Why do some allies get dragged into military operations led by the United States, while others

Transcription

1 Destined to Join US-led Missions: Asia-Pacific s Regional Perspectives on International Security Yoon Jin Lee ISA Asia-Pacific Conference 2016 Why do some allies get dragged into military operations led by the United States, while others are able to avoid such costly enterprises? The U.S. led operations in Iraq in the period clearly demonstrate that variation in the joining behavior exists across allied partners in Europe and the Americas. For instance, among NATO member countries, Denmark, Poland, Spain, and United Kingdom committed troops to the U.S. led coalition from the beginning months of the operations, while Belgium, Canada, France, and Luxembourg stayed out of the mission throughout the entire period. 1 Similarly, among Rio Treaty allies, Dominican Republic, El Salvador, Honduras, and Nicaragua deployed troops to the U.S. led mission in Iraq, whereas Argentina, Chile, Peru, and Uruguay did not. 2 An interesting puzzle observed in Asia-Pacific is that there is no similar variation in the joining behavior across U.S. allies in the region. Australia, Japan, South Korea, and the Philippines all participated in the U.S. led operations in Iraq in the period. One can argue that the sample size of the U.S. allies in Asia-Pacific is much too small to show any variation in joining behavior. However, such argument still does not illuminate the causal factors behind entanglement in military conflicts initiated by an ally. As it is highly likely that future 1 See Appendix - Force Contributions by Nations in Stephen A. Carney, Allied Participation in Operation Iraqi Freedom (Washington, D.C.: Department of the Army, 2012) for the number of ground troops deployed, deployment date, unit designation, primary deployment location, and casualties by participant country. 2 See Appendix - Force Contributions by Nations in Carney (fn. 1). 1

2 U.S. administrations will continue to operate in ad hoc coalitions under current unipolarity, 3 it is useful to know whether all U.S. allies in Asia-Pacific will follow the United States lead in the future, and why. Three factors highlighted in conflict joining literature may explain why some allied partners join military operations led by an ally, while others stay out. Threat perception, material gain, and historical learning may affect a state s joining behavior. However, in this paper, I argue that a state s level of dependence on the alliance with the United States is key to joining U.S. led missions. Specifically, I expect that states with higher levels of dependence on the alliance with the U.S. are more likely to become entangled in military conflicts initiated by the United States. I employ the linear probability model to run a statistical analysis on a sample of 50 U.S. allies for the case of the U.S. led operations in Iraq in the period. The results of this medium-n analysis show that levels of alliance dependence effectively account for the change in the probability of a state s participation in the U.S. led mission in Iraq. The rest of this paper is organized as follows. First, I discuss three different factors highlighted in conflict joining literature and explain what these variables cannot explain. Second, I present my argument based on a state s level of dependence on the alliance with the United States. Third, I describe research design and show operationalization of data. Then, I conduct statistical analyses and present findings as well as robustness test results. Finally, I conclude and discuss the implications of the results. 3 Walt, Stephen M. Alliances in a Unipolar World. World Politics 61, no.1 (2009):

3 Threat Perception, Material Gain, and Historical Learning Threat Perception As much of the alliance formation and duration literature suggests (Walt 1987, 1997; Morrow 1993; Leeds and Morgan 2012), one line of argument in the conflict joining literature maintains that a state s threat perception of the target of the use of force is the driving force behind its joining behavior. Bennett et al. (1994) ask why some states contributed to the U.S. led Operation Desert Storm in 1991, while other states did not. Bennett et al. (1994) argue that those states facing the highest threat from the target, in which case this was Saddam Hussein s Iraq, are not only more likely to join but also more likely to contribute in proportion to national product. 4 In the Persian Gulf case, distance from Iraq, perceived intentions of Saddam Hussein, and dependence on the Gulf oil affected the state leaders threat perceptions. Indeed, Egypt, Syria, and the Gulf States all provided substantial military support to the Operation Desert Storm. Similarly, when examining why Britain, France, and Italy provided military support for the U.S. led missions in some cases but refused to contribute in other cases, Davidson (2011) demonstrates that the variability in the state leaders threat perceptions can explain the variation. Davidson (2011) reasons that states are more willing to incur the costs of providing military support to the U.S. led mission when states perceive high threat from the target than when states perceive low target threat. Likewise, while explaining the variation in the joining behavior of Central and Eastern European countries in military operations in Afghanistan, Iraq, and Libya, Marton and Eichler (2013) argue that the political construction of threat matters. Marton and 4 Bennett, Andrew, Joseph Lepgold, and Danny Unger. Burden-sharing in the Persian Gulf War. International Organization 48, no.1 (1994): 43. 3

4 Eichler (2013) show that when the governing elites engaged in threat construction, as was the case for Afghanistan and Iraq, Central and Eastern European countries were more ready to provide support for these missions. In contrast, for the Libya intervention, the case was casted as a just cause than a threat-balancing one, which dwindled the military support of the Central and Eastern European countries for the operation. Overall, the variability in the level of threat perceived from the target of the use of force may explain the variation in the state decisions to join the U.S. led mission. Material Gain Another line of reasoning focuses on the causal relationship between material gains and joining behavior of states. As works on economic interests and war initiation suggest (Russett 1990; Vasquez 1993; Hensel 1996), studies on economic benefits and joining behavior of states emphasize on the material gains as an important driving force behind the state decisions to participate in the U.S. led mission. When examining the role of American economic linkage in shaping the U.S. led coalition for the Iraq War in 2003, Newnham (2008) finds that material linkages, ranging from foreign and military aid to access to U.S. markets, successfully led some countries to join the U.S. led coalition. As well, Wivel and Oest (2010) show that even though the Pacific microstates faced low threat from Saddam Hussein s Iraq, the security and economic gains provided by the United States were a critical factor behind the microstates decisions to give diplomatic support for the U.S. led mission in Indeed, Wivel and Oest (2010) find that all of the Pacific microstates receiving bilateral U.S. aid expressed diplomatic support. Similarly, Tago (2008) emphasizes on the importance of material interests in the state decisions to join the U.S. led coalition but argues that negative linkages are far more effective than positive ones. Tago (2008) points out that rewarding foreign aid for participation in a 4

5 coalition is not a general phenomenon that can be applied to middle-high income countries. Rather, Tago (2008) reasons that when a state expected to participate in the U.S. led coalition fails to do so, the United States could impose punishment by reducing the amount of military aid. By examining 15 U.S. led coalition forces in the post-wwii period, Tago (2008) finds that while the United States punished states for unexpected nonparticipation, U.S. did not always reward states for their support. Whether it is punishment or reward in the form of economic or military aid, the variability in the material interests of states could explain the variation in the state decisions to participate in the U.S. led operation. Historical Learning Another explanation sheds light on the ability of state leaders to make the joining decisions based on learning from the past. For instance, Long (2015) stresses on the initiator state s past military performance as a crucial factor in the state decisions to participate in the current war. Long (2015) argues that when the initiator state s past military record is strong, states are more likely to join the initiator state in the ongoing war, because military effectiveness signals that the cost of joining will be lower than when the initiator state is less effective. Indeed, Long (2015) finds that the initiator state s past military victories are associated with a higher likelihood that its allies will join the initiator state in the ongoing war. However, a critical shortcoming of this explanation is that the initiator state s past military record is a fixed constant, which fails to account for the variation in the joining behavior of its allies. Reiter (1994, 1996) focuses on a more flexible variable based on learning from the past. Reiter (1994, 1996) contends that when state leaders decide to join alliances, they tend to ignore the level of threat in the international environment and instead focus on avoiding past mistakes while continuing to repeat past successes. Reiter (1994, 1996) tests such learning effect on 5

6 the alliance choices of small states in the 20 th century and finds that state experiences in the previous war are a powerful indicator of alliance choices after the war. As well, Reiter (1994, 1996) shows that the level of threat had only marginal effects on alliance choices. Accordingly, the variability in the lessons drawn from the states past experiences in the U.S. led operations may explain the variation in the state decisions to join the current U.S. led coalition. The three factors highlighted in conflict joining literature certainly empower the agency of state leaders. If decision-makers perceive high levels of threat from the target of the use of force, then they will decide to participate in the U.S. led mission. Yet, if they perceive low target threat, then they are not likely to commit troops to the coalition. As well, if there are material benefits to reap from joining the U.S. led mission, then state leaders are more likely to join the coalition than when there are no gains to collect. Likewise, if state leaders deem that their past participation in the U.S. led mission was a success, then they are more likely to deploy troops to the current operation. On the other hand, if they evaluate their past participation in the U.S. led mission to be a mistake, then they are less likely to join the current coalition. However, there are cases where state leaders decide to join the U.S. led mission, even if their own preferences would dictate otherwise. As well, there may be cases where state leaders have to stop short from committing troops to the U.S. led operation, even if they prefer to do so. The three factors cannot account for such cases where state leaders make decisions against their own preferences, because these variables concentrate on the agency of decision-makers while neglecting the structural constraints within the decision-makers environments. 6

7 Degrees of Dependence on the Alliance with the United States When introducing the opportunity and willingness framework of conflict diffusion, Siverson and Starr (1990) stress that both components of state decisions need to be examined. While opportunity represents the environmental and structural factors, willingness represents the choice processes that occur on the micro level 5. When state leaders make decisions to join the U.S. led mission, they are embedded in an environment where there is already some established form of relationship with the United States. Out of many areas, states alliance relations with the U.S. can function as the chief environmental constraint in the decision-making process. Specifically, if a state s degree of dependence on the alliance with the U.S. is high, then state leaders are more likely to join the U.S. led mission. On the other hand, if a state s level of alliance dependence is insignificantly low, then state leaders are less likely to incur the costs of committing troops to the U.S. led operation. When a state s level of dependence on the alliance with the U.S. is high, decision-makers may still choose to participate in the U.S. led mission even in the absence of high target threat, material gains, and learning from the past. As well, when a state s level of dependence on the alliance with the U.S. is insignificantly low, state leaders may choose not to participate in the U.S. led operation even in the presence of high target threat, material gains, and historical learning. Thus, the level of alliance dependence can explain such cases where state leaders are compelled to make decisions against their own preferences. Indeed, much of the literature on alliance security dilemma suggests that states with higher levels of alliance dependence are more prone to entanglement in military conflicts initiated by their ally, because the costs and risks of 5 Siverson, Randolph M. and Harvey Starr. Opportunity, willingness, and the diffusion of war. The American Political Science Review 84, no.1 (1990): 48. 7

8 abandonment outweigh the costs and risks of entrapment. 6 Accordingly, the following expectation holds. Hypothesis: States with higher levels of dependence on the alliance with the United States are more likely to join U.S. led missions. To measure a state s level of alliance dependence, I argue that one must macroscopically examine a state s official security arrangements. Official security arrangements can take any form of inter-governmental organization or agreement, as long as there is a security guarantee for all participating members. Official security arrangements include military alliances, but they do not always have to be in the form of alliances. For example, European Union (EU) and Union of South American Nations (UNASUR) are considered as official security arrangements, because these institutions provide a common defense and security policy for all member states 7. I identify three aspects of a state s official security arrangements that may influence its level of dependence on alliance with the United States. The first aspect concerns the number of partners in the alliance with the United States. Specifically, it considers whether the U.S. is the only, exclusive alliance partner or not. For instance, in US-Japan bilateral alliance, the United States is the only alliance partner of Japan. Japan would solely depend on the alliance with the United States for assistance in times of crises. However, in multilateral alliances, such as NATO, U.S. is not the only alliance partner for the rest of NATO member countries. In such multilateral 6 Snyder, Glenn H. The Security Dilemma in Alliance Politics. World Politics 36, no.4 (1984): ; Walt, Stephen M. Origins of Alliances. Ithaca: Cornell University Press, 1987; Cha, Victor D. Abandonment, entrapment, and neoclassical realism in Asia: The United States, Japan, and Korea. International Studies Quarterly 44, no.2 (2000): Posen, Barry R. European Union security and defense policy: Response to unipolarity? Security Studies 15, no.2 (2006): ; and Riggirozzi, Pia. Region, Regionness and Regionalism in Latin America: Towards a New Synthesis. New Political Economy 17, no.4 (2012):

9 settings, the alliance with the United States is not as central and weaker allies may rely on other NATO partners. Hence, states with bilateral alliances would have higher levels of dependence on the alliance with the United States than those with multilateral alliances. The second aspect considers the availability of an alternative security arrangement besides the alliance with the United States. It is important that the alternative security arrangement does not include U.S. as its member. Otherwise, such arrangement would only increase a state s level of dependence on the alliance with the U.S. and not decrease it. If a state lacks such an alternative security arrangement, then it would solely depend on the alliance with the United States. On the other hand, if a state has alternative security arrangements, then the alliance with the United States is not as central to its security and it may rely on these other arrangements for assistance during crises. Thus, allies without any alternative security arrangements would have higher degrees of dependence on the alliance with the United States than those with alternative arrangements in place. Finally, the third aspect involves the relative age of the official security arrangements. Here, both the state s years of alliance with the United States and its years of participation in an alternative security arrangement matter. Three cases can be considered. First, both the years of the alliance with the U.S. and the years of participation in alternative security arrangements are equal in length. In this case, relative age is not too informative about the state s level of dependence on the alliance with the United States. Second, the age of the alliance with the U.S. is older than the age of the state s participation in alternative security arrangements. Third case is where the age of the state s participation in alternative security arrangements is older than the age of the alliance with the United States. Given the sticky influences of organizational and 9

10 bureaucratic structures, 8 a state s level of dependence on the alliance with the U.S. would be higher in the second case than in the third case. Table 1. Levels of Dependence on the Alliance with the United States US: Exclusive Partner Alternative Arrangement No Yes (1) High Dependence Yes (3) Low-Medium Dependence No (2) Medium Dependence (4) Low Dependence Table 1 summarizes the relationship between the first two aspects of a state s official security arrangements and its degree of dependence on the alliance with the United States. States in Box (1) are in a bilateral alliance with the United States and have no other alternative security arrangements. In practice, such set of U.S. allies includes Australia, Japan, South Korea, and the Philippines, as they have bilateral alliances with the United States but do not have any other alternative security arrangements. For these states, U.S. is the only alliance partner, and without the United States, no other state is obliged to come to their aid at times of crises. Therefore, the degree of dependence on the alliance with the United States would be the highest. States in Box (2) are in a multilateral alliance with the United States and have no other alternative security arrangements. Examples of such set of countries include NATO and Rio 8 Keohane, Robert O. After Hegemony: Cooperation and Discord in the World Political Economy. Princeton, New Jersey: Princeton University Press, 1984; and Leeds, Alliance Reliability in Times of War: Explaining State Decisions to Violate Treaties, p

11 Treaty member countries that do not have enjoy alternative security arrangements. These states are non-eu, NATO member countries as well as non-unasur, Rio Treaty member countries. Although these states do not have alternative security arrangements to rely on, they do have other alliance partners besides the United States to depend on at times of crises. This is why the level of dependence on the alliance with the United States is not as high as in Box (1). States in Box (3) are in a bilateral alliance with the United States but also have alternative security arrangements that exclude the United States. In practice, however, there is no state that is in a bilateral alliance relationship with the U.S. but has other security arrangements in place. Potentially, the Philippines could belong in Box (3), as it is in a bilateral alliance with the U.S. and is concurrently a member of ASEAN. However, because ASEAN does not have a common defense and security mechanism for its member states, the Philippines would not expect or receive any security assistance from other member states. Nonetheless, theoretically, for this set of countries, the degree of dependence on the alliance with the United States would be low to medium. States in Box (4) are in a multilateral alliance with the United States but also enjoy alternative security arrangements that exclude the United States. Amongst NATO and Rio Treaty member states, such set of countries includes concurrent NATO and EU member states as well as concurrent Rio Treaty and UNASUR member states. These states not only have other alliance partners within the multilateral alliance to depend on but also have entirely separate security arrangements to rely on for assistance at times of crisis or war. Because their security relationship with the U.S. is not exclusive in both aspects, the degree of dependence on the alliance with the United States would be the lowest. 11

12 Finally, the third aspect, which concerns the relative age of the official security arrangements, is considered. For states in Boxes (1) and (2), the age of the alliance with the United States is the sole concern and the age of other security arrangement is irrelevant. However, for the countries in Boxes (3) and (4), both the years of the alliance with the U.S. and the years of participation in alternative security arrangements matter. Of the states in Box (4), those with longer relative years of participation in alternative security arrangements would have lower levels of dependence on the alliance with the U.S. than those with shorter relative years of participation in alternative security arrangements. The same logic applies for the theoretical set of weaker allies in Box (3). Research Design To test if states with higher levels of dependence on the alliance with the United States are more prone to entanglement in U.S. led operations, I analyze the joining behavior of 50 U.S. allies in the case of the 2003 Iraq war. 9 The case of the 2003 Iraq war is the most appropriate instance possible in a nonexperimental, single-case setting, where one can examine which of the U.S. allies are susceptible to entanglement in U.S. led missions and why. In this medium-n analysis, the variation of interest is not in the types, or amounts, of contributions made to the U.S. led coalition. Rather, the variation of interest is binary. I examine whether an ally committed troops to the coalition or not. If an ally deployed troops to the U.S. led mission, then the state is considered to have joined the coalition. On the other hand, if an ally did not dispatch troops to the U.S. led mission, then it is considered not to have joined the

13 coalition. It does not matter whether or not the deployed troops are combatant or non-combatant in nature. Even if a state sends non-combatant troops, composed of only a handful of construction engineers, it is the country s national flag, which these two to three military engineers bear, that counts. In the same vein, verbal diplomatic support, financial contribution, or allowing of the U.S. led forces to use air and naval bases do not count as joining the U.S. led mission, because these actions do not add a national flag to the poles on the ground. To measure my explanatory variable, I systematically assign weights to the three aspects of a state s official security arrangements to measure its degree of dependence on the alliance with the United States. The variables employed in this medium-n analysis are operationalized as follows. Data DEPENDENT VARIABLE Decision to join the U.S. led mission The binary dependent variable takes on two values of either 1 or 0. If an ally committed troops to the U.S. led coalition in , then it takes a value of 1. If an ally did not commit troops to the U.S. led coalition in , then it takes a value of 0. I use Baum s (2013) troops per capita variable to code 1 for countries with nonzero troop contributions and 0 for countries with zero troop contribution. 10 INDEPENDENT VARIABLE Levels of alliance dependence as measured by variation in security arrangements I assign weights to the three aspects of a state s official security arrangements. To begin with the first aspect, if a state is in a bilateral alliance with the United States, and the U.S. is its only 10 Baum, Matthew A. The Iraq Coalition of the Willing and (Politically) Able: How Party Systems, the Press and Public Influence on Foreign Policy. American Journal of Political Science 57, no.2 (2013): Replication dataset: 13

14 exclusive partner, I assign a weight of 1/2, or 0.50, for the alliance. If a state is in a multilateral alliance with the United States, and the U.S. is not the state s only partner, then I assign a weight of 1/3, or 0.33, for the alliance. The weight values of 0.50 for bilateral alliances and 0.33 for multilateral alliances do not necessarily convert to a numerical reality of the levels of alliance dependence; rather, these weights simply capture the different degrees of dependence on the alliance with the United States. States with bilateral alliances would have higher degrees of dependence on the alliance with the United States than those with multilateral alliances. Next, I assign weights for alternative security arrangements. If a state s alternative security arrangement is bilateral in form, then I assign a weight of 1/3, or 0.33, for the bilateral arrangement. If a state s alternative security arrangement is multilateral in form, then I assign a weight of 1/4, or 0.25, for the multilateral arrangement. Again, these weight values do not necessarily correspond to a numerical reality of the levels of dependence on alternative arrangements. Yet, the weight assigned for bilateral alternative arrangements is larger than the weight assigned for multilateral alternative arrangements, because bilateral arrangements induce stronger commitment and dependence. Here, I note why the weight values of 0.33 for bilateral alternative arrangements and 0.25 for multilateral alternative arrangements are smaller than the weight values assigned for the bilateral and multilateral alliances with the United States. Because the U.S. is the strongest power in the current unipolar structure, the states levels of dependence on alternative security arrangements that exclude the U.S. are expected to be lower than their levels of dependence on the alliance with the United States. Hence, even if a state has an alternative security arrangement to rely on, its level of dependence on the other security arrangement would still be lower than its level of dependence on the alliance with the United States. Therefore, the weight values of

15 for bilateral alternative arrangements and 0.25 for multilateral alternative arrangements are smaller than those assigned for the bilateral and multilateral alliances with the United States. Finally, I assign weights for the duration of security arrangements. I do not use the raw years of security relationships, because states with 45 years of multilateral alliance with the U.S. may not necessarily be more dependent on the alliance with the U.S. than states with 44 years of the same alliance relationship. However, states with 45 years of multilateral alliance with the U.S. may be more dependent on the alliance with the U.S. than states with 35 years, 15 years, and 5 years of the same alliance relationship with the United States. Thus, I specify one unit of time as ten years. I count the years from a state s accession year to 2003, which is the year when the U.S. led operations in Iraq began, and then divide the years by ten. I round up the number of years to the next decade. For instance, if a state was in an alliance with the United States for 41 years, I round it up to 50 and assign 5 units of time. For certain countries that were in the process of joining a security arrangement in 2003 (e.g. Central and Eastern European countries joining NATO and EU, South American countries forming UNASUR, etc.), I still assign 1 unit of time, because all of these states were clearly aware of their security arrangements and options in Indeed, the Central and Eastern European countries completed the accession processes for both NATO And EU by the year For certain countries that have temporarily withdrew its membership from a currently existing security arrangement (e.g. France from NATO), I just count the years in which these countries have belonged to the currently existing security arrangements. I exclude security arrangements that have already terminated in the past (e.g. ANZUS, SEATO, etc.). 15

16 The weights assigned to the three aspects of a state s official security arrangements is combined in the following way to compose a state s level of dependence on the alliance with the United States: Level of alliance dependence = [Weight of dependence on alliance with US * Time] - [Weight of dependence on alternative security arrangement * Time] Alternative Explanatory Variables Victims of 9/11 States may be more willing to incur the costs of providing military support to the U.S. led operation when they perceive high levels of threat from the target of the use of force (Bennett et al. 1994; Davidson 2011; Marton and Eichler 2013). I measure the states levels of threat perceptions on Saddam Hussein s Iraq with the number of victims from the 9/11 tragedy. After the 9/11 terrorist attacks, President George W. Bush framed prospective U.S. military operations in the Middle East as a war on terror and encouraged collective support from the free world 11. Therefore, states levels of threat perceptions may be in part affected by their losses in the 9/11 tragedy. Data is from the Memorial Mapping Project 12. Economic aid Material linkages, ranging from foreign and military aid to access to U.S. markets, may influence states to join the U.S. led coalition (Newnham 2008; Tago 2008; Wivel and Oest 2010). I use U.S. economic disbursements to recipient states to measure the amounts of U.S. foreign aid received by countries in Data is from the USAID Country Summary dataset

17 Military aid Similarly, I use U.S. military disbursements to recipient states to measure the amounts of U.S. military aid received by countries in Data is from the USAID Country Summary dataset 14. Gulf War When state leaders make decisions to join the U.S. led coalition, they may focus on avoiding past mistakes while repeating past successes (Reiter 1994, 1996). Given the similarities in the context of the war, the states past participation in the U.S. led Vietnam War would be an ideal explanatory variable. However, as those countries that participated in the Vietnam War (i.e. Australia, New Zealand, South Korea, Philippines, and Thailand) have also joined the U.S. led coalition of the willing in , the Vietnam War variable predicts success perfectly and STATA automatically dropped the variable. This is why I use the U.S. led Gulf War in 1991 to test if state decisions reflect learning from the past. Although UN Security Council authorized the use of force in the 1991 Gulf War, the U.S. led Operation Desert Storm was the biggest commitment of U.S. military power since Vietnam 15. As well, the U.S. led operation was swift and efficient. In a matter of six weeks since the intervention, Saddam Hussein ordered retreat from Kuwait, and in terms of casualties, the cost of the conflict was kept low 16. As Long (2015) argues, such past military effectiveness of the U.S. may influence the state decisions to join the U.S. led operation in , so I use the Gulf War variable. If states have participated in the Operation Desert Storm in 1991, then they take on a value of 1, while non-participant states take on a value of Sobel, Richard. The Impact of Public Opinion on U.S. Foreign Policy since Vietnam. New York: Oxford University Press, p Sobel, Richard. The Impact of Public Opinion on U.S. Foreign Policy since Vietnam. New York: Oxford University Press, p

18 Control Variables Miles to Baghdad As geographic proximity to the war zone may lead states to join the conflict, I control for the distance between the states capitals and Baghdad (Starr and Most 1983; Siverson and Starr 1990; Buhaug and Gleditsch 2008). The miles are calculated by Google map. Inflation rate As rationalist explanations of war highlight, states capabilities and expected costs of war may influence state decisions to engage in military conflicts (Fearon 1995; Morrow 1999). High levels of inflation rate in 2003 may raise the costs of war. Therefore, countries with high levels of inflation rate in 2003 may be less likely to join the U.S. led operation in I use Baum s (2013) 2003 inflation rate variable. GDP per capita Countries with high levels of GDP per capita may have more capabilities to participate in the U.S. led mission. Nonetheless, countries with high levels of GDP per capita may be less pressured by the U.S. to join the mission than countries with low levels of GDP per capita. I use Baum s (2013) 2003 GDP per capita variable. GDP change per capita Countries with high levels of GDP change per capita may have more capabilities to participate in the U.S. led mission. Again, however, countries with high levels of GDP change per capita may also be less pressured by the U.S. than countries with low levels of GDP change per capita. I use Baum s (2013) annual change in GDP per capita variable. 18

19 Muslim population States with higher percentages with Muslim population may be less likely to join the U.S. led mission and send troops to Iraq. I control for states Muslim populations by using the Muslim population by country in 1990 data from the Pew Research Center 17. Empirical Evidence As my dependent variable is not a continuous variable, a linear regression model is not appropriate in this case. In fact, because the probabilities that I am modeling are binary values of 1 and 0, a logistic regression model seems appropriate. The logistic regression equation is as follows. Decision to Join = β0 + β1 Level of Alliance Dependence + β2 Victims of 9/11 + β3 Economic aid + β4 Military aid + β5 Gulf War + Control Variables Table 2 presents the results of the logistic regression. The coefficient in front of the independent variable, Level of alliance dependence, is statistically significant at the 5% level. If the level of a state s dependence on the alliance with the United States, as measured by the three aspects of the state s security arrangements, increases by one unit, the log odds of the state s participation in the U.S. led mission in Iraq increases by As well, the positive sign in front of the statistically significant coefficient supports the expectation that allies with higher degrees of dependence on the alliance with the U.S. are more likely to join the U.S. led misison. On the other hand, the coefficients in front of the alternative explanatory variables are not statistically significant. The coefficients of the 9/11 victims, Economic aid, and Military aid

20 variables do not show any statistical significance. Only the coefficient in front of the Gulf War variable is statistically significant at the 10% level. Such finding may indicate that states do take the past military effectiveness of the U.S. into consideration, since the Operation Desert Storm in 1991 is evaluated as a successful U.S. led mission. However, there is a caveat in that the Gulf War variable may be capturing the same set of countries that join the U.S. led missions under all circumstances. For instance, countries that committed troops to the U.S. led mission in 1991 are also more likely to join the U.S. led coalition in , not because of the effects of learning from the past but rather due to the same underlying mechanism that leads certain states to join the U.S. led missions. The coefficients in front of the control variables make sense. Specifically, the coefficients in front the Miles to Baghdad and GDP per capita variables are statistically significant at the 1% level. As expected, countries that are further away from Baghdad are less likely to join the U.S. led mission, because these countries face lower levels of risk and uncertainty from the danger posed by Saddam Hussein s Iraq. Likewise, countries with higher levels of GDP per capita are less likely to join the U.S. led coalition. The coefficient in front of Inflation variable is statistically significant at the 10% level. Countries with higher levels of inflation rate in 2003 are less likely to join the U.S. led mission due to the higher costs of war. Finally, GDP change per capita, and Muslim population variables do not show any statistical significance. Overall, the results of this medium-n analysis indicate that the levels of states alliance dependence, as measured by the variation in their official security arrangements, are an effective predictor of which states are more likely to participate in the U.S. led missions and why. Allies with higher degrees of dependence on the alliance with the United States are more likely to join the U.S. led operation than those with lesser degrees of dependence on the alliance with the 20

21 United States. In contrast, changes in a state s level of threat perception and the amounts of economic linkages do not seem to explain the change in the probability of a state s participation in the U.S. led mission in Iraq. Robustness Test Results To check if the findings are robust, I run a series of robustness tests. First, I check if my findings are model dependent. Instead of using the logit model, I re-run the regression using the probit model, where the dependent variable can also be binary. Then, I change various weights on the three aspects of states official security arrangements. Specifically, I change three values: 1) weights on the number of partners in the alliance with the United States; 2) weights on alternative security arrangements; and 3) units of time. Table 3 shows the results of the probit regression. As was the case in Table 2, the coefficient in front of the independent variable is statistically significant at the 5% level. If the level of a state s dependence on the alliance with the United States, as measured by the variation in its security arrangements, increases by one unit, the z-score increases by The other explanatory variables show no statistical significance, except for the Gulf War variable, which is statistically significant at the 10% level. Again, however, the caveat here is that the Gulf War variable may be capturing the same set of countries that join the U.S. led missions under all circumstances. Next, I change the weights assigned to the number of partners in the alliance with the United States. So far, I assigned the weight values of 1/2, or 0.50, for bilateral alliances with the U.S. and 1/3, or 0.33, for multilateral alliances with the United States. Although the two values capture the essential difference between states in a bilateral alliance with the U.S. and states in a multilateral alliance with the United States, I check to see if different weights produce different 21

22 results. Specifically, I assign new weight values of 0.60 for bilateral alliances with the U.S. and 0.40 for multilateral alliances with the United States. Again, the weight value for bilateral alliances is larger than the weight value for multilateral alliances, because states with bilateral alliances would have higher degrees of dependence on the alliance with the United States than those with multilateral alliances. I rerun the equation using the logit model. The results are summarized in column (1) of Table 4. The coefficient in front of the newly measured independent variable is statistically significant at the 5% level. If the level of a state s dependence on the alliance with the United States, as measured by its variation in security arrangements, increases by one unit, the log odds of the state s participation in the U.S. led mission in Iraq increases by I also change the weights for alternative security arrangements. Instead of the weight values of 0.33 for bilateral alternative arrangements and 0.25 for multilateral alternative arrangements, I assign new weight values of 0.50 for bilateral alliances and 0.33 for multilateral alliances. Here, I do not differentiate between the weight values for bilateral alliances with the U.S. and those assigned for bilateral alternative arrangements. Similarly, I do not differentiate between the weight values for multilateral alliances with the U.S. and those for multilateral alternative arrangements. Even though the U.S. is the unipolar power, the states levels of dependence on alternative security arrangements that exclude the U.S. may be similar to their levels of dependence on the United States. Thus, by assigning these new weights for alternative security arrangements, I double check for two points. One, I check to see if different weights produce different results. Two, I check to see if the underlying assumption that states levels of dependence on the alliance with the U.S. are systematically higher than their levels of dependence on alternative security arrangements produces biased results. 22

23 I rerun equation using the logit model. The results are summarized in column (2) of Table 4. The coefficient in front of the newly measured independent variable is statistically significant at the 5% level. If the level of a state s dependence on the alliance with the United States, as measured by its variation in security arrangements, increases by one unit, the log odds of the state s participation in the U.S. led mission in Iraq increases by Finally, I checked to see if the main findings were dependent on the measurement of the unit of time. Instead of grouping a period of 10 years as one unit of time, I change the unit of time to 15 years. Column (3) of Table 4 shows that such change in the unit of time does not make a difference. The results are consistently similar. The coefficient in front of the newly measured independent variable is still statistically significant at the 5% level. If the level of a state s dependence on the alliance with the United States, as measured by its variation in security arrangements, increases by one unit, the log odds of the state s participation in the U.S. led mission in Iraq increases by Grouping a shorter period of time, such as 5 years, as one unit of time does not change the results. The coefficients in front of other control variables in Tables 3 and 4 are similar to those in Table 2. Conclusion and Policy Implications The findings of this paper indicate that states with higher degrees of dependence on the alliance with the United States carry higher risks of entanglement in U.S. led operations. States with higher levels of alliance dependence are more likely to participate in U.S. led missions than those with lower levels of alliance dependence. A series of robustness tests demonstrate that the main findings are robust to modeling choice and weight choices. 23

24 This paper also emphasizes on how one understands and measures a state s level of alliance dependence is critical. When measuring the levels of states alliance dependence, one should consider the entire configuration of a state s official security arrangements, such as the number of partners in the alliance with the United States, the availability of alternative security arrangement, and relative duration of security arrangements. In fact, all U.S. allies in Asia-Pacific committed troops to the U.S. led mission in Iraq in the period, because these states are in a bilateral alliance with the United States and have no other alternative security arrangements. Thus, they have the highest degrees of dependence on the alliance with the United States. Unless Australia, Japan, South Korea, and the Philippines arrange a multilateral alliance with the United States or form a separate security arrangement excluding the United States, these states are likely to become entangled in future U.S. led operations. 24

25 Table 2. Logistic Regression of Levels of Alliance Dependence on State Decisions to join the U.S. led Mission in Iraq Decision to Join Level of alliance dependence 4.761** [2.063] 9/11 victims [0.0467] Economic aid -5.76e-09 Military aid [9.46e-09] 2.55e-09 [2.02e-08] Gulf war 3.185* [1.751] Miles to Baghdad *** [ ] Inflation * [0.111] GDP per capita *** [ ] GDP change per capita [7.112] Muslim population [0.198] R N 50 Note: Standard errors in brackets *** p<0.01, ** p<0.05, * p<

26 Table 3. Probit Regression of Levels of Alliance Dependence on State Decisions to join the U.S. led Mission in Iraq Decision to Join Level of alliance dependence 2.533** [1.035] 9/11 victims [0.0275] Economic aid -3.41e-09 Military aid [5.57e-09] 1.29e-09 [1.34e-08] Gulf war 1.854* [0.990] Miles to Baghdad *** [ ] Inflation * [0.0607] GDP per capita *** [5.84e-05] GDP change per capita [4.048] Muslim population [0.113] R N 50 Note: Standard errors in brackets *** p<0.01, ** p<0.05, * p<

27 Table 4. A Series of Robustness Tests (1) (2) (3) Decision to Join Decision to Join Decision to Join Level of alliance dependence 3.249** 5.696*** 5.364** [1.588] [2.191] [2.368] 9/11 victims [0.0465] [0.0477] [0.0460] Economic aid -5.94e e e-09 [9.39e-09] [1.04e-08] [9.01e-09] Military aid 3.16e e e-09 [2.31e-08] [1.75e-08] [1.78e-08] Gulf war 2.485* 4.693** 2.557* [1.511] [2.197] [1.477] Miles to Baghdad ** *** *** [ ] [ ] [ ] Inflation * [0.102] [0.135] [0.110] GDP per capita ** *** *** [9.80e-05] [ ] [8.54e-05] GDP change per capita [6.420] [8.902] [6.650] Muslim population * [0.182] [0.237] [0.188] R 2 N Note: Standard errors in brackets *** p<0.01, ** p<0.05, * p<

28 References Bennett, Andrew, Joseph Lepgold, and Danny Unger. Burden-sharing in the Persian Gulf War. International Organization 48, no.1 (1994): Buhaug, Halvard, and Kristian Skrede Gleditsch. Contagion or Confusion? Why Conflicts Cluster in Space. International Studies Quarterly 52, no.2 (2008): Carney, Stephen A. Allied Participation in Operation Iraqi Freedom. Washington, D.C.: Department of the Army, Cha, Victor D. Abandonment, entrapment, and neoclassical realism in Asia: The United States, Japan, and Korea. International Studies Quarterly 44, no.2 (2000): Davidson, Jason W. America's allies and war: Kosovo, Afghanistan, and Iraq. Palgrave Macmillan, Hensel, Paul R. Charting a Course to Conflict: Territorial Issues and Militarized Interstate Disputes, Conflict Management and Peace Science 15, no.1 (1996): Keohane, Robert O. After Hegemony: Cooperation and Discord in the World Political Economy. Princeton, New Jersey: Princeton University Press, Leeds, Brett Ashley. Alliance Reliability in Times of War: Explaining State Decisions to Violate Treaties. International Organization 57, no.4 (2003): Long, Stephen B. A winning proposition? States' military effectiveness and the reliability of their allies. International Politics 52, no.3 (2015): Marton, Peter and Jan Eichler. Between willing and reluctant entrapment: CEE countries in NATO's non-european missions. Communist and Post-Communist Studies 46, no.3 (2013): Morrow, James D. Arms versus Allies: Trade-offs in the Search for Security. International Organization 47, no.2 (1993): Newnham, Randall. Coalition of the Bribed and Bullied? US Economic Linkage and the Iraq War Coalition. International Studies Perspectives 9, no.2 (2008): Posen, Barry R. European Union security and defense policy: Response to unipolarity? Security Studies 15, no.2 (2006): Reiter, Dan. Learning, Realism, and Alliances: The Weight of the Shadow of the Past. World Politics 46, no.4 (1994):

29 Reiter, Dan. Crucible of beliefs: Learning, alliances, and world wars. Cornell University Press, Riggirozzi, Pia. Region, Regionness and Regionalism in Latin America: Towards a New Synthesis. New Political Economy 17, no.4 (2012): Russett, Bruce. Economic decline, electoral pressure, and the initiation of interstate conflict. Prisoners of war (1990): Siverson, Randolph M. and Joel King. Attributes of National Alliance Membership and War Participation, American Journal of Political Science 24, no.1 (1980): Siverson, Randolph M. and Harvey Starr. Opportunity, willingness, and the diffusion of war. The American Political Science Review 84, no.1 (1990): Snyder, Glenn H. The Security Dilemma in Alliance Politics. World Politics 36, no.4 (1984): Starr, Harvey and Benjamin A. Most. Contagion and border effects on contemporary African conflict. Comparative Political Studies 16, no.1 (1983): Tago, Atsushi. Is there an aid-for-participation deal? US economic and military aid policy to coalition forces (non)participants. International Relations of the Asia-Pacific 8, no.3 (2008): Vasquez, John A. and Marie T. Henehan. The War Puzzle. Cambridge University Press, Walt, Stephen M. Origins of Alliances. Ithaca: Cornell University Press, Walt, Stephen M. Alliances in a Unipolar World. World Politics 61, no.1 (2009): Wivel, Anders and Kajsa Ji Noe Oest. Security, profit or shadow of the past? Explaining the security strategies of microstates. Cambridge Review of International Affairs 23, no.3 (2010):