This is the third contribution in our mini–symposium on what policymakers can learn from recent academic research into nuclear weapons. Colin H. Kahl is an associate professor at Georgetown University’s Edmund A. Walsh School of Foreign Service and a senior fellow and director of the Middle East Security Program at the Center for a New American Security. From 2009 to 2011, he was the U.S. deputy assistant secretary of defense for the Middle East.
In recent years, a new generation of scholars has increasingly turned to sophisticated statistical methods to tackle decades-old questions regarding the causes and consequences of nuclear proliferation, including nuclear deterrence, compellence and prospects for crisis escalation. Two of the most widely discussed “large-n” quantitative studies — Todd S. Sechser and Matthew Fuhrmann’s “Crisis Bargaining and Nuclear Blackmail” and Matthew Kroenig’s “Nuclear Superiority and the Balance of Resolve: Explaining Nuclear Crisis Outcomes” — appeared in the January 2013 issue of the flagship political science journal International Organization. Both studies use statistical methods to assess whether nuclear weapons demonstratively provide states with coercive power during international crises.
Based on an analysis of a dataset of 200 “compellent threats” from 1918-2001, Sechser and Fuhrmann argue that nuclear weapons states are no more successful at coercing adversaries than non-nuclear powers. Sechser and Fuhrmann contend that nuclear weapons are useful for deterrence purposes, but because atomic arms are useless for conquest and are extraordinarily costly to employ, these weapons have little utility in compelling adversaries. Nuclear blackmail, in other words, is a myth.
Kroenig disagrees. Analyzing a data set of 52 “nuclear crises dyads,” Kroenig concludes that states enjoying nuclear superiority more often than not succeed in backing weaker nuclear-armed adversaries down. It is worth noting that while the two studies appear to be diametrically opposed, they may not be. Indeed, it is theoretically possible that Sechser and Fuhrmann are right that nuclear weapons are poor tools of coercion, in general, because it is difficult to credibly threaten or use them to compel others, but in crises between nuclear powers, where the inherent risk of nuclear use is higher, nuclear superiority may still matter, as Kroenig suggests.
A detailed critique of these studies by Frank Gavin serves as the jumping off point for a new online H-Diplo/ISFF forum, “What We Talk About When We Talk About Nuclear Weapons,” in which several of the nation’s leading nuclear security scholars square off on the best methods for studying the nuclear revolution. Gavin suggests that a number of shortcomings plague Sechser and Fuhrmann and Kroenig’s studies, and quantitative nuclear studies more broadly, including: coding errors (both in terms of measuring key causal variables and appropriately defining the crisis outcomes under investigation); “selection effects” (i.e., the notion that many of the potential coercive effects of nuclear weapons are not captured in the data sets because compellence and deterrence manifest themselves before parties enter into a crisis or make explicit threats); the existence of potentially important causal variables that are omitted from the models (typically because they are difficult to quantify); and a failure to convincingly demonstrate that the posited causal dynamics in the models — as opposed to the reported correlations — are actually at work in important historical cases of nuclear crises.
Because Gavin believes that many of these flaws are inherent to the methodology employed, not just the particular studies, he is skeptical about the general usefulness of large-n statistical models for studying the effects of nuclear weapons on international relations. Elsewhere in the H-Diplo/ISFF forum, Scott Sagan and Vipin Narang also voice skepticism of quantitative approaches to nuclear studies. Sechser and Furhmann and Kroenig, in both joint and separate pieces, offer spirited rejoinders to Gavin, while Erik Gartzke’s contribution to the forum contends that, for all their shortcomings, large-n methods are still preferable to qualitative ones for studying nuclear politics.
I am sympathetic to many of the critiques of large-n approaches to studying the impact of nuclear weapons on coercion and crises outcomes (as well as Marc Trachtenberg’s critique here), although I also see merit in creative research designs that combine multiple methods. But instead of wading into this methodological quagmire, I want to raise another issue under-explored by the forum: even if Sechser and Fuhrmann and/or Kroenig’s findings are valid, should their conclusions change the way U.S. decision makers view, and respond to, the perceived dangers associated with emerging nuclear states?
Sechser, Fuhrmann, and Kroenig claim that their research has clear policy implications. Both studies, but especially the Sechser and Fuhrmann piece, seem to suggest that U.S. policy makers should be more sanguine about proliferation by relatively weak states than they currently are. Most obviously, if Sechser and Fuhrmann’s findings are valid, the United States should not worry that nuclear acquisition will fundamentally enhance the coercive power of regional adversaries such as Iran or North Korea, including their ability to blackmail American allies. For Kroenig, nuclear acquisition may complicate the projection of U.S. conventional power (which is one reason Kroenig’s other work remains generally pessimistic about the consequences of further nuclear proliferation, especially with regard to Iran). At the same time, so long as the United States continues to maintain nuclear superiority over potential nuclear-armed adversaries, such as North Korea or a future nuclear-armed Iran, Kroenig concludes that “Washington will frequently be able to achieve its basic goals in nuclear confrontations” short of war.
But here’s the problem from a policy-making perspective: regardless of whether nuclear weapons actually provide nuclear-armed states with greater capabilities and opportunities to engage in effective coercion, new nuclear states appear to believe they do, at least for some period of time, and act accordingly. At least some nuclear-weapons states appear to think a nuclear deterrent shields them from large-scale conventional retaliation from targets of coercion, tempting them to engage in more assertive military behavior below the nuclear threshold, including conventional aggression, low-level violence, proxy attacks, terrorism and the initiation of crises. And this pattern appears to hold even against stronger adversaries that enjoy nuclear superiority.
During the Cold War, for example, nuclear deterrence discouraged large-scale conventional or nuclear war, but the superpowers engaged in several direct crises, as well as proxy wars throughout the so-called Third World. Scholars posited that this was the result of a “stability-instability paradox”in which the very “stability” created by mutually assured destruction (MAD) generated greater “instability” by making superpower provocations, disputes and conflict below the nuclear threshold seem “safe.” More recently, nuclear weapons have similarly made the Indian-Pakistani rivalry more crisis-prone even as they discouraged large-scale war or a nuclear exchange.
The historical record also strongly suggests that states with “revisionist” aims become more aggressive — both directly and through the use of proxies — after acquiring nuclear weapons, at least for some period of time. Less than six months passed between the August 1949 testing of the first Soviet atomic bomb and Stalin’s green light to North Korean plans to invade South Korea. And, shortly thereafter, Moscow encouraged Ho Chi Minh to intensify his offensive against the French in Indochina. And, as Gavin’s research suggests, the development of thermonuclear weapons in 1955 and intercontinental ballistic missiles in 1958 also appear to have made Khrushchev more assertive, culminating in the 1958-1961 Berlin crisis. Similarly, five years after China became a nuclear power, Mao Zedong authorized Chinese troops to attack Soviet border forces in 1969. Archival evidence also suggests that Iraq’s quest for nuclear weapons was in part driven by Saddam Hussein’s desire to use them as a cover for conventional aggression against Israel. And, more recently, Pakistan’s emboldened support of anti-Indian terrorism and militancy and North Korea’s escalating provocations provide additional illustrations of the possible incentives and opportunities nuclear weapons create to advance a revisionist agenda.
Large-n quantitative studies on the emboldening effects of nuclear weapons have produced mixed results. On average, nuclear weapon states appear no more (or less) likely to become involved in international militarized disputes, or to initiate these disputes. But, with regard to interactions between nuclear states, Robert Rauchhaus finds that nuclear status increases the likelihood of low-level militarized disputes, including threats and the limited use of force, even as it reduces the chances of large-scale war. Time and learning may also play a key role. Michael Horowitz finds that the longer a state possesses nuclear weapons, the less likely it is to become involved in disputes. But new nuclear powers appear to be more prone to involvement in militarized disputes in the initial period of time after developing nuclear weapons against all types of states (including nuclear ones).
In short, regardless of whether nuclear weapons are objectively useful — or not — in coercion, at least some nuclear states — especially those with revisionist ambitions — seem to believe they are and act accordingly, even toward more nuclear-armed powerful adversaries. And, in many cases, it is precisely this type of adventurism by adversaries that so worries U.S. policymakers.
This was my experience observing the Obama administration’s deliberations on the potential dangers of Iranian nuclearization. U.S. officials believe Iranian nuclear acquisition would embolden Tehran — a state with both defensive and ideologically revisionist motivations — to be even more assertive in supporting terrorism, militancy and making coercive threats against its neighbors. They also fear that Iranian nuclearization would spark conventional and nuclear arms racing by other regional powers. Together, these dynamics would make an already volatile Middle East even more difficult to police and manage, requiring costly and complex U.S. deterrence and reassurance strategies and increasing the risk of irregular and conventional war.
The non-trivial risk of nuclear escalation in this environment also looms large in the calculations of U.S. policy makers. The worry is not so much that “mad mullahs” in Tehran would use nuclear weapons intentionally, but rather that high levels of mistrust, poor channels of communication, short missile flight times, uncertain command-and-control systems and destabilizing nuclear postures could contribute to inadvertent escalation between Iran and Israel or Iran and the United States in the midst of a crisis. The history of the nuclear age — and the powerful deductive logic of MAD emphasized by “proliferation optimists” — seem to suggest that the risks of this happening are low. But there are also enough examples of close calls and organizational mishaps throughout the nuclear age to suggest that the risks of miscalculation, unauthorized use or accidental escalation are not zero.
This last point raises an important difference in the way that political scientists and policy makers typically conceptualize risk in the nuclear context. Political scientists understand that all causal relationships are probabilistic. Statistical models search for regularities, but there are always outliers, and the best theories only account for broad patterns of behavior and outcomes most of the time. Yet, when it comes to catastrophic threats such as nuclear war, policymakers are simply not comforted by scholars’ claims that the emergence of new nuclear states will not result in higher levels of instability or risks of crisis escalation in the vast majority of instances. Instead, policymakers tend to see even miniscule risks of extraordinarily bad outcomes as compelling reasons to prevent additional nuclear proliferation.
As Peter Lavoy noted nearly 20 years ago, “even if . . . nuclear weapons do lower the likelihood of war by a considerable margin, can policymakers afford to allow the one or two exceptions? . . . [T]he 1 percent of exceptional cases of nuclear proliferation is what U.S. policymakers must worry about.”