Roundtable 12-7 on Arguing About Alliances:  The Art of Agreement in Military-Pact Negotiations50 min read

The politics of alliance formation is central to the study of international relations.  Many prominent alliances have been forged since the end of the Napoleonic Wars, including the North Atlantic Treaty Organization (NATO) in 1949 and the Warsaw Pact in 1955.  Less well known is the fact that many treaty negotiations ended in failure.  On these occasions, two or more countries wanted to form an alliance treaty but ultimately could not agree on the terms.  The book featured in this roundtable, Paul Poast’s Arguing About Alliances: The Art of Agreement in Military-Pact Negotiations, seeks to explain why some treaty negotiations fail while others succeed.

H-Diplo | ISSF Roundtable XII-7

issforum.org

Paul Poast.  Arguing About Alliances:  The Art of Agreement in Military-Pact Negotiations.  Ithaca:  Cornell University Press, 2019.  ISBN:  9781501740244 (hardcover, $49.95).

29 January 2021 | https://issforum.org/to/ir-12-7
Editor: Diane Labrosse | Commissioning Editor: Matthew Fuhrmann | Production Editor: George Fujii

PDF Version

Contents

Introduction by Matthew Fuhrmann, Texas A&M University

The politics of alliance formation is central to the study of international relations.  Many prominent alliances have been forged since the end of the Napoleonic Wars, including the North Atlantic Treaty Organization (NATO) in 1949 and the Warsaw Pact in 1955.  Less well known is the fact that many treaty negotiations ended in failure.  On these occasions, two or more countries wanted to form an alliance treaty but ultimately could not agree on the terms.  The book featured in this roundtable, Paul Poast’s Arguing About Alliances: The Art of Agreement in Military-Pact Negotiations, seeks to explain why some treaty negotiations fail while others succeed.

Poast’s theory highlights two variables: the compatibility of war plans and the presence of attractive outside options.  Alliance negotiations are likely to succeed when all parties agree about which countries should be targeted in the event of war, and how.  When war plan compatibility is low, however, failure is likely, especially if countries have attractive outside options.  Poast uses statistical analysis to test this theory, drawing on a unique dataset that includes information about failed negotiations in addition to successful ones.  He also conducts two in-depth case studies: the 1901 Anglo-German negotiations (a failure) and NATO (a success).  The new quantitative and qualitative evidence assembled in the book shows that joint war planning plays a key role in the solidification of alliances.

All three reviewers in this roundtable agree that this book makes important contributions to the study of alliance politics.  Brett Ashley Leeds refers to it as “an important new book” that should be “essential reading for all security scholars.” Tongfi Kim calls it “an excellent example for social science researchers” and adds that it is an “excellent platform for arguing about past and future alliance negotiations.” Scott Wolford characterizes the book as “a clever, useful, and engaging piece of scholarship.”

This roundtable raises a number of thought-provoking questions about Poast’s book and alliance politics more generally. It is worth highlighting some of those issues here.  Kim points out that countries sometimes form alliances for non-security reasons.  While acknowledging that states argue about war plans, he wonders how disagreements in the non-security domain could influence the success of negotiations.  Kim further points out that many alliances after World War II are asymmetric, meaning that is a large imbalance of power among the members (think, for example, of the U.S. bilateral alliance with the Philippines).  He questions the degree to which the war planning views of a weaker country would matter in the context of negotiations with a much more powerful state.  Wolford suggest that the theory might be unnecessarily complex.  In particular, he argues that Poast’s emphasis on cognitive biases is not needed to produce essentially the same predictions.  Leeds points out that countries disagree about which countries they will fight and where, in addition to war plans. She suggests that it would be useful to think further about how differing views on the costs of war and the importance of the issues at stake interact with doctrinal issues relating to war planning to influence negotiation outcomes.  Leeds also wonders whether Poast’s theory can explain the heterogeneity that exists among alliances, not just whether or not they are formed. All of the issues raised by the reviewers present interesting opportunities for future research.

Poast constructively engages the issues raised by the reviewers in his response.  He clarifies why he places emphasis on the issue of warfighting, succinctly putting it this way: “military alliances are ultimately about the deployment of military forces in battle.” Poast goes on to clarify how he conceptualizes war planning in the book, and transparently discusses some limitations that stem from his approach.

Overall, this roundtable offers engaging and insightful discussions on issues of central importance for alliance politics and security studies more generally.  In the introduction to Arguing About Alliances, Poast expresses hope that his book will sit between Stephen Walt’s Origins of Alliances and Glenn Snyder’s Alliance Politics on the bookshelf.[1]  There are lots of good reasons, which are highlighted by the reviewers, to believe that it will.  Poast has written an important book, and the issues raised in this roundtable should facilitate further thinking, discussions, and research.

Participants:

Paul Poast is an associate professor of political science at the University of Chicago.

Matthew Fuhrmann is a professor of political science at Texas A&M University.  He is the author of Atomic Assistance: How ‘Atoms for Peace’ Programs Cause Nuclear Insecurity (Cornell University Press, 2012) and coauthor of Nuclear Weapons and Coercive Diplomacy (Cambridge University Press, 2017).

Tongfi Kim is an assistant professor of international affairs at Vesalius College and a KF-VUB Korea Chair Senior Researcher in Brussels, Belgium.  He received his Ph.D. in political science from the Ohio State University and previously worked at Purdue University, Griffith University, and Peace Research Institute Frankfurt.  His research centers on security studies and the international relations of East Asia.  He is the author of The Supply Side of Security: A Market Theory of Military Alliances (Stanford University Press, 2016).

Brett Ashley Leeds is Professor of Political Science and Chair of the Department of Political Science at Rice University.  Much of Leeds’s research has focused on the design and effects of military alliances.  In 2008, Dr. Leeds was the recipient of the Karl Deutsch award, which is awarded annually by the International Studies Association to a scholar in IR under age 40 who is judged to have made, through a body of publications, the most significant contribution to the study of International Relations and Peace Research, and in 2019, Leeds won the Lifetime Achievement Award from the Conflict Processes section of the American Political Science Association.  Leeds served as President of the International Studies Association during 2017-2018 and President of the Peace Science Society in 2018-2019.

Scott Wolford is a Professor of Government at the University of Texas at Austin and the Editor of Conflict Management and Peace Science.  He is the author of two books, The Politics of the First World War: A Course in Game Theory and International Security (Cambridge University Press, 2019) and The Politics of Military Coalitions (Cambridge University Press, 2015).

 

 

Review by Tongfi Kim, Vesalius College, Vrije Universiteit Brussel

Under what conditions are negotiations to form an alliance more likely to reach an agreement?  Paul Poast answers this question thoroughly and convincingly, although I have some disagreement with him when he writes that “when states are arguing about an alliance, they are arguing about joint war plans” (172).  Arguing About Alliances not only makes an important contribution to the literature on military alliances but also sets an excellent example for social science researchers interested in topics that are fraught with measurement and other research design problems.  Poast persuasively makes his case with a careful and systematic investigation of the subject, which paves the ground for further research.  In this review, I first explain the book’s argument and contribution.  In the following parts, I will critically examine and challenge some of Poast’s research design choice and arguments.  I conclude this review with a brief speculation about the role of ideal war plans for future alliances.

Poast argues that the likelihood of successful alliance negotiations depends on two variables: compatibility of ideal war plans, and number of participants with attractive outside options.  Based on the two independent variables, he categorizes negotiations into four types: “Same Page,” “Pleasant Surprise,” “Standard Bargaining,” and “Revealed Deadlock,” in a decreasing probability of agreement (17, 33, 42).  Poast develops and tests two hypotheses throughout the book.  First, “[n]egotiations where the participants have high compatibility in their ideal war plans are more likely to end in agreement than negotiations where the participants have low compatibility in their ideal war plans” (43).  Second, a “negotiation reaching agreement is influenced by the participants having attractive outside options only when the compatibility of the participants’ ideal war plans is low” (43-44), and attractive outside options do not significantly damage the prospect of agreement when the compatibility of ideal war plans is high.  He tests the two hypotheses with various methods and meticulously explains their relationships with competing explanations.

Poast’s statistical analysis focuses on negotiations, held between 1815 and 1945, for pacts that included offensive or defensive obligations, and involved at least one European state and not more than five participants (46, 50).  He justifies his sampling choice well, and the book also deals with negotiations that are not included in the statistical analysis.  For instance, an entire chapter is devoted to the 1948-49 negotiations for the North Atlantic Treaty.  The 1941 Anglo-American negotiations, which led to the Declaration of the United Nations and a large alliance that fought the Axis powers, also feature prominently as an example of “Same Page” negotiations, although not in the statistical analysis.

Poast clearly defines the scope conditions for his theory.  The book explains what happens once negotiations to form an alliance begin, not why such negotiations do or do not begin.  This still leaves researchers with a large empirical task.  Poast develops fertile ground for research by shedding light on failed negotiations, which are often neglected in international relations research.[2] As he aptly paraphrases Sherlock Holmes, he identifies “the dogs that thought about barking but decided not to” (47).  The data Poast makes available in the replication material is a useful complement to datasets on successfully formed alliances such as those of the Alliance Treaty Obligations and Provisions (ATOP) and the Correlates of War (COW) projects.[3]

The book extends alliance literature by unpacking the specifics of the so-called capability aggregation models of alliances, in which states form alliances to combine their capabilities to fight or deter other states.[4] In my own work, I have criticized the capability aggregation models and instead emphasized the importance of non-military policy concessions and efficiency gains from exchanges between allies, but Poast shows us an important and specific way for efficiency gains of alliances to be created;[5] after all, if the war plans of allies are not compatible, military efficiency gains are likely to be small. Poast’s theory is parsimonious and yet goes beyond abstractly discussing costs and benefits of alliances.  Readers should also keep in mind that Poast himself has previously written on the importance of economic side payments for alliance formation and his theoretical framework can accommodate the inclusion of side payments, especially for “Standard Bargaining” negotiations.[6]

Before I turn to my disagreement with Poast’s arguments, let me briefly point out an inconsistency in the book’s discussion of a key variable, the number of participants with attractive outside options.[7] In his development of the theory and case studies, Poast distinguishes negotiations into two categories: those in which “all participants perceive themselves as having an attractive outside option” and those in which “not all participants do” (31).  In his quantitative analysis, however, Poast’s distinction is between negotiations in which any participant has an outside option and those in which no participant has an outside option, operationalized by the existence of an outside alliance.  Although Poast offers justifications for both ways of dichotomization, and he is correct in pointing out the difficulty of “empirically measuring outside option attractiveness in general,” the coexistence of the two different dichotomizations is confusing to readers (100-101).

Either approach to the number of participants with attractive outside options seems reasonable, but using both in a study creates a discrepancy between the large-N analysis and case studies. For instance, Poast devotes a chapter each to the 1901 Anglo-German negotiations (Chapter 4) and the 1948-49 North Atlantic Treaty negotiations (Chapter 5) and writes that they are both “Standard Bargaining” negotiations (134, 135), but they do not fit this category according to his operationalization in the statistical analysis.  In the case of the Anglo-German negotiations, all parties had outside options (alliances) and low compatibility of war plans, so they should be categorized as “Revealed Deadlock” negotiations and must have been treated as such in the large-N analysis.  NATO is not in his dataset, but it cannot be “Standard Bargaining” negotiations if we adopt the dichotomization used for statistical analysis because some of the negotiation participants had outside alliances.

In terms of my disagreements with some of the arguments in the book, first, although military coordination is important to alliance politics, arguing about alliances is not just arguing about joint war plans, and states form alliances for non-security goals as well.  Poast emphasizes the importance of war plans and also defines the attractiveness of an outside option by “how a state’s decision makers perceive, relative to their ideal war plan, an outside option’s ability to achieve the state’s security goals” (30). With the exclusive focus on security goals, however, how do we explain alliances that involve militarily weak states whose military contributions are much smaller than the costs of the alliances for their allies?  Some small states do offer valuable military bases, but there are many asymmetric alliances that cannot be explained away by basing arrangements.[8]

Non-security goals are particularly important to great powers, who tend to be the suppliers of military protection (and the recipients of policy concessions) in alliances.  The expansion of NATO is one example, which Poast also briefly discusses (175-176).  Reasonable arguments can be made on both sides of the debate over NATO enlargement in the post-Cold War era, but it is difficult to argue that new members of NATO brought more military value to the United States than what Washington had to incur as a result of its expanded security commitment.[9] I argue in my work that the United States pushed for the expansion of the alliance in order to discourage the emergence of another security supplier in Europe—not because a new pole in Europe presented a threat to U.S. security but because the monopolistic position of the United States in the alliance market is conducive to its non-security goals.[10] Through examination of U.S. and British declassified documents and interviews of former policymakers, Liviu Horovitz and Elias Götz find that the United States pursued the expansion to maintain stability in and influence over Europe primarily for economic reasons.[11]

Second, and relatedly, some alliances are formed in very asymmetric relations, and the ideal war plan of the participants with weak bargaining power might not matter much or is falsely presented to improve the chance of alliance formation.  For example, the 1951 Security Treaty Between the United States and Japan was negotiated under U.S.-led occupation of Japan.[12] Although Japan had military reasons to ally with the United States, the negotiations for the alliance were linked with the end of occupation.[13] The U.S.-Japan alliance, which went through a revision in 1960, is an interesting case also because Japan’s pacifist constitution was interpreted, until July 2014, to present an obstacle to collective self-defense.[14]

There are always some differences between states’ ideal war plans, and a negotiation participant who is desperate for an alliance likely will propose something not too far from the other participants’ ideal war plan.  Therefore, negotiations which involve desperate participants might appear to have higher compatibility of ideal war plans than is accurate.  Such negotiations are more likely to succeed because desperate participants are more willing to make concessions, not because their interests are in harmony.  Poast writes that allies have incentives to reveal their ideal war plans once negotiations begin (25-26), but some states have incentives to (for both the 1954 Southeast Asia Collective Defense Treaty and the 1959 bilateral executive agreement), probably downplayed its focus on countering India as opposed to countering Communism.[15] Relatedly, states have incentives to hide their desire to entrap allies during the negotiations.

Third, I have strong reservations about Poast’s assertion that the multipolar period of 1815 to 1945 “is likely more relevant” to “understanding and forecasting state behavior during the remainder of the twenty-first century…than the majority of the Cold War and immediate post-Cold War period” (12).  Setting aside the likelihood of continuing American unipolarity in military affairs or the emergence of Sino-American bipolarity in the coming decades, there are some important trends in alliance politics that limit the similarity between the period Poast examines and the future.  Related to my earlier point, there are more asymmetric alliance ties after 1945, which weakens the capability aggregation rationale for military alliances.  Nuclear weapons have played an important role in alliance politics since 1945.[16] After World War II, colonial empires became widely illegitimate, and this might have transformed the relationship between trade and alliance formation.[17] In their analysis of alliances formed between 1815 and 2003, Brett Ashley Leeds and Michaela Mattes find that Cold-War alliances and post-Cold War alliances had significant differences from pre-1945 alliances.[18] They observe that standing alliances, which are formed for general deterrence, have been increasing over time as opposed to reactive alliances, which are “formed to deal with a specific crisis or threat,” and the compatibility of war plans may play different roles in the two different kinds of alliances.[19]

Fourth, how important really is the compatibility of ideal war plans even in the pre-1945 era or the negotiations for the North Atlantic Treaty, which Poast examines?  The compatibility of war plans definitely seems important in many cases, but further research is necessary to determine how widely important it is. Among the seven “Same Page” negotiations identified in Poast’s statistical analysis, for instance, three show a questionable compatibility of strategic interests.[20] In May 1849, Saxony and Hanover formed a defense pact with Prussia (ATOPID 1135), but they withdrew from it in October in order to switch from the Prussian camp to the Austrian sphere.[21] Bulgaria and Serbia signed an alliance with both defensive and offensive obligations in February 1912 (ATOPID 1470), but the alliance collapsed in June 1913, with the outbreak of the Second Balkan War in June 1913, where the two fought against each other. The defense pact between Bulgaria and Greece that was signed in May 1912 (ATOPID 1475) ended in the same manner in June 1913.[22]

NATO is an important alliance where allies have argued extensively about war plans, but details of military coordination were not as important as the political meaning of the alliance.  In the early negotiations for the alliance, “officials from both the [U.S.] State Department and the [British] Foreign Office agreed that military capabilities were essentially secondary.”[23] As the Director of Policy Planning Staff at the U.S. Department of State George Kennan pointed out, the fundamental threat to Europe was Soviet “political conquest” and defense coordination was not the most important issue, because “military force plays a major role only as a means of intimidation” in the context.[24] Poast is absolutely right that war plans are important for deterrence (2), but the role of military coordination may be significantly different when the allies’ focus is primarily on deterrence rather than actual fighting.

Notwithstanding these objections, Poast convincingly demonstrates the importance of ideal war plans in alliance negotiations, and his book made me wonder about the theory’s application to potential future alliances.  When Vietnam or India think about forming an alliance with the United States, for example, how important will their ideal war plans be? The Asian states would be fearful of eventual abandonment by the United States and the risk of provoking China.  In terms of the compatibility of ideal war plans, the key therefore seems to be how credibly the United States can commit to the protection of the Asian states in the long term rather than the specifics of war plans.  When we see the United States as the target of a potential offensive or defensive pact, the risk to the alliance members is enormous.  China and Russia already have consultation pacts and have been increasing military coordination, but their upgrade to an alliance that commits members to use of active military force may require the desperation of both Beijing and Moscow.[25] Although I tend to focus on aspects of military alliances other than war plans, I still believe that Poast’s exemplary work offers us an excellent platform for arguing about past and future alliance negotiations.

 

 

Review by Brett Ashley Leeds, Rice University

Paul Poast’s Arguing About Alliances is an important new book that enters the conversation about alliance formation at a different point than most previous alliance research[26] and thus is essential reading for all security scholars.  Poast studies the conditions under which leaders who express a desire to form an alliance are successful in agreeing upon a treaty.  Poast is clear that he is not explaining which countries enter alliance negotiations, but instead, he examines which countries among those who enter alliance negotiations come away with a treaty.  He develops a parsimonious theory and evaluates that theory in a large N test, made possible by his impressive data collection of alliance negotiations that failed.  Poast also provides detailed case studies, one of failed negotiations and one of successful negotiations.

Poast argues that whether an alliance forms given that negotiations have begun depends on (a) whether the ideal war plans of the potential member states are compatible (both in terms of military doctrine and identification of threats), and (b) the outside options available to the states.  This is a good application of standard bargaining theory ideas to alliance politics.  The theory is quite parsimonious, and Poast explains his argument well and conducts careful empirical analysis, considering a variety of potential threats to inference and demonstrating the robustness of his results.  The frequent references to particular negotiations throughout make the book an enjoyable read and also make the argument more compelling.

As all good books do, and particularly ones that encourage consideration of a new phase in a broader process, Arguing About Alliances leaves lots of room for future research.  In the remainder of my discussion I want to focus on additional areas that I hope will be explored as a result of this book.

First, in considering the compatibility of war plans, Poast focuses on which countries states will fight where (the identification of threats) and how they will fight (military doctrine).  An additional question allies need to ask one another is what they will fight for.  From the perspective of a bargaining approach to war, for example, allies need to know what demands they prefer to concede rather than bear the costs of war and which demands they are willing to resist through the use of force. In more technical terms, when bargaining with an adversary, the allies need to have a shared reservation point.  Scott Wolford’s The Politics of Military Coalitions focuses on bargaining over exactly this issue, and I will be interested in how scholars are able to combine Wolford’s insights about bargaining over demands with Poast’s insights about compatibility of war plans when thinking about alliance formation.[27] Poast’s book opens with a discussion of the British-French-Soviet failed negotiations prior to the Second World War.  It seems to me that this debate might not have been just about how the parties would fight, but also under what conditions they would fight rather than concede.  In the end, what we observe is that the Soviets were willing to make a deal to divide Poland and the western powers were not.  Perhaps this was driven primarily by differences in war plans, or perhaps it was driven by different weighing of the value of the issues at stake versus the costs of war, or some combination of the two.  Poast does the field a service by reminding us that at their core, alliances are military agreements, and compatibility of war plans is central to their formation.  Understanding how war plans interact with and influence decisions about positions in bargaining with adversaries is an important area for future research.

Second, I fully agree with Poast’s observation on page 2 that much more theorizing has been done about alliance treaties as signals versus alliance treaties as coordination mechanisms.  Thinking about alliance treaties as coordination mechanisms, however, raises additional questions, some of which Poast notes in his conclusion.  Why is it that sometimes the treaties include detailed discussion of war plans and shared threats and sometimes they do not?  Are the cases in which these discussions are not included in the treaties ones in which war plans are naturally compatible (i.e., in Poast’s terminology, same page and pleasant surprise negotiations)?  If so, one might ask Jim Morrow’s famous question, “Why Write Them Down?”[28] It is important for us to think about the role of formal treaties for this coordination mechanism, and the conditions under which coordination alone can explain the need for formal treaties.

Third, while Poast focuses on the extent to which war plans are detailed in the treaties in his discussion of future research regarding alliance design, there are at least two other areas that could profitably be explored.  Poast appropriately focuses this first study on whether an alliance is concluded or not.  There is also significant variance in the design of alliances that are concluded.  First, some alliances are explicitly limited to some threats, even when others exist for the states.  Some treaties even go so far as to indicate that they will not be activated by a particular threat or limit their applicability to a particular conflict or a particular location.  Under what conditions do states that do not have similar threat portfolios but share an individual threat form these more limited alliances?  Second, why do consultation pacts exist?  Consultation pacts have sometimes been viewed as partial or halfhearted alliance commitments.[29] Given that this theory is about the formation of defense and offense pacts, are negotiations that end in a consultation pact or a neutrality pact failed negotiations or successful ones?  More generally, can this theory help us to understand not only which negotiations end in a treaty, but why treaties vary?

Understanding variance in treaties will involve opening up the black box of negotiations and thinking more about relative bargaining power. Poast notes that his theory is “too sparse” to predict more than the fact that multiple outcomes—“capitulation, compromise, or collapse”— are possible when some participants have attractive outside options (39). Moving forward will involve theorizing more about the relative attractiveness of outside options to understand when we expect compromise outcomes and what those look like.  Understanding alliance commitments and treaty design will also involve incorporating aspects of bargaining power that come from issue linkage, an area where Poast’s past work has been instrumental in our current knowledge.[30] Sometimes states that have less desirable outside options will entice a partner through offering concessions on another issue, for example trade or economic aid or arms sales (Poast notes this on page 30, but it does not play an explicit role in the theory).  Sometimes states have enough military power to change the outside options available to their potential alliance partners, for instance by credibly threatening to overthrow the government or control strategic territory by force if a partner does not agree to the alliance.  There is much opportunity for future research to understand how and under what conditions parties to negotiations offer issue linkage or attempt to change the outside option expected by a partner.  Future scholars should also investigate how relative bargaining power affects not only the probability of an agreement, but the content of the agreement.

Turning now to the large N empirical work, I am very sympathetic to the difficulties in matching concepts to data that are available across a wide spatial-temporal domain.  Poast undertook an enormous data collection effort to provide the sample for this study, and I understand why developing additional operationalizations of the key concepts for the independent variables was left to future work.  I hope, though, that future research will refine and improve some of these measures.

For example, in Poast’s theory, outside options play a very important role, and theoretically, Poast suggests that there are three main outside options that are possible for a state—unilateral action, an alternative alliance, or buck-passing.  In the empirical work, this is operationalized only as a dummy variable for whether a party to negotiations already has another ally.  More can be done here in the future.  For example, given that threats have been identified, why not operationalize the unilateral option by looking at the capability ratio between the potential member state and the threat?[31] If a state can defeat the potential adversary without help, that is an attractive outside option.  Why not consider the aggregate capabilities of existing alliances rather than simply code if an alliance exists?[32] Entering into alliance negotiations suggests that a state thinks it might be interested in more allies, so why not investigate under what conditions additional allies are crucial for dealing with the threat?  It would also be useful to think about other potential alliance partners that have not yet been engaged.  For example, Poast mentions that sometimes a state might be simultaneously negotiating with multiple potential partners (29).  Could those data be leveraged for understanding outside options?

Arguing About Alliances should be required reading for all scholars of alliance politics and security policy more generally.  It deals with a part of the chain of alliance politics that has rarely been centered before—how do leaders move from wanting an alliance to getting an alliance.  We needed this piece of research, facilitated by this creative and detailed data collection on failed alliance negotiations, to be able to pursue the additional questions that I have raised.  We should all be excited both by what we have learned thus far from Poast’s work and by the potential for important research that builds on this study.

 

 

Review by Scott Wolford, University of Texas at Austin

Whatever else they may be, alliances are instruments of war.  Paul Poast’s Arguing About Alliances does important work to remind us of that fact, centering its analysis of alliance formation around the realities of war-planning: identifying threats, defining military doctrines, and choosing partners.  A great deal of empirical work on alliance formation emphasizes the relationships between potential partners,[33] even as theoretical work on extended deterrence focuses on both partners and their potential enemies.  Poast manages to bring the former in alignment with the latter.[34]  The text also leverages extensive quantitative and qualitative evidence to establish patterns of success and failure in treaty negotiations as a function of disagreements over how and against whom potential allies might one day fight.  In the process, Poast also makes a compelling case for the exploration data on the pre-1945 international system, noting that eight decades of postwar bi- and unipolarity may be giving way to the historical norm of a multipolar international system.

Poast’s key conceptual move is to think about alliance negotiations as efforts at joint war planning.  This recovers an important element, the alliance’s enemies, that is often elided by studies of alliance formation that focus on partners’ preferences, political institutions, locations, or military power.  We often associate these explanatory factors with “cooperation theory,” and while Poast’s approach is certainly about cooperation, the thing over which allies cooperate—waging war—is front and center.  Signing an alliance requires agreement on which countries members may have to fight and how they are likely to fight them, a conceptual scheme that doesn’t always map cleanly onto cooperation theory’s isolated bilateral incentives for cooperation. Leveraging unique data on successful and failed alliance negotiations, Poast shows first that compatible “ideal war plans,” (17) measured by shared threats and compatible military doctrines, are associated with greater chances of successful alliance negotiations.  Second, measures of the presence of attractive outside options (i.e., extant alliances) are associated with greater chances of failed negotiations, but only when ideal war plans are not compatible.  Two case studies, one of failed Anglo-German negotiations in 1901 and another of the successful North Atlantic Treaty of 1949, show that leaders indeed argued about the nuts and bolts of potential war plans when judging whether the terms required to secure an alliance were worth the price.

This is an impressive array of evidence, meticulously collected and analyzed, and it indicates that Poast’s key variables of ideal war plans and the attractiveness of outside options carry important explanatory weight.  Chapter 2, in which Poast makes the case for his focus on pre-1945 data, is singularly impressive.  The identification of the failed alliance negotiations that make up half the sample for quantitative analysis is no mean task, and these efforts at new data collection sit comfortably alongside simple and innovative measurement schemes based on established data.  For example, Poast’s measure of shared threat, the ratio of all shared rivals to total rivals among negotiation partners, uses William Thompson’s data on strategic rivalries.[35] All told, the empirical analyses are all the more impressive for the collection of an appropriate sample and the construction of measurements with good face validity. The relationships are clear and robust, and Poast takes great care to explain the methods behind not only his main analyses but also the robustness checks that bolster his claims.

What engaged me most, though, was the theory.  Poast’s argument, in a nutshell, is that when states agree on which countries to fight and how to fight them, agreement is easy.  But when plans are less compatible, i.e. when there is some nontrivial distance between ideal war plans such that compromise is required, the chances of success depend on the attractiveness of outside options, whether those are alternative alliances or simply going it alone. Absent viable outside options, states are inclined to compromise, but the presence of an attractive alternative—like Britain’s “splendid isolation” from Europe and sticking with Japan in Asia in 1901— makes negotiations less likely to succeed.  Figure 1.1 represents pretty cleanly how these two variables interact to generate predictions over the probably of agreement, which is high when ideal war plans are compatible, low when they are incompatible and outside options are attractive, and mixed when plans are incompatible and outside options are unattractive.

I spent most of my time thinking about the thorny problem of how to get from the theory’s first principles to the predictions in Figure 1.1. I don’t want to overstate the critique, because I do think there is a way to connect the hypotheses deductively to the first principles. However, the solution is simpler than the one proposed in the book.  To preview my argument, the theory rests on an assumption that states can credibly reveal their ideal war plans once negotiations begin.  The story as given is that cognitive biases and misapplied heuristics cloud judgments before negotiations, but incentives to be honest can clear up those psychological processes once negotiations begin (24-26).  But the assumption about biases, especially how they’re cleared away, is ultimately unnecessary.

In other words, this theory needs fewer moving parts; it can explain the same amount with less, all while avoiding the thorny issue of what resolves cognitive biases, over what issues, and when.  All told, I think the theory is more parsimonious than it appears.  First, states choose negotiation partners based on how observable factors, like material capabilities and reliability, shape the eventual agreement.  And as Poast acknowledges in a very clear discussion, the result of this selection-into-sample problem (9, 69) is that things known before the negotiations should not have much of an observable effect on the prospects for success; negotiators will have priced observables in by the time they sit down at the table.  If, for example, a country rules out potential partners because they would be unreliable, the sample of negotiations would be made up mostly of reliable partners, and measures of reliability would have no bearing on the observed rate of success.  As a result, information revealed during negotiations should have a greater observable impact on their chances of success than information known beforehand.  For the present theory, then, states must enter negotiations with a non-trivial amount of uncertainty about each other’s (a) ideal war plans and (b) views of their outside options, lest the effects of both factors be strategically selected out of the sample.

Second, the hypotheses require that ideal war plans are commonly known during negotiations, while the attractiveness of outside options remains privately known. It’s straightforward to write down a model of negotiation[36] that shows this to be the case: the probability of agreement increases in the similarity of commonly-known ideal policy positions (here, the compatibility of ideal war plans). If both or neither are uncertain, however, the predictions Figure 1.1 no longer follow cleanly from the theory’s assumptions.  It is important, then, that ideal war plans be either commonly known or, at a minimum, better known than the attractiveness of outside options when states argue about alliances, despite being privately known by participations before negotiations begin.

Poast approaches this challenge thoughtfully, noting that the attractiveness of outside options is something that a negotiator would not want to reveal, because a believable threat to walk away can induce other states to compromise.  Therefore, claims of attractive outside options are not believable because of the incentive to misrepresent just how much one needs the agreement.  Before and after negotiations, this uncertainty is likely to be present because there is never a strong incentive to be honest about needing an agreement badly.

When it comes to ideal war plans, though, the argument for credible revelation is less clear (see 22-26).  The first part invokes cognitive biases and misapplied heuristics that prevent accurate judgments about others’ war plans before negotiations begin, and the second rests on the notion that, once negotiations begin, states credibly and honestly reveal the rest of their plans.  Poast notes that states cannot adopt each other’s plans without knowing what they are; this is surely true.  But this only establishes that information is useful to have, not that states have incentives to be honest in what information they give.

Further, this argument does not explain why (a) an incentive to be honest about one’s war plans makes the cognitive biases and heuristics of one’s partner fall away and why (b) those biases and heuristics do not extend to other factors based on information, like reliability, that is equally private that is still treated as known heading into negotiations (9).  Negotiations are a threshold beyond which cognitive biases no longer cloud inferences thanks to incentives to be honest (25), but it is not clear why this should be the case.  This story does not firmly establish either that claims of compatibility or incompatibility should be credible, which is key for the common-knowledge assumption.  Further, these biases should in principle operate all the time; one could even imagine that something psychological would stand in the way of recognizing when interlocutors are being (or have incentives to be) honest.[37] The circle needs to be squared, but I would argue that this psychological story is ill-suited to the task.

I think there is a solution, though, and it is already present in the theory.  Suppose that there is always a chance that one’s proposed war plan will be accepted.  If states have compatible ideal war plans, they have every incentive to be honest about it; exaggerating differences can only lead to a regrettable treaty or a subsequent admission of bluffing when asking for an adjustment.  And claiming false compatibility runs the same risk, so there is an incentive to be honest about divergences as well.  That is a straightforward incentive to represent war plans accurately, and it relates not to the value of having a proposal to implement but to the value of making an honest proposal.  This argument can (a) square the circle in a way that the manuscript’s description of the theory does not and (b) avoid reliance on cognitive biases existing (for some issues) and disappearing solely as a function of incentives to be honest.  Before negotiations, states have no reason to announce ideal war plans publicly, but inside them, when an agreement might be struck and a policy implemented, it is more valuable to be honest about both similarities and divergences.

Ultimately, the theory does not need to rely on cognitive processes or any claim about when and whether incentives to be honest can help them fall away; nor need it claim that the effects of such bias are limited to war plans and not, say, reliability, which the theory requires players to know before negotiations begin (see chapters 3 and 4).  Adjusted in this way, the theory can generate the same predictions with fewer moving parts; pre-negotiation incentives to misrepresent war plans can be replaced by in-negotiation incentives to reveal by virtue of the costs of proposing an unattractive plan.  Incentives to doubt can be replaced with incentives to believe without any reference to cognitive processes.  In that sense, the theory is much stronger than it appeared.  This is not to say that psychological theories cannot tell us anything about the success and failure of alliance negotiations, but they do not need to be introduced to the framework that is already in place for it to serve its intended purpose.

Arguing About Alliances is a clever, useful, and engaging piece of scholarship, and it does an important service by re-centering shared enemies in our understanding of the origins of alliances.  Much of the quantitative work on alliance formation examines the features of potential allies without reference to partners’ potential enemies, but Poast shows that there is much to be gained from keeping this fuller strategic setting in mind.  The literature on extended deterrence, as noted above, deals with tripartite interactions between two allies and a potential attacker, yet the aforementioned work on alliance formation looks only at the allies themselves, and not at the states they might be fighting should deterrence fail. This strategic disconnect is not without costs if our goal is to explain why some states form alliances and others don’t, and Poast’s excellent, and timely, book goes a long way towards filling this gap in our understanding.

 

 

Response by Paul Poast

Many thanks to Matthew Fuhrmann for organizing this outstanding panel.  I could not ask for a better group of scholars to review my book.  Ashley Leeds’s work is foundational to alliance studies.  She led the creation of the Alliance Treaty Obligations and Provision dataset,[38] enhanced our understanding of why states write alliance treaties, and demonstrated how military alliance treaties shed light on more general questions of international cooperation.  Scott Wolford’s book The Politics of Military Coalitions was hugely influential in the crafting of this book.[39] Since Wolford was clear that his book was about informal coalitions and not alliance treaties, I hoped to offer a book on alliance treaties that would complement his work.  Tongfi Kim has led recent efforts to apply a political economy approach to the study of alliances, with his book The Supply Side of Security adopting a market analogy to alliance formation.[40] It is humbling and gratifying to receive praise from this group.

Of course, they also raise questions and critiques.  Rather than address their critiques item by item or author by author, I will use this space to address a general concern that their comments, as a collective, highlight: how should we think about military alliances?  In so doing, I hope this symposium serves as a useful tool for scholars who are thinking about the meaning and purpose of alliance treaties.

The book has a central idea driving the theory and analysis: that military alliances are ultimately about the deployment of military forces in battle.  The word ‘ultimately’ is critical, as it is the hope (if not the expectation) of the parties that actual deployment in battle is unnecessary; they hope the alliance effectively deters.  The phrases ‘allies’ or ‘ally partner’ are bandied about policy circles to refer to everything from unofficial cooperative military partnerships with non-state actors to long standing trade partners.  That is why I focus on military alliance treaties.  The central assumption of the book is that states are negotiating such a pact because the use of military force is of central concern to at least one (and usually all) parties.

Of course, alliances are not solely about military cooperation.  Indeed, as someone who has written extensively on the political economy of international security, I am all too aware that security policies are frequently pursued in the hope of achieving non-security aims.  Some of my earliest work into alliance politics was on how states will link non-security goals (namely trade cooperation) to alliance pacts.[41] Since at least the seminal paper by Morrow,[42] we recognize that states can and do indeed have a multitude of reasons for wanting to form alliances, especially non-security goals.  Some participants might only seek security, while others could be using the alliance to gain influence over a junior partner.  Still others might join an alliance in expectation of receiving goodies, such as a share of ‘spoils’ after a war or economic aid.

Just as military alliances are not only about security goals, the negotiations to form military alliances do not focus solely on war plans or security goals. But if security against a threat is not an objective of any of the parties to the negotiation, then it is unlikely (let’s just say impossible) that the signed treaty will have military cooperation elements. The parties will instead negotiate something else, perhaps a trade treaty or a treaty of friendship.  But by viewing security acquisition as central to military pacts, I surmised that joint war planning must be central to the negotiations to form such pacts.

While one might think of war planning as officers moving pieces across a map (an image which is depicted on the book’s cover), war plans, as I write in the book, have three broad components: (1) The strategic component, which identifies who the plan is directed and where; (2) the operational component, which specifies the general guidelines and logistical aspects for militarily engaging the target; and (3) the tactical component, which delineates the exact military methods by which to engage the target’s forces in battle. While alliance negotiations can (and have) addressed all three components, they nearly always address the first component: the ‘who’ of the alliance.  This is the area of grand strategy, national interests, and geostrategic priorities.  In other words, the strategic component deals with the realm of “high politics” that is the domain of the political leadership and diplomatic staff (who are involved in such treaty negotiations).  The “how” and “when” (i.e. the second and third components) are more in the area of military expertise that can be delegated (sometimes after signing the formal alliance treaty) to military planners.

Admittedly, there are potential limits to this way of thinking about military alliance negotiations.  I will highlight and address three.

First, an aspect of how I theorize military alliance negotiations that may not be found in a  fare number of alliance treaties is the high-level operational component, namely the doctrine guiding the application of military forces. Moreover, while I outline a host of reasons in the text for why this should be considered when evaluating military alliance negotiations, it could be the case that even this general level of war planning is not raised in many negotiations.  This can especially be true in situations where the parties either (1) can’t even agree on the threat, or (2) are so desperate for an agreement – perhaps due to a lack of alternative options – that the issue is set aside for later. Relatedly, it might be the case that the states negotiate less over the exact military approach to handling the threat and more about what they want from the threat (as pointed out by Leeds in her review).  Do the allies have a `shared reservation point,’ meaning do they simply want `no troops in our territory’ or `dismantling of a weapons program’?  Or do the allies want more, such as possession of the target’s territory?

Second, the lower-level elements of a war plan for the junior partner may not have much influence on actual military operations carried out by the alliance.  But even here, the “high level” elements of a junior partner’s war plan – the threat identification and the location in which that country is a threat – still matter.  NATO is a good example.  As Kim points out in his review, concerns over Soviet political incursions, rather than straight military invasion, were most acute in the late 1940s.  This is why, to quote John Hickerson, the Prague coup, “Scarred the living bejesus out of every” (141). But the parties still needed agreement on the high-level elements of the threat –who and, particularly in the case of NATO, where to direct their support.

Third, war plans are not often found in the alliance treaty itself (assuming the parties reach agreement on the treaty).  While there are indeed alliance treaties where the direct threat to be opposed and even schedules of troop movements and placements are written in the text, there are famous treaties where no such information is to be found (see the North Atlantic Treaty).  Hence, while studying the text of alliance treaties can and has offered a host of insights into the operation and role of alliances in international politics, the text, I claim, leaves out key subtext.  Such variation in text is still intriguing and, as Leeds rightfully points out, merits further exploration.  But my book theorizes what takes place in the “room where it happens,” but not what is written into the final document.

If we are to theorize what is written, possibilities abound.  Treaties with detailed descriptions of war plans could be due to the negotiations having less of a ‘same page’ quality, thereby requiring written clarification of expectations.  This is consistent with Wolford’s extremely helpful clarification for why states have incentives to reveal plans during the negotiation: “exaggerating differences can only lead to a regrettable treaty or a subsequent admission of bluffing when asking for an adjustment.” There are times where the details must be written down in the treaty to ensure that the parties are perfectly clear on what is to be done and when.  Or more textual specificity could be related to concerns over reliability (perhaps due to perceiving a partner as having an attractive outside option): ‘I need to have this written down so that you know that I will know when you haven’t fulfilled your promise,’ or ‘I need to have this written down so that you are perfectly clear on what is the behavior I consider to be compliant with the treaty’s obligations.’

I am again humbled and thankful that Kim, Leeds, and Wolford took the time to not only read my book but to offer such detailed and extensive comments.  From new possibilities of military cooperation between the United States and India, renewal of NATO’s mission to counter Russian aggression, and China and Iran seeking substantial multi-domain cooperation, alliances and alliance pacts, once viewed as obsolete relics of bygone age, are again recognized as pillars of international politics.  Their prominence not only means that states will continue to argue about alliances, but that scholars will continue arguing about how we argue about alliances.

 

Notes

[1] Stephen Walt, The Origins of Alliances (Ithaca: Cornell University Press, 1987); Glenn Snyder, Alliance Politics (Ithaca: Cornell University Press 1997).

[2] On the importance of examining failed negotiations, also see Paul Poast, “Does Issue Linkage Work?  Evidence from European Alliance Negotiations, 1860 to 1945,” International Organization 66:2 (April 2012): 277-310.  DOI: https://doi.org/10.1017/S0020818312000069.

[3] On the ATOP dataset, see Brett Ashley Leeds, Jeffrey M. Ritter, Sara McLaughlin Mitchell, and Andrew G. Long. “Alliance Treaty Obligations and Provisions, 1815-1944,” International Interactions 28:3 (2002).: 237-260, DOI: https://doi.org/10.1080/03050620190047599.  On COW’s formal alliances, see, for example, Douglas Gibler, International Military Alliances, 1648-2008 (Washington, D.C.: CQ Press, 2009).

[4] To date, the capability aggregation models remain the dominant approach to studying military alliances.  See, for example, George Liska, Nations in Alliance: The Limits of Interdependence (Baltimore: Johns Hopkins University Press, 1962); Stephen Walt, The Origins of Alliances (Ithaca: Cornell University Press, 1987); Glenn Snyder, Alliance Politics (Ithaca: Cornell University Press 1997).  For alternative rationales for forming alliances, see, for example, Paul Schroeder, “Alliances, 1815–1945: Weapons of Power and Tools of Management.” In Historical Problems of National Security, edited by Klaus Knorr.  (Lawrence: University of Kansas Press, 1976); James Morrow, “Alliances and Asymmetry: An Alternative to the Capability Aggregation Model of Alliances.” American Journal of Political Science 35:4 (November 1991): 904-933.  DOI: https://doi.org/10.2307/2111499; and Michael Barnett and Jack Levy, “Domestic Sources of Alliances and Alignments: The Case of Egypt, 1962-73,” International Organization 45.  3 (Summer 1991): 369-95.  DOI: https://doi.org/10.1017/S0020818300033142.

[5] Tongfi Kim, The Supply Side of Security: A Market Theory of Military Alliances (Stanford: Stanford University Press, 2016).

[6] Poast, “Does Issue Linkage Work?”

[7] Poast acknowledges this inconsistency in the text (100).

[8] Kim, 34-36.

[9] Kim, 7.

[10] Kim, 14, 89-92.

[11] Liviu Horovitz and Elias Götz, “Defending Globalization: The Rationale behind the US Decision to Enlarge NATO.” Unpublished manuscript.

[12] Technically speaking, Poast can dismiss the 1951 treaty as outside his scope because the United States accepted the “right” rather than obligations to use forces “in and about Japan.”  See Article 1 of “Security Treaty Between the United States and Japan” on the Avalon Project https://avalon.law.yale.edu/20th_century/japan001.asp.  Both ATOP and COW projects, however, consider the treaty to obligate the United States to defend Japan.

[13] Kazuya Sakamoto, Nichibei domei no kizuna: Anpo–joyaku to sogosei no mosaku (The Japan–U.S. alliance nexus: The Security Treaty and the Search for Mutuality) (Tokyo: Yuhikaku, 2000).

[14] Joint military planning, therefore, has long been a politically sensitive topic in Japan.

[15] To be sure, Pakistan’s focus against India was clear to U.S. policymakers.  Americans accordingly guarded themselves against entrapment by Pakistan.  For SEATO, the United States’ understanding is that the defense obligation applies “only to communist aggression.” See “Southeast Asia Collective Defense Treaty (Manila Pact)” on the Avalon Project . https://avalon.law.yale.edu/20th_century/usmu003.asp.  On the conditions for activation of the 1959 executive agreement, see, for example, Dennis Kux, The United States and Pakistan, 1947-2000: Disenchanted Allies (Woodrow Wilson Center Press with Johns Hopkins University Press, 2001), 102.  After the September 11 attacks against the United States in 2001, Pakistan downplayed its focus against India and pretended to care about terrorist threats more than it did.  A former Pakistani ambassador to the United States writes that “the alliance with the United States has been more about securing weapons, economic aid and diplomatic support in its confrontation with India.” See Husain Haqqani, “To Win Afghanistan, Get Tough on Pakistan.” New York Times, 6 July 2017, https://www.nytimes.com/2017/07/06/opinion/to-win-afghanistan-get-tough-on-pakistan.html.

[16] Michael Mandelbaum, The Nuclear Revolution: International Politics before and after Hiroshima (New York: Cambridge University Press, 1981).

[17] Benjamin Fordham, “Trade and Asymmetric Alliances.” Journal of Peace Research 47:6 (March 2010): 685-696.

[18] Brett Ashley Leeds and Michaela Mattes.  “Alliance Politics During the Cold War: Aberration, New World Order, or Continuation of History?” Conflict Management and Peace Science 24:3 (July 2007): 183-199.

[19] Leeds and Mattes, 197.

[20] I commend Poast for helping this kind of criticism by keeping everything transparent and making the replication material available online.

[21] Gibler, 141.

[22] Gibler, 223-226.

[23] Timothy Andrews Sayle.  Enduring Alliance: A History of NATO and the Postwar Global Order (Ithaca: Cornell University Press, 2019), 14.

[24] Quoted in Sayle, 15-16; Memorandum by the Director of the Policy Planning Staff (Kennan), November 24, 1948, Foreign Relations of the United States, 1948, Western Europe, Volume III, 285.  https://history.state.gov/historicaldocuments/frus1948v03/d182.

[25] I argue that “the decline in the demand and supply of security in many parts of the world” led to the reduction of offensive or defensive alliances and increase of consultation pacts after the end of the Cold War, but I also argue that consultation pacts are an important form of military alliances.  Kim, 94. On recent Sino-Russian military cooperation, see, for example, Alexander Korolev.  “On the Verge of an Alliance: Contemporary China-Russia Military Cooperation.” Asian Security 15.  3 (2019): 233-252.  DOI: https://doi.org/10.1080/14799855.2018.1463991.

[26] For example, Glenn H. Snyder, Alliance Politics (Ithaca: Cornell University Press, 1997); Patricia A. Weitsman, Dangerous Alliances: Proponents of Peace, Weapons of War (Stanford: Stanford University Press, 2003); Stephen M. Walt, The Origins of Alliances (Ithaca: Cornell University Press, 1990).

[27] Scott Wolford, The Politics of Military Coalitions (New York: Cambridge University Press, 2015).

[28] James D. Morrow, “Alliances: Why Write Them Down?” Annual Review of Political Science 3 (2000): 63-83.

[29] James D. Fearon, “Signaling Foreign Policy Interests: Tying Hands Versus Sinking Costs,” Journal of Conflict Resolution 41:1 (1997): 68-90.

[30] Paul Poast, “Does Issue Linkage Work?  Evidence from European Alliance Negotiations, 1860-1945,” International Organization 66:2 (2012): 277-310.

[31] Jesse C. Johnson, “External Threat and Alliance Formation,” International Studies Quarterly 61 (2017): 736-745.

[32] Johnson, Brett Ashley Leeds, and Ahra Wu, “Capability, Credibility, and Extended General Deterrence,” International Interactions 41:2 (2015): 309-336.

[33] See, inter alia, Mark J.C. Crescenzi, Jacob D. Kathman, Ketja B. Kleinberg, and Reed M Wood, “Reliability, Reputation, and Alliance Formation.” International Studies Quarterly 56:2 (June 2012): 259-274, DOI: https://doi.org/10.1111/j.1468-2478.2011.00711.x; Douglas M. Gibler and Scott Wolford, “Alliances, Then Democracy: An Examination of the Relationship Between Regime Type and Alliance Formation.” Journal of Conflict Resolution 50:1 (February 2006): 129-153, DOI: https://doi.org/10.1177/0022002705281360, Benjamin Fordham and Paul Poast, “All Alliances Are Multilateral: Rethinking Alliance Formation.” Journal of Conflict Resolution 60:5 (August 2016) 840-865, DOI: https://doi.org/10.1177/0022002714553108; Brian Lai and Dan Reiter, “Democracy, Political Similarity, and International Alliances,” Journal of Conflict Resolution 44:2 (April 2000): 203-227, DOI: https://doi.org/10.1177/0022002700044002003.

[34] See, inter alia, Stephen L Quackenbush, “Not Only Whether but Whom: Three-Party Extended Deterrence.” Journal of Conflict Resolution 50.4 (August 2006): 562-583, DOI: https://doi.org/10.1177/0022002706290431: Amy Yuen, “Target Concessions in the Shadow of Intervention.” Journal of Conflict Resolution 53.5 (October 2009): 745-773, DOI: https://doi.org/10.1177/0022002709339046.

[35] If I have a concern about measurement strategies, it involves the measure of outside options (the existence of an outside ally), which is based on information known to the actors in the data, even as the theory assumes that the attractiveness of these outside options is privately known.  That need not be much of an issue, though, if we understand the absence of an outside alliance as a situation in which states should find it quite difficult to bluff at all about whether they can do without the alliance they are attempting to negotiate.

[36] For one example, see http://www.scott-wolford.com/uploads/2/5/2/4/25249202/arguing_review_proofs.pdf.

[37] If that is the case, then selection into the sample wouldn’t be a problem, but it is also not clear how states could learn ideal war plans, either.

[38] Brett Ashley Leeds, Jeffrey Ritter, Sara Mitchell, and Andrew Long. “Alliance treaty obligations and provisions, 1815-1944.” International Interactions 28:3 (2002): 237-260.

[39]Scott Wolford.  The Politics of Military Coalitions (New York: Cambridge University Press, 2015).

[40] Tongfi Kim.  The Supply Side of Security: A market theory of military alliances (Stanford: Stanford University Press), 2016.

[41] Paul Poast, “Does Issue Linkage Work?  Evidence from European Alliance Negotiations, 1860 to 1945.” International Organization 66:2 (2012): 277-310.

[42] James D. Morrow, “Alliances and Asymmetry: An Alternative to the Capability Aggregation Model of Alliances.” American Journal of Political Science 35:4 (1991): 904-933.