Abstract
This article asks the simple question of why East and Southeast Asian rivals have experienced a sharp increase in military activity despite their decisive shift to democracy. To explain this puzzle, I develop a “theory of perverse political accountability” in East and Southeast Asian rivalries. Central to my theory is that perverse political accountability, in which a hawkish leadership adopting a more conflictual policy toward foreign rivals is rewarded by a hard-line constituency or selectorate, makes democratic leaders in rival states highly conflict-prone. In addition, the theory claims that democratic rivals will likely initiate military conflicts against autocratic rivals due to their leaders’ hawkish preferences and to perverse political accountability, and that autocratic rivals will likely initiate conflicts against democratic rivals and against each other due to their leaders’ hawkish preferences and to the absence of perverse political accountability. The article finds strong empirical evidence for these theoretical arguments. The findings contribute to an enhanced understanding of the causes of increased militarism in East and Southeast Asian rivalries.
Keywords
Introduction
For years following the end of the Cold War, there have been a growing number of studies about the prospects for war and peace in Asia, more specifically in East and Southeast Asia. 1 The potential for competition over regional hegemony, the worsening of tension in territorial disputes, the varying mixes of autocracy and democracy, and the rise of exclusive nationalism all lead to growing scholarly attention to East Asia (Acharya, 2003; Berger, 2000, 2012; Betts, 1993; Calder, 2004; Friedberg, 1993; Fukuyama, 2005). Amongst other factors is a sharp rise in military expenditure and, consequentially, an increased chance of military conflict in East and Southeast Asia, which is of concern to scholars of international and Asian security.
The countries in East and Southeast Asia have experienced an increased militarism since the mid-1980s. Rising China, Japan’s ardent effort to check China, routinized militarized competition between China and Taiwan and between North and South Korea, maritime disputes between China, Vietnam, and Southeast Asian countries, and ongoing border disputes between Thailand and Cambodia have all prompted the countries’ efforts to fortify their military capabilities, thereby increasing the chance of military conflict in the region (Calder, 2006; Cha and Kang, 2003; Guan, 1998; Hughes, 2009; Mearsheimer, 2011; Scobell, 2011).
With regard to the causes of such increased militarism in East and Southeast Asia, the article gives primary attention to East and Southeast Asian “rivalries” because they are the main vehicle of interstate conflict. 2 Most states do not engage in military conflict with one another. Only rivals have a strong tendency to be involved in series of military conflicts. This is why most studies on Asian security devote considerable attention to pairs of rivals such as China and Taiwan, China and Japan, China and Vietnam, South and North Korea, and Cambodia and Thailand (Calder, 2006; Cha and Kang, 2003; Christensen, 1999; Guan, 1998; Hughes, 2009; Scobell, 2011).
What is more interesting, however, is that military activity between East and Southeast Asian rivals has gradually been increasing even as many of these states have become increasingly democratic. Since the mid-1980s, the so-called “third wave of democratization” has swept Asia (Huntington, 1991) and many East and Southeast Asian rivals have experienced a systematic development of democracy (Chan, 1995; Diamond, 2012; Fukuyama, 2012). According to scholars of rivalries, such democratic developments reduce the risk of military conflict and increase the chance of peace (Bennett, 1998; Hensel et al., 2000). Contrary to the expectation, however, East and Southeast Asian rivals have become more bellicose in dealing with each other as they have become more democratic. This leads to the question of why East and Southeast Asian rivals increase their military activities despite their shift to democracy.
To answer this empirical puzzle, I develop a “theory of perverse political accountability” in which the democratic leadership adopting a policy of confrontation and conflict against foreign rivals is electorally rewarded while the leadership appearing to be conciliatory and weak in dealing with rivals is electorally punished (Colaresi, 2004; Nincic, 1989). The theory claims that, with the recent democratic developments in East and Southeast Asian countries, such distorted political accountability has become a major driver of military conflict.
The article is organized as follows: the first section critically reviews prior studies of Asian international relations, emphasizing both the strengths and weaknesses of the studies. The second section develops a novel theoretical framework—i.e. a theory of perverse political accountability—and presents some testable hypotheses drawn from the theory. The third section conducts a series of empirical analyses and offers evidence commensurate with the expectations that the theory generates. The final section provides both the conclusion and possible lines for future research, followed by some policy implications.
Literature on East Asian security
Competing theories of international relations strive both to explain and to predict an emerging security environment in post-Cold War East Asia. Scholars working in the tradition of realism, for instance, predict that a full-blown arms race and major military conflict will replace the now defunct US–Soviet rivalry in a multipolar Asia (Betts, 1993/4; Friedberg, 1993/4). The scholars stress in particular that the military adventurism of China, spurred on by its increasing power and corresponding territorial disputes with Japan and other Southeast Asian countries, leads to an increased chance of major military conflict in East Asia, as does Japan’s bid to become a normal state and North Korea’s nuclear gamble (Calder, 2006; Cha and Kang, 2003; Christensen, 1999; Matthews, 2003; Mearsheimer, 2010).
Drawing on the Kantian tripod, liberal scholars argue that a growing economic interdependence, the spread of democracy, and thriving regional regimes such as the Association of Southeast Asian Nation (ASEAN), the ASEAN Regional Forum (ARF), the Asia-Pacific Economic Cooperation (APEC), and the East Asian Summit (EAS) propel states in East and Southeast Asia to enter a zone of peace (Acharya, 2003/4; Duffield, 2003; Goldsmith, 2007; Xinbo, 2008). Rejecting realist pessimist prediction of East Asian security, these studies assert that East Asia is following the same path that the West has since the Second World War, i.e. that of durable peace and stability.
Constructivists seize a middle ground by asserting that, as long as identity conflict persists between heterogeneous political regimes in East and Southeast Asia, the prospect for stability and peace is not bright. By focusing on the potentially volatile nature of competing identities between the US and China, for example, Nau (2003) asserts that the two hegemons will magnify the power rivalry and the chances of military conflict. Some also argue that the clash between increasingly divergent constructions of national identity in Taiwan and Mainland China, and in Japan and China, as well as the drive for national unity by North Korea, might serve as tipping points for future conflict in East and Southeast Asia (Berger, 2000, 2012; Moon and Suh, 2007).
The security landscape in East Asia, which has evolved since the dissolution of the Soviet Union, however, does not support the predictions that these schools of thought have produced. Realists’ alarmist prediction, that a fully-fledged arms race and major war will soon replace the Cold War rivalry system, has yet to materialize (Kang, 2003). Despite liberals’ roseate predictions of the incoming Kantian peace, most East and Southeast Asian countries have continued to militarily confront each other over various issues in disputes (Ganguly and Thompson, 2011). The constructivist approach, which asserts that irreconcilable identities will make the interstate relations in East Asia more volatile and increase the chance of military conflict, fits better in the explanation of the East Asian security landscape but it does not explain why some identity conflicts escalate to military conflict while others do not.
Noting these weaknesses, the article develops a more nuanced approach to an increased militarism in East and Southeast Asia, which might be called a theory of perverse political accountability. Drawing on the large literature of international rivalries and of democratic peace, the theory postulates that rivalries are at the center of the increased militarism in East and Southeast Asia, and a general advancement of democracy in the countries makes military conflicts more likely by reinforcing perverse political accountability. The following section offers a full definition of perverse political accountability with some testable hypotheses.
Theoretical framework: Perverse political accountability and the “onset” of military conflicts
To construct a theoretical framework for understanding the conflict behavior of East and Southeast Asian rivals, I draw on the large literature on international rivalries and democratic peace. The conventional wisdom in the literature is that democracy leads to peace and stability in world politics. Bruce Bueno de Mesquita and his colleagues, for instance, demonstrate that democratic leaders tend to be highly cautious about their choice of military conflict and war because such risky policies are likely to lead to domestic replacement (Bueno de Mesquita et al., 1999). In a similar vein, Huth and Allee claim that ex post political accountability makes democratic leaders more peaceful in dealing with foreign rivals (Huth and Allee, 2002a, 2002b). Russett and Oneal also found that the check and balance system institutionalized in democracy inhibits democratic leaders from adopting a reckless foreign policy (Russett and Oneal, 2001). Drawing on the concept of audience costs, Fearon and colleagues also show that democracies tend to signal their intentions to other states more credibly and clearly, ameliorating the security dilemma between democratic states (Fearon, 1994; Tomz, 2007). Mansfield and Snyder (2002, 2005), in contrast, found that democracy, especially a rapid transition to democracy, has an incendiary effect on military conflicts by preventing leaders from effectively regulating mass political competition and belligerent nationalism.
However, the theories delineated above are not without shortcomings. First, they say little about the relationships between democracy and international rivalries. Either the democratic peace theory or the theory of democratizing initiator of war has been developed to explain the impact of democracy or democratization on the chance of military conflicts between all states in international systems. Therefore, the theories remain silent about the connection between democracy and military conflict in international rivalries. In addition, the theories primarily give attention to all world regions, thereby dismissing the possibility that some region-specific factors may have an impact on the likelihood of military conflicts.
Noting these gaps, I focus on East and Southeast Asia, and especially on rivalries in this region. A voluminous literature on international and Asian security has claimed that, unlike most European counterparts, East and Southeast Asia has been at the center of increased militarism. The tension and conflict backed by old-fashioned feuds and animosity, in particular, have rapidly risen to the surface in the post-Cold War setting, thereby contributing to an increased chance of military conflicts both at the regional and global level (Betts, 1993/4; Calder, 2004; Friedberg, 1993/4; Fukuyama, 2005). Among others are the rivalries with which the scholars are primarily concerned because they believe these rivalries to be the main vehicles for military conflict in the region. Accordingly, pairs of rivals such as China and Japan, China and Taiwan, North and South Korea, South Korea and Japan, China and Vietnam, and Thailand and Cambodia have received considerable scholarly attention (Cha and Kang, 2003; Christensen, 1999; Guan, 1998; Hughes, 2009; Scobell, 2011; Wiegand, 2015). Table 1 offers a sketch of the connection between rivalries, war, and militarized interstate disputes (MIDs) short of war in East and Southeast Asia.
War and militarized interstate disputes (MIDs) in East and Southeast Asia, 1945–2001.
Source: Maoz (2005).
As shown in Table 1, East and Southeast Asian rivalries have been a major vehicle for inter-state military conflicts in the region. Forty-eight out of 74 wars (about 65 percent) have occurred in the rival countries during the period 1945–2001. The rivalries also have been at the center of MIDs short of war. One hundred and seventy-four of 200 MIDs (about 87 percent) have occurred in the rivalry dyads. Simply put, the rivalries have been at the center of military conflicts in East and Southeast Asia.
What strikes me the most, however, is that many of the East and Southeast Asian rivalries have experienced a systematic development of democracy over the past decades but the development has increased the likelihood of militarized conflicts rather than decreased it. Applying the democratic peace theory to international rivalries, some argue that rivalries are more likely to be peaceful if they become more democratic (Bennett, 1998; Hensel et al., 2000). However, the situation with East and Southeast Asian rivalries appears to evolve in the opposite direction. Since the mid-1980s, the “third wave of democratization” has swept Asia, especially in East and Southeast Asia (Huntington, 1991). The introduction of democracy in the Philippines and South Korea was followed by that of Taiwan, Thailand, Cambodia, and then Malaysia (Diamond, 2012; Fukuyama, 2012; Peou, 2014). What we witness, however, is that these rivals have increased the chance of militarized disputes, despite their decisive shift to democracy (Calder, 2004, 2006; Choi, 2005; Fukuyama, 2005; Hughes, 2009; Wiegand, 2015). The question then becomes: Why do the rivals who experience a general advancement of democracy become more willing both to engage in, and to initiate, conflict? 3
To answer this puzzle, I develop a “theory of perverse political accountability” in which East and Southeast Asian rivals’ conflict behavior receives primary attention. Perverse political accountability refers to a political reward mechanism in which state leaders adopting a policy of confrontation and conflict are electorally better off while leaders adopting a policy of conciliation and cooperation are electorally worse off. The theory of the accountability constitutes two major building blocks, i.e. the incipient nature of democracy and the hawkish preference of the leaders and their hard-line constituency. The theory claims that these two elements are inextricably intertwined to increase the chance of military conflict between rivals in East and Southeast Asia.
First, the “young” and “incipient” nature of democracy in most East and Southeast Asian states matters (Diamond, 2012; Peou, 2014). 4 The leaders’ legitimacy in such young democracies is still weak but their stakes for domestic politics are very high. The party system, which is designed to mitigate conflicting interests among societal actors, has yet to be consolidated (Kapstein and Converse, 2008: 58). Under these circumstances, the democratic leaders’ legitimacy can be more easily contested by a majority constituency or selectorate, often causing governance problems that could lead to premature collapse or a nearly successful attempt (Hays et al., 2003; Huntington, 1991; Kapstein and Converse, 2008; Keefer, 2007; Peou, 2014). 5 When the leaders fail to satisfy the demands of the constituency, therefore, they will suffer a series of political backlashes such as coups, riots, and public demonstrations. In this sense, the core constituency tends to have a greater sway over their leaders’ policies in younger democracies than in older ones.
Given the incipient nature of democracies, the theory claims that the leaders and constituencies with a larger majority (or selectorate) in East and Southeast Asian rival countries will be hard-liners. Scholars of international rivalries cogently demonstrate that hostility from a rival, in the form of verbal threats and military actions, serves to reinforce the position of hawkish leaders and their hard-line constituencies (Colaresi, 2004; Nincic, 1989; Merolla et al., 2007; Vasquez, 1993). East and Southeast Asian rivals are not an exception to this trend. Most leaders and constituencies in rival states have been a hostage of the enmity and distrust that decades-long rivalry relationships have created, and have therefore become hard-liners (Berger, 2012; Calder, 2004, 2006; He, 2007; Matthews, 2003; Moon and Suh, 2007; Scobell, 2011). It is true that the Cold War muted such rivalry-driven antagonism by creating a new fault line, i.e. capitalism vs. communism. However, the abrupt dissolution of the Soviet Union has led to a renewal of feuds in the rival states by encouraging old-fashioned antagonism to return and fill the ideological void left by the end of the Cold War. Grounded in such renewed feuds, the hard-line leaders and constituencies rapidly expand their influence in their societies.
The incipient nature of democracy in East and Southeast Asian rival countries, coupled with the renewed hostility backed by the hard-line constituencies, reinforces perverse political accountability such that the democratic developments may directly contribute to increased militarism. On the one hand, the very young nature of democracy makes the rival leaders’ legitimacy very vulnerable to premature collapse if they fail to meet the needs of the core constituency. On the other hand, the hard-line groups coalesce into a powerful selectorate and rapidly expand their power because the democratic development creates a growing public space for bottom-up venting of the groups’ hawkish preferences. In this circumstance, potentially hard-line democratic leaders face a distorted political reward mechanism in which their conflict-based policies are electorally rewarded by the constituencies while their conciliatory policies are electorally punished (Colaresi, 2004; Hensel, 1996; Nincic, 1989). Accordingly, democratic leaders in these countries will become more willing to play the military card against foreign rivals as their polity becomes more democratic.
The examples abound showing that perverse accountability has prompted East and Southeast Asian rival leaders’ aggressive policies against their foreign rivals. Pressured by a nationalist public whose influence increased markedly during the period of democratic development in the 1990s, Thai’s hawkish democratic leader Mark Abhisit Vejjajiva has rapidly militarized the dispute with Cambodia around the Preah Vihear Temple (Hughes, 2009). The democratic advancement in Taiwan since 1991 also triggered a vigorous process of rivalry outbidding between competing hawkish parties, which prompted the incumbent hawkish head of state Lee Teng-hui’s launch of a series of tough foreign policies on the PRC (Scobell, 2011). South Korean leaders such as Kim Young-Sam and Roh Mu-Hyun have also been pressured by a number of hard-line groups to play the military card in dealing with the Dokto/Takeshima Islands disputes with Japan as their polity has become more democratic (Choi, 2005; Wiegand, 2015).
My theory postulates, given the varying degrees of democracy in East and Southeast Asian rival states, that the more democratic the setting a country operates in, the more likely is military conflict. This is because the leaders are more likely to be constrained by the hard-line constituency or selectorate as their regime moves further toward democracy and, as a result, are more likely to engage in conflict. The argument leads to a hypothesis as follows: Hypothesis 1
East and Southeast Asian rivals with higher scores of democracy will experience an increased chance of the onset of MIDs.
Perverse political accountability and the initiation of military conflicts
Although a general advancement of democracy in East and Southeast Asian rival states increases the likelihood of the onset of military conflicts, the likelihood that a democratic rival will initiate the conflicts varies with the regime type of the other rival. The conventional wisdom in the literature on democratic peace asserts that democracies are less likely to engage in conflict (Huth and Allee, 2002a, 2002b; Russett and Oneal, 2001) and that the most dispute-prone pairs of states are those that contain one democracy and one autocracy. This is the so-called “cats-and-dogs” effect (Doyle, 1986, 2005; Russett and Oneal, 2001). If the two states have very opposite systems, they would have much to fight about, leading to an increased chance of conflict.
Given the highly conflict-prone nature of democracy-autocracy dyads, however, the literature diverges on which regime is the major culprit in military conflict. Doyle (1986, 2005) and Peceny et al. (2002), for instance, show that the probability of a militarized dispute between a democracy and an autocracy, notably a personalist regime, increases because militarily strong and capable democracies pick weak personalist regimes as targets. Some, in contrast, argue that the leaders in an autocracy are more likely to initiate a military conflict against a democracy, primarily because the political opposition in an autocracy is less capable of contesting state politics and, as a result, the leaders are unlikely to lose power, even if they launch a losing conflict short of catastrophic defeat (Chiozza and Goemans, 2004; Reiter and Stam, 2003).
However, my perverse political accountability theory expects that, at least in East and Southeast Asian rivalries, democratic rivals are highly likely to “initiate” military conflicts toward autocratic rivals. Given the potentially hard-line preference of the leaders in the rivalries, the incipient nature of democracy and aggregated powers of the hard-line constituency, which acts as the selectorate, are intertwined to reinforce a perverse political accountability through which the leaders are pushed into the initiation of military conflicts against their autocratic counterparts. The leaders, in particular, would believe that their autocratic counterparts are less likely to be constrained by the hard-line constituency due to the absence of political accountability itself and, as a result, are more likely to back down although they are involved in the conflicts (Huth and Allee, 2002a, 2002b). With this expectation, the democratic leaders in the rivalries will likely initiate conflicts against autocratic opponents.
It is also important to note that autocratic leaders in East and Southeast Asian rivalries also have a strong incentive to initiate a military conflict against democratic rivals. Many autocratic leaders in East Asian rivalries are personalistic leaders such as Mao, Pol Pot, Sukarno, Suharto, Marcos, and Kims, who have extremely belligerent preferences (Weeks, 2012). Moreover, these leaders have faced no ex post perverse accountability mechanism. Although they initiate a military conflict toward their democratic rivals and lose the conflict short of catastrophic defeat, therefore, they are least likely to be punished by the hard-line constituency or selectorate (Chiozza and Goemans, 2004; Reiter and Stam, 2003; Weeks, 2012). The autocratic leaders may expect that their democratic counterparts will stand firm if they are involved in the conflict, due to perverse political accountability. Expecting that they will be insulated from any ex post perverse political accountability, however, the leaders will act upon their aggressive preferences. Accordingly, the autocratic leaders are likely to initiate the conflicts against their democratic counterparts.
If democratic rivals face off with each other, my theory expects, they are not more likely to initiate military conflicts against each other. Given the presence of a perverse political accountability on both sides, the democratic leaders on one side would believe that the democratic leaders on the other side will act as tough as they do. So, the democratic leaders in East and Southeast Asian rivalries are prevented from initiating conflicts against each other. In a similar vein, pairs of anocracies or mixed regimes in the rivalries are not more likely to initiate conflicts against each other. In the mixed regimes in the rivalries, perverse political accountability has yet to be firmly installed (Mansfield and Snyder, 2002, 2005). Accordingly, the leaders in these regimes have little incentive to initiate conflicts against each other.
Pairs of autocracies in East and Southeast Asian rivalries will likely initiate a military conflict against each other. As noted previously, autocratic leaders in East and Southeast Asian rivalries are mostly personalistic leaders whose preferences are extremely belligerent (Weeks, 2012). Hostility from foreign rivals, in the form of verbal or military threat, would aggravate them. Given their belligerent preferences, the absence of ex post perverse political accountability will make the conflict between the autocratic rivals more likely. Because the leaders are unusually insulated from the consequences of foreign policy failure, they will act on their preferences and take gambles against each other (Chiozza and Goemans, 2004; Weeks, 2012). These theoretical arguments lead to a set of hypotheses as follows: Hypothesis 2
A democratic rival (an autocratic rival) is more likely to initiate MIDs against an autocratic rival (or a democratic rival) in East and Southeast Asia. Hypothesis 3
Pairs of democracies and of mixed regimes in East and Southeast Asian rivalries are not more likely to initiate MIDs against each other. Hypothesis 4
Autocratic rivals are more likely to initiate MIDs against each other in East and Southeast Asia.
Research design
Population and dependent variables
I analyze all post-war East and Southeast Asian rivals, from 1945 to 2001, using Klein, Goertz, and Diehl (2006)’s new rivalry dataset. The updated dataset treats a rivalry as the same pair of states: 1) engaging in three or more militarized disputes and; 2) experiencing an interrelation of issues across repeated conflict (Klein et al., 2006: 335–336). Several MIDs between two states therefore only signals the existence of a rivalry when there are common issues at stake that link each instance of conflict together. Drawing on Klein et al.’s dataset, I generate 26 East and Southeast Asian rivalries and 621 rivalry dyad years during the period 1945–2001. 6
To test hypothesis 1, I create a dependent variable marking whether a rival i or j will experience the “onset” of MIDs. For this purpose, I generate a “non-directed rivalry dyad year” data set, in which the potential initiator and target between rival i and j are not differentiated. If the dyad year experiences a MID during the period 1945–2001 it is coded “1,” and “0” otherwise. Data on MIDs are drawn from Maoz’s (2005) dyadic MIDs dataset (v. 2.0).
To test hypotheses 2–4, I create the second dependent variable marking whether a rival i or j in East Asian rivalries “initiates” MIDs against another. For this purpose, I generate a “directed dyad year” data set, in which a potential initiator of a MID and a potential MID target are identified. In the data set, a dyad year between rival i and j appears twice, first as dyad i-j, in which rival i is the potential MID initiator and rival j is the potential MID target, and then as dyad j-i, in which rival j is the potential MID initiator and rival i is the potential MID target (Bennett and Stam, 2000; Reiter and Stam, 2003). I create a dichotomous variable coded “1” if the potential initiator in the direct dyad initiates a MID and “0” otherwise. Data on MIDs are drawn from Maoz’s (2005) dyadic MIDs dataset (v. 2.0).
Independent variables
My first independent variable is the higher score of democracy between rivals i and j in East and Southeast Asia, 1945–2001. So, I create Demo H variable by sorting out the higher score of 21-point democracy indices in each of the East and Southeast Asian rivalry dyads. Using this variable, I test a hypothesis that it is a rival with a higher democratic score that is the major culprit of the onset of MIDs, and that the likelihood of MIDs depends primarily on how strong the democratic constraints (the constraints from the hard-line constituencies) are on the “more” constrained rivals in the dyads. All data are derived from Polity IV dataset. For a simplification, I construct 21-point indices by adding 11 to the original Polity score, generating democracy scores ranging from 1 to 21. I expect a “positive” effect on the Demo H variable.
My second independent variable is a series of dummy variables marking the dyadic regime type in each East and Southeast Asian rivalry. The Demo/Auto ij variable is coded 1 if rivals i and j are a pair of a democracy and an autocracy, and 0 otherwise. The Auto/Demo ij variable is coded 1 if rivals i and j are a pair of an autocracy and a democracy, and 0 otherwise. The Demo/Demo ij variable is coded 1 if rivals i and j are a pair of democracies, and 0 otherwise. The Mix/Mix ij variable is coded 1 if both rivals i and j are a pair of mixed regimes or anocracies, and 0 otherwise. The Auto/Auto ij variable is coded 1 if rivals i and j are a pair of autocracies, and 0 otherwise. I use a threshold of Polity score +17 to identify a democracy, while using +5 to identify an autocracy. Any regime between these two scores is identified as mixed regime or anocracy. All data are derived from Polity IV dataset. I expect a positive effect on the Demo/Auto ij , Auto/Demo ij , and Auto/Auto ij variables, with little effect on the Demo/Demo ij and Mix/Mix ij variables.
Control variables
Throughout the analyses, the article controls for a set of variables drawn from the literature on international and Asian security. First, it controls for the Democrz ij variable to see if an increased chance of MIDs in East and Southeast Asian rivalries is the result of a rapid regime change toward democracy. For this analysis, I code both rival i and j as democratic, autocratic, or anocratic in every year, t−1, using a threshold of Polity score +6 and –6, and then measure rivals i and j’s regime types in year t−6. Democratization is defined as any change of regime type in a democratic direction between t−6 and t−1 (Mansfield and Snyder, 2002, 2005).
Many liberal scholars also suggest that economic interdependence and inter-governmental organizations (IGOs) have a strong conflict-reducing effect (Acharya, 2003/4; Duffield, 2003; Russett and Oneal, 2001). To test this liberal insight, the article includes the “Depend L ” variable, which is the bilateral trade-to-GDP ratio for the rival with the lower dependence score. The trade data are from the Correlates of War Bilateral Trade dataset (v. 3.0) while the data on the GDP derives from the Penn World Table (v. 7.1). I also include the IGOs variable. In East and Southeast Asia, a set of IGOs such as the UN, the ASEAN, the APEC, and the ARF are considered serving as the “potential kernel” of regional cooperation (Acharya, 2003/4; Duffield, 2003; Xinbo, 2008). The “IGO ij ” variable is created by marking whether the rivals i and j have joint membership in the above-mentioned IGOs. The data on IGOs and rivals’ membership in the IGOs are drawn from the Year Book of International Organizations. I expect a negative effect on these liberal variables.
The scholars of various realist stripes also claim that the balance of power and security ties of great powers with nuclear weapons affect the likelihood of both the onset and initiation of MIDs (Bennett, 1998; Paul, 2006; Weeks, 2012). For the measurement of the balance of power, I create the “BOP ij ” variable, which is the value of a stronger rival i’s (or j’s) Composite Index of National Capabilities (CINC) divided by the sum of both rivals i and j’s CINC. The CINC scores are drawn from the COW’s CINC index (v. 4.0). To control for the alliance ties between rivals and great powers, I create Alliance i and Alliance j variables, which is the weighted measure of the similarity of rivals i and j’s alliance portfolio with the system leader, i.e. the United States, during the period 1945–2001. I expect a negative effect on these variables. 7
Many studies in international security also illustrate that territorial disagreement and geographical contiguity affect the chances of military conflicts (Hensel, 1996; Huth and Allee, 2002a, 2002b; Vasquez 1993; Weeks, 2012). For measuring the effect of territorial dispute, I create a dummy variable “Territory ij ” marking whether the issue in dispute in East and Southeast Asian rival dyads is over contested territory. I use Issue Correlates of War (ICOW)’s updated list of territorial dispute cases during the period 1945–2001. If the dyad experiences a territorial dispute, it is coded 1, and 0 otherwise. I expect a positive effect on the variable. To control for the effect of geographical proximity, I include the Contiguity ij variable using Correlates of War (COW)’s contiguity scale ranging from 6 to 1. I expect a negative effect on the variable. The descriptive statistics of the variables mentioned here are presented in Table 2.
Descriptive statistics.
Model specification: Logit analyses with time dependency and fixed effects
Because the dependent variables are binary, I estimate ordinary logistic regression models with time dependency variables. In the time-series-cross-section analysis with a binary dependent variable, the observations are temporally related that the results of an ordinary logit may misleading. To control for temporal or time dependence, I follow Beck, Katz, and Tucker’s (1998) suggestion of using both “peace years” and “cubic splines” of the number of years since the last time rival i and rival j experienced an onset of a MID against each other (hypothesis 1) and of the number of years since the last time either rival i or rival j initiated a MID against each other (hypothesis 2–4). In addition, I also run the analyses with “fixed” effects to avoid the possibility of omitted variables that are specific both to undirected and to directed dyadic research design (Weeks, 2012).
Empirical findings and substantive interpretation
The “onset” of MIDs in East and Southeast Asian rivalries, 1945–2001
To test hypothesis 1 that the chance of the onset of MIDs will increase as a more democratic rival in East and Southeast Asia becomes even more democratic, I estimate the effect of the Demo H variable on the probability that rivals i and j will experience the onset of MIDs, with a set of control variables. To see if the increased chance of MIDs is the result of a rapid regime change toward democracy, I also estimate the effects of Democrz ij and compare them with the effects of Demo H . All results are reported in Table 3, with model-related statistics at the bottom of the table.
Logit analysis of the probability of MID onset in East Asian rivalries, 1945–2001 (non-directed rivalry dyads).
Notes: Cell entries are coefficient estimates and robust standard errors are in parentheses. Two-tailed significance levels of 0.10 or less are reported (***p < 0.01; **p < 0.05; *p < 0.10).
The first column of Table 3 (Model 1) presents the coefficient estimates of my base-line model. Crucial to testing my theoretical argument is the effect of Demo H on the probability that East and Southeast Asian rivals experience the onset of MIDs. The variable has a statistically significant and positive effect on the probability, indicating that as the rivals with the higher democracy scores become more democratic, the probability that they will experience the onset of MIDs increases. In addition, both the BOP ij and Contiguity ij variables have a statistically significant and negative effect on the likelihood of the onset of MIDs, indicating that the power imbalance and the greater distance between rivals i and j both decrease the chance of the onset of MIDs.
The second column of Table 3 (Model 2) tests the impact of the Democrz ij variable on the probability of the onset of MIDs in East and Southeast Asian rivalries. Interestingly enough, the Democrz ij variable does not have any statistically significant effects on the chances of MIDs, indicating that a rapid regime change toward democracy in the rivalries will not increase the probability in the rivalries. Regarding control variables, the BOP ij variable has a statistically significant and negative effect on the chance of the onset of MIDs, indicating that the power imbalance between i and j decreases the probability of the onset of MIDs.
The third column of Table 3 (Model 3) presents the coefficient estimates of my full model in which both the Demo H and Democraz ij are included. Not surprisingly, the Demo H variable has a statistically significant and positive effect on the likelihood of the onset of MIDs while the Democrz ij variable has little impact on the likelihood. The result confirms my prediction that it is a general advancement of democracy, not a rapid regime change toward democracy, that increases the likelihood of the onset of MIDs in East and Southeast Asian rivalries. Except for the Demo H variable, both the BOP ij and Contiguity ij variables have a statistically significant and negative effect on the chance of a MID in the rivalries, indicating that both the power imbalance between i and j and a greater distance between them decrease the chance of the onset of MIDs in the rivalries.
The fourth column of Table 3 (Model 4) tests the impact of both the Demo H and Democrz ij variables on the likelihood of the onset of MIDs in the rivalries with a set of control variables using a fixed-effect specification. Given the more demanding nature of fixed-effect analysis, especially the analysis’ correction of the possibility that omitted variables are specific to non-directed dyads, the Demo H variable has no statistically significant effect on the likelihood of MIDs onset, although its coefficient estimate is positive.
Statistical significance, of course, is not the same thing as substantive significance. To gauge the magnitude of the substantive effect of a key independent variable, i.e. Demo H , I calculate absolute change in predicted probability of the onset of MIDs associated with a change in the Demo H and in each of the other control variables when all other variables are held constant at their mean values. The results are reported in Table 4.
The substantive effect of Demo H on the probability of the onset of MIDs.
†Absolute change in probability of a MID onset associated with a change from one standard deviation below the mans to one standard deviation above the mean in the variables (from 0 to 1 for dichotomous variables) when all other variables are held constant at their mean values.
Table 4 presents the “absolute” change in the predicted probability of the onset of a MID associated with a change in a key independent variable, i.e. Demo H , while all other variables are held constant at their means. The result shows that the probability of the onset of MIDs will increase by 3.9%. Given the overall probability of the onset of MIDs in my sample is 0.179, such effect is substantive as well as statistically significant.
I also present a graph showing how the predicted probabilities of MID onset will vary with a change in the Demo H variable, holding all other variables constant at their mean. The results are presented in Figure 1.
Predicted probabilities of MID onset related to the change in the value of demo H .
The results are striking. Moving from an East and Southeast Asian rivalry dyad having the lowest value of the Demo H variable (= 2) to the dyad having the mean value of the Demo H variable (= 11.715) leads to about a 38-percentage increase in the predicted probability that the dyads will experience the onset of MIDs. Moving from the rivalry dyad having the lowest value of the Demo H variable (= 2) to the dyad having the highest value of the Demo H variable (= 21) leads to about an 83-percentage increase in the predicted probability that the dyads will experience the onset of MIDs. The results provide strong support for my argument that the more democratic the East and South East Asian rival is, the less constrained it will be from engaging in MIDs and the higher we can expect the chance of a MID to be.
“Initiation” of MIDs in East and Southeast Asian rivalries, 1945–2001
In what follows, I estimate the effect of dyadic regime types in East and Southeast Asian rivalries with a set of control variables on the probability that rivals i and j will “initiate” MIDs against each other. The non-directed research design that the Models 1–4 utilize does not allow us to identify which regime in each rivalry dyad initiates MIDs. So, I create a “directed dyad year” data in the rivalries in which a potential initiator of MIDs and a potential target are identified. In the data, a dyad year between rivals i and j appears twice, first as dyad i-j, in which rival i is the potential MIDs initiator and rival j is the potential MIDs target, and then as dyad j-i, in which rival j is the potential MIDs initiator and rival i is the potential MIDs target (Reiter and Stam, 2003). All results are reported in Table 5, with model-related statistics at the bottom of the table.
Logit analysis of probability of the initiation of a MID in East and Southeast Asian rivalries, 1945–2001 (directed dyadic years).
Note: Cell entries are coefficient estimates and robust standard errors are in parentheses. Two-tailed significance levels of 0.10 or less are reported (***p < 0.01; **p < 0.05; *p < 0.10).
The first column of Table 5 (Model 5) presents the effect of the regime type of the potential initiator (a democracy) and the potential target (an autocracy) on the probability that a rival will initiate MIDs against the other rival. The Demo/Auto ij variable has a statistically very significant and positive effect, indicating that democratic rivals are more likely to initiate MIDs against autocratic rivals. With regard to control variables, only the Contiguity ij variable has a statistically significant and negative effect on the probability, implying that a greater distance between rivals i and j will decrease the chance of the initiation of MIDs.
The second column of Table 5 (Model 6) presents the effect of the regime type of the potential initiator (an autocracy) and the potential target (a democracy) on the probability that a rival will initiate MIDs against the other rival. The Auto/Demo ij variable has a statistically significant and positive effect, indicating that autocratic rivals are more likely to initiate MIDs against democratic rivals. Regarding control variables, only the Contiguity ij variable has a statistically significant and negative effect on the probability, implying that a greater distance between the two regimes will decrease the chance of the initiation of MIDs.
The third column of Table 5 (Model 7) presents the effect of various dyadic regime types between the potential initiator and the potential target on the probability that a rival will initiate MIDs against the other rival. The Demo/Auto ij and Auto/Demo ij variables still have a statistically significant and positive effect, indicating that democratic (autocratic) rivals are more likely to initiate MIDs against autocratic (democratic) rivals. The Auto/Auto ij variable also has a statistically significant and positive effect, indicating that autocratic rivals are more likely to initiate MIDs against each other. But the Demo/Demo ij and Mix/Mix ij variables have little effect on the probability of the initiation of MIDs between the potential initiator and target. In addition, the capability variables of rivals i and j, denoted CINC i and CINC j , and the Contiguity ij variable have a statistically significant and negative effect, implying that these variables decrease the likelihood of the initiation of MIDs between the potential initiator and target.
The fourth column of Table 5 (Model 8) presents the effect of various dyadic regime types between the potential initiator and the potential target on the probability that a rival will initiate MIDs against the other rival using a fixed-effect specification. The Demo/Auto ij and Auto/Demo ij variables still have a statistically significant and positive effect, indicating that these regime combinations of the potential initiator and target make the initiation of MIDs more likely. But the Demo/Demo ij , Mix/Mix ij , and Auto/Auto ij variables have little effect on the probability of the initiation of MIDs between the potential initiator and target.
To gauge the magnitude of the substantive effect of the key independent variables, i.e. Demo/Auto ij , Auto/Demo ij , and Auto/Auto ij , I calculate absolute change in the predicted probability that rivals i and j will initiate MIDs against each other associated with a change in the Demo/Auto ij , Auto/Demo ij , and Auto/Auto ij variables and in other control variables when all other variables are held constant at their mean values, relying on Model 7. The results are presented in Table 6.
The substantive effect of Demo/Auto ij , Auto/Demo ij , and Auto/Auto ij on the probability of the initiation of MIDs.
†Absolute change in probability of a MID onset associated with a change from one standard deviation below the mans to one standard deviation above the mean in the variables (of from 0 to 1 for dichotomous variables) when all other variables are held constant at their mean values.
Table 6 presents the “absolute” change in the predicted probability of the initiation of MIDs associated with a change in the key independent variables. The result shows that the probability of the initiation of MIDs between potential initiator (a democracy) and potential target (an autocracy) increases by about 22%. In a similar vein, the probability of the initiation of MIDs between potential initiator (an autocracy) and potential target (a democracy) increases by about 19.9%. The probability of the initiation of MIDs between potential initiator (an autocracy) and potential target (an autocracy) also increases by about 7%. The results show that the effect of these variables on the probability of the initiation of MIDs is substantive as well as statistically significant.
To summarize, my analyses show that democracies in East and Southeast Asian rivalries are not as peace-prone as theorists of democratic peace have assumed. Not only are the rivals with the higher democracy score more likely to engage in a military conflict as they become more democratic, but they are more likely to initiate military conflicts against autocratic rivals. It is also worth noting that autocratic rivals in South and East Asia will be likely to initiate MIDs against democratic rivals and against each other. The results provide strong support for my argument that democratic rivals are quite conflict-prone due to their leaders’ hawkish preferences and to a perverse political accountability, while autocratic rivals are also conflict-prone because of the leaders’ hard-line preferences and the absence of accountability.
Conclusions and directions for future research
The article asks a simple question of why Asian rivalries have experienced a sharp increase in military activities despite their decisive shift to democracy? To explain this puzzle, I develop a theory of perverse political accountability. Central to my theory is that perverse political accountability, in which a hawkish leadership adopting a more conflictual policy toward foreign rivals is rewarded by the hard-line constituency or selectorate, makes the democratic leaders in the rivalries highly conflict-prone. In addition, the theory claims that democratic rivals will likely initiate military conflicts against autocratic rivals due to their leaders’ hawkish preferences and due to perverse political accountability, and that autocratic rivals will likely initiate conflicts against democratic rivals and against each other due to their leaders’ hawkish preferences and to the absence of perverse political accountability.
The value added by this article is two-fold. First, the article develops a novel approach to military conflicts in East and Southeast Asia. Drawing on the large literature on international rivalries and on democratic peace theory, the article develops a theory of perverse political accountability in which a general advancement of democracy is agued to increase the chances of military conflict in East and Southeast Asian rivalries. Most studies on democratic peace claim that democracy has a strong pacifying effect and that such effect is universal. By focusing on the unique historical conditions that have pervaded East and Southeast Asian rivalries, however, the article offers such a counter-intuitive theoretical approach.
Second, the article provides systematic evidence confirming the hypotheses that the theory of perverse political accountability generates. Controlling for a series of realist and liberalist variables, which offer intensive information about the international and domestic political situation in East and Southeast Asia rivalries, the article finds that the chance of the rivals experiencing military conflict increases as they become more democratic. Using a directed-dyad research design, the article also finds that democratic rivals are more likely to initiate conflicts against autocratic rivals due to accountability. In addition, it finds that autocratic rivals, which are insulated from accountability, are more likely to initiate conflict against democratic rivals and against other autocratic rivals.
Despite such contributions, however, the article is the initial step for a complete understanding of the effect of democratic development on militarized disputes in international rivalries. Since my study limits its focus to the post-war East Asian rivalries, future research must expand its focus to rivalries across regions and over time. It also may be suggested that future research examine the effect of “incipient” and “consolidated” democracies on militarized disputes in rivalries separately. Given the relative strengths of democratic institutions in a consolidated democracy, their leaders may be less constrained by the hard-line constituency or selectorate and more peaceful even against their foreign rivals.
The study also has an important policy implication. That is, a gradual and steady introduction of democracy into East and Southeast Asian rivalries has a strong incendiary effect. This is mainly because democratic development in the rivalries creates more public space for bottom-up venting of popular hostility toward foreign rivals, which may result in the installment and consolidation of a perverse political accountability. Keenly aware of this, the leaders in the rivalries need to check the powers of hard-line constituencies armed with jingoistic rhetoric and preferences as they become more democratic. Unless properly controlled, the constituencies will likely press the leaders to be unnecessarily bellicose in dealing with their foreign rivals, increasing the chances of military conflicts in the region.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
This research is supported by National Research Foundation of Korea (Grant No., 2016S1A3A2924409).