Roundtable 5-3, “Weapons of Mass Migration: Forced Displacement, Coercion, and Foreign Policy”65 min read

Weapons of Mass Migration coverBy the accounts of the three reviewers below, Kelly Greenhill has hit a home run. Their collective view substantiates the judgment of the International Studies Association (ISA), which gave Weapons of Mass Migration the Association’s Best Book of the Year Award for 2011. In turn, the reviewers and the ISA have confirmed my judgment of four years ago that this is an especially important book in the field of security studies.



H-Diplo | ISSF Roundtable, Volume V, No. 3 (2013)
H-Diplo/ISSF Editors: Thomas Maddux and Diane Labrosse
H-Diplo/ISSF Web and Production Editor: George Fujii
Commissioned for H-Diplo by Thomas Maddux
Kelly M. Greenhill. Weapons of Mass Migration: Forced Displacement, Coercion, and Foreign Policy. Ithaca, NY:   Cornell University Press, 2010.   ISBN: 978-0-8014-4871-3 (hardcover, $35.00).
Published by H-Diplo/ISSF on 15 April 2013

Introduction by Robert J. Art, Brandeis University

By the accounts of the three reviewers below, Kelly Greenhill has hit a home run. Their collective view substantiates the judgment of the International Studies Association (ISA), which gave Weapons of Mass Migration the Association’s Best Book of the Year Award for 2011. In turn, the reviewers and the ISA have confirmed my judgment of four years ago that this is an especially important book in the field of security studies. For purposes of full disclosure, I co-edit, along with Robert Jervis and Stephen Walt, the series in which this book appears (the Cornell Series in Security Studies), and I was the editor, along with an outside reviewer, who assessed the manuscript for publication. I thought then, and am happy to have confirmed since, that Weapons of Mass Migration is an innovative, original project.

Weapons of Mass Migration examines an apparently widespread and, by international standards, reasonably frequent phenomenon — “coercive engineered migration” (CEM), or “migration-driven coercion” — but one that had largely escaped systematic scholarly notice until Greenhill’s study. Greenhill has uncovered 56 clear-cut cases of attempted CEM and 8 other possibles from 1951 — the advent of 1951 Refugee Convention — until 2006 — the cutoff point of her research. (See Table 1.1 in the book for a list of the sixty four cases.) More often than not, CEM is a strategy employed by weak states or non-state actors (the coercer or challenger) against stronger ones (the state target) to force the target to bend to the will of the coercer. CEM therefore involves the deliberate creation and manipulation of migration and refugee crises to achieve a political end.

The CEM strategy is especially useful against democratic states because of their susceptibility to what Greenhill terms “hypocrisy costs” (4-5): the gap between a democracy’s commitment to liberal norms and its concrete actions to support them. If democracies preach the importance of protecting human rights and rescuing people who have had theirs savagely denied, then such governments find themselves particularly hard pressed when presented with the choice of taking in large numbers of refugees and migrants who have been forcibly sent there, or of rejecting them. They are torn by their split publics — between those who want to uphold the liberal norms and those who do not want to pay the costs of upholding them. Consequently, democratic states are highly susceptible to blackmail by such coercers. Herein, Greenhill asserts, lies a validation of the constructivist claim that norms can be powerful in international politics. Norms, not material power, can be leveraged by weak states, which, by definition, are inferior in material power, against strong ones.

For readers of the realist persuasion, Greenhill also has interesting things to say about CEM. Two points, in particular, stand out. The first has to do with how CEM works; the second, with the technique’s success rate.

Ever since Robert Pape’s path breaking study on the coercive use of airpower, the conventional wisdom has been that coercion works best by military denial, not by actual punishment or by the threat (risk) of punishment.[1] That is, what works best to compel an adversary to change its behavior is the coercer’s ability to convince the adversary that the latter’s theory of military victory will not work. The way to achieve this is to demonstrate that one has the military capability to thwart the adversary’s military forces from achieving their objectives. Coercion achieved through punishment inflicted by conventional bombing, or coercion achieved though the threat of such punishment, were not nearly as successful as coercion achieved through military denial. Punishing civilians was less successful than preventing the adversary’s military forces from attaining their objectives.

In contrast, Greenhill finds that when CEM works, it does so, not through denial, but through punishment or the threat of punishment. Punishment (or the threat of punishment) is inflicted (or threatened) not only on the migrants and refugees who have been forced to move, but, more importantly, on the state target(s) to which they have been forced to move. The latter are punished by having to accept the costs of taking the displaced people, or by having to bear the costs of engaging in political contortions to explain why they are not taking them. Having the courage of one’s convictions (taking the migrants and refugees), or suffering the hypocrisy costs of not taking them (international public shaming and torn domestic publics), are both forms of punishment. In a military-to-military confrontation, the weaker coercer could never hope to compel the stronger target to bend to its will. Through the threat of punishment or inflicted punishment via forced migration, however, it can. This calls for a revision of Thucydides’ famous dictum from the Melian Dialogue: “the weak do what they can and the strong suffer what they must.”[2]

Second, and equally noteworthy, the success rate of CEM is quite high, especially considering that many of Greenhill’s cases involve a weak coercer against a strong target. According to Greenhill’s coding, based upon the number of clear-cut CEM cases she located, CEM achieved complete success (achieved all of its objectives) 57% of the time, and partial success (achieved some but not all of its objectives) 73% of the time.

How do we interpret these two success rates? That is, how does this newly-studied strategy or instrument of statecraft compare to other instruments of statecraft — coercive diplomacy, sanctions, the coercive use of airpower, and extended deterrence – for which we have some quantitative data about success rates?

First, we need to note that Greenhill follows the convention set by Thomas Schelling.[3] Coercion comes in two forms, argues Schelling:   deterrence and compellence. The purpose of the former is to prevent an adversary from undertaking a certain action or set of actions; the purpose of the latter, to get an adversary to change the action or set of actions that it is currently engaged in. Deterrence seeks to prevent a change to the status quo by stopping the adversary from undertaking an action; compellence, to produce a change in the status quo by getting the adversary to stop doing what it is doing.[4] Following Schelling’s convention, Greenhill’s CEM data include three types of cases: instances where only deterrence was tried, instances where only compellence was tried, and, finally, instances where both deterrence and compellence were tried. The 57% and 73% success rates lump these three types together, and, as a consequence, we cannot determine what the success rates are, respectively, for the deterrence only, and the compellence only, cases, nor whether compellence worked in those cases where deterrence was tried first and failed, unless we sort Greenhill’s cases by these three types, something which can be done but which neither she nor I have done.

How do CEM success rates stack up against those other instruments and strategies for which we have some quantitative data? For extended deterrence, pretty well. The best quantitative data we have on extended deterrence comes from the studies by Bruce Russet and Paul Huth. They found a success rate of 57% for extended deterrence.[5] For coercion through airpower, Pape located 40 cases of coercive air campaigns from 1917 through 1991 that involved military denial, punishment, or both, and, for our purposes, 13 cases where the threat of military denial or actual military denial through conventional bombing was high or very high. In those 13 cases, conventional military denial was successful 10 times, for a success rate of 77%.[6] This rate is probably too high since in 4 of these 10 denial successes, punishment was also high or very high. If we take out the 4 cases where both punishment and denial were high or very high, conventional military denial has a success rate of 46% (6 military denial only successes out of 13 total cases where denial was high or very high). This is a little higher than the results of the study by Michael Horowitz and Dan Reiter of the coercive use of airpower. Out of 53 cases from 1917-1999, they found a success rate of 36% (and a failure rate of 64%).[7]

Coercive diplomacy is an instrument that involves threats to use force, or that uses only small amounts of force, to get an adversary to change its objectionable behavior. In their landmark study of coercive diplomacy, Alexander L. George, David K. Hall, and William E. Simons found, in a small number of cases, a success rate of about 30%.[8] Patrick M. Cronin and I found similar results on a slightly larger set of cases: coercive diplomacy worked about one third of the time.[9] In the most comprehensive study to date, Todd Sechser concluded that coercive diplomacy worked 41.4% of the time.[10] All three coercive diplomacy data sets, as well as Pape’s airpower study, are concerned only with compellence, not deterrence.[11] Finally, the literature on economic sanctions tells us that they work roughly one third of the time to produce a change in the target’s behavior, a success rate roughly in line with the success rates we have for coercive diplomacy. Sanctions worked 33% of the time to produce a modest change in the target’s behavior and 25% of the time to produce a major policy change.[12] Thus, compared to the success rates of other strategies and instruments of statecraft for which we have some data, the two success rates for CEM stack up quite well.

Finally, over half a century ago, Schelling told us in Arms and Influence that deterrence is easier to achieve than compellence. That may still be true, but if Sechser is correct and if the Huth-Russett data is sound, then the difference in success rates between coercive diplomacy, an especially hard form of compellence since it relies on threats or only small amounts of force if force is used, on the one hand, and extended deterrence, which is tougher to make work than homeland deterrence, on the other, is smaller than one might have expected –41% versus 57%. Greenhill could do us a great favor by sorting her cases among the three types noted earlier in order to provide us with more evidence to bear on the relative ease of compellence versus deterrence and in order to test whether CEMs differ markedly is this respect from these other strategies.

I have dwelt on how Greenhill’s work fits into the literature on coercion in order to situate her work in the field of security studies. The three reviewers concentrate on other aspects of her work. Saleyhan and Rudolph provide more detailed descriptions of what the book is about, plus sharing reservations about the frequency of the CEM phenomenon and the lack of a more rigorous quantitative analysis. Finally, Crawford focuses on three issues:   whether CEM is really useful for the weak, what particular problems democratic targets of CEM face, and what validity Greenhill’s recommendations have to lessen democracies’ vulnerability to CEM campaigns. Whatever their criticisms, however, all three reviewers note the significance, rigor, and originality of Greenhill’s book.



Kelly M. Greenhill is Associate Professor at Tufts University and Research Fellow at Harvard Kennedy School of Government’s Belfer Center for Science and International Affairs. In addition to Weapons of Mass Migration, Greenhill is co-author and co-editor of Sex, Drugs and Body Counts: The Politics of Numbers in Global Crime and Conflict (Cornell University Press, 2010). Her research has also appeared in a variety of other venues, including in the journals International Security, Security Studies, Civil Wars, and International Migration, in media outlets such as the New York Times, the Los Angeles Times, the International Herald Tribune, and the British Broadcasting Company, and in briefs prepared for the U.S. Supreme Court and other organs of the U.S. government. She is currently completing a new monograph, a cross-national multi-method study that explores why, when, and under what conditions, contested sources of political information—such as rumors, conspiracy theories, myths and propaganda—materially influence the development and conduct of states’ foreign and defense policies.
Robert J. Art is Herter Professor of International Relations at Brandeis University and Director of MIT’s Seminar XXI Program. He is the author of a number of articles and books including “The United States and the Rise of China: Implications for the Long Haul,” Political Science Quarterly 125:3 (2010): 359-391; A Grand Strategy for America (Ithaca: Cornell University Press, 2003); and America’s Grand Strategy and World Politics (New York: Routledge, 2009).
Timothy W. Crawford is associate professor of political science at Boston College. He is the author of Pivotal Deterrence: Third-Party Statecraft and the Pursuit of Peace (Ithaca, NY: Cornell University Press, 2003), and co-editor, with Alan J. Kuperman, of Gambling on Humanitarian Intervention: Moral Hazard, Rebellion, and Civil War (Routledge, 2006). His most recent work, which explores the role of wedge strategies in alliance politics, includes “Preventing Enemy Coalitions: How Wedge Strategies Shape Power Politics,” International Security (Spring 2011) and “Powers of Division: From the Anti-Comintern, to the Nazi-Soviet and Japanese-Soviet Pacts, 1936-1941,” in Jeffrey Taliaferro, Steven Lobell, and Norrin Ripsman, eds., The Challenge of Grand Strategy: The Great Powers and the Broken Balance Between the World Wars (Cambridge University Press, 2012).

Christopher Rudolph (PhD UCLA 2001) is Associate Professor of Political Science and International Relations at the School of International Service at American University (Washington, DC).  He is the author of National Security and Immigration (Stanford University Press, 2006), and his work has appeared in the American Political Science Review, International Organization, Security Studies, and International Studies Review, among others.

Idean Salehyan is an Associate Professor of Political Science at the University of North Texas. In addition, he is an Associate at the John Goodwin Tower Center for Political Studies at Southern Methodist University, at the Peace Research Institute Oslo, and at the University of Texas’ Robert S. Strauss Center for International Security and Law. Dr. Salehyan is the author of Rebels Without Borders: Transnational Insurgencies in World Politics (Cornell University Press, 2009), and his articles appear in journals such as the American Journal of Political Science, Journal of Politics, International Organization, International Studies Quarterly, and World Politics.  His current research interests include environmental security, external intervention in civil wars, and forced migration. Dr. Salehyan holds a Ph.D. in Political Science from the University of California, San Diego.

Review by Timothy W. Crawford, Boston College

Professor Idean Salehyan’s review (which also appeared in International Studies Review[13] ) nicely summarized the main arguments of Weapons of Mass Migration and outlined the strengths and limitations of its quantitative and qualitative evidence. He also rightly applauds the book’s considerable theoretical and analytical advances.[14] Rather than retrace Salehyan’s steps, I shall move directly to three major themes arising from the book.

The first concerns the notion that “coercive engineered migration” [hereafter CEM] has particular utility as a ‘weapon of the weak’(2,13).   The second concentrates on the particular strategic problems facing liberal democratic states targeted by CEM, especially those confronting multiple and potentially simultaneous challenges. The third addresses the question raised in Greenhill’s conclusion, as to what liberal democratic states in particular might do to mitigate their vulnerability to such campaigns.

Weapon of the Weak or Weapon against the Weak?

Greenhill emphasizes the general claim that CEM tends to be a weapon of the weak. In particular, “generator” governments, and non-state agents provocateurs, use it as a coercive tool against target states that are much stronger in relative military and economic terms (23, 27-29). This generalization is derived from Greenhill’s data, which shows that most CEM attempts are undertaken by actors much weaker than their target in gross measures of relative power. Greenhill buttresses the general claim with a theoretical logic that expresses a comparative utility calculus. She argues that CEM has many serious limitations—it is a “blunt instrument” of coercive statecraft and a “pretty poor method of persuasion,” because it is hard to control the imposition of costs and to avoid strategically perverse consequences (36, 5). Still, it may have utility for actors with an impoverished menu of options: as Greenhill puts it, “the instrumental generation or manipulation of migration crises can be an attractive method of influence for those with limited resources and few other options at their disposal” (30). But keep in mind that this comparative utility logic does not point to high, or even average, rates of coercive success. Indeed, the thrust of the logic lies in its ability to explain why an instrument of statecraft with poor prospects of success may still be rationally used.[15] Here lies a tension in Greenhill’s theoretical account, for she then goes on to demonstrate empirically—and argues at considerable length—that CEM manipulators actually do pretty well, with a surprising success rate (73% if you include “partial successes” and 57% if you only count clear successes) that matches or beats other techniques of coercive pressure (32). And so the picture shifts from an understanding of CEM as a kind of desperate, grasping at straws, to a cost-effective tool of influence with “an impressive rate of success,” given a certain kind of target and policy-contingency context.[16]

Describing and explaining the nature of that context is, in fact, a central focus of Greenhill’s theoretical enterprise and indeed the book overall. What this boils down to is a constellation of factors in the target’s domestic politics and international legal commitments that make its leaders prone to cave-in to CEM attempts. Specifically, Greenhill argues that CEM is more likely to work against targets that are liberal in both the political institutional and normative senses.[17] These “soft-liberal” targets are “particularly vulnerable” to CEM, because their institutional and normative characteristics interact—and can be “actively manipulated”—in ways that “offer would-be coercers powerful bargaining leverage” (64, 63). Thus, CEM practitioners can put liberal democratic leaders in a domestic political vice—with mobilized immigration opponents and NIMBY resistors on one side and mobilized refugee supporters and human rights activists on the other—from which the leaders cannot escape without concessions to the active manipulators.   The coercive pressure is “multiplied” by liberal leaders’ exposure to “hypocrisy costs” that inevitably arise when their actual dealings with refugees fall short of the humanitarian norms their states have strenuously espoused, and indeed have done much to enshrine in international society. Clever CEM challengers “exploit these norms” and manipulate “hypocrisy costs for their own benefit.” (271). For this reason, Greenhill argues, liberal democracies are “particularly (but not exclusively)” vulnerable to CEM and “have been most frequently (and most successfully) targeted” by it (63,13).[18]

Although Greenhill is ambivalent about which of the two facets of liberalism—pluralistic political institutions or normative commitments—matters more in creating this vulnerability, I would argue that the main driver is institutional, for it is the democratic leader’s exposure and sensitivity to escalating and conflicting societal pressures (including those that express liberal normative concerns) that causes it to knuckle under. The more leaders are insulated from domestic political pressures, the more easily hypocrisy costs—like other kind of audience costs—are borne, or even erased, by authoritative ‘spin control’.[19] That said, however, it may be easier for liberal states to adjust the normative drivers of the vulnerability than the political institutional ones and for that reason, the normative dimensions of the dynamic may matter more from an operational policy perspective—a point to which I will return later.

In the political economy literatures that conceptualize liberal states that are highly exposed to conflicting, pluralistic, “societal” pressure, those polities are called “weak states.”[20] In this light, one could say that Greenhill has identified a coercive instrument of particular utility not so much for weak states as against weak states. This is not merely a semantic jab: the point is that state weakness (in both the state vs. domestic society and state vs. international society senses) makes all soft-liberal democratic states vulnerable, regardless of where they stand in the distribution of power internationally (or, to be more specific, in a challenger-target dyad). Furthermore, the particular vulnerability of soft-liberal states should, in theory at least, make them attractive targets for CEM attempts not just by the underdogs of international politics but by more powerful states as well.   If CEM is particularly efficacious against soft-liberals—more so than, say, military or economic sanctions—then it should be an attractive tool even for countries that have a wider range of tools of influence at their disposal. Indeed, at least in theory, it could have particular utility as a weapon that large and populous states with little sensitivity to liberal norms can use against smaller, less populous, soft-liberal states.

Because the major theoretical thrust of Weapons of Mass Migration is to elucidate the particular vulnerabilities of soft-liberal states to CEM, and the agitation mechanisms that challengers use to exploit these vulnerabilities, a useful starting point for extending Greenhill’s framework is to categorize soft-liberal states as the main target of CEM, and to then think about the things that might, given their commonalities, increase or decrease their vulnerability.[21] On the first matter, we should look to number and proximity of potential CEM threats. On the second, to the opportunity for soft liberal states to engage in a mutually supportive project of normative recalibration, since hypocrisy costs are primarily self-inflicted or fratricidal wounds. To those two matters we now turn.

Hypocrisy Costs and the N+1 Challenger Problem

One important way to further gauge the vulnerability of soft-liberal targets is suggested in Greenhill’s reference to the potential for target concessions to encourage subsequent challenges (129, 273), what she has elsewhere referred to as the problems of “recidivists” and “copycats”.[22] While these two problems are related and may overlap, they are also distinct. Obviously, states that face more than one potential CEM threat are more vulnerable than those that face more than one. But going beyond the obvious additive point here, we may posit that soft-liberal targets face significantly greater levels of vulnerability to CE, when they are exposed to multiple potential CEM generators. For targets that only face one potential challenger, the recidivist risk presents an inter-temporal trade-off—reduce costs now through appeasement, but risk higher costs latter by emboldening subsequent blackmail efforts by the same generator.   For the targets that face more than one, the problem has both longitudinal and latitudinal dimensions. Appeasing challenger A in round 1 may encourage challengers A, or B, or A and B both, to try blackmail in round 2.[23] Furthermore, just seeing challenger A initiate a crisis may be enough to trigger challenger B to do the same, because simultaneous moves may escalate the agitation mechanisms within soft-liberal states that increase the challengers’ leverage. In short, piling on confers advantages.

Thus, as Greenhill’s accounts of the CEM crises generated by Fidel Castro and Jean-Betrand Aristide in 1994-95 demonstrate, such challenges intersect not just temporally but politically and strategically, and create an especially nasty problem for soft-liberal states. That is because the transparent domestic politics of soft-liberal states sharply curtail their leaders’ ability to deal with the separate generators in different, strategically expedient (and alas morally inconsistent) ways. Here is where “hypocrisy costs” really bite.   Thus, both Castro and Artistide got leverage over U.S. presidents by forcing into the spotlight the contradiction between U.S. policy on Haiti’s “economic” refugees and Cuba’s “political” refugees (119, 127, 201-205). Both Castro and Artistide played off of the deep refugee fatigue concentrated in Florida that they together had created (112, 206).[24]

We can see in these circumstances how soft-liberal states—with their wide-open domestic politics putting leaders betwixt angry anti-immigrant activists and strident humanitarians advocates and immigrant supporters—might be particularly vulnerable to swarm-ball CEM campaigns. In sum, if the N+1 challenger problem is worse than the N problem for all potential targets, it is especially so for soft-liberal states.

How a common understanding of this context would influence the incentives and bargaining power of the parties is an interesting question. On the one hand, it might increases the upfront leverage of the “first move” generator, in so far as the target’s leaders have high incentives to avert an escalation that attracts others. On the other hand, and more importantly I think, it should ratchet-up the target’s incentives to demonstrate resolve by defying the pressure and stimulate efforts to refashion the normative framework that cedes so much power to challengers. And this is an effort other soft-liberal states would have a strong interest in supporting.

The Demand for a Regime against Migration Coercion?

Let’s assume, then, that Greenhill’s insights will become widely known. And let’s posit, as Greenhill suggests, that there are many other potential CEM generators and agents provocateurs out there looking to emulate Fidel Castro’s and the Kosovo Liberation Army’s successful tactics. The targets most likely to bear the brunt of this diffusion of coercive practices are, of course, the soft-liberal states facing N+1 potential generators. Here it is useful to return to Greenhill’s hypocrisy cost argument and note that it is, in at least one respect, a kind of reputation argument familiar to students of security studies and diplomatic history—it shows how compromising and counterproductive state policies can be that are heavily influenced by fears of a loss of prestige.[25]

The tension between concerns about hypocrisy costs and concerns about rewarding CEM blackmail, then, is a contest between reputation concerns. It is a tension between making a bad impression on a friendly audience, by adopting a tough and normatively ‘inappropriate’ response, and giving potential CEM challengers the perverse impression that one is an easy mark (179). A key dimension of this scenario is that soft liberal states and the international civil society they support are the primary generators of the hypocrisy cost pressure. It is inter-subjective pressure. We should not exaggerate, as some have, the ability of illiberal despots to ‘appropriate’ liberal norms and use them against liberal states. Liberal publics and their leaders are much more sensitive to their in-group reputations than to judgments about the ‘appropriateness’ of their policies coming from authoritarians in politically backward corners of the world. It is also the case that the members of the liberal community are the major innovators and custodians of liberal norms and regimes in the international community.

If Greenhill is right, then in her assessment that these kinds of attempts are on the upswing, we should expect then that there will emerge among liberal states increasing demand for change in the international norms and regimes that make them vulnerable to and magnets for CEM. This points to a kind of response that is not addressed in Greenhill’s menu of prescriptions for states “faced with the threat of coercive engineered migration” (276-280). Greenhill highlights the possibility that leaders can promote a more tolerant public attitude toward migrants, thus softening one half of the domestic political vice that CEM practitioners exploit. She also notes the possibility that individual states will try to “abrogate relevant humanitarian norms, either by underlining national security concerns or by refusing to recognize those fleeing as worthy of protection” (278). But Greenhill largely treats the constraints of international liberal norms, regimes, and activists as given. What is missing is the multilateral intergovernmental option. Yet it is that kind of approach, undertaken by leading members of the liberal security community, that offers some purchase on the problem of hypocrisy costs.

If liberal states and their publics are the major generators of the normative scorn that activates hypocrisy costs, many of them are also in the same boat, as attractive targets of multiple potential CEM threats, and desired destinations for economic and political refugees. Liberal states are unlikely to remain passive and trapped in this position: over time, they are likely to try to adjust the balance of normative forces and regime constraints that have made them particularly vulnerable to CEM. Looking ahead, we may see liberal state leaders develop a new inter-subjective architecture that enables them to describe the strategic interaction of CEM in a way that reduces the power of the agitators, by enabling targets to ‘name’ CEM when it occurs, and ‘shame’ not just the practitioners, but those who abet them by pressuring leaders to capitulate to CEM blackmail.

To be more concrete, if we think about the relationship between CEM practitioners and targets as an evolving strategic relationship, it is reasonable to foresee that leaders in the community of liberal states will work toward a shared recognition of migration coercion as a common threat, and promote a norm that recognizes the collective value of standing firm against migration blackmail. Soft-liberal states, in this view, would be doing the liberal community a favor when they resist CEM, and they will be more likely to give each other normative cover when they defend their policies to each other in these terms. In this way, soft-liberal states might begin to inoculate each other against some of the threat of CEM by making it not only acceptable to publicly describe a situation as one involving migration coercion, but praise-worthy to define one’s policy as refusing to reward such blackmail. In this vein, we might even find that liberal state leaders claim to be defending the broader interests of the true victims, since suppressing the market for CEM will in the long run discourage others from manipulating and exploiting the poor and suffering people that make up the “human demographic bombs” (3).

In sum, this new normative framework, emerging in response to the growing threat of CEM, might consist of three planks. The first would the recognition that there is a problem: CEM blackmail is a widespread problem and it tends to be targeted at liberal states. Indeed, perhaps one the most important contributions of Greenhill’s book will be to give the policymaking strata of soft-liberal states a conceptual framework for identifying and thinking systematically about the nature of this shared strategic problem.[26] The second plank would assert that CEM should be discouraged, because its tendency will otherwise be to escalate and spread. The third plank would hold that standing firm against CEM is praiseworthy because it helps to establish expectations that CEM does not pay, which in turn will help to protect both innocents and the broader community of liberal states from exploitation. These normative adaptations would reduce the ability of CEM generators to leverage hypocrisy costs in their coercive strategies. They would also push those within the targeted countries whose agitation supports CEM attempts to reconcile their position with its larger political and strategic implications. Finally, they may even lay the normative foundations for mobilizing multilateral sanctions against the practitioners of CEM.

Of course, such hypothetical developments are unlikely to occur if CEM does not, in fact, become an increasing challenge for liberal states. And even if they did occur, they would hardly eliminate CEM because, as Weapons of Mass Migration makes clear, the sources and drivers of it go beyond the peculiar vulnerabilities of soft liberal states. Moreover, if they did occur, there could also be an unfortunate pattern in which liberal state leaders cynically abuse the anti-CEM normative frameworks to stop migrant flows that are not propelled by coercive manipulators. But it is conceivable that this will not often be the case and that the liberal states that are the strong magnets for immigration—of both “normal” and coercive provenance—will better be able to distinguish between those kinds of population flows and calibrate their policy responses accordingly.

Review by Christopher Rudolph, American University

Given the significant and complex effects migration has on issues of interest to international relations scholars, it’s surprising (and lamentable) how little attention it has received in the sub-field. Fortunately scholars have started to address that lacuna over the past decade and Kelly Greenhill’s book, Weapons of Mass Migration, makes a solid contribution to the emerging International Relations (IR) literature on the subject.   In addition to broadening the scope of IR scholarship, it justifies the unique contribution IR frameworks can have on migration and refugee studies.

Greenhill sets out to explain a phenomenon that many, including migration experts, did not believe existed: the use of state-engineered migration as a tool of coercive statecraft in the post-WWII era. Indeed, as she rightly points out, this rather insidious means of political suasion has been used numerous times over the relatively short period examined, and with a striking degree of success. The book recounts sixty-four instances where, as she puts it, “displaced people have been used as pawns” in international statecraft, involving as many as 10 million people in a single occurrence. Moreover, Greenhill’s relatively conservative calculus includes only those instances where documentation clearly supports both intent and execution of this tactic, although there is considerable anecdotal evidence to suggest that its use is even more widespread. While she notes that those states which engage in these acts of coercive diplomacy (and are often weaker than their targets) have succeeded in achieving at least some of their political aims seventy-one percent of the time, it is the displaced who suffer the human costs involved. This is an issue with clearly significant implications for interstate relations, human rights, and human security.

The book addresses two central questions, the first largely descriptive, the second more theoretically grounded: how often is forced displacement used for coercive diplomacy, and how, why, and under what conditions does it succeed or fail? Providing a thorough descriptive account of a largely overlooked phenomenon would itself make a contribution to the literature. The book provides both an extensive accounting of the cases where migration was used as a tool of coercive diplomacy (in the appendix) as well as in-depth case studies of several prominent examples (chapters 2-5).   These cases include migration from Cuba, the Balkans, Haiti, and North Korea. These historical chapters offer a rich and detailed account of the events, the context in which they took place, and their wider implications.   Theorists might prefer that the large-n dimension of the book would be dedicated to a broader testing of the hypotheses generated instead of the more descriptive accounting of the instances where coercive engineered migration was identified by the author. Instead, the theoretical arguments are tested against a rather small sample of cases using the structured, comparative case method.   While the scope of testing is thus limited, the book offers an exquisitely detailed account of events.

Greenhill responds to her second set of questions by arguing that “coercive engineered migration can be usefully conceived of as a two-level game, generally asymmetric, coercion by punishment strategy, in which challengers on the international level seek to influence the behavior of their targets by exploiting the existence of competing domestic interests within the target state(s) and by manipulating the costs or risks imposed on their civilian population(s)” (3). Coercion is exercised either by threatening to overwhelm the capacity of the target country to logistically accommodate a large influx of migration and/or by using migration to create domestic political instability by exacerbating existing tensions between domestic interest groups. To these causal mechanisms, Greenhill adds a dimension of “hypocrisy costs”(4), which are created when treatment of migrants and refugees by receiving states contradicts their professed commitment to liberal norms and values. Not surprisingly, this should make liberal democracies more vulnerable to coercive engineered migration—a hypothesis supported by the empirical evidence drawn from the post-WWII era presented in the book.

Though not suggested in the book, one might also consider how succumbing to coercive engineered migration would affect domestic perceptions of the target state’s capacity to regulate migration, and hence, to its very sovereignty.   In National Security and Immigration [27], I showed how public perceptions of a weakening capacity of state control heightened insecurities regarding immigration/refugee flows in Europe and the United States over the past half-century and contributed to growing political mobilization of groups favoring more restrictive policy. Indeed, concerns regarding an erosion of what I referred to as “societal sovereignty” could produce a strong shift in domestic political coalitions since its appeal is not limited to the right wing.[28] Mikhail Alexseev documented a similar effect on domestic politics, and explained the effect using the framework of the security dilemma to account for the spiral logic of threat perception.[29]

There are also other likely candidates for factors that strongly affect a state’s vulnerability (in terms of political sensitivity). For example, kinship ties may factor prominently. A large population of ethnic kin present in the target could significantly impact its willingness to accept the migrants/refugees. Conversely, we would expect the opposite to be the case if there were historical grievances between a sub-population of the target state and the incoming migrants.   Receptivity/vulnerability to engineered migration may also be strongly affected by the skill composition of the migrants and/or the economic conditions and economic needs of the target country. In short, there are numerous likely factors that shape the politics of migration policy domestically that can affect the type of statecraft outlined in the book.

One might also wonder why geography did not figure more prominent as a variable.   Although decreasing transportation costs partly explain the increasing size and scope of migration flows (Massey et al. 1998), geography still plays a major role in any migration equation.[30] In the sense used here, we might think of geography as having a strongly constraining effect on the “force projection” of threats regarding coercive engineered migration.

Greenhill categorizes the actors who engage in coercive engineered migration into three groups: generators, agents provocateurs, and opportunists. If we are to speak of coercive engineered migration sui generis, the inclusion of the latter category seems inherently problematic.   Greenhill defines “opportunists” as states which “play no direct role in the creation of the migration crisis, but simply exploit for their own gain the existence of outflows generated or catalyzed by others” (30). While these actors may seek to exploit a situation that has presented itself, they do not, in fact, engineer either the migration or the crisis. Given that more than one third of the cases (twenty-six out of sixty four 64) fit this category, this might reduce our perception of the significance of coercive engineered migration since it magnifies the degree of frequency of the phenomenon. In terms of both international relations and human rights, it would seem that generators and agents provocateurs are the more insidious actors, since they are in fact producing the migration flow.

In addition to adding to the IR literature on statecraft more generally, the book offers a glimpse of the international politics of smaller states, and in particular, their relations with larger, more powerful countries.   As Greenhill notes, “Crisis generation represents one of the few areas in which weak actors may possess relative strength vis-à-vis their targets—and, in the case of migration crises, also vis-à-vis their even weaker victims” (28).

Weapons of Mass Migrations is innovative, well-written, rigorously researched, and timely. It is both theoretically innovative and policy-relevant, and will likely spur several new paths for IR research and migration studies.

Review by Idean Salehyan, University of North Texas

Refugees as Weapons of the Weak (This review appeared originally in International Studies Review (2010) 12, 640-642 and is reissued with the author’s permission)

Until recently, International Relations (IR) scholars hardly paid attention to migration and refugee flows as serious subjects of study. Refugees were considered to be the unfortunate by products of war and political violence rather than central to the dynamics of conflict within and between states. In Weapons of Mass Migration, Kelly Greenhill offers a superbly written and thoroughly researched analysis of a phenomenon that has barely received scholarly attention: coercive engineered migration. As opposed to refugee flows that are rooted in domestic conflicts, without regard for their external consequences, coercive engineered migration is a deliberate attempt by a state or nonstate actor to extract concessions from a foreign target(s), using forced displacement as a weapon. Hosting refugee communities can be quite financially or politically costly for governments, and weak actors can threaten to impose these costs by creating migration crises.

Greenhill is not the first to point out that migration has often been used as a weapon of the weak or as a tool of foreign policy. Myron Weiner was among the first to recognize that migration has important consequences for international security and can be used strategically by sending states.[31] James Hollifield further urged IR scholars to pay greater attention to the security consequences of migration flows.[32] However, Weapons of Mass Migration is the most theoretically developed and well-researched study of the strategic uses of emigration to date. Migration scholars and those interested in coercive bargaining will find this book to be a welcome addition to their bookshelves.

In Chapter one, Greenhill acknowledges that she may be facing a skeptical audience. She writes, ‘‘accepted wisdom suggests this kind of coercion should be rare at best’’ (13, n. 3). Therefore, her first task is to document the extent of this phenomenon. According to her research, there were at least fifty-six attempts to use migration crises as bargaining chips since 1951, with another eight cases that possibly fit the mold. Cases include Fidel Castro’s strategic use of the 1980 Mariel crisis and Libya’s promise to stop illegal immigration from its shores in exchange for the lifting of European sanctions. The book’s appendix very usefully offers short narratives of each case along with the sources consulted during the research. While some cases are more clear-cut than others, it is undeniable that coercive engineered migration is an important tool used in international politics.

The book’s next task is to develop a theoretical framework, analyzing which actors are most likely to use coercive engineered migration and when are they are most likely to succeed. Beyond stating that this tactic is often used by the weak, Greenhill offers a three-part typology of the types of actors who attempt to use migration crises to their advantage: generators, agents provocateurs, and opportunists (23-31). Generators directly create population displacement while agents provocateurs are often nonstate actors that deliberately invite refugee-generating government repression to draw international attention to their cause (for a similar argument, see Kuperman and Crawford).[33] Opportunists, on the other hand, bear no responsibility for the migration outflow but attempt to use it to their advantage nonetheless. Macedonia, for example, was able to extract concessions from NATO countries during the Kosovo campaign in exchange for agreeing to host Albanian refugees.

When and how is coercive migration effective? First, Greenhill argues that as the capacity and resources of the target to deal with the refugee influx are ‘swamped,’ they are more likely to give in to demands. This argument does not entirely fit the empirical record, however, because her research shows that the size of the flow is not a good predictor of success. The analysis does not offer a way to assess, ex ante, when the host’s carrying capacity is reached. Second, and more importantly, migration crises are more likely to lead to concessions as the flow becomes politically unpalatable. Refugee flows often create tensions between pro- and anti-immigrant⁄refugee camps, and target governments are often forced to act to resolve domestic political turmoil. Actors can use domestic politics and public opinion in the target state—which is often xenophobic—to their advantage. Additionally, refugee crises may impose what Greenhill terms ‘‘hypocrisy costs’’ (4) on the target. Hypocrisy costs arise when leaders are caught between self-professed adherence to liberal human rights norms and the hard political realities of coping with immigration. According to the argument, liberal democracies are the most vulnerable to this type of pressure, and indeed, the empirics show that democracies are more likely to be targets of coercive engineered migration.

The book focuses on four cases in order to delve into the mechanisms that drive the argument. First, Greenhill looks at Fidel Castro’s multiple attempts to use migration as a coercive tool against the United States. Second, Slobodan Milosevic attempted, unsuccessfully, to use the threat of mass migration to break the NATO alliance’s will to act during the Kosovo conflict. Interestingly, the Kosovo Liberation Army was able to successfully use the same migration crisis to its advantage by strengthening international resolve for humanitarian action. Third, through broadcasts to his supporters from exile, Haitian President Jean-Bertrand Aristide was able to manipulate Haitian refugee flows to win U.S. backing for his reinstatement after a military coup. Finally, a network of NGOs attempted, but failed, to use refugee flows from North Korea as pressure on China and South Korea to act decisively against the Pyongyang regime.

These case studies are very well written and are full of rich detail. Greenhill successfully uses the case evidence to highlight the causal mechanisms at work and devotes part of each case study chapter to assessing her theoretical framework. While her process tracing methodology is sound, however, the empirical research raises a couple of issues. First, given the number of variables in the theoretical framework, these four case studies are not able to capture all of the relevant dimensions of the theory. For instance, Greenhill uses only one example of a nondemocratic target, China, whose cozy relationship with North Korea somewhat confounds the analysis. Secondly, given that she finds fifty-six cases of coercive engineered migration, it is surprising that the book does not offer a more rigorous quantitative analysis. A regression table controlling for alternative explanations, or even a few cross-tabulations, would have greatly aided in testing the book’s major hypotheses.

Weapons of Mass Migration is not only a valuable contribution to the literature on forced migration but speaks to broader themes in IR. Greenhill highlights how weak actors use forced migration as a coercive tool, how humanitarian norms interact with tangible costs, and how actors use political discord within their targets to their advantage. Therefore, the great strength of this book is that it places the study of migration squarely within the field of International Relations.

Author’s Response by Kelly M. Greenhill, Tufts and Harvard Universities

Many thanks to H-Diplo for organizing a roundtable review of Weapons of Mass Migration, to Timothy Crawford, Christopher Rudolph and Idean Salehyan for their thoughtful and thought-provoking reviews of the book, and to Robert Art for his comprehensive and insightful introduction. The distinct areas of expertise of these four scholars make them ideal reviewers for a book that focuses on the use of real and threatened migration crises as instruments of state-level coercion. I therefore consider myself particularly fortunate to have been the beneficiary of their collective wisdom, comments, and critiques. I will not respond to all of the points raised in the reviews, particularly where I largely agree with them. I concur, for instance, with many of the suggestions made regarding possible extensions of the book’s theory and empirics and am quite intrigued by Crawford’s thought experiment about how international collective action might stave off future attempts at migration-driven coercion.[34]

Instead I will confine my response to instances where the reviewers’ comments suggest that there may be some confusion about certain aspects of my theory and those few places where points of disagreement appear to exist. Before proceeding, however, I want to amend slightly the description of the book offered by one of the reviewers. Specifically, Weapons of Mass Migration poses three, rather than two, key questions, which serve three distinct analytical purposes: first, how often does this kind of unconventional coercion occur (measurement of incidence), second, how often does it work (evaluation of success and failure), and third, how and why does it work (description of the phenomenon)? I seek to answer these questions by drawing on data gathered through extensive primary and secondary source research and dozens of interviews, in the U.S. and abroad, with government and non-governmental officials as well as with migrants and refugees, in the U.S., the Caribbean and the Balkans.

The reviewers have provided excellent outlines of my answers to the second and third questions. In terms of the issue of frequency, the 56-64 cases I have conclusively identified indicate that this kind of coercion was attempted on average at least once a year between 1951 and 2006—which is to say, rather more often than has been generally understood (12, 15-17). Indeed, it was long claimed in some circles that Cuban President Fidel Castro was the only actor ever to have employed coercive engineered migration (CEM) (12, fn 1). However, Castro was neither the first to engage in migration-driven coercion—in fact, CEM appears to extend back to Biblical times—nor was he the last. While I completed active data gathering for the book in 2006, incidents of this kind of coercion unfortunately did not.[35]

Several of the reviewers’ comments suggest that some confusion exists regarding the precise nature of my argument and evidence. The most significant points of apparent misunderstanding surround the efficacy of this kind of coercion, the nature of the actors who employ it, and who is actually most vulnerable to it. I address each of these issues in turn.

First, Crawford appears to have misconstrued slightly, but materially, a piece of my argument about the relative utility of CEM. While I am indeed unequivocal in my assertion that CEM is “no superweapon” and “rarely a weapon of first resort” (5), I do not in fact say it is a “pretty poor method of persuasion.” Instead I observe that although CEM has been relatively successful when actually employed (32-35), skeptics nevertheless might be inclined to discount its apparent efficacy. Specifically, I indicate that the previous “discussion notwithstanding, one might still conclude, for selection effects-related reasons, that this kind of coercion is still a pretty poor method of persuasion, undertaken only by highly resolved challengers and only when they believe there is a relatively high probability of success” (36). That is to say, it could be the case that would-be coercers are careful to focus solely on low-hanging fruit and to initiate coercive attempts only against particularly vulnerable targets, thus inflating the apparent rate of coercive success, and it is important to acknowledge that possibility. However, in point of fact, we cannot know one way or the other whether this is the case, because we cannot ascertain how often, metaphorically speaking, dogs in this context could have barked, but chose not to—i.e., how often CEM is entertained but ultimately rejected.

So although I do warn readers of potential selection effects, I do not infer from the possible existence of these effects that CEM is perforce a poor method of persuasion—quite the contrary, in fact. I conclude that despite the fact that exercises of CEM can be fraught with dangers, it “can still appear to be a strategy worth pursuing, [particularly] for challengers seeking to influence the behavior of potentially vulnerable targets disinclined to accede to their demands under normal circumstances—powerful advanced liberal democracies” (37). Though this distinction may seem minor, it is anything but when considered in the context of Crawford’s further observations about the relative utility of CEM. Specifically, a tension does not in fact exist between the theory and the (available) data: a coercive tool may have limitations—the same may be said of military force in many contexts—and may not be an instrument of first choice, but may still fare well from both a relative and comparative utility perspective.[36]

Second, I am not in truth “ambivalent” about whether pluralistic politics or normative commitments are more significant in accounting for CEM’s efficacy against liberal democracies. Rather, I find that both of these interrelated and self-reinforcing factors are important in affecting target vulnerability (60-65). Moreover, which factor appears to have mattered most varies across cases. Sometimes the nature of domestic politics seems to have mattered more (e.g., the US vis-à-vis Cuba); sometimes human rights commitments and other normative factors seem to have mattered more (e.g., the US vis-à-vis Kosovo); and sometimes both variables appear to have been equally consequential (e.g., the EU vis-à-vis Libya). In any case, even when the principal threat being leveled is the imposition of domestic political costs, because pro-refugee/migrant camp arguments tend to be rooted in claims about target states’ (violations of) humanitarian commitments, institutional and normative factors tend to be inexorably intertwined and concomitantly important.

Third, I do not in fact take the norms and rules governing target responses as “given” or fixed—far from it (see, for instance, 272, 278 and 281-83, as well as the concluding sections of several of the case study chapters). Further, I observe that norms matter more in some places and in some contexts than in others. I also take quite seriously the possibility of target blowback and backlash against human rights norms, lamentable though such developments may be. I note explicitly that whether the relevant norms and human rights commitments will be sustainable “is an open question,” especially as many targets “are actively tightening their immigration laws and asylum policies, and otherwise reducing their explicit normative commitments to the protection of the most vulnerable populations in the world” (272). I further cite several tragic incidents where we have already seen moves in this direction (e.g., Australia’s actions during the infamous Tampa incident).[37]

Fourth, I am not convinced that rebranding what I refer to as “soft” states as “weak” states is semantically or analytically helpful. While, as Crawford notes, some international political economy (IPE) literature refers to decentralized and constrained states as “weak”, other strands employ, as I do, the term “soft.”[38] In any case, the concepts to which the terms refer are essentially synonymous (see 63-65), and since relative material capabilities [as well as institutional constraints] both come squarely into play in my argument and analysis, using the term “weak” to refer to both raw capabilities and institutional freedom and capacity could serve to obscure more than it elucidates.

Regardless of the terminology employed, however, Crawford could well be correct that CEM might offer the most utility “for large and populous states with little sensitivity to liberal norms” against “small, less populous, soft-liberal states.” However, the available data do not support such a claim. Instead, in 54 of the 64 possible cases, and 49 of 56 conclusive cases, coercers were weaker in raw power terms than their target(s) (see Table 1.3, 33-35).

Could there be a bevy of hidden cases where strong actors have used this tool against soft ones? This is certainly possible.[39] But for Crawford’s hypothesis to be correct, not only would there have to be a goodly number of as yet unidentified additional cases, but also a disproportionate number of them would have to look demonstrably different from those cases already identified through nearly a decade’s worth of primary and secondary source research. Is this possible? Yes. Is this probable? I am skeptical. On the other hand, could it be the case that CEM is an as yet largely untapped tool, which might yet come into greater use by the powerful against the soft? This too is certainly possible, but for the sake of the potential victims, I very much hope not.

At the same time, while the reviewers mostly focus on the use of CEM as a weapon of the weak, there is a bit more to the story. For one thing, although the reviewers are correct that weak actors [in the sense of material power] comprise a disproportionate—albeit not exclusive—share of thus far identified generators and agents provocateurs, such is not the case for opportunists, who have been “both weak and strong, both democrats and demogogues” (31). Indeed, in 13-16% of heretofore identified cases, the coercer has been the more powerful actor, an actor who was usually—but not always—confronting a situation in which traditional coercive tools (like the use of military force) were considered too costly or too dangerous.[40] In such circumstances, CEM can be quite attractive to the relatively strong as well as the relatively weak, for use against targets both strong and weak.

For instance, in 1954, the U.S. (acting as a generator) employed CEM against North Vietnam to, among other things, try to deter then North Vietnamese leader Ho Chi Minh from demanding promised nationwide elections (287-88).[41] In the early 2000s, the U.S. (acting as an opportunist) supported the efforts of a network of non-governmental organizations (NGOs) attempting to bring down the North Korean regime by encouraging flows into China (see Chapter 5). And, in 1961, the U.S. (again acting as a generator) quite possibly threatened CEM against the USSR to persuade the Soviets to back down on Berlin (292-93)—although, in this case, the evidence remains ambiguous enough that I code the case as indeterminate. In sum, while in the aggregate CEM appears to have heretofore been used more often by the weaker against the stronger, there are some important and noteworthy exceptions (see the Appendix for brief outlines of all identified cases), and additional research may uncover different patterns and lead us to draw different and/or more expansive conclusions.[42]

As noted at the outset, I agree with much of what the reviewers say about the book. There are just a few areas where points of disagreement seem to exist. Two in particular stand out. The first involves the significance of differing manifestations of CEM; the second, the explanatory power of alternative explanations.

First, I do not understand why Rudolph discounts the importance of opportunists relative to other types of coercers. For one thing, coercion is coercion, however and by whomever it is produced. Further, as the case summaries in the Appendix indicate, whether coercers are passive or active does not necessarily correlate with how many displaced people are victimized in the process. CEM by opportunists can result in large numbers of displaced people, while coercion by generators and agents provocateurs can result in small outflows (Table 1.4, 67-69). It is not necessarily the kind of coercer that determines the size of a crisis, but rather the particular circumstances of the case.

In addition, even if one does view opportunists’ behavior as being fundamentally different from that of generators and agents provocateurs, if 40% of identified coercers have been opportunists, the vast majority (60%) still falls in the other two “more insidious” [as Rudolph puts it] categories (Table 1.2, 24-26). Finally, and somewhat ironically in light of the previous discussion, although opportunists by their nature are not responsible for creating the original (actual or threatened) outflows, by in turn threatening to (re-) expel or displace those already displaced by others, they in effect become generators themselves, albeit generators who may try to justify their actions by claiming they are simply responding to a situation thrust upon them by others (30-31).

Second, while Rudolph is spot on when he notes that issues such as kinship and history may play important roles in determining whether a particular target will be vulnerable to migration-driven coercion, how these variables affect coercive outcomes is less clear than he seems to imply. A prior affinity or historical (e.g., colonial) relationship with a particular group might indeed be expected to affect the response of a target state to attempted coercion. But in which direction? In favor of the group or against it? On the one hand, target states in which particular immigrant communities have become well established may have significant influence over their leaders, which can engender enhanced support and heightened mobilization within pro-refugee/migrant camps. Likewise, it is certainly true that asylum burdens are strongly (positively) correlated with historical links between countries of origin and countries of destination.

On the other hand, research has shown that, historically, hostility and envy have not been highest vis-à-vis entirely foreign groups but, rather, groups “who have some ethnic or other affinity to that host country—such as Algerian pieds noirs forced to return to France after the war of independence, displaced Germans resettling in West Germany after World War II, Ugandan Asians with British passports admitted to England, and Afghan Pathans moving into ethnically-related areas of Pakistan.” (71) Thus, prior relationships with migrant groups could just as well enhance the strength and size of anti-refugee/migrant camps as their pro-group counterparts.

Furthermore, both situations—highly developed affinity in one segment of society and highly developed hostility in another—could simultaneously obtain, making coercive success still more likely. As Robert Art rightly notes, “previous immigration into a target state and its immigration policies [toward that group can] play an important role, [however] that role only has significant effects for its disruptive (as opposed to absorptive) effects for democracies” (71).

In sum, as Rudolph intimates, existing historical and/or kinship relationships with particular migrant/refugee groups can and often do play a measurable role in determining outcomes. However, whether those effects make coercive success more or less probable is case-specific, rather than systematically correlated (either directly or inversely) with the nature of the preexisting relationship or policies. Moreover, while prior relationships will indeed heighten potential effects in cases in which crises become salient to pro- or anti-camps (or to both), such relationships are not universal prerequisites for coercion to be either attempted or effected—i.e., in the aggregate, their effects are additive rather than independently determinative.

Similarly, with respect to the potential explanatory power of geography, propinquity can indeed make certain targets particularly attractive relative to other states (66); however, as the data show, geography is neither necessary nor sufficient on its own and is not an independent determinant of coercive success or failure (66-70). Consider, for instance, that while a target’s vulnerability to migration-driven coercion can and often does shift dramatically over time, geographic proximity/distance between particular coercers and their targets is something that tends to remain constant. To cite a concrete example: although the distance between Haiti and the US did not change between 1994 and 2004, in 1994 , through the threat of increasing population outflows from the island nation, then exiled Haitian President Jean-Bertrand Aristide successfully coerced the U.S. into launching a domestically unpopular military operation to help restore him to power, while in 2004, similar threats were met with his escorted exit from power (and a flight to the Central African Republic). (See Chapter 4 for details.).

Moreover, we have witnessed numerous cases of successful coercion from very far afield, and corresponding cases of propinquitous failure (see Figure 1.4, 70). Overall, in fact, geographical proximate cases comprise 57% of all successful heretofore identified cases of coercion, while non-proximate cases comprise 43%. However, geographically proximate cases comprise an even higher percentage of failures, namely 71% (vs. 29%). Thus, while attempted coercion appears more likely to occur within the neighborhood, it has been more likely to succeed if conducted from farther afield.

Thus, geography has been far less important than the degree to which targets are held responsible for, and thus are compelled to respond to, particular crises—whether for historical, domestic constituency-driven, or geopolitical reasons. For example, given the root culpability of the United States for what ultimately became known as the Vietnamese boatpeople crises, it twice found itself vulnerable to coercion from afar by Hong Kong and a core group of Association of South East Asian Nations (ASEAN) member states (see, for instance, 299-302).

Finally, the case of how China has responded to the threat of a North Korean influx does not, as I see it, muddy the analysis proffered in the book. By examining the mechanism in a different context, the China-North Korea case demonstrates precisely how and why it works and where it works best. Put another way, as a non-democracy, China has more freedom of maneuver—and is less susceptible to coercion—than its liberal democratic counterparts (63-65). In short, as a liberalizing but still deeply centralized and autonomous authoritarian state absent a tradition of embracing liberal norms, China has relatively wide latitude in its policy responses to CEM. Consistent with that position, when faced with repeated demands in the early 2000s to treat fleeing North Koreans as refugees, China took full advantage of its concomitant institutional and normative freedoms by resisting calls for the creation of refugee camps, tightening its border controls, declaring all those fleeing “illegal economic migrants” and repatriating many North Koreans living in China (236, 237; see Chapter 5 for further details).[43]

At the same time, however, some of the Chinese government’s responses—the installation of barbed wire around embassies and the transfer to safe-asylum countries of individuals who managed to find a (highly publicized) way to claim asylum—suggest that the government was acutely aware of, and somewhat sensitive to, the imposition of hypocrisy costs (240). Moreover, the Chinese government’s staunch adherence to the idea that all North Koreans were to be treated as economic migrants strongly suggests a recognition that acknowledging that some North Koreans might be legitimate asylum seekers would impose serious constraints on its behavior (259-60). Likewise, reports suggest that officials in the Ministry of Foreign Affairs were deeply concerned about “’the image projected by their strong-arm tactics,’ even if they felt they could rationalize them” (250). The China-North Korea case thus helps to illustrate that the behavior of even illiberal states is still subject to potentially costly, external scrutiny, and, as a consequence, such states are somewhat vulnerable to the imposition of hypocrisy costs, albeit rather less so than their liberal democratic counterparts.

Many thanks again to the reviewers for providing such constructive food for thought and to H-Diplo for arranging this roundtable and for giving me an opportunity to respond to the reviewers. I hope to pursue many of the fruitful suggestions made regarding extensions of my research as well as to publish further results drawn from data already in hand. I hope others too will subject the book’s theory and empirics to serious scrutiny and build upon them to help further our knowledge and understanding of this unconventional instrument of coercion, which—given its very real, real world consequences—was “hiding in plain sight” for far too long (14).


© Copyright 2013-2015 The Authors.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 United States License.



[1] Robert Pape, Bombing to Win: Airpower and Coercion in War (Ithaca: Cornell University Press, 1996).

[2] Thucydides’ phrasing is: “…the strong do what they have the power to do and the weak accept what they have to accept.” See Thucydides, History of the Peloponnesian War, Rex Warner translation (New York: Penguin Classics, 1972), 402.

[3] Thomas C. Schelling, Arms and Influence (New Haven: Yale University Press, 1966),

[4] Schelling, Arms and Influence, 71-72.

[5] Paul Huth, and Bruce Russett, ‘What Makes Deterrence Work? Cases from 1900-1980,” World Politics 36:4, 497-526; and Paul Huth and Bruce Russett, ”Testing Deterrence Theory: Rigor Makes a Difference,” World Politics 42:4, 466-501.

[6] Pape, Tables 1, 2 and 3, pp. 49-53. Pape does not report his findings in this manner because his theory is dealing with different issues than what concerns us here. The figures and calculations are mine, derived from his tables.

[7] Horowitz and Reiter have analyzed Pape’s date much more rigorously than I have and extended the cases to 1999. They confirm Pape’s conclusion that the target’s military vulnerability, and hence the chances for the coercer to engage in military denial, is a better predictor of success than civilian vulnerability. The success rate for their 53 cases of airpower coercion from 1917-1999 is 36% and the failure rate 64%. See Michael Horowitz and Dan Reiter, “When Does Aerial Bombing Work: Quantitative Empirical Tests, 1917-1999,” Journal of Conflict Resolution, 45:2, 148 and 160.

[8] Alexander L. George, David K. Hall, and William E. Simons, The Limits of Coercive Diplomacy – Laos, Cuba, and Vietnam (Boston, Little Brown, 1971; and Alexander L. George and William E. Simons, The Limits of Coercive Diplomacy, 2nd ed. (Boulder: Westview Press, 1994.

[9] Robert J. Art and Patrick M. Cronin, The United States and Coercive Diplomacy (Washington, DC: United States Institute of Peace Press, 2003.

[10] Todd S. Sechser, “Militarized Compellent Threats, 1918-2001,” Conflict Management and Peace Science 28:4, 389.

[11] There are actually two different meanings of the term “coercion” in the literature and that leads to some confusion. One is the path Schelling took: coercion has two faces – deterrence and compellence. For Pape and the coercive diplomacy students, following Alexander George, coercion and compellence are used interchangeably and mean the same thing. When Schelling and his followers speak of coercion, then, they can mean either deterrence or compellence. For Pape and the coercive diplomacy students, there is coercion/compellence, on the one hand, and deterrence on the other.

[12] Gary Clyde Hufbauer, Jeffrey J. Schott, and Kimberly Ann Elliot, Economic Sanctions Reconsidered: History and Current Policy, 2nd ed. (Washington, DC: Institute for International Economics, 1990), 93, 51. The third edition added more cases but the two success rates are roughly the same.

[13] International Studies Review (2010) 12, 640-642.

[14] For students of coercive diplomacy, the book also represents a considerable conceptual advance: before Weapons of Mass Migration few had conceived of migration manipulation as an instrument of statecraft let alone grasped its potential efficacy.

[15] On this general point, the best treatment remains David A. Baldwin, Economic Statecraft (Princeton, NJ: Princeton University Press, 1985), 123-128. For another recent illuminating application of this approach, see Amy Oakes, “Diversionary War and Argentina’s Invasion of the Falkland Islands,” Security Studies, Vol. 15, No. 3 (July-September 2006), 431-463; Amy Oakes, Diversionary War: Domestic Unrest and International Conflict (Palo Alto, CA: Stanford University Press, 2012).

[16] On the need to consider policy-contingent context when explaining cases that appear to reflect a paradoxical power outcome—that is, cases where the weak prevail over the strong—see David A. Baldwin, “Power Analysis and World Politics: New Trends versus Old Tendencies,” World Politics, Vol. 31, No. 2 (January 1979), 161-194, esp. 164. Greenhill does not invoke Baldwin’s vocabulary and conceptual framing, but the book is in many ways an exemplar of the approach Baldwin advocated.

[17] Greenhill conceptualizes two mechanisms of CEM cost-infliction: (1) “capacity swamping” and (2) “political agitation” (40-41). The focus of her theoretical effort, however, is on political agitation: it is the “lynchpin” (9, 40) against targets in the developed world, where the bulk of CEM occurs. Here, the targets are mostly liberal democracies, and their domestic politics and international normative attachments make them especially vulnerable to CEM via political agitation.

[18] In this stress on the particular vulnerability of liberal democracies to coercion-by-the-weak, Greenhill’s argument is akin those made concerning democracies’ vulnerability to terrorism. See Robert Pape, “The Strategic Logic of Suicide Terrorism” American Political Science Review, Vol. 97, No. 3 (August 2003), 349-350; Andrew Kydd and Barbara Walter, “The Strategies of Terrorism,” International Security, Vol. 31, No. 1 (Summer 2006), 61-63.

[19] For Greenhill’s discussion of the hypocrisy costs, and the similarities and differences between them and audience costs, see: 10, 263-264. Although the two kinds of costs have opposite impacts on the democratic leader’s resolve, both hypothesized effects depend on leaders being highly constrained by public perceptions of a gap between their publicly pledged commitments and actual behavior. The strength of such effects has recently been challenged. See Jack Snyder and Erica D. Borghard, “The Cost of Empty Threats: A Penny, Not a Pound,” American Political Science Review, Vol. 105, No. 3 (August 2011), 437-456, Marc Trachtenberg, “Audience Costs: An Historical Analysis,” Security. Studies, Vol. 21, No. 1 (Autumn 2011), 3-42.

[20] See, e.g., Stephen D. Krasner, “US Commercial and Monetary Policy: Unraveling the Paradox of External Strength and Internal Weakness,” International Organization (Autumn 1977), 635-671, esp. 641-650; J. P. Nettl, “The State as a Conceptual Variable,” World Politics, Vol. 20, No. 4 (July 1968), 559-592, esp. 564.

[21] Greenhill is somewhat resistant to this logic, suggesting that there is too much significant variation among liberal states, and too many similarities between them and non-liberal states, to warrant the exclusive focus (63-65).

[22] Kelly M. Greenhill, “Weapons of Mass Migration.” Talk and Slides Presented at Boston College, Political Science Department, 17 April 2012, slide 29.

[23] For an excellent explication and application of the “entry deterrence” logic invoked here, see Barbara F. Walter, Reputation and Civil War: Why Separatist Conflicts are So Violent (Cambridge University Press, 2009). By potential generators, I mean those that are likely to be motivated to attempt such coercion, and have the proximity and capability to do so.

[24] If one wanted to critique the book’s research design, this interdependence of two major cases would be something to highlight, because the causal effect of one coercive attempt is getting a big boost from the other while the conditioning variable that this relationship represents is only obliquely accounted for in the model (as a generic domestic political hypocrisy cost). On the general problem in case study research, see Robert Jervis, “International Politics and Diplomatic History: Fruitful Differences,” H-Diplo/ISSF Essays, No. 1 (2010), 4 . But to focus on the methodological problem is to miss the more interesting thing about this example of the “multiple, simultaneous challenger” problem, which is the opening for conceptual and theoretical extension.

[25] For major variations on this theme see: Patrick Morgan, “Saving Face for the Sake of Deterrence,” Psychology and Deterrence, eds. Robert Jervis, Richard Ned Lebow, and Janice Gross Stein (Baltimore: Johns Hopkins University Press, 1985), 125-152; Robert Johnson, “Exaggerating America’s Stakes in Third World Conflicts,” International Security, Vol. 10, No. 3 (Winter 1985/86), 40-41; Jonathan Mercer, Reputation in International Politics (Ithaca: Cornell University Press, 1996); Robert Jervis, “Domino Beliefs and Strategic Behavior,” Dominoes and Bandwagons: Strategic Beliefs and Great Power Competition in the Eurasian Rimland, eds. Robert Jervis and Jack Snyder (New York: Oxford University Press, 1991), 20-50, esp. 32.

[26] On the value of this kind of policy relevant research, see Alexander L. George, Bridging the Gap: Theory and Practice in Foreign Policy (Washington DC: USIP Press, 1993), esp. chaps. 2 and 10.

[27] Christopher Rudolph, National Security and Immigration: Policy Development in Western Europe and the United States since 1945 (Palo Alto: Stanford University Press, 2006).

[28] Christopher Rudolph, “Sovereignty and Territorial Borders in a Global Age,” International Studies Review 7:1 (Spring 2005 ): 1-20.

[29] Mikhail Alexseev, Immigration Phobia and Security Dilemma: Russia, Europe and the United States (Cambridge: Cambridge University Press, 2005).

[30] Douglas Massey et al., World in Motion. Oxford: Oxford University Press, 1998.

[31] Myron Weiner, The Global Migration Crisis: Challenge to States and Human Rights (New York: Addison-Wesley, 1995).

[32] James Hollifield, ‘‘The Politics of International Migration,’’ in Migration Theory: Talking across Discipline, edited by Caroline B. Brettell and James F. Hollifield (New York: Routledge, 2000).

[33] Alan Kuperman and Timothy Crawford, Gambling on Humanitarian Intervention (New York: Routledge, 2008).

[34] Although I am somewhat skeptical that such a scheme is practicable in the current political environment. The halting and uneven adoption and implementation of EU border and migration policy harmonization schemes, despite more than 15 years of efforts, does not inspire confidence. Neither does the fact that the first impulse of many EU members when confronted with an influx from North Africa, for instance, is to leave Italy holding the bag. Nor do recent decisions by individual EU member states—and contrary to the wishes of Brussels— that they reserve the right to re-establish national border control in the case of strong and unexpected migratory pressure, or the failure of a state to protect the external EU borders. And, as we are talking about harmonization within an established supranational entity, this should be a relatively easy case for cooperation and burden sharing. Still, to be clear, I would be delighted to be wrong and to discover I am unduly pessimistic in this regard.

[35] Although I have not been systematically monitoring and/or investigating all potential new cases for some time, several recent (but not unique) cases from the Middle East (in the context of the uprising in Libya in 2011, and in the context of the ongoing Syrian civil war) will be examined in a paper I will present at the 2012 ISA meeting in San Francisco in early April.

[36] Even if we should in the fullness of time uncover myriad additional cases in which the nature of the challengers and targets are substantially different than those heretofore identified, this statement will still hold. Powerful (and even democratic) actors sometimes used this tool—namely, when more traditional tools were likely to be too costly or the consequences of their employment too uncertain. (See my fourth point below for a discussion of the occasion use of CEM by the powerful.)

[37] I also go to some trouble to emphasize that while such actions may temporarily reduce a target’s vulnerability, abrogation of such commitments is not a recommended policy response. They can lead to competitive races to the bottom, leaving relevant actors equally exposed, while also undermining key protections for the most vulnerable (278).[38] See, for instance, Michael Mastanduno, David A. Lake, G. John Ikenberry, “Toward a Realist Theory of State Action,” International Studies Quarterly, vol. 33, no. 4 (December 1989): 457-74.

[39] See, for instance, pp. 18-20 of the book, in which I discuss factors that impede identification of cases of CEM.

[40] Conclusive (13%) v. all possible cases (16%)

[41] At the time, Ho Chi Minh held both the positions of president and prime minister.

[42] I welcome further research in this regard and notification of any documented cases missing from my dataset and/or suspected cases that bear further investigation.

[43] This term is still actively employed by the Chinese; see, for instance, “China Says N. Korean Escapees Are Illegal Migrants,”, February 21, 2012.