The Survival and Breakdown of Victorious War Coalitions

Transcription

1 The Survival and Breakdown of Victorious War Coalitions Scott Wolford August 5, 2014 Abstract States often take on coalition partners to secure victory in war, but afterwards their members may fall into conflict over the terms of the peace just secured, often at tremendous cost. What distinguishes those victorious coalitions that manage to sustain cooperation from those that disintegrate into hostility? To answer this question, I gather data on the characteristics of victorious wartime coalitions from 1859 to the present and estimate the duration of cooperation between their members in an event history framework. I explore the effects of coalition size, the diversity of preferences within the coalition, wartime alliance commitments, military strength, and several features of the war itself on the sustainability of peace amongst the members of victorious coalitions. While large coalitions are uniquely likely to break down, so are those with (a) diverse preferences and, surprisingly, (b) those comprised of a larger proportion of allies. Draft version; comments welcome, but please do not cite without permission. Prepared for presentation at the 2014 Annual Meeting of the American Political Science Association, 28 August, Washington, DC. Draft version; comments welcome, but please do not cite without permission. Thanks to Dan McCormack and Toby Rider for helpful comments and suggestions.

2 Mere months after the end of the First Balkan War in 1913, the members of the Balkan League fell into open conflict over the spoils of their victory. Unable to agree over the division of newly captured territories in Ottoman Europe, Bulgaria fought its former partners, Serbia and Greece, in the Second Balkan War (Hall 2000), heightening the regional rivalries that would ultimately play a role in the opening of the First World War (Clark 2012, Ch. 5,6). Many coalitions, on the other hand, win their wars and manage to cooperate much longer in maintaining the peace just secured; for example, Brazil, Argentina, and Urugray have avoided war with one another since defeating Paraguay in the Lopez War of 1870 (Abente 1987, Bethell 1996), and the broad coalition that won the Gulf War of 1991 (Atkinson 1993, Freedman and Karsh 1991) has also yet to see its members go to war. What accounts for these differences? Why do some winning coalitions fall apart, descending into open warfare soon after defeating a common enemy, while others share the spoils of victory peacefully for years or decades afterwards? Coalitions have participated in 40% of interstate wars in the past two centuries (Sarkees and Wayman 2010), fighting to preserve or restore of the status quo, to overturn the balance of power, or to maintain the balance against rising challengers. Multilateral wars tend to be longer (Slantchev 2004), deadlier (Shirkey 2012), and more consequential than bilateral wars, and after their common victory coalition members face a unique challenge that single victors do not: cooperating in the maintenance of the peace just secured (Wagner 2004, 2007). Fighting over the spoils of victory is doubly tragic, due to both the inherent costs of fighting and the fact that victory had been, until recently, a common goal. States undoubtedly have incentives to choose partners with whom they are unlikely to fall into conflict after the war, but the exigencies of securing victory in the first place often lead them to take on partners that, in less desperate times, they would surely avoid (Starr 1972, Wolford 2014a). Unforseen problems might also arise after the war that turn formerly reliable partners into enemies. However erstwhile victors turn 1

3 into adversaries, hard-won peace can be undermined, and, at the extreme, global orders constructed after major systemic wars can distintegrate in new conflagrations. The stability of postwar peace is generally studied in the context of recurrent war between former belligerents (see, e.g., Fortna 2003, Lo, Hashimoto and Reiter 2008, Werner 1999, Werner and Yuen 2005), but, as attested by the example of the Balkan Wars above, peace among the victors of a multilateral war can be especially precarious and its breakdown no less consequential. Just as war can change the environment in which erstwhile enemies interact (Morey 2011b), it can do the same for former partners (Wagner 2004, 2007). To explore this link between shared victory and the stability of postwar peace, I analyze data on victorious coalitions in war over the last two centuries and the duration of peace between their members, which proxies for the stability of cooperation over the postwar settlement. Using an event history framework, I estimate the effects of coalition size (Riker 1962), preference diversity (Wolford 2014b), and formal alliance commitments on the durability of post-victory peace, finding that all three perhaps surprisingly, in the case of alliance commitments tend to shorten the time between victory and subsequent war between former partners in victorious coalitions. Thus, in addition to its immediate chances of winning wars (Morey 2011a) and provoking counter-coalitions (Wolford 2014b), as well as its share of the subsequent spoils (Starr 1972), a state s choice of coalition partner can also affect the subsequent durability of whatever new peace follows the defeat of a common opponent. All else equal, small coalitions with homogeneous preferences and a minimum of formal alliance commitments appear to be the most conducive to durable peace amongst the victors of war. The rationales behind the effects of size and preference diversity are straightforward, but the destabilizing effect of alliances is surprising. However, since allied pairs are also the most likely amongst former partners to fall into conflict, the relationship is consistent with the notion that states tend to bind one another into formal commitments 2

4 when they expect that cooperation will be difficult. Therefore, treaties of alliance may be symptomatic of an underlying obstacle to cooperation (Gibler and Rider 2004), a latent problem that becomes manifest once an immediate common threat is eliminated by victory in war. Broadly, the empirical model shows that the stability of peace after wars is a function of the postwar settlements reached not just between erstwhile belligerents, as the literature shows, but also between former partners a common and consequential but heretofore unexplored problem in the study of international war and peace. War and Coalitional Durability Whether they pursue votes, the passage of legislation, or the submission of their enemies in war, coalitions of various stripes have one goal in common: winning (Riker 1962). However, building a coalition is expensive, as potential partners must be compensated for their cooperation with side payments or shares of the spoils of success, and their price can be quite high when cooperation entails participating in a war. States generally try to minimize the compensation they must pay by choosing cheap partners, but when desperate for military help, war partners will be willing to forego greater shares of the spoils to assure victory (Starr 1972, p. 53). The Allies in the Second World War, for example, needed Soviet support to defeat Nazi Germany, and despite its recent alignment with Germany and the certainty of fraught postwar peace negotiations, they allied with Stalin s Soviet Union, their military purposes mortgaged in advance (Kennan 1984, p. 77), in hopes of securing victory in the short-term. On balance, durable coalitions i.e., those that won t fall into open conflict after victory are quite attractive: why fight a war with a partner today that one will only go on to fight in the future? However, durable coalitions can also represent such substantial threats to potential future enemies that fearful third parties join today s war, siding 3

5 with a coalition s target in order to eliminate a future threat before it grows too powerful (Wolford 2014b). Indeed, states often try to choose partners to avoid precisely that problem; the United States, for example, worked hard to keep Taiwan out of its Korean War coalition (Stueck 1995, 2004) and Israel out of its 1991 Gulf War coalition (Atkinson 1993) in order to reassure potentially fearful third-party states that a coalitional victory would not represent subsequent threats to their interests. We should thus expect that coalitions will vary in their ability to cooperate in maintaining the peace secured by victory in war. Among those factors proposed to accelerate the dissolution of coalitions, three are particularly relevant in the context of interstate war. First, Riker (1962) argues that size, often critical to secure an initial victory, becomes a liability once the victors must divide the spoils amongst themselves. His size principle implies that states should take on only as many partners as required to secure victory, and coalitions that turn out to be too large after victory will be winnowed down, adjusted to ensure that the winning coalition divides the spoils among as few members as possible. Thus, larger winning coalitions should be associated with more conflict over the fruits of victory than smaller winning coalitions. Second, Wolford (2014b) argues that coalitions with diverse preferences are less likely to survive intact after war than more homogeneous coalitions. Coalitions with diverse foreign policy preferences, often built out of short-term military necessity, are more likely to come into conflict over the disposition of the spoils of victory than those with similar preferences, because the latter are better able to find an agreeable division of the spoils that saves the costs of war and more likely to agree on any given issue in the future. 1 Finally, Wagner (2004) links victory in war to shifts in the distribtuion of power between coalition members. To the extent that victors can credibly commit not to use 1 When powerful, diverse coalitions are also uniquely likely to provoke counter-coalitions aimed at facilitating their future dissolution (Wolford 2014b). 4

6 newfound power against one another, the coalition should survive its post-victory peace intact. However, if renegotiation is both attractive and cheap, former partners may fall easily into conflict over the newly established peace. Measuring the credibility of commitments, however, is difficult. Homogeneous preferences, for example, might mean that states are less concerned about each other s gains in power or influence; this is consistent with the above-proposed effects of preference diversity. Those states that enter a war as allies, on the other hand, might see the formality of their commitments cut both ways. 2 Allies might have a great deal of interests in common, which makes alliances attractive to form in the first place. However, alliances are also most attractive when they tether otherwise unreliable states to one s own foreign policy (Gibler and Rider 2004); why use an alliance to signal or ensure behaviors that other states already expect? To the extent that alliances are formed in the shadow of a common threat, however, defeating that threat might bring otherwise dormant conflicts into the open by eliminating the cause for the alliance. In other words, alliances are costly instruments of commitment, most likely to be used when commitment is not automatic that is, when the costs paid to ensure commitment are justified due to a partner s potential unreliability (Morrow 1994, 2000). At the other end of the spectrum, closely aligned partners that do not need an alliance to cement cooperation might be far less prone to subsequent conflict after defeating their common enemy. A similar rationale supports each of these arguments: states fight wars, sometimes in cooperation with one another, to impose their preferences on the terms of peace agreements that will follow (Wagner 2007). The terms of these settlements, quite intuitively, affect the duration of peace between former belligerents (cf Werner and Yuen 2005). However, when states work together to defeat an opponent, the divergent preferences 2 Alliances are distinguished from informal relationships, e.g. alignments, by their formalization in treaties; alliances are written down, but alignments are not (Morrow 2000). 5

7 subordinated to a tacit neutrality (Gamson 1961, p. 374) in order to secure victory may come to the surface once that goal has been achieved, meaning that distributive issues between former partners can also lead to subsequent war. Nonetheless, the literature s focus on recurrent war between former belligerents (Fortna 2003, Lo, Hashimoto and Reiter 2008, Werner 1999, Werner and Yuen 2005) has come at the cost of understanding the crucial dyanmics of bargaining and war between the very states that cooperated to win the war. In the following section, I specify and estimate an empirical model designed to provide the first systematic analysis of this problem. What kinds of coalitions manage to cooperate in sharing the spoils of victory, and what kinds of coalitions descend into intramural conflict after defeating their common enemy? Research Design In this section, I discuss the construction of the relevant sample of victorious war coalitions, the collection and coding of key theoretical and control variables, and the specification of the Cox proportional-hazards model used to estimate the effect of the independent variables on the time between victory in war and conflict erupting between former coalition partners. The event-history framework is ideal for analyzing the durability of coalitions, because it models the probability that a coalition stays intact or that it survives given that it has survived until the present time (see Box-Steffensmeier and Jones 2004). When any of a coalition s members fall into war with one another, the coalition fails, or exits the sample. The technique has been widely used to study the duration of war (Bennett and Stam 1996, Goemans 2000, Slantchev 2004) and the duration of postwar peace between former belligerents (Fortna 2003, Lo, Hashimoto and Reiter 2008, Werner 1999, Werner and Yuen 2005), but I apply it here to the unique problem of postwar peace between former partners. 6

8 Assessing the durability of post-victory cooperation requires, first, identifying those coalitions that win interstate wars. To do so, I use the Correlates of War project s list of interstate wars from (Sarkees and Wayman 2010), which identifies twenty victorious coalitions in interstate wars, starting with France, Austria, and the Kingdom of the Two Sicilies victory in the War of the Roman Republic in 1849 and ending with the topping of Iraq s government during the interstate phase of the Iraq War in 2003 by the United States, United Kingdom, and Australia. 3 Table 2, which can be found in the Appendix, lists the coalitions and their members that COW identifies as winning their wars, generated by eliminating all bilateral wars, as well as those in which coalitions lose to single belligerents. The remaining wars are those won by coalitions, as opposed those in which coalitions lose or see the war end in stalemate. 4 Some wars transition to another type of war, often intra-state if conquest opens the door to insurgency; in these cases, if the interstate phase had a clear winner like the Iraq War, in which the coalition successfully toppled the target government before the emergence of the insurgency I coded the war s end at the transition point and recorded the outcome accordingly. 5 Eleven coalitions, roughly half the sample, see a war break out amongst at least two of their members after victory and before the end of the observation period in 2010, the first of which for a coalition marks the end of cooperation and serves as the indicator of failure by which coalitions exit the data. 6 To be sure, coalitions may see cooperation break down in a number of ways short of full-scale war, but using war as a measure of coalitional dissolution has two advantages. First, all states can, in principle, go to war 3 COW s operational definition of coalition membership differs from Wolford s (2014b), which includes basing and staging contributions, meaning that Kuwait, from which the 2003 invasion of Iraq was launched, would count as a member of the Iraq War coalition. 4 See Starr (1972) for an analysis of the distribution of payoffs between both winning and losing coalitions. 5 This leads to the inclusion of the War Over Angola (COW #186), Phase Two of the Second Ogaden War (COW #187), the 2001 Invasion of Afghanistan (COW #225), and the 2003 Iraq War (COW #227). 6 There are also no ties, in that no wars erupt the same number of days after victory, making the choice of how to handle ties irrelevant. 7

9 with one another, and they may do so over a variety of issues. Therefore, intramural war as a measure of coalitional breakdown does not depend on the issues involved in the war just ended or the specific terms imposed after victory; in contrast, indicators short of war say, abrograted alliances (Leeds and Savun 2007, Leeds, Mattes and Vogel 2009) might. Second, as a relatively coarse indicator, war is also less prone to measurement error than some more inclusive alternative variables, such as lower-level disputes or diplomatic tensions (e.g. Ghosn, Palmer and Bremer 2004, Wilkenfeld and Brecher 2010); when war erupts between two states, there is little doubt that security cooperation between them has broken down. There are three key independent variables of interest: the number of states in the coalition (Riker 1962), the diversity of preferences amongst its members (Wolford 2014b), and extent to which the members of the coalition are tied to one another by formal alliances at the outbreak of war. Collecting data on the size principle is straightforward; the variable Size is simply the number of states that COW identifies as members of the victorious coalition (Sarkees and Wayman 2010). Measuring the diversity of preferences inside coalitions is more difficult. Ideally, we would like a measure of revealed preferences, such as ideal points estimated from voting patterns in the UN General Assembly (Reed et al. 2008, Strezhnev and Voeten 2013), but the limited temporal coverage such data is nonexistent before 1946 eliminates too much of the sample. Therefore, as a rough proxy of the diversity of preferences inside the coalition, I create the variable Diversity by calculating the variance of the Polity IV combined democracy scores (Marshall and Jaggers 2009) amongst members in the first year for which data is available after the war, on the assumption that, in general, states with similar domestic institutions will hold similar preferences over foreign policy; the greater the variance in Polity scores, the more diverse the coalition s preferences. Given the large number of European wars in the data, linking Polity scores to what Braumoeller (2012) calls the distribution of ide- 8

10 ology is far from inappropriate; in both the 19th and 20th Centuries, democracies and various forms of autocracy contested the relative sizes of the democratic and either legitimist or Communist communities, making domestic political similarity a workable, if imperfect, indicator of preference similarity. Third, as an indicator of the credibility of commitments to sharing the spoils of victory, I measure the percent of pairs of coalition members that share formal alliance commitments in the first year of the war. If, for example, a coalition includes three states and two of the three possible pairs are allied in the year the war begins, then Percent Allied equals 2/3; if all three states are allied with one another, then Percent Allied equals one. I use two versions of the variable, one based on the Alliance Treaty and Obligations Project (ATOP) data (Leeds et al. 2002), where I consider states allied if they share a defensive, offensive, or consultation pact, and another based on the Correlates of War data (Gibler and Sarkees 2004), where I code allies as those sharing a defense pact or an entente. However, as noted above, expectations over the behavior of a simple measure of alliance membership are not strong: alliances may signal algined preferences or latent cooperative difficulties that might emerge after the elimination of a common threat. Whether the effect is to shorten or lengthen the peace between victors will depend, in large part, on the relative balance of these two types of allies in the sample. Finally, I include several controls the coalition s total military capabilities, logged values of battle deaths and the length of the war in days, as well as an indicator for the Second World War to account for its extreme values on duration, deaths, size, and diversity all derived from the Correlates of War data (Sarkees and Wayman 2010, Singer, Bremer and Stuckey 1972). The full specification of the Cox proportional-hazards (PH) model is h (t X ) = h(t)exp [ (Size)β 1 + ( Diversity ) β 2 + (Percent Allied)β 3 + Xβ ], 9

11 where the hazard function h (t X ) characterizes the probability that a coalition survives intact at time t, given that it has survived up to time t, as a function of a set of regressors and their coefficients. The coefficients β 1 -β 3 are associated with the three theoretical variables identified above, and they are accompanied by a vector of control variables and their coefficeints Xβ. The Cox model makes no assumptions about the underlying or baseline hazard rate, but it does assume that the variables in X have a similar effect on the probability of survival at any time after victory that is, that the harzards are proportional. Following Box-Steffensmeier and Jones (2004, pp ), I examine the Schoenfeld residuals as a function of time to look for possible violations of the proportional hazards assumption. I find none, which indicates that the model does not need to be adjusted for time-varying effects. I turn now to a discussion of the results of estimating this model on my sample of victorious war coalitions. Results Table 1 presents the results of estimating four different versions of the Cox proportionalhazards model, based on each possible combination of alliance measures (ATOP and COW) and the Second World War dummy variable. Rather than coefficients for each variable, I present hazard ratios, which indicate the multiplicative effect on the hazard rate that is, on the probability that a coalition dissolves at time t of a one-unit increase in the variable (however scaled). If an increase in a given variable has no effect on the hazard, then the hazard ratio is 1; accordingly, an increase in the hazard rate (here, war sooner rather than later) implies a ratio greater than one, while decreases in the hazard rate (war later rather than sooner) are indicated by a ratio less than one. Before discussing the theoretical variables, it is worth noting that the direction, magnitude, and statistical significance of all variables are generally robust to both alternate 10

12 Table 1: Cox PH models of time until coalitional breakdown, h(t X ) Variable Model 1 Model 2 Model 3 Model 4 ATOP ATOP/WWII COW COW/WWII Theoretical Variables Size Diversity Percent Allied Control Variables Total Capabilities Log Battle Deaths Log War Duration WWII Model Statistics Subjects/failures 20/11 20/11 20/11 20/11 Log Likelihood LR χ 2 (d.f.) Significance levels : : 10% : 5% : 1% 11

13 alliance codings and the inclusion of the World War II dummy variable. Beginning with the control variables, coalitions that are more powerful in the aggregate appear less prone to breakdown than weaker coaltions, as indicated by a hazard ratio consistently les than one; note that this ratio represents the effect of power alone, as two other factors generally correlated with power size and preference diversity are also included in the model. Next, there is no systematic relationship between either the total battle deaths or the duration of the war and the post-victory durability of the victorious coalition, and despite World War II s own extreme values in coalition size, diversity, duration, and severity, its victors are not statistically distinguishable from other coalitions in the durablity of their own post-war cooperation. 7 Turning to the theoretical variables, all three of which are stripped of possible correlation with aggregate military power, each exhibits a hazard ratio greater than one. Thus, increases in coalitional size, preference diversity, and the percentage of allied members are each associated with an increased risk of coalitional breakdown. Put differently, coaltions that are larger, more diverse, and with a larger share of allied members will tend to dissolve into conflict sooner than coalitions that are small, homogeneous, or less composed of allies. In the case of Size, each additional coalition partner for example, increasing from two to three members makes a coalition roughly 8.5 times more likely to break down at a given time. This is effect is substantively large and statistically disernible at p < 0.05; the probability that this relationship is due to chance alone is less than 5%. This is consistent with Riker s (1962) size principle, and the relationship holds up even in the presence of the a measure of preference diversity, which might confound 7 The COW data codes the WWII coalition breaking down with the onset of the Korean War, which ultimately pitted the United States against China. This poses some problems of agency by the time of the Korean War, China was controlled by the Chinese Communist Party but two factors suggest that this is not an unreasonable coding. First, the United States and Soviet Union did work with the CCP to varying degrees during the war. Second, North Korea s invasion of the south occurred with the explicit blessing of both China and the Soviet Union (Stueck 1995, 2004); the latter was unambiguously a part of the winning coalition in the previous war. 12

14 Figure 1: Hazard of coalitional breakdown by preference diversity, Smoothed hazard function analysis time Low Variance Mean Variance High Variance the effects of the size principle itself. However similar their members preferences, then, larger coalitions tend to shed members violently, in this case after victory in war. Next, the hazard ratio on Diversity is greater than one, though the substantive effect is difficult to assess given the scaling of the variable, whch ranges from zero (the least diverse coalition) to 85 (the most diverse). A one-unit increase in the variance here makes little sense, so Figure 1 plots the estimated hazard function, which tracks the probability of coalitional breakdown over time as a function of a minimally diverse coalition, a coalition of average diversity, and a maximally diverse coalition. Beginning with the lowvariance coalition, the probability of war between the victors begins low and increases only slightly over time, but moving to coalitions of average and sample-maximum diversity, each substantially increases the risk that victorious coalitions will break down into subsequent conflict. Consistent with Wolford s (2014b) hypothesis, homogeneous coalitions regardless of size or aggregate strength are uniquely durable compared to 13

15 Figure 2: Hazard of coalitional breakdown by percent of states sharing ATOP alliances, Smoothed hazard function analysis time 25% Allied 50% Allied 100% Allied their more diverse counterparts. This may be good news for the victors down the line but potentially bad news for other states that worry about how long the coalition might stay together and take advantage of its aggregate power. Finally, an increase in Percent Allied is also associated with a statstically significant increase in the risk of coalitional breakdown; in this case, the variable ranges from 0 to 100, such that a coalition with 51% of its members allied will be 1.1 times more likely to break down than a coalition with 50% of its members allied. To get a better sense of the substantive meaning of the relationship, Figure 2 plots the hazard function for coalitions that are 25%, 50%, and 100% allied, across which the differences in the hazard are quite dramatic: coalitions with few allied members are substantially less likely to break down at a given time than those with even half of their members sharing alliances. With Diversity capturing the degree of shared preferences in the coalition, Percent 14

16 Allied captures all those effects of alliances not tied to similar preferences. These remaining effects point to more formally committed coalitions being associated with more fragile post-victory cooperation. If alliances represent particularly costly forms of commitment (Morrow 1994, 2000), especially relative to informal pledges of cooperation, then Percent Allied may capture the underlying cooperative problems that necessitated the formation of the alliance in the first place. It is also possible, as well, that coalitions with a larger fraction of allies are comparatively more willing to take on partners as the Western allies were with respect to the Soviet Union that will be unreliable after the war, because the allies expect to be able to work together in defense of the postwar status quo against unreliable partners. However, the average survival time for non-allied dyads is roughly 20 thousand days by the COW measure and 19 thousand for ATOP, while the average survival time for non-allied dyads is roughly 11 thousand and 13 thousand for COW and ATOP, respetively suggesting that it is, indeed, allied states who turn against one another in competition over the spoils of victory. Conclusion Roughly half of all victorious war coalitions have seen their members go on to fight one another in subsequent years, and for some, like the Balkan League in 1913, the time between victory and intramural war is startlingly short. The empirical analysis above identifies three features of victorious coaltions that make them uniquely prone to violent breakdowns in the future: size, diversity of foreign policy preferences, and the extent to which coalition members were formally allied at the outbreak of war. While the first two results are consistent with extant predictions about coalitional durability (Riker 1962, Wolford 2014b), the latter is more surprising. Alliances can bolster general deterrence (Leeds 2003, Morrow 1994), increase trade (Gowa and Mansfield 1993, 2004, Mansfield 15

17 and Bronson 1997), as well as provide some degree of control over states that would otherwise pursue undesirable policies (Morrow 1991, Pressman 2008, Weitsman 2004), but these benefits may come at the cost of an increased risk of future conflict once allies eliminate or subdue a common threat on the battlefield. States often choose their partners in war coalitions under a tight constraint the desire to win which leads to compromises over both future shares of the spoils and, as shown here, future conflicts over those spoils. While the present analysis above explores only a few pre-war characteristics of coalitions, future work might explore intra-war decisions as well, such as the assignment of geographical areas of responsibility, compensation schemes (i.e., whether partners are rewarded up front or expect to receieve concessions after the war), and perhaps the timing with which members join the coalition during the war. Each of these factors likely shapes the credibility of commitment to postwar cooperation, perhaps mitigating or exacerbating issues caused by the initial choice of partner, and uncovering their effects will shed yet further light on the potential downstream problems caused by belligerents overriding goal of winning their wars. References Abente, Diego The War of the Triple Alliance: Three Explanatory Models. Latin American Research Review 22(2): Atkinson, Rick Crusade: The Untold Story of the Persian Gulf War. New York: Houghton Mifflin. Bennett, D. Scott and Allan C. Stam The Duration of Interstate Wars, American Political Science Review 90(2): Bethell, Leslie The Paraguayan War ( ). University of London Institute of Latin American Studies Research Papers, No. 46. Box-Steffensmeier, Janet M. and Bradford S. Jones Event History Modeling: A Guide for Social Scientists. Vol. New York Cambridge University Press. 16

18 Braumoeller, Bear F The Great Powers and the International System: Systemic Theory in Empirical Perspective. New York: Cambridge University Press. Clark, Christopher The Sleepwalkers: How Europe Went to War in New York: HarperCollins. Fortna, Virginia Page Scraps of Paper? Agreements and the Durability of Peace. International Organization 57(2): Freedman, Lawrence and Efraim Karsh How Kuwait Was Won: Strategy in the Gulf War. International Security 16(2):5 41. Gamson, William A Review 26(3): A Theory of Coalition Formation. American Sociological Ghosn, Faten, Glenn Palmer and Stuart Bremer The MID3 Data Set, : Procedures, Coding Rules, and Description. Conflict Management and Peace Science 21: Gibler, Douglas M. and Meredith Reid Sarkees Measuring Alliances: The Correlates of War Formal Interstate Alliance Dataset, Journal of Peace Research 41(2): Gibler, Douglas M. and Toby J. Rider Prior Commitments: Compatible Interests versus Capabilities in Alliance Behavior. International Interactions 30(4): Goemans, H.E War and Punishment: The Causes of War Termination and the First World War. Princeton: Princeton University Press. Gowa, Joanne and Edward D. Mansfield Power Politics and International Trade. American Political Science Review 87(2): Gowa, Joanne and Edward D. Mansfield Alliances, Imperfect Markets, and Major- Power Trade. International Organization 58(4): Hall, Richard C The Balkan Wars, : Preude to the First World War. London: Routledge. Kennan, George F American Diplomacy, Expanded Edition. Chicago: University of Chicago Press. Leeds, Brett Ashley Do Alliances Deter Aggression? The Influence of Military Alliances on the Initiation of Militarized Interstate Disputes. American Journal of Political Science 47(3): Leeds, Brett Ashley and Burcu Savun Terminating Alliances: Why Do States Abrogate Agreements? Journal of Politics 69(4):

19 Leeds, Brett Ashley, Jeffrey M. Ritter, Sara McLaughlin Mitchell and Andrew G. Long Alliance Treaty Obligations and Provisions, International Interactions 28: Leeds, Brett Ashley, Michaela Mattes and Jeremy S. Vogel Interests, Institutions, and the Reliability of International Commitments. American Journal of Political Science 53(2): Lo, Nigel, Barry Hashimoto and Dan Reiter Ensuring Peace: Foreign-Imposed Regime Change and Postwar Peace Duration, International Organization 62(4): Mansfield, Edward D. and Rachel Bronson Alliances, Preferential Trading Agreements, and International Trade. American Political Science Review 91(1): Marshall, Monty G. and Keith Jaggers Polity IV Project: Political Regime Characteristics and Transitions, Polity IV dataset version Accessed 30 July 2009 from URL Morey, Daniel S. 2011a. Military Coalitions and the Outcome of Interstate Wars. Presented at the 2011 Annual Meeting of the American Political Science Association. Morey, Daniel S. 2011b. When War Brings Peace: A Dynamic Model of the Rivalry Process. American Journal of Political Science 55(2): Morrow, James D Alliances and Asymmetry. American Journal of Political Science 35(4): Morrow, James D Alliances, Credibility, and Peacetime Costs. Journal of Conflict Resolution 38(2): Morrow, James D Alliances: Why Write Them Down? Annual Review of Political Science 3: Pressman, Jeremy Warring Friends: Alliance Restraint in International Politics. Ithaca: Cornell University Press. Reed, William, David H. Clark, Timothy Nordstrom and Wonjae Hwang Power, and Bargaining. Journal of Politics 70(4): War, Riker, William H The Theory of Political Coalitions. New Haven: Yale University Press. Sarkees, Meredith Reid and Frank Wayman Resort to War: Washington: CQ Press. 18

20 Shirkey, Zachary C When and how many: The effects of third party joining on casualties and duration in interstate wars. Journal of Peace Research 49(2): Singer, David J., Stuart Bremer and John Stuckey Capability Distribution, Uncertainty, and Major Power War, In Peace, War, and Numbers, ed. Bruce Rusett. Beverly Hills: Sage pp Slantchev, Branislav How Initiators End Their Wars: The Duration of Warfare and the Terms of Peace. American Journal of Political Science 48(4): Starr, Harvey War Coalitions: The Distributions of Payoffs and Losses. Lexington, MA: Lexington Books. Strezhnev, Anton and Erik Voeten United Nations General Assembly Voting Data. Available at Stueck, William The Korean War: An International History. Princeton: Princeton University Press. Stueck, William Rethinking the Korean War. Princeton: Princeton University Press. Wagner, R. Harrison Bargaining, War, and Alliances. Conflict Management and Peace Science 21(3): Wagner, R. Harrison War and the State: The Theory of International Politics. Ann Arbor: The University of Michigan Press. Weitsman, Patricia A Dangerous Alliances: Proponents of Peace, Weapons of War. Stanford: Stanford University Press. Werner, Suzanne The Precarious Nature of Peace: Resolving the Issues, Enforcing the Settlement, and Renegotiating the Terms. American Journal of Political Science 43(3): Werner, Suzanne and Amy Yuen Making and Keeping Peace. International Organization 59(2): Wilkenfeld, Jonathan and Michael Brecher International Crisis Behavior Project. v.10 ed. Ann Arbor: ICPSR Study #9286. Accessed February 7, 2011 at URL: http: // Wolford, Scott. 2014a. The Politics of Military Coalitions. Book manuscript, University of Texas at Austin. Wolford, Scott. 2014b. Power, Preferences, and Balancing: The Durability of Coalitions and the Expansion of Conflict. International Studies Quarterly 58(1):

21 Appendix Table 2: Victorious Coalitions and their Member States COW # War Name End Year Coalition 16 Roman Republic 1849 France Austria Two Sicilies 22 Crimean 1856 United Kingdom France Italy Turkey 28 Italian Unification 1859 France Sardinia-Piedmont 46 Second Schleswig-Holstein 1864 Germany Austria 49 Lopez War 1870 Brazil Argentina 52 Naval War 1866 Peru Chile 55 Seven Weeks War 1866 Prussia Mecklenburg Schwerin Italy 58 Franco-Prussian 1871 Bavaria Prussia Baden Wurttemburg 82 Boxer Rebellion 1900 United States United Kingdom France Russia Japan 100 First Balkan 1912 Serbia Greece Bulgaria 103 Second Balkan 1913 Serbia Continued on next page 20

22 Table 2 Continued from previous page COW # War Name End Year Coalition Greece Romania Turkey 106 World War I 1918 United Sates United Kingdom Belgium France Portugal Italy Serbia Greece Romania Russia Japan 107 Estonian Liberation 1920 Estonia Finland 108 Latvian Liberation 1920 Estonia Latvia 112 Hungarian Adversaries 1919 Czechoslovakia Romania 136 Nomonhan War 1939 Soviet Union Mongolia 139 World War II 1945 United States Canada Brazil United Kingdom Netherlands Belgium France Poland Italy Yugoslavia Greece Bulgaria Romania Soviet Union Continued on next page 21

23 Table 2 Continued from previous page COW # War Name End Year Coalition Norway Ethiopoia South Africa China Mongolia Australia New Zealand 155 Sinai War 1956 United Kingdom France Israel 158 Ifni War 1958 France Spain 186 Angola 1976 Cuba Angola 187 Second Ogaden, Phase II 1978 Cuba Ethiopia 211 Gulf War 1991 United States Canada United Kingdom France Italy Morocco Egypt Syria Saudi Arabia Kuwait Qatar United Arab Emirates Oman 221 Kosovo 1999 United States United Kingdom Netherlands France Germany Italy Turkey Continued on next page 22

24 Table 2 Continued from previous page COW # War Name End Year Coalition 225 Invasion of Afghanistan 2001 United States Canada United Kingdom France Australia 227 Invasion of Iraq 2003 United States United Kingdom Australia 23