclear weapons. Sec. Stud. 2(3/4):37–78Frankel B, Davis ZS, eds. 1993. The Proliferation Puzzle: Why Nuclear Weapons Spread and What Results. New

York: RoutledgeFuhrmann M. 2009a. Spreading temptation: proliferation and peaceful Nuclear Cooperation Agreements. Int.

Secur. 34(1):7–41Fuhrmann M. 2009b. Taking a walk on the supply side: the determinants of civilian nuclear cooperation. J.

Confl. Resolut. 53(2):181–208Harney R, Brown G, Carlyle M, Skroch E, Wood K. 2006. Anatomy of a project to produce a first nuclear

weapon. Sci. Global Secur. 14(3):163–82Hecker SS. 2010. Lessons learned from the North Korean nuclear crisis. Daedalus 139(1):44–56Hymans JEC. 2001. Of gauchos and gringos: why Argentina never wanted the bomb, and why the United

States thought it did. Secur. Stud. 10(3):153–85Hymans JEC. 2006. The Psychology of Nuclear Proliferation: Identity, Emotions and Foreign Policy. Cambridge,

UK: Cambridge Univ. PressHymans JEC. 2008. Assessing North Korean nuclear intentions and capacities: a new approach. J. East Asian

Stud. 8(2):259–92Hymans JEC. 2009. Proliferation implications of Atoms for Peace: a case study of the Yugoslav experience. Presented

at Nobel Inst. Symp. Peace, Stability, and Nuclear Order, Oscarsborg, Drobak, Norway, June 25–27Hymans JEC. 2010. The study of nuclear proliferation and nonproliferation: toward a new consensus? See

Potter & Mukhatzhanova 2010, pp. 13–37International Atomic Energy Agency. 2002. Fact sheet: Iraq’s nuclear weapon programme.

http://www.iraqwatch.org/un/IAEA/iaea-facts-042502.htmInternational Panel on Fissile Materials. 2010. Limiting national fissile-material production capabili-

ties. http://www.fissilematerials.org/ipfm/pages_us_en/inventories/limiting_production/limiting_production.php

Japan Times. 2010. Tokyo sought out Bonn on going nuclear in ’69. Nov. 30.http://search.japantimes.co.jp/cgi-bin/nn20101130a3.html

Jo DJ, Gartzke E. 2007a. Determinants of nuclear weapons proliferation. J. Confl. Resolut. 51(1):167–94Jo DJ, Gartzke E. 2007b. Data notes for “Determinants of Nuclear Weapons Proliferation.” http://jcr.sagepub.

com/content/vol51/issue1/images/data/167/DC1/Jo&Gartzke_(Feb_07).zip

242 Sagan


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.

http://www.iraqwatch.org/un/IAEA/iaea-facts-042502.htm

http://www.fissilematerials.org/ipfm/pages_us_en/inventories/limiting_production/limiting_production.php

http://www.fissilematerials.org/ipfm/pages_us_en/inventories/limiting_production/limiting_production.php

http://search.japantimes.co.jp/cgi-bin/nn20101130a3.html

http://jcr.sagepub

http://jcr.sagepub


Kroenig M. 2009a. Importing the bomb: sensitive nuclear assistance and proliferation. J. Confl. Resol. 53(2):161–80

Kroenig M. 2009b. Exporting the bomb: sensitive nuclear assistance and proliferation. Am. Polit. Sci. Rev.103(1):113–33

Kroenig M. 2010. Exporting the Bomb: Technology Transfer and the Spread of Nuclear Weapons. Ithaca, NY:Cornell Univ. Press

Kuntzel M. 1995. Bonn and the Bomb: German Politics and the Nuclear Option. London: Pluto PressLavergeon A. 2009. The nuclear renaissance: an opportunity to enhance the culture of nonproliferation.

Daedalus 138(4):91–99Lavoy PR. 1993. Nuclear myths and the causes of nuclear proliferation. Secur. Stud. 2(3&4):192–212Lavoy PR. 2004. Predicting nuclear proliferation: a declassified history. Strategic Insights 3(1).

http://fas.org/man/eprint/lavoy.pdfLester RK, Rosner R. 2009. The growth of nuclear power: drivers and constraints. Daedalus 138(4):19–30Levite A. 2003. Never say never again: nuclear reversal revisited. Int. Secur. 27(3):59–88Liberman P. 2001. The rise and fall of the South African Bomb. Int. Secur. 26(2):45–86May M, ed. 2001. Verifying the agreed framework. Cent. Global Secur. Res., Lawrence Livermore Natl. Lab.,

Cent. Int. Secur. and Coop., Stanford Univ., UCRL-ID-142036, CGSR-2001–001Mearsheimer JJ. 1990. Back to the future: instability in Europe after the Cold War. Int. Secur. 15(1):5–56Meyer SM. 1984. The Dynamics of Nuclear Proliferation. Chicago: Univ. Chicago PressMiller M. 1979. The nuclear dilemma: power, proliferation and development. Technol. Rev. 81(6):18–20Miller SM, Sagan SD. 2009. Nuclear power without nuclear proliferation? Daedalus 138(4):7–18Montgomery AM. 2005. Ringing in proliferation: how to dismantle an atomic bomb network. Int. Secur.

30(2):153–87Montgomery AM, Sagan SD. 2009. The perils of predicting proliferation. J. Confl. Resolut. 53(2):302–28Montgomery AM, Sagan SD. 2011. The perils of predicting proliferation. In Causes and Consequences of Nuclear

Proliferation, ed. R Rauchhaus, M Kroenig, E Gartzke. New York: Routledge. In pressMoravcsik A. 2000 . The origins of human rights regimes: democratic delegation in postwar Europe. Int.

Organ. 54(2):217–52Muller H. 2003. German national identity and WMD nonproliferation. Nonproliferation Rev. 10(2):1–20Muller H, Schmidt A. 2010. The little known story of de-proliferation: why states give up nuclear weapons

activities. In See Potter & Mukhatzhanova 2010, pp. 124–58Nuti L. 1993. ‘Me too please’: Italy and the politics of nuclear weapons, 1945–1970. Diplomacy Statecraft

4(1):114–48Paul TV. 2000. Power versus Prudence: Why Nations Forgo Nuclear Weapons. Montreal: McGill-Queen’s Univ.

PressPerkovich G. 1999. India’s Nuclear Bomb. Berkeley: Univ. Calif. PressPotter WC. 2010. The NPT and the sources of nuclear restraint. Daedalus 139(1):68–81Potter WC, Mukhatzhanova G, eds. 2010. Forecasting Proliferation in the 21st Century: the Role of Theory. Palo

Alto, CA: Stanford Univ. PressPotter WC, Mukhatzhanova G. 2010. The role of theory, an introduction. See Potter & Mukhatzhanova

2010, pp. 1–12Purkitt HE, Burgess SF, Liberman P. 2002. Correspondence: South Africa’s nuclear decisions. Int. Secur.

27(1):186–94Reiss M. 1995. Bridled Ambition: Why Countries Constrain their Nuclear Capabilities. Washington, DC: Woodrow

Wilson Cent. PressRublee MR. 2009. Nonproliferation Norms: Why States Choose Nuclear Restraint. Atlanta: Univ. Georgia PressRussian Federation Foreign Intelligence Service. 1995. The nuclear potential of individual countries: Romania.

http://www.fas.org/irp/threat/svr_nuke.htm#romaniaSagan SD. 1996/1997. Why do states build nuclear weapons? Three models in search of a bomb. Int. Secur.

21(3):54–86Sagan SD. 1999. Rethinking the causes of nuclear proliferation: three models in search of a bomb. In The

Coming Crisis: Nuclear Proliferation, U.S. Interests, and World Order, ed. VA Utgoff, pp. 17–50



Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.

http://fas.org/man/eprint/lavoy.pdf

http://www.fas.org/irp/threat/svr_nuke.htm#romania


Sagan SD. 2009a. Introduction. In Inside Nuclear South Asia, ed. SD Sagan, pp. 1–24. Stanford, CA: StanfordUniv. Press

Sagan SD. 2009b. Shared responsibilities for nuclear disarmament. Daedalus 138(4):157–68Sagan SD. 2010. Nuclear latency and nuclear proliferation. See Potter & Mukhatzhanova 2010, pp. 13–37Sasikumar K, Way C. 2009. Testing theories of proliferation in South Asia. In Inside Nuclear South Asia, ed.

SD Sagan, pp. _______. Stanford, CA: Stanford Univ. PressScheinman L. 1987. International Atomic Energy Agency and World Nuclear Order. Washington, DC: Resour.

FutureSchiff BN. 1984. International Nuclear Technology Transfer: The Dilemmas of Dissemination and Control. Totowa,

NJ: Rowman & AllanheldShaker MI. 1980. The Nuclear Non-Proliferation Treaty: Origin and Implementation 1959–1979. New York:

OceanaSimmons BA. 2000. Compliance with international agreements. Am. Polit. Sci. Rev. 94(4):819–35Simmons BA. 2010. Treaty compliance and violation. Annu. Rev. Polit. Sci. 13:273–96Simmons BA, Hopkins DJ. 2005. The constraining power of international treaties: theory and methods. Am.

Polit. Sci. Rev. 99(4):623–31Singh S, Way CR. 2004. The correlates of nuclear proliferation: a quantitative test. J. Confl. Resolut. 48(6):859–

85Socolow RH, Glaser A. 2009. Nuclear energy and climate change. Daedalus 138(4):31–44Solingen E. 1994. The political economy of nuclear restraint. Int. Secur. 19(2):126–69Solingen E. 1998. Regional Orders at Century’s Dawn: Global and Domestic Influences on Grand Strategy. Princeton,

NJ: Princeton Univ. PressSolingen E. 2007. Nuclear logics: contrasting paths in East Asia and the Middle East. Princeton, NJ: Princeton

Univ. PressSpector LS, Berman DR. 2010. The Syrian nuclear puzzle. In Forecasting Nuclear Proliferation in the 21st Century:

A Comparative Perspective, ed. WC Potter, G Mukhatzhanova, pp. 100–30. Stanford, CA: Stanford Univ.Press

Stoll JR. 1996. World production of latent nuclear capacity. http://es.rice.edu:80/projects/Poli378/Nuclear/Proliferation/proliferation.html

Swango DE. 2009. The Nuclear Nonproliferation Treaty: constrainer, screener, or enabler. PhD thesis, Dep. Polit.Sci., Univ. Calif. Los Angeles.

Tannenwald N. 2007. The Nuclear Taboo: the United States and the Non-Use of Nuclear Weapons. Cambridge,UK: Cambridge Univ. Press

Thayer B. 1995. The causes of nuclear proliferation and the utility of the nuclear nonproliferation regime.Secur. Stud. 4(3):463–519

Tomz M. 2007. Domestic audience costs in international relations: an experimental approach. Int. Org.61(4):821–40

U.S. Department of State. 2010. Adherence to and compliance with arms control, nonproliferation, and disarmamentagreements and commitments. http://www.state.gov/documents/organization/145181.pdf

von Stein J. 2005. Do treaties constrain or screen? Selection bias and treaty compliance. Am. Polit. Sci. Rev.99(4):611–22

Walsh J. 1997. Surprise Down Under: the secret history of Australia’s nuclear ambitions. Nonproliferation Rev.5:1–20

Weeks JL. 2008. Autocratic audience costs: regime type and signaling resolve. Int. Organ. 62(1):35–64Wohlstetter A, Brown TA, Jones G, McGarvey DC, Rowen H, et al. 1977. Swords from Ploughshares: The

Military Potential of Civilian Nuclear Energy. Chicago: Univ. Chicago PressWorld Nuclear Association. 2010. Uranium enrichment. http://www.world-nuclear.org/info/

inf28.htmlZentner MD, Coles GL, Talbert RJ. 2005. Nuclear proliferation technology trends analysis. Pacific Northwest

Natl. Lab. http://www.pnl.gov/main/publications/external/technical_reports/PNNL-14480.pdf

244 Sagan


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.

http://es.rice.edu:80/projects/Poli378/Nuclear/Proliferation/proliferation.html

http://es.rice.edu:80/projects/Poli378/Nuclear/Proliferation/proliferation.html

http://www.state.gov/documents/organization/145181.pdf

http://www.world-nuclear.org/info/inf28.html

http://www.world-nuclear.org/info/inf28.html

http://www.pnl.gov/main/publications/external/technical_reports/PNNL-14480.pdf


CHI

ISR

IND

S AFR

PAK

UKR, BEL, KAZ

-KAZ

-UKR, -BEL

-S AFR

NK

6

8

10

12

14

Nu

mb

er o

f st

ates

USA

UKRUS

FRA

0

2

4

Year

Figure 1Nuclear weapons proliferation (Sagan 2010). Figure c© Scott D. Sagan based on data from Singh & Way(2004). For this graph, the dates of the following countries’ nuclear programs were modified from Singh &Way on the basis of new research: Israel (Cohen 1999), India (Perkovich 1999, Abraham 1998), North Korea(Hecker 2010).

-LIB

SYR

-S AFR

-ALG, -ROM

-IRAQ

-BRA

-TAI

TAI, -YUGO ROM

IRAN

ALG

-IRAN

-TAI

-SK

IRAN, YUGO

IRAQ,-AUS

LIB

S AFR, -SWI, -SWE

AUS, TAI, -EGY

-YUGO, -AUS

SK

-ITA, -WGER

ITA, WGER

AUS

EGY

SWE, YUGO

BRA

SWI

0

1

2

3

4

5

6

7

8

9

1945

1950

1955

1960

1965

1970

1975

1980

1985

1990

1995

2000

2005

2010

Year

Nu

mb

er o

f st

ates

Figure 2Nuclear exploration, program, and reversal (Sagan 2010). Figure c© Scott D. Sagan based on data fromSingh & Way (2004). The dates of the following countries’ nuclear programs were modified for this graphbased on new research: Yugoslavia (Hymans 2009), Argentina (Hymans 2001), Libya (Cirincione et al.2002), Iraq (International Atomic Energy Agency 2002), Romania (Russian Federation Foreign IntelligenceService 1995), Italy and West Germany (Kuntzel 1995, Muller 2003, Muller & Schmidt 2010), Syria(Spector & Berman 2010, U.S. Department of State 2010), and Norway (Forland 1997).

www.annualreviews.org • Causes of Nuclear Weapons Proliferation C-1


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.


0

10

20

30

40

50

60

19501955

19601965

19701975

19801985

1990/2000

Year

Nu

mb

er o

f S

tate

s

States withNuclearWeapons

NuclearCapableStates

Figure 3Stoll’s nuclear latency data for 1992 (Stoll 1996).

1940 1950 1960 1970 1980 1990 2000

Syria

Romania

Iran (1) + (2)

Algeria

Iraq

Libya

South Africa

Taiwan (1) + (2)

North Korea

Egypt

South Korea

(West) Germany

Italy

Australia (1) + (2)

Sweden

Yugoslavia (1) + (2)

Brazil

Switzerland

Autocracy (-10 – 0)

Semi-Democracy (1 – 6)

Democracy (7 – 10)

Transi�on Period

: Ra�fied NPT

: Ra�fied NPT as ROC

: Ra�fied NPT, Withdrew in 2003

Figure 4Regime type, date of ratification of the Non-Proliferation Treaty, and nuclear exploration and pursuit. Figure c© Scott D. Sagan basedon data from Sagan (2010).

C-2 Sagan


Ann

u. R

ev. P

olit.

Sci

. 201

1.14

. Dow

nloa

ded

from

ww

w.a

nnua

lrev

iew

s.or

gby

Sta

nfor

d U

nive

rsity

- M

ain

Cam

pus

- G

reen

Lib

rary

on

04/2

5/11

. For

per

sona

l use

onl

y.

Documents

The Causes of Nuclear Weapons Proliferation · PL14CH11-Sagan ARI 18 February 2011 17:14 R E V I E W S I N A D V A N C E The Causes of Nuclear Weapons Proliferation Scott D. Sagan