The Politics of Diplomacy: How the United States Builds Multilateral Military Coalitions

How does the United States build multilateral military coalitions? Conventional wisdom focuses on the role of formal alliance structures. Allies band together because they share threat perceptions, political ideology, norms, and values. I argue instead that US-led coalition-building efforts are influenced by the entirety of bilateral and multilateral ties that connects the United States with a third party. The breadth of institutions matters because it allows officials to gather information on the potential coalition partner’s deployment preferences beyond straightforward security considerations—such as what kind of economic and political considerations affect its willingness to join the coalition. Diplomatic embeddedness also helps American officials identify linkages between military and non-military interests. This facilitates the construction of side-payments. I find evidence for my argument by using an original dataset including all US-led multilateral coalitions in the pre– and post–Cold War era. I complement the quantitative analysis with a case study on US coalition-building efforts for the Korean War.

Ever since the Korean War, successive US governments have shown a remarkable preference for eschewing unilateral military interventions in favor of building multilateral coalitions. Indeed, since Operation Just Cause deployed to Panama in 1988, no Americans troops have deployed (overtly) to foreign theaters in the absence of allied forces. They have gone to battle repeatedly-in the Middle East, Europe, Africa, Latin America, and Asia-alongside an array of multilateral coalition partners. What political mechanics undergird the construction of these coalitions? Who joins them, and how does the coalition-building process unfold? Many scholars argue that US-led multilateral military coalition-building occurs without much diplomatic meddling: US allies coalesce around common interests, threat perceptions, political ideologies, norms, and values (Walt 1987;Weitsman 2003Weitsman , 2004Davidson 2011). They often consider American formal military alliances to be the key vehicles in this process. These alliances generate prior security commitments that then lead to coalition contributions (Morrow 2000;Snyder 1997;Bennett, Lepgold, and Unger 1997;Baltrusaitis 2009). Alternatively, scholars point to hegemonic orders-or "hierarchy"-to explain coalition participation (Lake 2009). According to this framework, the United States uses its asymmetrical power capabilities-political, military, financial, economic, and technological-to ensure coalition support.
I take a different approach. I argue that US coalitionbuilding involves intense diplomatic negotiations: a charged back-and-forth between high-level foreign and US government officials carefully debating the pros and cons of a particular deployment. Like any negotiation, this process involves arguing, persuasion, and often also side-payments (cf. Wolford 2015, 3). Overall, the key variable in this process is the accuracy of information about the underlying preferences potential coalition participants hold regarding the coalition deployment (cf. Olekalns and Smith 2009, 347-48;Pruitt and Lewis 1975;McKibben 2015). Without such information, American officials risk failure in their negotiations with potential coalition partners. With regard to most US coalitionbuilding efforts, these preferences include security and military concerns. Nevertheless, very frequently an array of other non-security-related factors also shapes the attitudes of possible partners: economic and financial interests, public opinion, and a range of other political concerns. Alliance ties often prove insufficient for providing US officials with access to state preferences on the full range of these issues. As a result, US officials frequently use a larger array of bilateral and multilateral diplomatic ties to gather such data and to structure effective sidepayments and issue-linkages. It is thus the entirety of bilateral and multilateral ties that connects the United States to potential coalition partners that matters-not just alliance ties alone. Moreover, while US diplomatic networks certainly form part of the American power infrastructure, hegemonic-order theory-at least in its simplest version-often cannot explain the outcomes of coalition negotiations. On many occasions, potential coalition participants enjoy considerable bargaining power at the negotiation table. They understand how much the United States desires their troop contribution and thus can drive a hard bargain (cf. Keohane 1971;McKibben 2015, 17-18;Moravcsik 1998, 62).
To illustrate these claims, I use a mixed-methods research approach. First, I conduct a statistical analysis including all US-led multilateral coalitions in the pre-as well as post-Cold War period. The statistical analysis tests whether an overall correlation between diplomatic Marina E. Henke is an Assistant Professor of Political Science working in international relations at Northwestern University. Her research focuses on multilateral military coalition-building, military interventions, and peacekeeping.
embeddedness-which I define as the sum of bilateral and multilateral institutional ties that link the United States to a third party-and coalition contributions exists. Second, I complement the statistical analysis with an indepth case study on the US-led multilateral coalitionbuilding effort for the Korean War. The Korean War broke out in the early days of the Cold War. As a result, one would assume that alliance concerns and US "hierarchy" significantly affected deployment decisions to Korea. The case study thus represents a crucial case (Eckstein 1975, 127;Gerring 2007) to illustrate the role that diplomatic networks and negotiations play in US-led coalition-building processes. Much of the evidence presented in this case study comes from archival research at the National Archives in Washington, DC, and the Truman Presidential Library Archives.
The statistical analysis strongly confirms the importance of diplomatic embeddedness in US coalition-building efforts. Across five different statistical models, diplomatic embeddedness strongly and positively correlates with coalition contributions. The qualitative analysis, in turn, shows how US officials relied on information gathered through diplomatic ties to entice countries to join the Korean War coalition. Moreover, they used diplomatic networks to construct side-payments and issue-linkages. On several occasions, the United States did not dominate the coalition negotiations. Instead, countries such as Ethiopia and South Africa were able to reduce American bargaining power by exploiting the fact that the United States desperately wanted troop contributions for the Korean War.
In times when the US administration pushes for an "America First" policy and questions the utility of international agreements and institutions, these findings are more significant than ever. Most importantly, they illustrate that "being networked" matters in international politics. Diplomatic embeddedness, if used purposefully, constitutes an important state resource. The United States can reap benefits from institutional connections that go way past the specific purposes the individual institutions were created for; bilateral and multilateral relationships matter beyond what they were set out to achieve; each institutional tie can be profitable beyond the purpose it was meant to address.
The remainder of the article proceeds as follows: it starts by offering a brief review of the existing literature on US coalition-building. Next, it presents a theory of the importance of diplomatic embeddedness in US-led coalition-building efforts. The third section tests this theory statistically. The fourth section complements the quantitative analysis with a case study of the US coalition-building process for the Korean War. The fifth section concludes.

What Do We Know About US-Led Coalition-Building?
Most scholars analyze US-led coalition-building through the lens of alliance theory (Davidson 2011;Von Hlatky 2013;Weitsman 2014;Bennett, Lepgold, and Unger 1997). Indeed, both policymakers and academics often use the terms "alliances" and "coalitions" as synonyms (Resnick 2010(Resnick /2011Walt 1987). Nevertheless, some writers see coalitions as shorter in duration, less institutionalized, and more task-specific than alliances (Morey 2016, 535;Dibb 2002, 132;Morrow 2000, 64;Snyder 1997, 12;Weitsman 2003, 20). Scholars suggest that US formal alliance agreements can impact American coalition-building efforts in a number of ways. First, US allies generally share US threat perceptions, norms, and values (Walt 1987;Sandler 1993;Snyder 1997;Davidson 2011;Weitsman 2003Weitsman , 2004. These factors most frequently account for alliance formation (Lai and Reiter 2000;Leeds and Savun 2007;Gibler and Wolford 2006;Fordham and Poast 2016). Second, US allies tend to further align their foreign policy interests with American ones. This may result from efforts to signal the credibility of their alliance commitment, socialization dynamics, or both (Adler and Barnett 1998;Morrow 2000, 69-70;Altfeld and Bueno de Mesquita 1979;Corbetta and Dixon 2004;Pilster 2011;Bearce and Bondanella 2007). Third, US alliance partners provide military support-or come under pressure to provide support-to American-led coalitions even when their threat perceptions and interests diverge from those of the United States. Most often this happens because of the Alliance Security Dilemma-a two-faced fear of either entrapment or abandonment by allies (Snyder 1984(Snyder , 1997. 1 With regard to coalition formation, a focus on the Alliance Security Dilemma predicts that American allies will join coalitions to enhance their status with the United States as a loyal and respectable ally (Von Hlatky and Trisko Darden 2015;Davidson 2011;Bennett, Lepgold, and Unger 1997, 13;Cox and O'Connor 2012;Davidson 2011). Fourth, US alliance partners participate in US-led coalitions to avoid facing audience costs. They worry that, if they do not join the coalition, they will gain a reputation for unreliability. This means that other states may prove reluctant to ally with them in the future (Crescenzi et al. 2012;Fearon 1994;Smith 1995;Gibler 2008;Morrow 2000, 71-72). Finally, US alliance partners usually engage in military coordination efforts (for example, establish common command structures, equipment interoperability, and exercises). This coordination can improve the ability of American allies to fight together and increase the chances of a successful intervention. This makes US allies even more favorably disposed to join American-led military coalitions (Morrow 1994).
In addition to alliance theory, hegemonic orders and related economic or political dependencies have also been used to explain US-led coalition contributions (Lake 2009;Baltrusaitis 2009). From this viewpoint, the United States uses its asymmetrical power capabilities to ensure that its foreign policies, which include US-led military interventions, are actively supported by those that benefit from US hegemonic benevolence (Ikenberry 2001, 27). Newnham (2008, 86), for instance, argues that the United States made use of its foreign aid policy to coerce countries to participate the 2003 US-led Iraq invasion (see also Tago 2008). In the same vein, Lake (2009, 173) argues that states that are "subordinate" to the United States are more likely to provide troops to US-led coalitions. With regard to the Iraq War, he writes: "It appears that states participated in the war as a costly signal of support for the United States and an acknowledgment of its authority" (Lake 2009, 173).
Finally, scholars have advanced normative explanations to account for multilateral coalition formation processes (Finnemore 2003;Wheeler 2000). These explanations 1 Entrapment refers to supporting an ally in a conflict in which the other ally has no interest. Abandonment, in turn, occurs when an ally realigns with one's adversary or fails to fight the adversary. The dilemma arises because reducing one of these risks increases the risk of the other: strong commitment to an ally's causes reduces the risk of abandonment but raises that of entrapment. are particularly prevalent with regard to humanitarian interventions and operations conducted under the umbrella of the United Nations (Bellamy and Williams 2013;Cunliffe 2013); although some US-led operations, such as Operation Restore Hope in Somalia in 1993, would certainly qualify. Participation in this latter type of operation is often attributed to shared regime type: democracies tend to fight alongside and for each other (Lebovic 2004;Gibler and Sarkees 2004;Leeds et al. 2002;Andersson 2000;Risse-Kappen 1995;Mousseau 1997). Democratic governance structures can, however, also inhibit troop contributions. Auerswald (2004) finds, for instance, that institutionally weak democratic executives are likely to provide only minimal support for US-led interventions, as they are too preoccupied with retaining office. The timing of elections also appears to matter: states are particularly reluctant to share US defense burdens if their leaders face the ballot box (Baltrusaitis 2009, 215;Gaubatz 1995). Tago (2014) also suggests that domestic economic variables impact coalition participation: if a country is in an economic recession or confronted with domestic riots, it is particularly unlikely to join a US-led coalition. Rathbun (2004), on the other hand, argues that partisan orientations matter: leftist parties are more likely to support participation in humanitarian interventions than rightist ones. Furthermore, participation in US-led coalitions has been attributed to state identities and socialization processes (Vucetic 2011;Adler and Barnett 1998).

What Are We Missing?
The existing literature has largely overlooked that US-led coalitions almost never fall into place automatically. 2 Instead, American officials have to engage in intense negotiations with potential coalition participants in an attempt to convince them to join a specific operation. I am interested in examining how these coalition negotiations unfold: Do these negotiations follow a systematic pattern? Is there a coherence and recurrence of activities? Who holds power in these negotiations? Thus far, the existing literature has failed to answer these questions.

The Role of Diplomatic Embeddedness in US Coalition-Building Efforts
This article suggests that diplomatic embeddedness between the United States and potential coalition participants greatly affects US-led multilateral coalition-building efforts. I define diplomatic embeddedness as the sum of bilateral and multilateral institutional ties that link the United States to a third party. Each of these ties leads to (more or less elaborate) institutional fora in which issues are discussed and bargained over, and through which resources can be channeled (cf. Hafner-Burton, Kahler, and Montgomery 2009). Each tie also establishes social interactions and thus creates relationships among US and foreign individuals (cf. Cooley and Nexon 2016, 78;Hardt 2014;Goddard 2012). I argue that these ties play a crucial role when it comes to determining the propensity of a country to join a US-led coalition. Why is this the case? US coalition-building involves intense negotiations: coalition partners need to be bargained into participating in a specific US-led coalition. In addition, on many occasions, coalition participants require side-payments to motivate their coalition contribution. In both of these instances, information about the underlying preferences of potential coalition participants with regard to the deployment is a crucial variable: the more and the better information the United States possesses on the exact preference constellations of potential coalition participants, the greater the likelihood that coalition negotiations will succeed. Moreover, the more ties exist, the easier the process of negotiating side-payments and issue-linkages (Martin 1992). Alliance ties provide information and issue-linkage opportunities mostly (though not exclusively) on securityand military-related matters. Nevertheless, coalition participation decisions do not only involve military and security concerns; on many occasions, participation decisions are linked to economic, financial, and other political interests. Other non-alliance-related bilateral and multilateral ties are thus often used to access such information and to facilitate the construction of side-payments. As a result, it is the entirety of bilateral and multilateral ties that connects the United States with a third party that positively affects coalition participation-not just alliance ties alone. In the following, I describe this theory in greater detail.

The Role of Information in US-Led Coalition-Building Efforts
International Relations theory often overlooks agency and process and reduces outcomes to structural features (Rathbun 2014, 1;Goddard and Nexon 2016, 8). This is particularly true when it comes to US-led multilateral military coalition-building. Many accounts depict these coalitions as emerging instinctively without much political labor involved (Walt 1987;Weitsman 2003Weitsman , 2004. This is, however, a fallacy. US coalition-building involves extensive political maneuvering. Over the course of several months, US officials work tirelessly to recruit coalition participants. The coalition-building process typically unfolds in two steps: first, US officials establish so-called "prospect lists" of countries that might be good candidates for coalition negotiations. Second, once these lists are established, US officials start negotiations with this select number of countries. In both of these steps, diplomatic embeddedness is a decisive factor. Indeed, US selection criteria for coalition partners are usually threefold: (1)  The following assets are often considered: (1) size of the contingent a state can contribute; (2) quality of troops; (3) provision of specific military services or military assets (e.g., helicopters); (4) specific geographical assets (proximity to the theater of operation); and (5) specific political, ideological, or religious assets. A memo written by John T. McNaughton, the US Assistant Secretary of Defense for International Security Affairs, to McGeorge Bundy, the US National Security Advisor in the Lyndon B Johnson administration, presenting a prospect list for the Vietnam War coalition illustrates the presence of these three criteria quite nicely: the memo lists three categories of states that might be potential contributors to the Vietnam War. The first category lists three states that already have combat troops in Vietnam: South Korea, Australia, and New Zealand. McNaughton notes that it would be possible to obtain two more battalions (2,000 troops) from Australia without "cumshaw" since it has enough capability and motivation to do so. The second category then lists states that might consider deploying troops if the United States delivered "cumshaw." McNaughton even estimates a possible dollar amount: US$300-500 million. These countries include Greece, Taiwan, the Philippines, and embeddedness to gather data on these three criteria-especially on criteria (2) and (3), which are in essence subjective political considerations. Without access to private information, these considerations cannot be determined. Diplomatic embeddedness, nevertheless, opens information channels to access such information (Sending, Pouliot, and Neumann 2015, 223). Effective US diplomats in overseas posts with which the United States has established a tight net of bilateral and multilateral ties can usually provide detailed information on personalities, policy preferences, and parochial interests of their host state governments that is not readily available from public media accounts (Neumann 2012, 35;Rozental and Buenrosto 2013, 230-32;Bull 1977, 181). Such data is often retrieved from one-on-one interactions, which further increases the reliability of the information (Holmes 2013;Hall and Yarhi-Milo 2012, 562;Yarhi-Milo 2014, 18). Much of its quantity depends on diplomatic embeddedness: the more diplomatic ties that exist, the more officials can be accessed and the more information can be gathered. 5 In the multilateral context, international summits are also critical events to study other governments' worldviews, their stances on broad issues, as well as the boundaries in which they operate (Devin 2013, 92-93;Pouliot 2016, 63): on which topics do foreign officials speak up? When do they stand on the sidelines? How do they interact with other members? As a result, both bilateral and multilateral diplomatic ties generate information. This information can then be used to construct a good argument to persuade potential coalition participants to join a given coalition (cf. Risse 2000, 10-11). Furthermore, on occasion, a potential participant needs more than just a good argument. In these instances, diplomatic embeddedness can provide information on what type of side-payment the coalition participant is interested in. Indeed, without access to such information, side-payment negotiations are likely to fail even in the presence of joint gains, as both parties have an interest in dissimulating their true red lines in an attempt to get a better deal (De Dreu, Koole, and Steinel 2000;O'Connor and Carnevale 1997;Olekalns and Smith 2009, 347-48;Risse 2000, 21;Rathbun 2014).

Other Benefits of Diplomatic Embeddedness
In addition to accessing information, diplomatic embeddedness also has several other positive effects on coalition negotiations. Diplomatic embeddedness can lead to trust among US and foreign counterparts. This trust can make coming to the negotiating table even more likely, as diplomats are less afraid that the information they reveal during the negotiations will be used against them in the event the talks break down (Rathbun 2014, 28;Schultz 2005;Larson 1997). Moreover, diplomatic embeddedness may also lead to the convergence of negotiation practices, which further facilitates the bargaining process (Pouliot 2008;Adler-Nissen and Pouliot 2014). Diplomatic embeddedness can also be used to channel side-payments through already-existing bilateral or multilateral ties, thus avoiding public scrutiny and/or the necessity to create new budget lines, which might require congressional or other political approval (Henke 2012). Finally, of course, diplomatic embeddedness can also play an important role in generating "coercive cooperation." In other words, the United States can manipulate diplomatic ties to the detriment of the potential participant, thus changing the incentive structures of a potential coalition participant. The more bilateral and multilateral ties that exist, the greater the range of linkages and also the greater the credibility of these linkages (Martin 1992, 39-40;Axelrod 1984;Keohane 1984).
Alliance ties alone can often not provide the same breadth of services. While military and security preferences do impact coalition negotiations, a range of other topics frequently gets tied to coalition bargains, such as economic aid and trade agreements, atomic energy agreements, sanction relief, debt relief, World Bank loans, high-level political appointments, cash payments, and so on (Henke 2016;Sarantakes 1999;Wolford 2015). As a result, US officials frequently rely on diplomatic embeddedness instead of alliance ties when bringing coalition negotiations to a close. The next section examines statistically whether these hypotheses of the importance of diplomatic embeddedness in US coalitionbuilding efforts are indeed accurate.

Statistical Analysis: Does Diplomatic Embeddedness Affect US Coalition-Building Efforts?
This section presents a statistical analysis of the importance of diplomatic embeddedness in US coalitionbuilding processes. I use two different datasets to test this hypothesis. I start the quantitative analysis by replicating a statistical study conducted by Lake (2009). 6 The dataset includes 15 US-led coalitions that were constructed between 1950 and 2000. 7

First Statistical Analysis
Lake (2009) uses "Coalition Participation" as a dependent variable. It has a binary value, that is, 0 (¼ no participation) or 1 (¼ participation). The key independent variables of the study are two "hierarchy indexes," which attempt to measure how dependent a state is on the United States. The first index, the "security hierarchy index," measures a state's dependence on the US security umbrella. The second index, the "economic hierarchy index," intends to capture a state's economic dependence on the United States. Each index is composed of two indicators. For the "security hierarchy index," the index components are (1) an indicator that measures the presence of US forces on the territory of the potential coalition participant and (2) the number of non-US or "independent" alliances possessed by the potential coalition contributor. 8 Thailand. Finally, the third category includes states that the United States perceives as having very little intrinsic motivation to join the coalition and, thus, it would involve "selling our souls and raising hob in various ways" to encourage these states to commit. The latter group included Germany, Israel, Spain, Turkey, and the UK (McNaughton 1965). 5 Of course, each tie might not be of equal relevance (see for example Hafner-Burton, Kahler, and Montgomery 2009). Nevertheless, as Granovetter (1973) in his article "The Strength of Weak Ties" rightly observed, even "weak" ties can at times serve as an invaluable informational resource. 6 Lake's (2009) analysis is based on Tago (2007). For the "economic hierarchy index," the index components are (1) an indicator that measures monetary dependence on the United States and (2) an indicator measuring the relative trade dependence of the potential coalition participant on the United States. In addition, Lake (2009) controls for major power status of the potential participant, its annual military expenditure, its geographical contiguity with the target state of the intervention, joint primary language, joint regime type, involvement in an enduring rivalry, and a proxy variable to test whether the potential participant faces an internal government crisis at the moment of deployment. Furthermore, the analysis includes variables to test whether the type of coalition matters: is it a "war" (that is, military action against at least one sovereign state with more than 1000 battle deaths) or rather a domestic intervention (that is, military intervention that takes place inside a country's territory without the consent of the government)? Does the coalition operate under a UN or regional mandate?
I now add to this range of independent variables a new variable: Diplomatic embeddedness. Diplomatic embeddedness is operationalized by counting all bilateral and multilateral institutional ties the United States has established with a foreign country at the date of intervention. Data on cumulative joint international organization membership comes from the International Governmental Organization (IGO) Dataset (Pevehouse, Nordstrom, and Warnke 2004). Data on cumulative bilateral cooperation treaties comes from the World Treaty Index. 9

Model Specifications
I adopt Lake's model specifications in their entirety. I run logit regression models and employ panel-corrected standard errors to mitigate possible heteroskedasticity. I also use a lagged dependent variable and an AR (1) correction to address problems of serial autocorrelation (Lake 2009, 169).

Statistical Results
The statistical results of this first test are shown in Table 1. For comparability purposes, Model 1 represents the exact replication of the Lake study. Model 2 shows the results when the variable Diplomatic embeddedness is added to the analysis. We can see that Diplomatic embeddedness is positively correlated with troop contributions and highly statistically significant. One other noticeable change compared to Model 1 is the loss of statistical significance of the "security hierarchy index." All other results remain essentially the same. Model 3 now disaggregates the indexes into their individual components. The changes are as follows: Diplomatic embeddedness remains positively correlated with coalition contributions and highly statistically significant. The variable Exchange Rate Regime also strongly positively correlates with coalition contributions at the 0.01 level. The variable Independent Alliances positively correlates with coalition contributions at the 0.05 level. All other variables do not change substantially compared to the two previous models. Overall, this first statistical analysis thus provides preliminary evidence of the importance of diplomatic embeddedness in US coalitionbuilding efforts. In addition, it illustrates that exchange rate regimes affect coalition contributions. Alliance dependence also has an effect on coalition contributions, however, to a lesser degree. Overall, we learn that security hierarchy alone might not be able to explain the entirety of coalition contributions we observe.

Second Statistical Analysis
The second statistical analysis uses an original and much larger dataset of 41 US-led coalitions conducted between 1990 and 2005. 10 This dataset includes coalitions for This indicator is used to measure foreign policy autonomy from the United States. The more the potential participant has "independent alliances," the less hierarchical the security relationship between the potential participant and the United States (Lake 2009, 69-70). 9 I counted all bilateral treaties concluded since 1945 toward the overall number of bilateral ties. This includes all types of bilateral agreements, including security, socio-economic, and other. If a specific state did not exist in 1945, bilateral treaties are counted from the date of its creation. 10 I list these operations (and explain their inclusion) in the online appendix "Operations included in Dataset." which the United States served as a Pivotal State, defined as the state that either politically initiated and orchestrated the set-up of the coalition, 11 led, and/or made the most significant material or financial contribution to the operation. 12 During the time period under examination, the United States served as Pivotal State five times in an ad hoc manner and 36 times under the umbrella of an international organization such as the United Nations or NATO. 13

Dependent Variables
Lake (2009) uses "Coalition participation"-a binary variable-as a dependent variable. My analysis attempts to provide greater detail. Therefore, it uses three dependent variables: (1) a binary variable set to one if a country participated at all in a specific US-led coalition; (2) a binary variable set to one if a country deployed at least a company-sized contingent (minimum 100 troops); and (3) a continuous variable accounting for the exact number of troops deployed by a given country to a specific US-led coalition. While the first dependent variable examines overall participation, the second dependent variable focuses on substantive contributions, thus excluding symbolic or token contributions. The third dependent variable captures the exact level of troop commitment. Like Lake (2009)

Independent Variable
The key independent variable of this extended quantitative study of US coalition participation is again dyadic Diplomatic embeddedness. It is used to test the core hypothesis of this article: the higher the degree of diplomatic embeddedness between a potential contributing country and the United States, the more likely that country is to participate in/send a larger number of troops to a US-led operation.

Control Variables
The study uses control variables similar to Lake (2009). Nevertheless, its focus is less on "hierarchy." As a result, it does not employ any hierarchy indexes. Instead, it tests the explanatory power of variables such as formal alliance agreements between the United States and potential coalition participants; potential coalition participant-target state contiguity; potential coalition participant and target state common regional membership; the potential coalition participant's CINC score (Singer, Bremer, and Stuckey 1972); and its involvement in a dispute at the time of the intervention. 15 To capture the potential coalition participant's dependence (that is, hierarchy) on the United States, I use dyadic trade and aid data. I divide the aid and trade flows between the United States and the potential coalition participant by the participant's GDP to control for the relative importance of such flows to the participant. 16 In addition, I also control for whether the potential coalition participant has strong diplomatic, trade, and aid ties with the target state of the intervention. 17 To measure the relative importance of each particular economic tie from a potential participant to a target country, I again divide the overall bilateral aid and trade flows by the GDP of the participant. I also include the POLITY-2 score from the POLITY IV project (Marshall, Jaggers, and Gurr 2010) of the potential coalition participant to control for liberal motivations to contribute troops to US-led coalitions (cf. Lebovic 2004). Moreover, I include interest affinity between the United States and the potential coalition participant. This variable, which is based on the Affinity of Nation Index, tries to control for the overall similarity of a dyad's revealed preferences (Gartzke 2006;Strezhnev and Voeten 2013). 18 11 Cf. Bellamy and Williams (2010, 43) After the Cold War and, especially, after the Somalia intervention in 1993, the United States developed a policy (US Presidential Decision Directive 25) that US troop deployments to peacekeeping operations had to be kept to a strict minimum. This policy, however, did not stop the United States from orchestrating many (if not almost all) political aspects of such deployments in countries that were of interest to successive US administrations, notably, peacekeeping operations in Liberia, Sudan, Haiti, Somalia, and Ethiopia/Eritrea. There is ample empirical evidence that this latter group of peacekeeping operations would not have deployed without US leadership (see Henke (2016)  This dataset differs from Lake (2009) in the following ways: First, the United States is not required to provide the largest portion of coalition forces. Second, a US military officer does not need to command the coalition. However, under my definition the United States is required to politically initiate and orchestrate the set-up of the coalition. My dataset also differs from Wolford (2015, 14), who defines a military coalition as "a group of two or more states that makes a threat to use force together against another state (or states) in an international crisis." According to my definition, actual military forces need to be deployed abroad. A threat alone is not sufficient. 13 On the similarity between ad hoc and UN recruitment processes, see Henke (2016). 14 Codebook in author's possession. 15 More precisely, the variable captures the cumulative number of separate alliances in a particular year in which both states are members (overlapping for at least one day). Data on common alliance memberships has been retrieved from the Correlates of War (COW) Formal Alliance Dataset (Version 3.03) by Gibler and Sarkees (2004). Data on conflict involvement comes from the COW MID Dataset (4.1) by Palmer et al. (2015). 16 The aid data comes from the OECD DAC database. Trade data comes from the COW Trade Dataset (Version 2.01) by Barbieri and Keshk (2012). Aid and trade data is log transformed. 17 Embassies or consulates are usually only opened in places in which a country has something at stake. This is especially the case for developing countries for which the maintenance of an embassy/consulate constitutes a substantive financial burden. Diplomatic representation data has been retrieved from the COW Diplomatic Exchange Dataset (Version 2006.1) by Bayer (2006). 18 The inclusion of this variable might raise concerns of a causal relationship between dyadic UN affinity and diplomatic embeddedness. In other words, diplomatic embeddedness might cause UN affinity via socialization processes (Checkel 1999;Bearce and Bondanella 2007;Johnston 2001) or UN affinity might lead to dyadic diplomatic embeddedness. In both of these scenarios, the regression coefficients of both UN affinity and diplomatic embeddedness cannot be correctly interpreted. I don't think, however, that such concerns are warranted. Intervention decisions are quite peculiar. They are rooted in geography, history, economic, and other societal characteristics of individual states. Just think of why the United States intervened in Iraq (2003) or Haiti (1994). In both of these cases, a very specific set of domestic political dynamics drove the US intervention decision-making process: neoconservative ideas, the experience of 9/11, bureaucratic politics in the George W. Bush White House (among others) in the case of Iraq (Harvey 2012;Record 2010); humanitarian interests, ethnic, and business lobbies and mobilization by the Black Caucus in the US Congress (among others) (Malone 1998) in the case of Haiti. No other country was exposed to these dynamics to an equal degree, and it is highly unlikely that via diplomatic embeddedness these preferences Finally, I control for the potential coalition participant's level of economic development (in terms of GDP per capita). 19

Model Specification
I use the new dataset to run three different statistical models. Regarding the model specifications, I stick very closely to Lake's (2009) analysis above. Models 4 and 5 are both logit regression models. Model 4 uses the binary dependent variable set to one if a country participated at all in a specific military operation. Model 5 uses a binary dependent variable set to one if a country participated with at least 100 troops in the same operation. 20 Both models estimate robust standard errors clustered by target state to control for intraclass correlation among cross-sections. 21 I thus assume that observations are not independent within one specific conflict theater. 22 Both models also include year fixed effects and control variables for whether the operation is conducted ad hoc or under a NATO or UN umbrella. Finally, Model 6 employs a zero-inflated Poisson regression model. It uses a count variable as a dependent variable (that is, the log transformed number of troops deployed per country/ operation). It controls for an over-dispersed distribution of the dependent variable (that is, the dependent variable includes "excess zeros," countries that are counted as potential troop contributors but in actuality are not approached by the United States to serve in a specific coalition). I argue that diplomatically embedded states are more likely to be approached by the United States for troop contributions. I thus use diplomatic embeddedness to predict the excess zeros among the potential participants. All other model specifications stay exactly the same as in the previous models (all but time-static independent variables are lagged by one year). The statistical results are shown in Table 2 below.

Statistical Results
The second statistical analysis lends further probability to the claim that diplomatic embeddedness between the United States and potential coalition participants influences troop deployments. Across all three statistical models, using three different dependent variables, the variable Diplomatic embeddedness consistently positively correlates with troop deployments at highly statistically significant rates. Alliance ties do not appear to matter at all (cf. Wolford 2015 for a similar finding). Indeed, in two of the three models presented in Table 2, they are negatively correlated with coalition contributions. Nevertheless, in addition to Diplomatic embeddedness, other variables also appear to be important predictors of troop deployments.
Most importantly, the democracy score of a country seems to positively influence a country's decision to participate in a US-led coalition. In addition, countries that are involved in another conflict at the time of the US coalition-building effort are highly unlikely to join a USled coalition (cf. Tago 2007). The same effect occurs for dependence on trade with the United States. Figure 1 presents the predicted effects of Diplomatic embeddedness on the probability of participating in a US-led coalition while holding all other variables at their means. 23 The filled triangles show the impact of Diplomatic embeddedness on the probability of participating at all in a US-led operation. This probability amounts to roughly 15 percent at the mean of Diplomatic embeddedness (corresponding to about 71 bilateral and multilateral ties); 25 percent when we look at one standard deviation above the mean (about 129 ties); and 37 percent at two standard deviations (about 188 ties) above the mean. The hollow triangles illustrate the probability of contributing at least a company-sized contingent (minimum 100 troops) to a USled coalition. This probability amounts to roughly 3 percent at the mean of Diplomatic embeddedness. It doubles to 6 percent when we look at one standard deviation above the mean and increases to roughly 11 percent at two standard deviations above the mean.

Robustness Checks
I split my dataset into UN and non-UN operations to control for the possibility that the UN interferes in US-led coalition-building processes. 24 The results indicate that for both UN-only and non-UN operations Diplomatic embeddedness is positively associated with troop contributions and highly statistically significant: for only UN operations at the 0.01 level and for non-UN operations at the 0.01 level for deployments of minimum 100 troops and at the 0.05 level for all sizes of deployments. For only UN operations, the alliance coefficient is strongly negatively correlated with troop contributions. 25 For only non-UN interventions, the alliance variable is no longer statistically significant. In addition, I conduct fixed effects models. The models shown above might suffer under an omitted variable bias. One could argue that an unobserved factor -let's call it homophily, biases the estimations. In other words, Diplomatic Embeddedness, alliance agreements, interest affinity and even coalition participation does not occur randomly but could be the result of a US tendency to associate and bond with states that are similar to herself (i.e., homophily). The problem is that homophily cannot be measured. Nevertheless, if such tendency indeed exists, it does not change over time (Mouw 2006, Shalizi andThomas 2011). As a result, a fixed effects model can provide remedy. It produces unbiased estimates by using each variable's difference from its within-individual mean. Time-invariant variables, including unobserved one (i.e., homophily) thus drop out of the model (Halaby 2004). The results of the fixed effect models do not put into doubt any of the outcomes shown above. Finally, I test my results for multicollinearity. Alliances and diplomatic embeddedness might be highly "spilled over" into other countries and thus led to quasi-homogeneous preferences with regard to the intervention proposals. It is also unlikely that these specific interests led to the establishment of diplomatic ties. In short, two countries can cooperate intensely on economic, environmental, and even security issues and still have different preferences with regard to a particular intervention. While diplomatic embeddedness can have socialization effects and interest convergence can lead to diplomatic embeddedness, it is highly doubtful that both of these factors generate identical preferences on each single intervention proposal. 19 Data comes from the World Bank. I use the log of GPD per capita. 20 Model 5 thus attempts to exclude symbolic or token contributions. 21 See Wolford (2015, 77) for a very similar model choice. 22 Several target states receive multiple operations (e.g., the former Yugoslavia). A country that participates in the first operation in a specific target state might be more likely to stay on and also participate in all subsequent operations. I also run robustness checks controlling for intraclass correlation among coalitions and participants. The substantive results do not change. 23 Models 4 and 5 from Table 2 were used to perform the analysis. The estimation shows a confidence interval of 95%. 24 On US involvement in the initiation of UN operations and recruitment of UN peacekeepers, see Henke (2016). 25 At the 0.01 level for all contributions and at the 0.05 level for contributions equal to or larger than 100 troops. correlated, thus impacting the results of the various regression analyses. I find that Diplomatic embeddedness and the Index of Security Hierarchy (Table 1)  Overall, the statistical analyses thus strongly support the hypothesis that diplomatic embeddedness matters greatly in US coalition-building efforts. To further clarify the causal mechanisms at work and to examine, in particular, how diplomatic embeddedness and US power interact, I present an in-depth case study of the coalition-building process for the Korean War below.  (44). 26 This section attempts to illustrate how diplomatic embeddedness affected the recruitment of these coalition participants.  (Bohlin 1985, 153). Nevertheless, the Pentagon's dominance of the coalition-building process and its focus on military suitability was very short lived. Indeed, by early August 1950, the Soviet Union had launched an impressive propaganda machine denouncing the US-led intervention in Korea as lacking any substantive international support (Stueck 2005, 57). US Secretary of State Dean Acheson and his colleagues were alarmed. They felt that it would be debilitating to allow the Soviet Union to win the propaganda war on Korea. As a result, the number of participants in the Korean War coalition had to be increased no matter what. Acheson thus requested a direct repeal of the manner by which the recruitment process had been implemented by the Pentagon. He requested an accelerated recruitment process in which diplomatic considerations trumped military expediency. President Truman concurred. Thus, by early August 1950, the State Department had taken over political control of the coalition-building process for the Korean War. Military criteria would from now on only serve as a general guideline; they would no longer determine recruitment choices.

A Search Strategy for Coalition Participants Based on Diplomatic Embeddedness
Despite an initial political outburst of UN member support for the US-led intervention in Korea, very few countries volunteered any military forces (Stueck 2005, 57). Many countries thought that fighting in Korea was none of their business (Bohlin 1985, 161). As a result, US officials had to actively recruit and negotiate countries into the Korean War coalition. To do so, officials from the State Department 27 met on repeated occasions to establish so-called "prospect lists"-memos that would list countries to be approached for force contributions. One of these occasions was February 13, 1951 (Minutes of Meeting on Military Assistance for Korea, 1951). The first countries that officials discussed at this joint meeting as potential coalition contributors were all located in Latin America. Brazil and Uruguay were particularly high on the list, followed by Chile and Peru. None of these countries were chosen at random. Rather, State Department officials hand-selected these countries because bilateral diplomatic channels suggested that these countries might be willing to join the coalition if they were offered the right arguments and incentives. Negotiations with Uruguay had indeed been ongoing since August 1950. To be precise, on August 8, 1950, the Uruguayan President Luis Batlle Berres had informally communicated to the US ambassador to Uruguay, Christian Ravndal, that he was considering a deployment of 2,000 troops to Korea. Batlle, however, also indicated in further diplomatic communications that he was unhappy with the long-standing refusal of the United States to sell arms to his country. US officials then clarified that the only way to get these arms was via a deployment to Korea (Bohlin 1985, 177-80). Similarly, negotiations with Brazil had been ongoing since July 14, 1950, when the American government asked UN Secretary General Trygve Lie to advise Brazil that the Korean War coalition urgently needed troops. To endear the Brazilians to the American proposal, Washington immediately provided strong support of Brazil's candidacy for reelection to the UN Security Council in the fall of 1950. Moreover, the US government instructed the US Export-Import Bank to release US$25 million for steel equipment for usage in Brazil. When no positive reply followed, the State Department also proposed to open talks for further economic aid and the negotiation of a bilateral technical assistance agreement with Brazil (Hilton 1981, 606-7). The State Department also knew about Chile's preferences with regard to Korea via its diplomatic channels. In September 1950, the American embassy in Chile had reported that the Chilean cabinet had discussed a potential deployment. Back then, various cabinet members had, however, identified several obstacles that impeded action. For instance, the equipment of the Chilean Army, Navy, and Air Force was limited and antiquated, and public opinion was not yet prepared for a deployment On occasions, DOD officials participated as well in these meetings. (Bohlin 1985, 180). Thus, when listing Brazil, Uruguay, and Chile as key prospects for troop contributions, State Department officials used diplomatic embeddedness to overcome information asymmetries. They decided that they would again approach these countries based on the knowledge of their preferences with regard to Korea. Next, the team discussed the prospects for Canada, Australia, New Zealand, South Africa, and Sweden to deploy troops to Korea. They decided that Canada should be asked to increase its troop contribution to a brigade. Canada had signaled to US diplomats its willingness to help the Korean War effort. The only obstacle Canada faced was domestic opposition in Quebec, which was clamoring that Canada should "remain neutral for a couple of years." Prime Minister Pearson had thus pleaded for the United States to be understanding in that matter (Woodward 1950). In addition, the group concluded that Australia should increase its troops in Korea from one battalion to two, and that it "should be asked to furnish a replacement battalion in Japan for the battalion that had originally been transferred from Japan to Korea." The State Department was aware that Australia was seeking a defense agreement with the United States. Fearful of Britain's declining capacity to protect Australia and convinced of US unwillingness to extend its security umbrella voluntarily to Australia, the Australian government had been lobbying hard to gain a definite commitment from Washington (Stueck 2005, 73-74). Based on State Department officials' assessment, the group also concluded that "because of political considerations," Sweden should not be approached for troop contributions at this time (Minutes of Meeting on Military Assistance for Korea, 1951). Finally, the team analyzed troop prospects from the Middle East. The Deputy Secretary of State for Near Eastern Affairs, J. Lampton Berry, stated at the outset that there was practically no chance of securing any voluntary contributions of military forces to the Korean campaign from these countries; only serious "pump priming" would bring further contributions. He mentioned, for instance, that US diplomatic channels had gathered that Turkey "was requesting a direct security commitment from the United States Government and 100 million dollars to balance its budget" (Minutes of Meeting on Military Assistance for Korea, 1951). The Turkish prime minister had handed an aide-memoire to US Assistant Secretary McGhee already in August 1950 that included an urgent plea to include Turkey in NATO (McGhee 1950, August 14). Again in September 1950, the Turkish ambassador to the United States had raised the issue with Deputy Assistant Secretary Burton (Berry 1950, September 11, 1950, and even the Turkish president himself had mentioned the issue to the US ambassador to Turkey (Wadsworth 1950, September 12). The group considered the possibility of contributions from Iran, Iraq, Israel, India, and Pakistan last. Quickly, however, the panel concluded again that "political considerations" made it highly improbable that any assistance would be offered by these countries. The session concluded with the agreement that "the appropriate bureaus of the Department of State should, as soon as possible, take action in accordance with the consensus of view of the meeting in hope that some additional forces could be provided in early spring" (Minutes of Meeting on Military Assistance for Korea, 1951).
Several other of these types of meetings occurred throughout 1950-1951. Each of them showed that the US government undertook an explicit search process to recruit coalition partners. While military considerations mattered to some extent in the latter process, it was nevertheless political factors that were often decisive. These political factors mainly included how much a country cared to be a member of the coalition and what the US government could offer to make that same country join the coalition. The examples above show that coalition participation was tied to issues such as UN Security Council elections, export credits, bilateral technical assistance agreements, economic aid, arms deals, defense agreements, and cash. In addition, other factors also entered the equations such as public opinion concerns, notably in Canada (Quebec) and Chile. US government officials largely obtained information on these variables via US diplomatic channels. So what about individual coalition negotiation? His Imperial Majesty would wish to make the following inquiries: (a) to what extent would US be able to assist in equipping and arming a unit of approximately 1,000 officers and men to transport them and supply them at the fighting front? (b) To what extent is US prepared to assist in further equipping and strengthening the Ethiopian Armed Forces in order that Ethiopia may be able to increase her usefulness in the common defense of democratic institutions and her friends and allies? (c) What practical steps may be taken by Ethiopia to obtain US financial assistance to enable her to develop more rapidly her economic resources for her own needs and the need of the world in peace and in war?
The message also made a daring reference to the notion that Ethiopia felt unfairly treated when it came to US economic and military aid. It stated that "[h]is Imperial Majesty . . . does not wish to conceal his surprise at Ethiopia not having received comparable assistance from [the] US [than other countries]." He complained, in particular, about Italy having received "considerable military and economic assistance from America" despite its violation of "her obligations to the League of Nations" and "the war of aggression against Ethiopia" (Merrell 1950a, August 12 promised to keep the embassy informed (Acheson 1950b, November 22). In early December 1950, the US ambassador to Ethiopia again wrote to Acheson, this time reporting on significant political changes that had occurred in his host country. Eritrea had been officially restored to Ethiopia thanks to substantial US involvement. The ambassador thus informed Acheson that a wave of gratitude toward the United States swept Ethiopia. The ambassador mentioned that the emperor's private secretary had told him "it is all due to you [i.e., the United States]," while the emperor himself had requested "with obvious emotion that [the US ambassador] convey his deepest thanks to [the US] government for all its assistance." Also, the acting minister of foreign affairs and the minister of commerce "had expressed their warmest appreciation" and made it clear that in their opinion credit for the solution was due mainly to the United States (Merrell 1950b, December 6). In the end, Ethiopia agreed to send approximately 1,271 troops to Korea.
At the request of the United States, South Africa contributed approximately 826 troops to the Korean War coalition. Most of its military personnel was attached to an air squadron that fought alongside the air forces of other coalition participants. Similarly to Ethiopia, South Africa also bargained hard once US officials had made their request for troop contributions official. Via bilateral channels, the United States had learned that South Africa was interested in gaining US assistance with its atomic energy program (Noer 1985, 28). As a result, initial negotiations focused on finding an agreement in this area of cooperation. Nevertheless, South Africa soon asked for more. In October 1950, South Africa's defense minister, F. C. Erasmus, met with Dean Acheson and George Marshall to demand an increase in military aid from the United States. Erasmus argued that it had been a great burden for a "small state such as South Africa" to send troops to Korea (Memorandum of Conversation of the Secretary of State and the South African Defense Minister 1950, October 5). Moreover, during the course of 1951-1952, South Africa repeatedly asked the United States to replace South Africa's old F-51 Mustang fighter aircraft with new jet aircraft. The United States initially rebuffed South Africa, arguing that the South African request would be considered only once the US Air Force had converted all its own F-51 units in the Far East to jet aircraft, which was not programed prior to early 1953. South African officials were extremely discontent with this response. They thus tried to use the Korean War to put pressure on the Americans to reverse their position. In February 1952, the South African ambassador to the United States, G. P. Jooste, wrote to Dean Acheson announcing that South Africa would ground its air squadron in Korea for three months. The reason for this action was the American failure to heed the South African request for jet aircraft (Embassy of the Union of South Africa 1952, February 11). The ambassador pointed to recruitment problems in the case that no jet aircraft would be delivered. He wrote: "All regular fighter pilots in the South African Air Force have completed their respective tour of duty in Korea and that in consequence reliance will henceforth have to be placed on volunteer reserve pilots, active citizen force pilots and pupil pilots in order to maintain a regular flow of reinforcements to Korea." Nevertheless, the latter group of pilots required further training and thus could not deploy immediately. The only way out of this dilemma would be to provide South Africa with jet aircraft, which would "provide valuable operational training for a nucleus of South African pilots" and thus motivate them to volunteer for additional tours of duty in Korea (Embassy of the Union of South Africa 1952, February 11). A follow-up statement from the South African prime minister personally mentioned that "We do not want to act in a hurry but we must solve the recruitment problem. If we could get definite assurance of receiving jets within six months, we could reconsider the decision" (Hickerson 1952, February 18 Department requests the Department of Defense again to review the problem of equipping the South African squadron with jet aircraft, with the view to ascertaining the earliest date when conversion could be completed." The Pentagon's response was a promise to review the South African request in three months, which led to the South African squadron being kept in operation in Korea (Memorandum of Conversation "Maintenance of South African Air Squadron in Korea" 1952, March 31). In addition to the aircraft dispute, South Africa also tried to use its Korean War deployment to mellow the US public position on its repugnant apartheid regime. Prior to the Korean War, the US government had been openly very critical of the problems of racial discrimination in South Africa. But 1952 represented a turning point (Strong 1952, August 26). In October 1952, the South African ambassador to the United States, G. P. Jooste, met with Dean Acheson to request that the United States use its influence to avoid an "acrimonious [UN] debate" on apartheid as well as to defeat any UN resolution condemning South Africa. Acheson largely agreed, arguing that "in the present case there is very little that the UN can and should do vis-a-vis the policies of South Africa in terms of actual results" (Memorandum of Conversation with G. P. Jooste 1952, October 14)-thus effectively agreeing to protect the South African regime from international condemnation, particularly in the United Nations.
Most other coalition contributions to the Korean War were also only provided following explicit US requests. This includes the cases of the UK, Australia, Canada, France, Belgium, the Netherlands, and Luxembourg. In each case, the United States explicitly demanded their force contribution. For example, on August 11, Acheson sent a telegram to the US embassies in London, Paris, Ottawa, Brussels, and the Hague requesting troop contributions (Acheson 1950a, August 11). In each case, US diplomacy was critical in building the coalition.

Assessment of Case Study
The US-led search process for coalition contributions was far from random. Rather, US government officials mostly based in the State Department established very specific lists of countries to be approached for coalition contributions. US officials selected countries based on what the United States "knew about them." Military concerns played into the latter assessment but to a lesser degree, especially once the State Department had taken over full control of the recruitment campaign. For the most part, "political considerations" dominated the matter. It appears that US officials gathered most of the information on these political considerations via diplomatic embeddedness. In other words, via diplomatic relationships, US officials were able to access private information on deployment preferences, which was then used to construct arguments and side-payments to persuade countries to join the coalition. In the case of Turkey, for example, US diplomats were able to track via bilateral channels the financial needs of Turkey as well as its desire to become a member of NATO. Moreover, diplomatic embeddedness clearly helped in the construction of side-payments and issue-linkages, mostly by providing channels through which these deals could be transferred (for example the United Nations and Export-Import Bank). Finally, the coalition negotiations with Ethiopia and South Africa illustrate how weaker and "subordinate" states can exploit asymmetries in preference intensities and thus lower the bargaining power of the hegemon. The South African case is particularly instructive in this regard: it was fear of a chain reaction of troop withdrawals triggered by South Africa that led the US government to review the jet aircraft request despite the initial staunch US opposition to the deal. South Africa was able to exploit that anxiety and thus drive a very hard bargain.

Conclusion
Multilateral coalition-building has become the political standard of US intervention doctrine. Our understanding of the actual coalition-building process, however, remains limited. The existing literature has thus far failed to examine whether the political underpinnings of US coalitionbuilding efforts follow a systematic pattern. This article tried to remedy this shortcoming. It suggested that coalition-building is hard diplomatic work: coalition participants rarely show up "voluntarily." Instead, the US government needs to recruit them, which requires persuasion and often side-payments or issue-linkages. I argued that diplomatic embeddedness plays an important role in the latter process. The statistical evidence shows that, indeed, diplomatic embeddedness highly and positively correlates with coalition participation across five different statistical models. Furthermore, the case study of the Korean War illustrates some of my posited causal mechanisms. Most importantly, American officials use diplomatic embeddedness to collect information on other countries' deployment preferences. In addition, diplomatic embeddedness helps construct side-payments and link issues. Moreover, the case study also sheds some light on the interactions between, on the one hand, American power and diplomacy and, on the other hand, the potential for weaker states to tilt-at least to some degree-coalition negotiations in their favor. Indeed, coalition negotiations over Korea were affected by how much each government desired to reach a deal. This put the United States at times at a relative disadvantage.
This article carries with it a number of important implications. First, US-led multilateral military coalitions do not form automatically. Instead, they need to be constructed. In such scenarios, diplomatic embeddedness proves handy (cf. Nexon 2009, 52). 29 Diplomatic embeddedness eases coalition negotiations by providing information on deployment preferences and by channeling issue-linkages and side-payments through existing bilateral and multilateral ties.
Second, while alliances create very tight ties among military and defense officials, these ties often provide Washington with insufficient information about the broad array of issues that get linked to coalition negotiations. As a result, alliance ties by themselves often fail to explain the bulk of contributions to US-led military interventions. Instead, the entirety of bilateral and multilateral ties that connect the United States with a third party positively affects coalition participation. Does this mean that alliances are irrelevant? No. Not at all. But it is the sum of all diplomatic ties that helps generate coalition contributions. Thus, being "networked" matters in world politics. States can derive resources from being bilaterally and multilaterally connected. These benefits go way past the specific purposes the ties were initially created for. In particular, these ties provide access to private information, trust, and facilitate the negotiation of side-payments and issuelinkages.
Third, without a doubt, US coalition negotiations occur in the context of American hegemony. Nevertheless, at least in the Korean case, several troop contributions did not materialize out of "submissiveness"-or what Lake (2009, 139) terms "symbolic obeisance"-to the hegemon. Instead, the US government had to bargain third parties into US-led coalitions. Moreover, given that the United States "needed" potential coalition participants more than they cared about joining the coalition, "subordinate" states possessed substantive bargaining power. Aware of US fears to fight alone in Korea, Ethiopia and South Africa, for instance, bargained hard for generous sidepayments in exchange for their troop contributions. 30 The latter findings of this article thus compel us to move beyond simple carrot-and-stick models associated with some hegemonic-order theories. They also contribute to other ongoing efforts to rethink how hegemony really works. 31

Supplementary Information
Supplementary Information is available at www.marinahenke.com and the International Studies Quarterly data archive.