Territorial acquisition, commitment, and recurrent war

Territorial acquisition, commitment, and recurrent war Abstract This study investigates how territorial acquisition through war affects the durability of a successive ceasefire and determines what type of territorial acquisition is more detrimental to post-war peace. Despite the wealth of literature on recurrent war and on territory, the effect of territorial acquisition on war resumption has been understudied. This study shows that territorial acquisition creates expectations among adversaries for future power shifts, which results in a commitment problem that hinders peaceful revision of the existing ceasefire. Indeed, duration analysis of ceasefires following interstate wars since World War II shows that territorial change in war, especially acquisition of large and densely populated territories that have potential utility for greater power shifts, makes ceasefires more prone to failure. The analysis of Sino-Vietnamese ceasefires following militarized incidents over land and sea borders also illustrates the importance of territorial acquisition and the potential utility of the territory. 1 Introduction Ceasefires following interstate wars often break down after a short period and cause further devastation to people’s lives that have already been destroyed by war. Even ceasefires minimally defined by the absence of war have often failed: of the 56 pairs of countries engaged in full-scale war between 1946 and 2004, more than 40% reverted from a ceasefire to another full-scale war.1 This problem has cast a shadow over the Asia-Pacific region in which disputes that once caused wars between India and Pakistan and in the Korean Peninsula are yet to be resolved, and pairs of countries, such as China and Vietnam, have repeatedly engaged in military confrontations (Diehl and Goertz, 2000). Concerned over failed ceasefires and their consequences, scholars have made efforts to study why adversaries dishonor a ceasefire. Various factors have been identified ranging from the indecisive outcome of the preceding war, to compromises made in a negotiated ceasefire, to a post-war shift in adversaries’ power (e.g. Blainey, 1988; Werner, 1999; Quackenbush and Venteicher, 2008). Many studies investigated why the same set of states repeatedly engage in militarized disputes. However, another important question in this research field is, why do adversaries return to a full-scale war (Fortna, 2004; Werner and Yuen, 2005; Lo et al., 2008)? These are two different problems incurred because not all resumed disputes result in a war. Some of them are resolved short of war, while others escalate. Therefore, the problem of war resumption is worth exploring from normative and theoretical points of view. Although a handful of studies, such as Werner and Yuen (2005), have identified a mechanism specific to the resumption of war, the causes for recurrence of war remain understudied. This study examines the effect of territorial acquisition on the resumption of war. This factor has been largely ignored in the literature, even though territorial conflicts are consistently found to be violent and to recur (e.g. Goertz and Diehl, 1992; Vasquez, 1993, 1995; Quackenbush, 2010; Reed and Chiba, 2010). Specifically, this study highlights a commitment problem in the renegotiation of ceasefires as the cause for war resumption, and territorial acquisition through war identified as the source of the problem. This argument is tested against the data on (post-)interstate war ceasefires that occurred between 1946 and 1997, in which ceasefire is defined as ‘a period between two interstate wars, starting when the first war ends and failing when war resumes between the same adversaries’. The study also investigates what type of territorial acquisitions are more detrimental to ceasefires. Case studies of two Sino-Vietnamese conflicts at land and sea are presented as illustrative cases. The analysis reveals that territorial acquisition through war makes a subsequent ceasefire short-lived. The negative effect of territorial acquisition is not ascribed to the geographic contiguity of the adversaries, the contentiousness of territorial issues, or the observed post-war power shifts. Rather, it is argued that territorial change generates adversaries’ expectations for future power shifts, which causes a commitment problem when they engage in negotiation during ceasefire. Accordingly, it has been found that larger and more populated territories, that is, those with potential utility for greater power growth, are more likely to cause wars to resume when these are transferred in war. 2 Causes of recurrent war The literature on recurrent conflict can be divided into two camps according to the subject of research and the research question. Much of the literature focuses on settlement of disputes and solutions for underlying issues; the central question of this group of studies is, therefore, why do militarized interstate disputes occur between adversaries who have previously fought wars? Scholars have found that a military stalemate in the previous war (Blainey, 1988; Hensel, 1994; Grieco, 2001), compromises in reaching negotiated ceasefires (Senese and Quackenbush, 2003; Quackenbush and Venteicher, 2008), and third-party intervention and its withdrawal (Beardsley, 2008) cause disagreements and dissatisfaction with the ceasefire between adversaries. Some, however, as in the present study, ask a slightly different question: Why do adversaries return to a full-scale war, and not just a militarized dispute? In this framework focusing on wars causing a large number of casualties, a ceasefire is considered as maintained even if adversaries engage in small-scale military actions, such as a display of forces. Nevertheless, because dispute resumption is a part of the process in which war recurs, some studies also investigate mechanisms by which adversaries resume disputes and subsequent wars. Werner (1999) shows that large shifts in the post-war balance of power tend to cause war resumption because such shifts alter adversaries’ calculations and consequently encourage them to challenge the existing ceasefire. Lo et al. (2008) demonstrate that foreign-imposed regime change following war alters the preference and the institutions of the target country in favor of peace and reduces the risk of recurrent war. However, war resumption involves a mechanism distinct from that of dispute resumption, namely, failure of renegotiation. Even if adversaries challenge an existing ceasefire, it does not necessarily result in war, as it might be resolved through renegotiation of ceasefire terms. This issue is articulated in the bargaining theory of war, which posits that adversaries can usually settle their disputes without fighting, but they sometimes fail to do so and fight a war due to lack of credible information and commitment to a negotiated deal (e.g. Fearon, 1995). Applying this insight to the resumption of war, adversaries engage in another war because there are obstacles to peaceful renegotiation of the existing ceasefire. The literature on recurrent war explicitly or implicitly addresses this problem. Werner (1999) recognizes that post-war shifts in power, while encouraging adversaries to resume disputes, might not necessarily cause them to return to war. Adding to Werner’s study from an institutionalist perspective, Fortna (2004) shows that strong ceasefire agreements, for instance, those that are formally accepted and introduce demilitarized zones, contribute to post-war peace by increasing the cost of waging another war, by reducing uncertainty about each other’s actions and intentions, and by preventing accidental wars. Responding to Fortna, Werner and Yuen (2005) offer a new theory introducing an informational problem in reaching and maintaining ceasefires. Specifically, they argue that ‘unnatural settlements’ achieved after wars with inconsistent battle outcomes or under pressure from a third party prevent adversaries from acquiring accurate information and hence are more likely to become ‘obsolete’. Rephrased in the words of the bargaining theory of war, such ceasefires conclude despite significant uncertainty remaining unresolved, which makes peaceful renegotiation of ceasefires difficult. The present study adds to the literature on recurrent war, investigating the effect of territorial acquisition through war on the likelihood of war resumption. In contrast to the informational explanation offered by Werner and Yuen (2005), this study argues that territorial acquisition hinders peaceful revision of ceasefires by creating a situation in which adversaries are unable to credibly commit to a negotiated agreement. Moreover, by highlighting the consequence of territorial acquisition, the implication of this study also extends to the literature studying the association between territories and conflict, as is discussed in the next section. 3 War and territory It is well established that disputes over territory are violent and recurrent. This tendency toward escalation and recurrence is often attributed to two key characteristics of territorial disputes: contiguity and issue contentiousness. The geographic contiguity of adversaries provides them more opportunities to interact, a greater number of potential issues to fight over, and an environment more favorable to military deployment (e.g. Bremer, 1992; Quackenbush, 2006). Reed and Chiba (2010) also found that neighboring states not only differ in their ‘observable’ environment, such as the degree of economic interdependence, but also behave differently – namely, they respond more aggressively to a shift in the environment than do noncontiguous states. In a similar vein, Quackenbush (2010) reported that geographical contiguity increases the probability of dispute resumption after war, although the effect fades over time. Another key reason for the violent nature of territorial disputes is that territorial issues are salient because of the natural tendency of humans to occupy and defend territory. These territorial disagreements engender grievances among people, which help political leaders mobilize domestic support for militarized actions (e.g. Vasquez, 1993, 1995; Hensel, 1994, 2001; Walter, 2003; Senese and Vasquez, 2005). According to Senese and Vasquez (2003) and Senese (2005), territorial claims are important in explaining war onset, whereas contiguity makes militarized disputes more likely but not war. The same logic explains why militarized disputes over territory are more likely to recur (Hensel, 1996; Quackenbush, 2010). Acknowledging that ceasefires between contiguous parties and over territorial issues are precarious, this study argues that territorial transfers add yet another dimension to the problem. Early works investigating this issue was studied by Goertz and Diehl (1992), who argued that territorial changes spur losing parties’ grievances over lost territory and motivate them to engage in another militarized dispute. As Goertz and Diehl (1992) relied on factor analysis and did not treat censored data properly, Tir (2003) provided a more comprehensive and appropriate analysis. Comparing the effect of violent and peaceful territorial transfers on a prospective territorial dispute, Tir found that violent territorial transfers tend to be followed by shorter ceasefires. This association occurs because the party that obtained the territory seeks to acquire additional territory, especially when it is of strategic value, and the loser nurtures grievances over the lost territory, particularly if it has economic value. However, adversaries can establish a less hostile relationship even after violent territorial transfers if new borders are drawn in accordance with previous administrative frontiers (Carter and Goemans, 2011, 2014). Although violent territorial transfer appears to have a negative impact on peace, its effect on the resumption of war, especially its relation to failure of renegotiation, has not yet been properly investigated. To fill the gap, this study analyzes the durability of ceasefires following wars resulting in territorial acquisition as opposed to those without, an approach contrasting that of Tir (2003), who compares violent territorial transfer with a peaceful one and focuses on their impact on militarized disputes. Moreover, no studies have yet explicitly decomposed the impact of multiple mechanisms by which territorial acquisition could possibly affect ceasefire duration, for instance, the contiguity of adversaries, the salience of issues, and future power shifts. Without such an analysis, it is difficult to identify what type of territories pose threats to post-war peace. 4 Territorial acquisition and recurrent war War resumes because adversaries fail to revise their ceasefire peacefully. According to the literature on war, lack of commitment to a negotiated settlement is one of the reasons why adversaries resort to force rather than resolve their dispute through negotiation. The problem arises when adversaries are in a situation where they are unable to commit credibly to a negotiated agreement. One such situation is when the balance of power between adversaries shifts. The growing party is unable to make a credible commitment because it will be able to coerce the opposing party to accept a new deal that is more favorable to the growing party in the future, and the opponent knows it (e.g. Fearon, 1995; Powell, 2006). Consequently, the relatively declining party may choose to wage a preventive war rather than accept a negotiated deal that may be dishonored in the future. The same logic applies to the resumption of war; adversaries fail to revise their ceasefire peacefully if their power balance is expected to change with the passage of time. Territorial acquisition is one such factor generating shifts in the balance of power during a ceasefire. Territories seized in war are commonly believed to tip the post-war power balance in favor of the party that has gained them (hereafter, the victor). They encourage the victor’s power growth in various ways: by serving strategic purposes, empowering the victor through an endowment of natural resources, and offering the victor industries and markets. In each case, the acquired territory benefits the victor in terms of post-war power growth. The expectation of future power shifts caused by territorial acquisition then reduces the window of opportunity for a negotiated revision of the ceasefire and renders the ceasefire a failure. Anticipating the victor’s power growth, the loser is motivated to wage a war and recapture the lost territory before the victor fully exploits its value, which concomitantly stimulates the victor’s preventive actions. The key here is that territorial acquisition creates expectations among adversaries for future power shifts. Therefore, a commitment problem due to expectations of future power shifts is distinct from incentives to renegotiate, provoked by power shifts that have actually occurred, as was proved by Werner (1999). Not only causing a commitment problem, territorial acquisition also makes war a very attractive option because a successful restoration of territory will considerably change the tide of power shifts. Thus, the first hypothesis is as follows: Hypothesis 1 (Territorial Acquisition):Ceteris paribus, when a disputant has seized territory from the opponent during a preceding war, ceasefires tend to be more short-lived. Territories vary in their potential to empower the victor depending on their size and nature. Ceasefires are generally more vulnerable when territories provide the victor with greater potential for power growth. The victor that has seized territories with higher utility is expected to become more threatening as time passes; consequently, adversaries become less patient in revising an existing ceasefire. Thus, conflict is more likely to resume when the victor seizes a greater amount of resources and becomes more powerful over the course of a ceasefire. Moreover, territories differ in their exchangeability with regard to military power and potential benefits. Specifically, seized territory causes long-term power shifts if it contains industrial urban areas. Territory with value in industrial cities is not readily exchangeable with military power, but it does offer great potential benefits in the long term, as investments of capital, labor, and technology yield increasingly greater economic growth, which can be eventually converted into military power. Although the acquisition of strategically important territory also stimulates a shift in the military balance between adversaries, seizure of industrial cities is key to a long-term shift in power balance. Thus, more densely populated territories comprising industrial cities are expected to be more detrimental to ceasefires when acquired in the preceding war. Hypothesis 2a (Potential Utility of Acquired Territories I):Ceteris paribus, the greater the amount of territory acquired from the opponent during a preceding war, the more short-lived the ceasefires. Hypothesis 2b (Potential Utility of Acquired Territories II):Ceteris paribus, the more densely populated the acquired territory, the more short-lived the ceasefires. The following sections test these hypotheses against data on interstate ceasefires that have occurred since World War II and illustrate their underlying logic in relation to the Sino-Vietnamese conflicts. 5 Research design The hypotheses mentioned above are tested against data on ceasefires following interstate wars between 1946 and 1997, originally compiled by Fortna (2004).2 The analysis involves two parts. The first part tests the theoretical claim that territorial acquisition is one of the barriers to achieving durable ceasefires (Hypothesis 1). Next, it examines the link between ceasefire duration and characteristics of acquired territories (Hypotheses 2a and 2b). 5.1 Data Fortna’s dataset covers 48 ceasefires following 26 full-scale interstate wars between 1946 and 1997. In the dataset, a ceasefire is defined as ‘an end to or break in the fighting, whether or not it represents the final end of the war’ (Fortna, 2004, p. 45). Each case was based on the pair of countries involved in a conflict, dividing multilateral wars into several cases of principal belligerents. Each case rendered a series of observations representing each year or less after a ceasefire, so that time-varying covariates could be included in the analysis. In total, the 48 ceasefires generated 876 observations. Although new wars occurred and some ceasefires broke down after 1997, the analysis period was not extended beyond this point because some key variables were unavailable for the later wars. 5.2 Dependent variable The dependent variable of the analysis is the ceasefire duration, ‘measured from the date of the ceasefire to the start of another COW [Correlates of War] war between the same two belligerents, if there was one’ (Fortna, 2004, p. 48). Among the 48 ceasefires, war resumed in 22 cases, including the First Kashmir War and the First Turco-Cypriot War. In the other 26 cases, peace lasted beyond the time horizon of the analysis, and thus it was right-censored as of 1998. 5.3 Explanatory variables The first part of the analysis examines how territorial acquisition affects ceasefire duration. Each ceasefire was thus checked using the territorial change dataset of the COW Project to determine whether its preceding war resulted in any territorial acquisition (Tir et al., 1998). A binary variable, territorial change, was then created and coded as 1 when any violent territorial acquisition had been associated with a ceasefire, and as 0 if otherwise. In addition, two variables were constructed for capturing the varying degrees to which the seized territory is expected to empower the victor. The first factor considered is the size of the seized territory. The area of exchanged territory was included in the analysis based on the ‘Area of Unit Exchanged in Square Kilometers’ listed in Tir et al. (1998). Area was simply defined as the natural log of the land area obtained by the new owner through war. Observations from cases in which no territorial transfer occurred were assigned a value of 0. The second factor is the population density of the acquired territory. The density of exchanged territory was calculated by dividing the population residing therein by the area of the corresponding territory using the ‘Area of Unit Exchanged in Square Kilometers’ and the ‘Population of Unit Exchanged’ as listed in Tir et al. (1998). The natural log of the calculated density was used to create the variable density, thereby adjusting its diminishing marginal effect. Observations from cases in which no territorial change occurred were assigned a value of 0. 5.4 Other covariates The analyses also considered the following covariates. Related to territory, variables that represent the geographic proximity of adversaries and the issue at stake were included. The binary variable, contiguity, is coded 1 if adversaries are contiguous by land or separated by less than 150 miles of water. To measure whether the war was over contentious issues, a binary variable denoting whether adversaries’ existence was the major stake in the war was included. The variable was created by Fortna (2004) based on the ‘gravity of value threatened’ in the International Crisis Behavior (ICB) dataset (Brecher and Wilkenfeld, 2000). Because territory is one of the essential elements of a sovereign state, conflicts over existence necessarily involve territorial disagreements. The analysis also incorporated territorial integrity at stake, which is coded as 1 when one or both adversaries sought to seize a portion, but not the entirety, of the opponent’s territory. This variable was also derived by Fortna (2004) from the ICB dataset. In addition to these factors, the strength of ceasefire agreements was included, following Fortna (2004), who argued that a strong formal agreement contributes to durable peace. Fortna represented the strength of agreements using the agreement strength index, a composite measurement adding the following 10 factors: formal acceptance of a ceasefire proposal; withdrawal of military forces to status ante; formation of demilitarized zones; agreement on arms control procedure; involvement of peacekeeping forces; strength of enforcement measures by third-party involvement; strength of internal enforcement effort; number of paragraphs in the agreement text; agreement on confidence-building measures; and the level of dispute-resolution efforts. Following Werner and Yuen (2005), battle consistency and interrupted war were also incorporated into the analysis. The variable of battle consistency is measured by the duration over which one side is consistently winning battles toward the end of the war (the ‘last tide’ of the war) as a proportion of the total duration of the war. Interrupted war is a binary variable that takes 1 if ‘(1) there is clear evidence of significant third-party pressure to ceasefire, and (2) the ceasefire is invoked before the parties negotiate the final settlement’ (Werner and Yuen, 2005, p. 273), and 0 if otherwise. The indicator of actual shifts in power is represented by the post-war change in relative capabilities as created by Werner (1999) from the Composite Index of National Capability (CINC) score of the COW Project (Singer et al., 1972). This variable is measured by the absolute difference in adversaries’ growth rates. A country’s growth rate is obtained by the annual growth of its capability as a proportion of its capability the previous year. The variable is then obtained by taking the difference between the growth rates of the adversaries. Therefore, the larger the value of the change in relative capabilities, the more significant is the shift in the adversaries’ balance of power. Furthermore, a military tie was coded as 1 when a war ended in a tie and as 0 if otherwise. The cost of war was measured by the natural log of the total number of battle deaths. Last, adversaries’ history of conflict was measured by the number of militarized interstate disputes that occurred between the dyad’s members divided by the number of years that the dyad had existed in the interstate system when war started. These three variables were retrieved from the dataset provided by Fortna (2004). Descriptive statistics of all the variables are reported in Table 1. Table 1 Descriptive statistics Variables  Mean  Min  Max  Std. Dev.  Obs.  Territorial acquisition  0.15  0.00  1.00  0.36  891  Area  1.36  0.00  12.07  3.24  891  Density  0.68  0.00  7.36  1.69  891  Contiguous  0.66  0.00  1.00  0.48  876  Existence at stake  0.26  0.00  1.00  0.44  876  Territorial integrity at stake  0.17  0.00  1.00  0.38  876  Agreement strength  4.36  0.00  9.50  2.39  876  Battle consistency  0.42  0.00  1.00  0.43  876  Interrupted war  0.28  0.00  1.00  0.45  876  Change in relative capabilitiesa  0.12  0.00  5.43  0.28  770  Military tie  0.45  0.00  1.00  0.50  876  Cost of war  9.30  5.60  14.04  2.35  876  History of conflict  0.75  0.00  3.41  0.83  876  Variables  Mean  Min  Max  Std. Dev.  Obs.  Territorial acquisition  0.15  0.00  1.00  0.36  891  Area  1.36  0.00  12.07  3.24  891  Density  0.68  0.00  7.36  1.69  891  Contiguous  0.66  0.00  1.00  0.48  876  Existence at stake  0.26  0.00  1.00  0.44  876  Territorial integrity at stake  0.17  0.00  1.00  0.38  876  Agreement strength  4.36  0.00  9.50  2.39  876  Battle consistency  0.42  0.00  1.00  0.43  876  Interrupted war  0.28  0.00  1.00  0.45  876  Change in relative capabilitiesa  0.12  0.00  5.43  0.28  770  Military tie  0.45  0.00  1.00  0.50  876  Cost of war  9.30  5.60  14.04  2.35  876  History of conflict  0.75  0.00  3.41  0.83  876  aTime-varying covariate. Agreement strength might change over time if follow-up agreements are reached. Table 1 Descriptive statistics Variables  Mean  Min  Max  Std. Dev.  Obs.  Territorial acquisition  0.15  0.00  1.00  0.36  891  Area  1.36  0.00  12.07  3.24  891  Density  0.68  0.00  7.36  1.69  891  Contiguous  0.66  0.00  1.00  0.48  876  Existence at stake  0.26  0.00  1.00  0.44  876  Territorial integrity at stake  0.17  0.00  1.00  0.38  876  Agreement strength  4.36  0.00  9.50  2.39  876  Battle consistency  0.42  0.00  1.00  0.43  876  Interrupted war  0.28  0.00  1.00  0.45  876  Change in relative capabilitiesa  0.12  0.00  5.43  0.28  770  Military tie  0.45  0.00  1.00  0.50  876  Cost of war  9.30  5.60  14.04  2.35  876  History of conflict  0.75  0.00  3.41  0.83  876  Variables  Mean  Min  Max  Std. Dev.  Obs.  Territorial acquisition  0.15  0.00  1.00  0.36  891  Area  1.36  0.00  12.07  3.24  891  Density  0.68  0.00  7.36  1.69  891  Contiguous  0.66  0.00  1.00  0.48  876  Existence at stake  0.26  0.00  1.00  0.44  876  Territorial integrity at stake  0.17  0.00  1.00  0.38  876  Agreement strength  4.36  0.00  9.50  2.39  876  Battle consistency  0.42  0.00  1.00  0.43  876  Interrupted war  0.28  0.00  1.00  0.45  876  Change in relative capabilitiesa  0.12  0.00  5.43  0.28  770  Military tie  0.45  0.00  1.00  0.50  876  Cost of war  9.30  5.60  14.04  2.35  876  History of conflict  0.75  0.00  3.41  0.83  876  aTime-varying covariate. Agreement strength might change over time if follow-up agreements are reached. 5.5 Statistical model Ceasefire duration was examined through duration analysis to enable proper treatment of the right-censored processes. The current study employed the Weibull model, assuming that the baseline probability of war resumption monotonically changes over time. The same analyses using the Cox model were run here to reduce the risk of arriving at a conclusion sensitive to assumptions about the shape of the baseline hazard, although a residual analysis suggests that the Weibull model better fits the data. Corresponding results are reported in the Supplementary Material. 6 Results The results of the analysis are reported in the form of coefficients that show the marginal effects of variables on the hazard function, that is, the instantaneous probability of failure at a certain point in time conditional on survival up to that point. A factor with a negative coefficient decreases the instantaneous probability of ceasefire breakdown and thus extends its duration. For substantive interpretation, the hazard ratio is also reported: the instantaneous probability of ceasefire failure increases by the factor of the ratio, given survival up to that point. Predictors of the hazard ratio exceeding a value of 1 thus increase the likelihood of war resumption. 6.1 The effect of territorial acquisition The main results of the duration analysis on Hypothesis 1 are reported in Table 2. Model 1 examines the effect of territorial acquisition by controlling for factors such as the characteristics of war, the form of ceasefire agreements, and post-war shifts in the balance of power. Territorial acquisition is found to have a disturbing effect on ceasefire duration. When one or both adversaries acquire new territory through war, a succeeding ceasefire, when achieved, faces a higher risk of failure. The effect of territorial change is robust across different models and model specifications, as is discussed later and in the Supplementary Material. In all models, territorial acquisition decreases ceasefire duration. Table 2 Ceasefire duration and territorial acquisition (Weibull model)   (1)  (2)  (3)  (4)    coef.  coef.  coef.  coef.  Territorial acquisition  2.16***  2.57***      (0.23)  (0.70)      Area      0.29***        (0.05)    Density        0.38*        (0.15)  Contiguous    1.29  1.10  1.60    (1.10)  (0.98)  (1.10)  Existence at stake    2.50*  2.45*  2.60*    (1.22)  (1.11)  (1.30)  Territorial integrity at stake    −3.53***  −3.84***  −2.96***    (0.66)  (0.54)  (0.86)  Agreement strength  −0.23  −0.35  −0.32  −0.37  (0.13)  (0.23)  (0.22)  (0.29)  Battle consistency  −1.85  −7.31*  −7.69**  −6.78*  (1.03)  (2.85)  (2.89)  (3.09)  Interrupted war  2.39***  6.94***  7.13***  6.02***  (0.49)  (0.95)  (1.04)  (1.30)  Change in relative capabilities  0.88***  1.29***  1.24***  1.33***  (0.15)  (0.31)  (0.29)  (0.38)  Military tie  2.25*  4.52  4.17  4.40  (0.92)  (2.81)  (2.36)  (2.69)  Cost of war  −0.36*  −0.87***  −0.87***  −0.94***  (0.16)  (0.13)  (0.14)  (0.15)  History of conflict  0.78***  0.60  0.68  0.73  (0.17)  (0.62)  (0.54)  (0.61)  P  0.18  0.33**  0.32**  0.26*  (0.13)  (0.12)  (0.13)  (0.12)  N  770  770  770  770  BIC  137.9  130.0  130.2  141.0  Log-likelihood  −35.7  −31.8  −31.9  −34.0    (1)  (2)  (3)  (4)    coef.  coef.  coef.  coef.  Territorial acquisition  2.16***  2.57***      (0.23)  (0.70)      Area      0.29***        (0.05)    Density        0.38*        (0.15)  Contiguous    1.29  1.10  1.60    (1.10)  (0.98)  (1.10)  Existence at stake    2.50*  2.45*  2.60*    (1.22)  (1.11)  (1.30)  Territorial integrity at stake    −3.53***  −3.84***  −2.96***    (0.66)  (0.54)  (0.86)  Agreement strength  −0.23  −0.35  −0.32  −0.37  (0.13)  (0.23)  (0.22)  (0.29)  Battle consistency  −1.85  −7.31*  −7.69**  −6.78*  (1.03)  (2.85)  (2.89)  (3.09)  Interrupted war  2.39***  6.94***  7.13***  6.02***  (0.49)  (0.95)  (1.04)  (1.30)  Change in relative capabilities  0.88***  1.29***  1.24***  1.33***  (0.15)  (0.31)  (0.29)  (0.38)  Military tie  2.25*  4.52  4.17  4.40  (0.92)  (2.81)  (2.36)  (2.69)  Cost of war  −0.36*  −0.87***  −0.87***  −0.94***  (0.16)  (0.13)  (0.14)  (0.15)  History of conflict  0.78***  0.60  0.68  0.73  (0.17)  (0.62)  (0.54)  (0.61)  P  0.18  0.33**  0.32**  0.26*  (0.13)  (0.12)  (0.13)  (0.12)  N  770  770  770  770  BIC  137.9  130.0  130.2  141.0  Log-likelihood  −35.7  −31.8  −31.9  −34.0  Note: Standard errors are clustered by conflict and reported in parentheses. * P < 0.05, **P < 0.01, ***P < 0.001. Table 2 Ceasefire duration and territorial acquisition (Weibull model)   (1)  (2)  (3)  (4)    coef.  coef.  coef.  coef.  Territorial acquisition  2.16***  2.57***      (0.23)  (0.70)      Area      0.29***        (0.05)    Density        0.38*        (0.15)  Contiguous    1.29  1.10  1.60    (1.10)  (0.98)  (1.10)  Existence at stake    2.50*  2.45*  2.60*    (1.22)  (1.11)  (1.30)  Territorial integrity at stake    −3.53***  −3.84***  −2.96***    (0.66)  (0.54)  (0.86)  Agreement strength  −0.23  −0.35  −0.32  −0.37  (0.13)  (0.23)  (0.22)  (0.29)  Battle consistency  −1.85  −7.31*  −7.69**  −6.78*  (1.03)  (2.85)  (2.89)  (3.09)  Interrupted war  2.39***  6.94***  7.13***  6.02***  (0.49)  (0.95)  (1.04)  (1.30)  Change in relative capabilities  0.88***  1.29***  1.24***  1.33***  (0.15)  (0.31)  (0.29)  (0.38)  Military tie  2.25*  4.52  4.17  4.40  (0.92)  (2.81)  (2.36)  (2.69)  Cost of war  −0.36*  −0.87***  −0.87***  −0.94***  (0.16)  (0.13)  (0.14)  (0.15)  History of conflict  0.78***  0.60  0.68  0.73  (0.17)  (0.62)  (0.54)  (0.61)  P  0.18  0.33**  0.32**  0.26*  (0.13)  (0.12)  (0.13)  (0.12)  N  770  770  770  770  BIC  137.9  130.0  130.2  141.0  Log-likelihood  −35.7  −31.8  −31.9  −34.0    (1)  (2)  (3)  (4)    coef.  coef.  coef.  coef.  Territorial acquisition  2.16***  2.57***      (0.23)  (0.70)      Area      0.29***        (0.05)    Density        0.38*        (0.15)  Contiguous    1.29  1.10  1.60    (1.10)  (0.98)  (1.10)  Existence at stake    2.50*  2.45*  2.60*    (1.22)  (1.11)  (1.30)  Territorial integrity at stake    −3.53***  −3.84***  −2.96***    (0.66)  (0.54)  (0.86)  Agreement strength  −0.23  −0.35  −0.32  −0.37  (0.13)  (0.23)  (0.22)  (0.29)  Battle consistency  −1.85  −7.31*  −7.69**  −6.78*  (1.03)  (2.85)  (2.89)  (3.09)  Interrupted war  2.39***  6.94***  7.13***  6.02***  (0.49)  (0.95)  (1.04)  (1.30)  Change in relative capabilities  0.88***  1.29***  1.24***  1.33***  (0.15)  (0.31)  (0.29)  (0.38)  Military tie  2.25*  4.52  4.17  4.40  (0.92)  (2.81)  (2.36)  (2.69)  Cost of war  −0.36*  −0.87***  −0.87***  −0.94***  (0.16)  (0.13)  (0.14)  (0.15)  History of conflict  0.78***  0.60  0.68  0.73  (0.17)  (0.62)  (0.54)  (0.61)  P  0.18  0.33**  0.32**  0.26*  (0.13)  (0.12)  (0.13)  (0.12)  N  770  770  770  770  BIC  137.9  130.0  130.2  141.0  Log-likelihood  −35.7  −31.8  −31.9  −34.0  Note: Standard errors are clustered by conflict and reported in parentheses. * P < 0.05, **P < 0.01, ***P < 0.001. Model 2 in Table 2 incorporates the geographic proximity of adversaries and the issues at stake into Model 1. The effect of territorial acquisition remains consistent even when these factors are controlled for, that is, territorial change results in shorter ceasefires regardless of adversaries being contiguous or of existential and territorial issues at stake. Moreover, in Models 1 and 2, the effect of territorial acquisition is observed with controlling for post-war shifts in relative capabilities. According to the results of Model 2, among variables relating to territories, the geographical contiguity of adversaries does not affect ceasefire durability in a significant manner. In contrast, it was found that ceasefires are less likely to endure when an adversary’s existence as a sovereign state is threatened, although the magnitude of the variable’s substantive effect might be moderate, as is explained later. On the other hand, ceasefires are maintained for a longer period if their territorial integrity, but not existence, is challenged. The contrasting effects of these variables suggest that the more contentious the issue at stake is, the more likely the adversaries are to return to war. Additionally, interrupted wars are found to be associated with shorter ceasefires and wars with consistent battle outcomes tend to be followed by a more durable peace, as Werner and Yuen (2005) demonstrated. However, the analysis presented here is unable to determine whether third parties caused unstable ceasefires or whether particularly unstable ceasefires attracted third-party intervention. Moreover, consistent with Werner (1999), the post-war change in relative capabilities increases the risk of ceasefire breakdown. Furthermore, the greater the cost of the preceding war, the more durable the subsequent ceasefire, which is consistent across all model specifications. Figure 1 visually presents the hazard ratio of each variable estimated from Model 2. It shows that the occurrence of territorial acquisition considerably increases the probability of war resumption. The instantaneous probability of war resumption following territorial acquisition is estimated to be 13 times higher (95% confidence interval [CI] [3.29, 51.66]) than observed in cases involving no territorial change, ceteris paribus. Moreover, the resumption of war appears more likely when existence is at stake, showing the estimated hazard ratio of 12.2 (95% CI [1.12, 131.77]). However, from a substantive point of view, the existential issue might have only a moderate effect on ceasefire duration, taking into account its large standard error. The influence of a shift in relative capabilities appears moderate, considering that a one-unit increase of the variable signifies a very large shift in the balance of power such that the power of one side is doubled in a span of one year. Figure 1 View largeDownload slide Effects on the likelihood of war resumption. The circles indicate the hazard ratio of each predictor and the line around them shows the 95% confidence interval. The scale of the horizontal axis is adjusted so that the behavior of all variables is substantively and intuitively understandable. Curtailed parts are denoted by vertical double lines. Created from Model 2 in Table 2. Figure 1 View largeDownload slide Effects on the likelihood of war resumption. The circles indicate the hazard ratio of each predictor and the line around them shows the 95% confidence interval. The scale of the horizontal axis is adjusted so that the behavior of all variables is substantively and intuitively understandable. Curtailed parts are denoted by vertical double lines. Created from Model 2 in Table 2. Looking at the left side of the figure, territory at stake and battle consistency greatly contribute to stable ceasefire outcomes. In other words, war is less likely to resume if it was fought over issues of territorial integration, annexation, and separatism, but not the existence of country. Adversaries are also less likely to return to war if they are able to infer their balance of power accurately from consistent battle outcomes of the preceding wars. The cost of war also has a non-negligible effect. To give an example of substantive interpretation, the figure shows the effect of the variable when the total number of battle deaths increases from 2,980 to 8,100. 6.2 The potential utility of exchanged territory Hypotheses 2a and 2b are then examined to analyze how the effects of territorial acquisition vary depending on the characteristics of territorial change. Model 3 in Table 2 provides the estimation results of the duration analysis incorporating the size of the acquired territories in occurrences of territorial acquisition. Consistent with the hypothesis, ceasefire breakdown is positively related to the size of the area exchanged between adversaries. The hazard ratio for area is 1.33, with 95% CI [1.20, 1.47]. To give an example, a one-unit increase in area is roughly equivalent to the difference between the Golan Heights (7.4) and West Bank (8.7) acquired by Israel from Syria and Jordan, respectively, through the Six-Day War. Considering that the observed value of area ranges from 5.6 to 12.1 with a mean value of 8.8 for cases involving territorial change, the result suggests that the effect of the variable is substantively important. Model 4 in Table 2 incorporates the density of exchanged territory while dropping area because of their high collinearity (the correlation is 0.91). The result shows that ceasefires are less durable when there is an exchange of more densely populated territory. In cases involving territorial change, the observed value of density varies from 2.1 to 7.4, with a mean value of 4.4. For instance, the density of the Golan Heights was 3.6 and that of the West Bank was 4.7. Hence, the hazard ratio of 1.47, with 95% CI [1.09, 1.98], indicates the variable’s non-negligible effect on ceasefire duration. Next, a graphical representation in Fig. 2 depicts how the population density of the acquired territory affects the likelihood of war resumption over time. The plots report the estimated hazard function (the left column) and the estimated survivor function (the right column) for three scenarios: ceasefires following (i) no territorial acquisition, (ii) acquisition of territories with average density, and (iii) acquisition of territories with the highest density recorded in the dataset. The 95% CIs are not reported for the sake of readability. According to the illustrations, supposing that 10 years (3,650 days) have passed since the war terminated without any territorial change, the likelihood of the ceasefire enduring until then is slightly higher than one following the acquisition of most densely populated territories. After 20 years (7,300 days), the possibility of a ceasefire preceded by the transfer of most densely populated territories remaining intact is much smaller than one involving no territorial change. Figure 2 View largeDownload slide Estimated hazard function and survivor function for density. All other variables are held at their mean values. The 95% CIs are not shown in the plots for the sake of readability. Created from Model 4 in Table 2. Figure 2 View largeDownload slide Estimated hazard function and survivor function for density. All other variables are held at their mean values. The 95% CIs are not shown in the plots for the sake of readability. Created from Model 4 in Table 2. 6.3 Discussion The first part of the analysis demonstrates that territorial acquisition increases the likelihood of war resumption and that this effect is not attributed to the contiguity of adversaries, the importance of territorial issues, and the actual shift in relative capabilities. Therefore, the findings suggest that territorial acquisition increases the likelihood of war resumption because it generates a commitment problem by shaping adversaries’ expectations for future power shifts. Potential concerns with the above analysis would be that the results depend on methodological assumptions and influential observations in the dataset. To address the first concern, the same analyses are run using the Cox model. A series of robustness checks are also conducted to control for the effect of possible outliers and influential observations, such as the Sinai War between Israel and Egypt and the First Turco-Cypriot War. The results remain unchanged, and are reported in the Supplementary Material. The second half of the analysis shows that the effect of territorial acquisition varies by territories’ potential utility to empower the victor. In other words, the seizure of large and densely populated territories through war causes less durable ceasefires. An alternative interpretation of the results may be that governing the local population inhabiting a newly acquired territory is more difficult than occupying an underpopulated land. The local population may organize rebellion(s) or work as operatives when the territory’s former ruler wages a counter-offense. More heavily populated territories are thus more difficult to secure control over and are more susceptible to war resumption. Although this study alone cannot reject this possibility, this alternative argument serves as a limited account of failed ceasefires, considering that they are not durable when democratic countries are victorious.3 When lack of effective governance over the occupied territory is the major barrier to a stable ceasefire, democratic countries should be able to promote post-war peace because they tend to have greater state capacity and to provide better governance. However, the supplemental analysis shows the opposite, implying that the main barrier is the victor’s potential for power growth rather than lack of governance. This statistical analysis illuminates the links between territorial acquisition, a commitment problem, and war recurrence. The next section applies this argument to cases in the Asia and Pacific region: the Sino-Vietnamese conflicts over the land border and in the South China Sea. The analysis also complements the statistical analysis, demonstrating that the patterns of their conflicts were affected by the utility of the seized territories in each region. 7 Illustration: Sino-Vietnamese conflicts The People’s Republic of China (hereafter, China) and the Socialist Republic of Vietnam (hereafter, the SRV or Vietnam) are known for their long-time rivalry. Although both countries uphold a communist ideology, this commonality has not implied shared interests even during the Cold War. Their disputes are most intensely militarized along their land border, over which they fought two full-scale wars in 1979 and 1987. They have also engaged in less intensified and sporadic militarized disputes over the South China Sea, where armed clashes have happened in 1979 and 1988, and then frequently since 2010. Without doubt, the ceasefires between 1979 and the late 1980s at land and sea greatly differ in terms of the intensity of military actions (wars on land and militarized disputes at sea) and are not readily comparable. However, contrasting them will shed light on the above-mentioned dynamics while holding other factors, such as the presence of the Soviet Union and the ongoing civil war in Cambodia, constant. The First Sino-Vietnamese War broke out on 14 February 1979, when China invaded Vietnam beyond the disputed land border. After China’s extensive invasion and firm resistance by the Vietnamese troops, war was terminated when China announced its withdrawal from the occupied territory in March 1979. Historical accounts disagree on who won the war. Some suggest that ‘[i]t was a total defeat for China’ (Bercovitch and Jackson, 1997, p. 189), whereas others claim that Vietnam lost in terms of war damage (Hood, 1992, p. 59). The war imposed considerable costs on both sides; more than 50,000 soldiers lost their lives and 500,000 people were killed in the war, although contradictory figures are provided by various sources (Hood, 1992, p. xv; Bercovitch and Jackson, 1997, p. 189). Although Beijing and Hanoi initiated peace negotiations immediately after the war, fighting along the border resumed soon after. By early 1981, the number of violent incidents committed by the Chinese allegedly exceeded 4,000 cases, and China accused Vietnam of instigating more than 2,000 armed provocations and the killing of 240 Chinese (Chang, 1986, pp. 79–80). The fighting culminated in January 1987 through a conflict referred to as the Second Sino-Vietnamese War. Confrontations lasted until March 1987, resulting in as many as 3,000, mostly Chinese, fatalities (Bercovitch and Jackson, 1997, p. 216). The dispute over the South China Sea contrasts with the failed ceasefire over the land border. The first confrontation at sea occurred between China and South Vietnam in 1974 when China seized the entire Paracel Islands, and South Vietnam in turn occupied part of the Spratly Islands. After reunification in 1975, the SRV inherited territorial claims and control over the Spratlys from South Vietnam. Then, the first incident between China and the SRV occurred in April 1979 when ‘China seized a Vietnamese reconnaissance boat allegedly conducting sabotage activities in the Paracels’ (Chang, 1986, p. 79). Fighting was limited in intensity and no significant military confrontation was reported in the South China Sea until 1988. Stability broke down in 1988, when a Chinese naval vessel attacked a Vietnamese gunboat near the Spratly Islands, resulting in more than 70 casualties. Thus, the ceasefire terminating the First Sino-Vietnamese War failed quickly, whereas stability was preserved after the armed clash over the South China Sea. As noted above, the two incidents in 1979 significantly differ in intensity and there might be room for argument regarding whether a ‘ceasefire’ occurred in the South China Sea. Nevertheless, it is remarkable that the South China Sea incident did not escalate or recur as opposed to that over the land border. This contrast is even more intriguing considering that a costly and uninterrupted war is commonly considered less prone to recurrence. This study focuses on the utility of seized territories at land and sea. During confrontations along the land border, no large territorial change was reported4; however, China and Vietnam exchanged small but important territories: strategically situated hills along the border. These hills served as strategically important locations from which the occupier shelled the opponent in conflicts that ‘often took the form of artillery duels’. For instance, in mid-1981, ‘the two sides engaged several thousand troops and fought … over a few hills in the Fakashan area of China’s Guangxi region’ (Chang, 1986, pp. 79–80). Similarly, in 1984, ‘Chinese troops proceeded to seize a couple of hilltops that China claimed to be inside Chinese territory but which had been occupied by Vietnam since the war and repeatedly used to shell Chinese border villages for five straight years.’ The offense invoked unprecedented levels of counterattacks by Vietnam ‘in an attempt to recapture the strategic high grounds that Hanoi claimed were inside Vietnamese territory’ (Chang, 1986, pp. 80–81). The main stakes in the South China Sea were underwater oil and mineral reserves. Although the islands also had strategic value, the archipelagos ‘have attracted new attention since the late 1960s when these areas became known as rich in undersea oil deposits’ (Chang, 1986, pp. 14–16). This ambition for oil and minerals was fueled by the global trend of setting exclusive economic zones (EEZs) to include maritime resources. Indeed, when the SRV took over Spratly Islands, ‘Hanoi reportedly had also started to conduct talks with foreign oil companies on the resumption of oil exploration and exploitation in areas adjacent to this archipelago’ (Chang, 1986, p. 28). Therefore, China and Vietnam exchanged two vital territories: strategically located hills and islands containing maritime resources. Although they did not exchange industrial cities as posed in the theoretical argument, the two territories had different implications for future power shifts during the period between 1979 and 1988 as described in the theory: strategically located hills had greater utility in terms of stimulating power shifts than maritime resources, which both China and Vietnam lacked the capacity to exploit. The value of these hills was readily consumable by both countries that could improve their military position by simply consolidating their control over the hills (Carter, 2010). The vitality of these hills was compounded by the underdeveloped economic and military capabilities of the adversaries. Vietnam’s economic performance was unsatisfactory during this period due to the long Vietnam War, poor administrative capacities, and corruption among the political elite. Its economic and military capacities mostly depended on assistance from the Soviet Union (Hood, 1992, p. 66). Although the new leadership that took office in 1986 decided to turn away from a centrally planned economy and toward a market economy through a process of ‘Doi Moi (economic renovation)’, its economic growth was limited during the 1980s. In China, similarly, the economy and society, at large, were devastated by the Cultural Revolution. Although conditions began to change in 1978 when Deng Xiaoping took leadership and applied economic reforms, more than half the population still lived below the poverty line as reported in 1981 (Ravallion and Chen, 2007). China’s gross domestic product (GDP) per capita remained at less than $2,000 (in 1990, in US dollars) until 1992.5 For adversaries with limited capacity, the hills were conceived of as an essential source of future power shifts. In contrast to the hills, both China and Vietnam lacked capacities to exploit offshore oil fields in the deep waters of the South China Sea. China was only able to exploit offshore oil from the shallow waters of the Gulf of Tonkin (Lo, 1989, pp. 60–61). To improve its capacity, under the new leadership of Deng Xiaoping, China established the China National Offshore Oil Corporation (CNOON) in 1982, and signed a number of contracts with foreign companies throughout the 1980s (Chang, 1986, p. 72). However, by the early 1980s, their activities were still limited to shallow sea areas in the Tonkin Gulf and off the coast of Hainan Island (Lo, 1989, p. 61). By the end of 1982, China had extended its focus to the South China Sea and installed oil rigs there. Nevertheless, it was not able to initiate actual oil production, even in the Gulf, until 1986 (Hood, 1992, p. 132). Similarly, Vietnam suffered from lack of technological capabilities to extract maritime resources. Although Vietnam and the Soviet Union allegedly installed three or four oil rigs in the Gulf as early as in 1975 and 1976, ‘it was learned that probably only two were actually working’ in 1984. Even in the shallow waters of the Gulf, it was not until April 1986 that ‘the Vietnamese-Soviet drill crews discovered their first major producer … in the Bach Ho (White Tiger) field’. Disappointed by the result, Vietnam considered forging a contract with Western companies, but the USSR did not allow Hanoi to do so (Hood, 1992, pp. 131–133). Interestingly, in contrast to the stability of the South China Sea, a couple of incidents occurred in the Gulf of Tonkin during this period. Quarrels in the Gulf were clearly related to oil exploration in the area: Hanoi warned that ‘if China continued to search for oil in the disputed waters in the Gulf of Tonkin, companies … would have to bear the consequences’ (Hood, 1992, p. 132). In fact, 14 armed Vietnamese boats clashed with Chinese fishing trawlers in the Gulf in 1982. Similarly, in 1983, two Chinese fishing boats operating in the Gulf were attacked by Vietnamese forces (Chang, 1986, pp. 76–77). Also interesting is the fact that the armed clash of 1988 over the South China Sea was preceded by China’s discovery of oil in the Spratlys. Although ‘[u]ntil late 1987, the Spratly potential was largely unknown and unexplored’, the situation drastically changed in November 1987 when ‘China announced it had discovered rich oil and gas deposits under the Spratlys after a survey of 180,000 square kilometers.’ Thus, the successful discovery of the oil reserve appears to have disrupted stability in the area. Hence, the above illustrations suggest that territorial acquisition and the utility of seized territories explain the dynamics of the Sino-Vietnamese conflicts occurring between 1979 and the late 1980s. 8 Conclusions This study demonstrated that territorial acquisition increases the risk of war resumption because it shapes expectations for future shifts in the balance of adversaries’ power and consequently makes peaceful renegotiation more likely to fail because of a commitment problem. The duration analysis of ceasefires following wars occurring between 1946 and 1997 shows that territorial acquisition renders likely resumption of war. Moreover, some types of territorial acquisition pose a more serious commitment problem than others based on the characteristics of the transferred territory: the size and the population density. In short, territorial acquisition that would likely produce a larger future shift in power was demonstrated to be significantly more detrimental to lasting ceasefire. The findings presented here thus contribute to the literature on recurrent war in several respects. First, although disputes over territory have been especially conflictual, the effect of territorial acquisition on recurrent war has been insufficiently studied in the literature. This study filled the gap, demonstrating that territorial acquisition has detrimental effects on post-war peace. Moreover, this study also found that the negative effect of territorial acquisition is not completely attributed to the contiguity of adversaries or the vitality of territorial issues. Another explanation, therefore, was offered. Territorial acquisition generates a commitment problem, which obstructs peaceful renegotiation of ceasefires and causes recurrent war. The study also showed that the effect of territorial acquisition varies depending on the utility of the acquired territory. These findings are somewhat contradictory to intuition-based results. The common belief is that a decisive victory contributes to post-war peace. However, if the size of a victory is defined by the volume of the seized territories as opposed to military outcomes, a significant victory may not lead to post-war peace. The implications of this research extend to current incidents between China and Vietnam over the South China Sea. After several years of peace since 1988, controversy reemerged in 2010 when Chinese forces arrested nine fishermen accused of carrying explosives close to the Paracel Islands (The Associated Press, 2010). This event has been followed by several militarized incidents that were apparently motivated by a desire for maritime resources. Although these incidents are similar to those occurring during the 1970s and 1988s, they have become more frequent than before. Viewed in the framework of this study, this increase in the number of armed clashes over the South China Sea is possibly affected by the economic and technological development of both countries in exploiting maritime resources in deep water areas. It is not argued that the mechanism identified in the current study serves as the sole and complete explanation of the Sino-Vietnamese rivalry to date. Rather, it sheds light on the dynamics of their rivalry revolving around issues of territorial acquisition and their economic development, hoping to advance our understanding of their ongoing dispute and to find a way to prevent it from further escalating. Acknowledgements For helpful comments on this project, the author is grateful to Miao-ling Lin Hasenkamp, David Leblang, Jeffrey Legro, Kazuto Ohtsuki, Anna Pechenkina, Todd Sechser, Allan Stam, Changhe Su, Atsushi Tago, three anonymous reviewers, the editors of International Relations of the Asia-Pacific, and all participants of the 2013 Joint Conference of International Studies Association and Peace Science Society, the 2013 Kobe Sakura Meeting, the 2017 Annual Conference of the Midwest Political Science Association, and the Workshop at Fudan University. The author also thanks Virginia Page Fortna for making her dataset available online. Special acknowledgement goes to Daichi Yamada for his research assistance. This research was supported by the JSPS KAKENHI 15K1699805. Supplementary material Supplementary material is available at International Relations of the Asia-Pacific online. Footnotes 1 The list of cases provided by Fortna (2004) is supplemented by the author in reference to the Expanded War Data (v.2.1) listed in Gleditsch (2004). 2 The dataset can be downloaded from http://www.columbia.edu/~vpf4/peacetime.htm (24 February 2018, date last accessed). 3 Corresponding results are reported in the Supplementary Material. 4 No territorial exchange was identified in the dataset. However, this omission does not seriously undermine the validity of the statistical analysis, because at worst it would only cause an underestimation of the effect of territorial change. 5 The Maddison-Project, available at http://www.ggdc.net/maddison/oriindex.htm, 2013 version (24 February 2018, date last accessed). References Beardsley K. ( 2008) ‘ Agreement without peace? International mediation and time inconsistency problems’, American Journal of Political Science , 52( 4), 723– 740. Google Scholar CrossRef Search ADS   Bercovitch J., Jackson R. ( 1997) International Conflict: A Chronological Encyclopedia of Conflicts and Their Management, 1945-1995 . Washington, DC: Congressional Quarterly. Blainey G. ( 1988) Causes of War , 3rd edn. New York: Free Press. Google Scholar CrossRef Search ADS   Brecher M., Wilkenfeld J. ( 2000) A Study of Crisis. Ann Arbor, MI: University of Michigan Press. Bremer S.A. ( 1992) ‘ Dangerous dyads: conditions affecting the likelihood of interstate war, 1816-1965’, Journal of Conflict Resolution , 36( 2), 309– 341. Google Scholar CrossRef Search ADS   Carter D.B. ( 2010) ‘ The strategy of territorial conflict’, American Journal of Political Science , 54( 4), 969– 987. Google Scholar CrossRef Search ADS   Carter D.B., Goemans H.S. ( 2011) ‘ The making of the territorial order: new borders and the emergence of interstate conflict’, International Organization , 65( 2), 275– 310. Google Scholar CrossRef Search ADS   Carter D.B., Goemans H.S. ( 2014) ‘ The temporal dynamics of new international borders’, Conflict Management and Peace Science , 31( 3), 285– 302. Google Scholar CrossRef Search ADS   Chang P. ( 1986) The Sino-Vietnamese Territorial Dispute . New York: Praeger. Diehl P.F., Goertz G. ( 2000) War and Peace in International Rivalry . Ann Arbor, MI: University of Michigan Press. Google Scholar CrossRef Search ADS   Fearon J.D. ( 1995) ‘ Rationalist explanations for war’, International Organization , 49( 3), 379– 414. Google Scholar CrossRef Search ADS   Fortna V.P. ( 2004) Peace Time: Cease-fire Agreements and the Durability of Peace . Princeton, NJ: Princeton University Press. Gleditsch K.S. ( 2004). ‘ A revised list of wars between and within independent states, 1816-2002’, International Interactions , 30( 3), 231– 262. Google Scholar CrossRef Search ADS   Goertz G., Diehl P.F. ( 1992) Territorial Changes and International Conflict . London: Routledge. Google Scholar CrossRef Search ADS   Grieco J.M. ( 2001) ‘ Repetitive military challenges and recurrent international conflicts, 1918–1994’, International Studies Quarterly , 45( 2), 295– 316. Google Scholar CrossRef Search ADS   Hensel P.R. ( 1994) ‘One thing leads to another: recurrent militarized disputes in Latin America, 1816-1986’, Journal of Peace Research , 31( 3), 281. Google Scholar CrossRef Search ADS   Hensel P.R. ( 1996) ‘ Charting a course to conflict: territorial issues and interstate conflict, 1816-1992’, Conflict Management and Peace Science , 15( 1), 43– 73. Google Scholar CrossRef Search ADS   Hensel P.R. ( 2001) ‘Contentious issues and world politics: the management of territorial claims in the Americas, 1816–1992’, International Studies Quarterly , 45, 81– 109. Google Scholar CrossRef Search ADS   Hood S.J. ( 1992) Dragons Entangled: Indochina and the China-Vietnam War . Armonk, MY: ME Sharpe. Lo C. ( 1989) China’s Policy Towards Territorial Disputes: The Case of the South China Sea Islands . London: Routledge. Lo N., Hashimoto B., Reiter D ( 2008) ‘ Ensuring peace: foreign-imposed regime change and postwar peace duration, 1914-2001’, International Organization , 62, 717– 736. Google Scholar CrossRef Search ADS   Powell R. ( 2006) ‘ War as a commitment problem’, International Organization , 60( 1), 169– 203. Google Scholar CrossRef Search ADS   Quackenbush S.L. ( 2006) ‘ Identifying opportunity for conflict: politically active dyads’, Conflict Management and Peace Science , 23( 1), 37– 51. Google Scholar CrossRef Search ADS   Quackenbush S.L. ( 2010) ‘ Territorial issues and recurrent conflict’, Conflict Management and Peace Science , 27( 3), 239– 252. Google Scholar CrossRef Search ADS   Quackenbush S.L., Venteicher J.F. ( 2008) ‘ Settlements, outcomes, and the recurrence of conflict’, Journal of Peace Research , 45( 6), 723– 742. Google Scholar CrossRef Search ADS   Ravallion M., Chen S. ( 2007) ‘ China’s (uneven) progress against poverty’, Journal of Development Economics , 82, 1– 42. Google Scholar CrossRef Search ADS   Reed W., Chiba D. ( 2010) ‘ Decomposing the relationship between contiguity and militarized conflict’, American Journal of Political Science , 54( 1), 61– 73. Google Scholar CrossRef Search ADS   Senese P.D. ( 2005), ‘ Territory, contiguity, and international conflict: assessing a new joint explanation’, American Journal of Political Science , 49( 4), 769– 779. Google Scholar CrossRef Search ADS   Senese P.D., Quackenbush S.L. ( 2003) ‘ Sowing the seeds of conflict: the effect of dispute settlements on durations of peace’, The Journal of Politics , 65( 3), 696– 717. Google Scholar CrossRef Search ADS   Senese P.D., Vasquez J.A. ( 2003) ‘ A unified explanation of territorial conflict: testing the impact of sampling bias, 1919–1992’, International Studies Quarterly , 47, 275– 298. Google Scholar CrossRef Search ADS   Senese P.D., Vasquez J.A. ( 2005) ‘ Assessing the steps to war’, British Journal of Political Science , 35( 4), 607– 633. Google Scholar CrossRef Search ADS   Singer J.D., Bremer S., Stuckey J. ( 1972) ‘Capability distribution, uncertainty, and major power war, 1820-1965’, in Russett B. (ed.), Peace, War, and Numbers , pp. 19– 48. Beverly Hills, CA: Sage Publication. The Associated Press ( 2010) Vietnam: seeking fishermen’s release (2010, October 7). The New York Times, Late Edition. http://www.nytimes.com/2010/10/07/world/asia/07briefs-Vietnam.html (24 February 2018, date last accessed). Tir J. ( 2003) ‘ Averting armed international conflicts through state-to-state territorial transfers’, Journal of Politics , 65( 4), 1235– 1257. Google Scholar CrossRef Search ADS   Tir J., Schafer P., Diehl P.F. et al.   ( 1998) ‘ Territorial changes, 1816–1996: procedures and data’, Conflict Management and Peace Science , 16( 1), 89– 97. Google Scholar CrossRef Search ADS   Vasquez J.A. ( 1993) The War Puzzle . London: Cambridge University Press. Google Scholar CrossRef Search ADS   Vasquez J.A. ( 1995) ‘ Why do neighbors fight? Proximity, interaction, or territoriality’, Journal of Peace Research , 32( 3), 277– 293. Google Scholar CrossRef Search ADS   Walter B.F. ( 2003) ‘ Explaining the intractability of territorial conflict’, International Studies Review , 5( 4), 137– 153. Google Scholar CrossRef Search ADS   Werner S. ( 1999) ‘ The precarious nature of peace: resolving the issues, enforcing the settlement, and renegotiating the terms’, American Journal of Political Science , 43( 3), 912– 934. Google Scholar CrossRef Search ADS   Werner S., Yuen A. ( 2005) ‘ Making and keeping peace’, International Organization , 59( 2), 261– 292. Google Scholar CrossRef Search ADS   © The Author(s) 2018. Published by Oxford University Press in association with the Japan Association of International Relations. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png International Relations of the Asia-Pacific Oxford University Press

Territorial acquisition, commitment, and recurrent war

Loading next page...
 
/lp/ou_press/territorial-acquisition-commitment-and-recurrent-war-k0ULyAfoA0
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press in association with the Japan Association of International Relations.
ISSN
1470-482X
eISSN
1470-4838
D.O.I.
10.1093/irap/lcy001
Publisher site
See Article on Publisher Site

Abstract

Abstract This study investigates how territorial acquisition through war affects the durability of a successive ceasefire and determines what type of territorial acquisition is more detrimental to post-war peace. Despite the wealth of literature on recurrent war and on territory, the effect of territorial acquisition on war resumption has been understudied. This study shows that territorial acquisition creates expectations among adversaries for future power shifts, which results in a commitment problem that hinders peaceful revision of the existing ceasefire. Indeed, duration analysis of ceasefires following interstate wars since World War II shows that territorial change in war, especially acquisition of large and densely populated territories that have potential utility for greater power shifts, makes ceasefires more prone to failure. The analysis of Sino-Vietnamese ceasefires following militarized incidents over land and sea borders also illustrates the importance of territorial acquisition and the potential utility of the territory. 1 Introduction Ceasefires following interstate wars often break down after a short period and cause further devastation to people’s lives that have already been destroyed by war. Even ceasefires minimally defined by the absence of war have often failed: of the 56 pairs of countries engaged in full-scale war between 1946 and 2004, more than 40% reverted from a ceasefire to another full-scale war.1 This problem has cast a shadow over the Asia-Pacific region in which disputes that once caused wars between India and Pakistan and in the Korean Peninsula are yet to be resolved, and pairs of countries, such as China and Vietnam, have repeatedly engaged in military confrontations (Diehl and Goertz, 2000). Concerned over failed ceasefires and their consequences, scholars have made efforts to study why adversaries dishonor a ceasefire. Various factors have been identified ranging from the indecisive outcome of the preceding war, to compromises made in a negotiated ceasefire, to a post-war shift in adversaries’ power (e.g. Blainey, 1988; Werner, 1999; Quackenbush and Venteicher, 2008). Many studies investigated why the same set of states repeatedly engage in militarized disputes. However, another important question in this research field is, why do adversaries return to a full-scale war (Fortna, 2004; Werner and Yuen, 2005; Lo et al., 2008)? These are two different problems incurred because not all resumed disputes result in a war. Some of them are resolved short of war, while others escalate. Therefore, the problem of war resumption is worth exploring from normative and theoretical points of view. Although a handful of studies, such as Werner and Yuen (2005), have identified a mechanism specific to the resumption of war, the causes for recurrence of war remain understudied. This study examines the effect of territorial acquisition on the resumption of war. This factor has been largely ignored in the literature, even though territorial conflicts are consistently found to be violent and to recur (e.g. Goertz and Diehl, 1992; Vasquez, 1993, 1995; Quackenbush, 2010; Reed and Chiba, 2010). Specifically, this study highlights a commitment problem in the renegotiation of ceasefires as the cause for war resumption, and territorial acquisition through war identified as the source of the problem. This argument is tested against the data on (post-)interstate war ceasefires that occurred between 1946 and 1997, in which ceasefire is defined as ‘a period between two interstate wars, starting when the first war ends and failing when war resumes between the same adversaries’. The study also investigates what type of territorial acquisitions are more detrimental to ceasefires. Case studies of two Sino-Vietnamese conflicts at land and sea are presented as illustrative cases. The analysis reveals that territorial acquisition through war makes a subsequent ceasefire short-lived. The negative effect of territorial acquisition is not ascribed to the geographic contiguity of the adversaries, the contentiousness of territorial issues, or the observed post-war power shifts. Rather, it is argued that territorial change generates adversaries’ expectations for future power shifts, which causes a commitment problem when they engage in negotiation during ceasefire. Accordingly, it has been found that larger and more populated territories, that is, those with potential utility for greater power growth, are more likely to cause wars to resume when these are transferred in war. 2 Causes of recurrent war The literature on recurrent conflict can be divided into two camps according to the subject of research and the research question. Much of the literature focuses on settlement of disputes and solutions for underlying issues; the central question of this group of studies is, therefore, why do militarized interstate disputes occur between adversaries who have previously fought wars? Scholars have found that a military stalemate in the previous war (Blainey, 1988; Hensel, 1994; Grieco, 2001), compromises in reaching negotiated ceasefires (Senese and Quackenbush, 2003; Quackenbush and Venteicher, 2008), and third-party intervention and its withdrawal (Beardsley, 2008) cause disagreements and dissatisfaction with the ceasefire between adversaries. Some, however, as in the present study, ask a slightly different question: Why do adversaries return to a full-scale war, and not just a militarized dispute? In this framework focusing on wars causing a large number of casualties, a ceasefire is considered as maintained even if adversaries engage in small-scale military actions, such as a display of forces. Nevertheless, because dispute resumption is a part of the process in which war recurs, some studies also investigate mechanisms by which adversaries resume disputes and subsequent wars. Werner (1999) shows that large shifts in the post-war balance of power tend to cause war resumption because such shifts alter adversaries’ calculations and consequently encourage them to challenge the existing ceasefire. Lo et al. (2008) demonstrate that foreign-imposed regime change following war alters the preference and the institutions of the target country in favor of peace and reduces the risk of recurrent war. However, war resumption involves a mechanism distinct from that of dispute resumption, namely, failure of renegotiation. Even if adversaries challenge an existing ceasefire, it does not necessarily result in war, as it might be resolved through renegotiation of ceasefire terms. This issue is articulated in the bargaining theory of war, which posits that adversaries can usually settle their disputes without fighting, but they sometimes fail to do so and fight a war due to lack of credible information and commitment to a negotiated deal (e.g. Fearon, 1995). Applying this insight to the resumption of war, adversaries engage in another war because there are obstacles to peaceful renegotiation of the existing ceasefire. The literature on recurrent war explicitly or implicitly addresses this problem. Werner (1999) recognizes that post-war shifts in power, while encouraging adversaries to resume disputes, might not necessarily cause them to return to war. Adding to Werner’s study from an institutionalist perspective, Fortna (2004) shows that strong ceasefire agreements, for instance, those that are formally accepted and introduce demilitarized zones, contribute to post-war peace by increasing the cost of waging another war, by reducing uncertainty about each other’s actions and intentions, and by preventing accidental wars. Responding to Fortna, Werner and Yuen (2005) offer a new theory introducing an informational problem in reaching and maintaining ceasefires. Specifically, they argue that ‘unnatural settlements’ achieved after wars with inconsistent battle outcomes or under pressure from a third party prevent adversaries from acquiring accurate information and hence are more likely to become ‘obsolete’. Rephrased in the words of the bargaining theory of war, such ceasefires conclude despite significant uncertainty remaining unresolved, which makes peaceful renegotiation of ceasefires difficult. The present study adds to the literature on recurrent war, investigating the effect of territorial acquisition through war on the likelihood of war resumption. In contrast to the informational explanation offered by Werner and Yuen (2005), this study argues that territorial acquisition hinders peaceful revision of ceasefires by creating a situation in which adversaries are unable to credibly commit to a negotiated agreement. Moreover, by highlighting the consequence of territorial acquisition, the implication of this study also extends to the literature studying the association between territories and conflict, as is discussed in the next section. 3 War and territory It is well established that disputes over territory are violent and recurrent. This tendency toward escalation and recurrence is often attributed to two key characteristics of territorial disputes: contiguity and issue contentiousness. The geographic contiguity of adversaries provides them more opportunities to interact, a greater number of potential issues to fight over, and an environment more favorable to military deployment (e.g. Bremer, 1992; Quackenbush, 2006). Reed and Chiba (2010) also found that neighboring states not only differ in their ‘observable’ environment, such as the degree of economic interdependence, but also behave differently – namely, they respond more aggressively to a shift in the environment than do noncontiguous states. In a similar vein, Quackenbush (2010) reported that geographical contiguity increases the probability of dispute resumption after war, although the effect fades over time. Another key reason for the violent nature of territorial disputes is that territorial issues are salient because of the natural tendency of humans to occupy and defend territory. These territorial disagreements engender grievances among people, which help political leaders mobilize domestic support for militarized actions (e.g. Vasquez, 1993, 1995; Hensel, 1994, 2001; Walter, 2003; Senese and Vasquez, 2005). According to Senese and Vasquez (2003) and Senese (2005), territorial claims are important in explaining war onset, whereas contiguity makes militarized disputes more likely but not war. The same logic explains why militarized disputes over territory are more likely to recur (Hensel, 1996; Quackenbush, 2010). Acknowledging that ceasefires between contiguous parties and over territorial issues are precarious, this study argues that territorial transfers add yet another dimension to the problem. Early works investigating this issue was studied by Goertz and Diehl (1992), who argued that territorial changes spur losing parties’ grievances over lost territory and motivate them to engage in another militarized dispute. As Goertz and Diehl (1992) relied on factor analysis and did not treat censored data properly, Tir (2003) provided a more comprehensive and appropriate analysis. Comparing the effect of violent and peaceful territorial transfers on a prospective territorial dispute, Tir found that violent territorial transfers tend to be followed by shorter ceasefires. This association occurs because the party that obtained the territory seeks to acquire additional territory, especially when it is of strategic value, and the loser nurtures grievances over the lost territory, particularly if it has economic value. However, adversaries can establish a less hostile relationship even after violent territorial transfers if new borders are drawn in accordance with previous administrative frontiers (Carter and Goemans, 2011, 2014). Although violent territorial transfer appears to have a negative impact on peace, its effect on the resumption of war, especially its relation to failure of renegotiation, has not yet been properly investigated. To fill the gap, this study analyzes the durability of ceasefires following wars resulting in territorial acquisition as opposed to those without, an approach contrasting that of Tir (2003), who compares violent territorial transfer with a peaceful one and focuses on their impact on militarized disputes. Moreover, no studies have yet explicitly decomposed the impact of multiple mechanisms by which territorial acquisition could possibly affect ceasefire duration, for instance, the contiguity of adversaries, the salience of issues, and future power shifts. Without such an analysis, it is difficult to identify what type of territories pose threats to post-war peace. 4 Territorial acquisition and recurrent war War resumes because adversaries fail to revise their ceasefire peacefully. According to the literature on war, lack of commitment to a negotiated settlement is one of the reasons why adversaries resort to force rather than resolve their dispute through negotiation. The problem arises when adversaries are in a situation where they are unable to commit credibly to a negotiated agreement. One such situation is when the balance of power between adversaries shifts. The growing party is unable to make a credible commitment because it will be able to coerce the opposing party to accept a new deal that is more favorable to the growing party in the future, and the opponent knows it (e.g. Fearon, 1995; Powell, 2006). Consequently, the relatively declining party may choose to wage a preventive war rather than accept a negotiated deal that may be dishonored in the future. The same logic applies to the resumption of war; adversaries fail to revise their ceasefire peacefully if their power balance is expected to change with the passage of time. Territorial acquisition is one such factor generating shifts in the balance of power during a ceasefire. Territories seized in war are commonly believed to tip the post-war power balance in favor of the party that has gained them (hereafter, the victor). They encourage the victor’s power growth in various ways: by serving strategic purposes, empowering the victor through an endowment of natural resources, and offering the victor industries and markets. In each case, the acquired territory benefits the victor in terms of post-war power growth. The expectation of future power shifts caused by territorial acquisition then reduces the window of opportunity for a negotiated revision of the ceasefire and renders the ceasefire a failure. Anticipating the victor’s power growth, the loser is motivated to wage a war and recapture the lost territory before the victor fully exploits its value, which concomitantly stimulates the victor’s preventive actions. The key here is that territorial acquisition creates expectations among adversaries for future power shifts. Therefore, a commitment problem due to expectations of future power shifts is distinct from incentives to renegotiate, provoked by power shifts that have actually occurred, as was proved by Werner (1999). Not only causing a commitment problem, territorial acquisition also makes war a very attractive option because a successful restoration of territory will considerably change the tide of power shifts. Thus, the first hypothesis is as follows: Hypothesis 1 (Territorial Acquisition):Ceteris paribus, when a disputant has seized territory from the opponent during a preceding war, ceasefires tend to be more short-lived. Territories vary in their potential to empower the victor depending on their size and nature. Ceasefires are generally more vulnerable when territories provide the victor with greater potential for power growth. The victor that has seized territories with higher utility is expected to become more threatening as time passes; consequently, adversaries become less patient in revising an existing ceasefire. Thus, conflict is more likely to resume when the victor seizes a greater amount of resources and becomes more powerful over the course of a ceasefire. Moreover, territories differ in their exchangeability with regard to military power and potential benefits. Specifically, seized territory causes long-term power shifts if it contains industrial urban areas. Territory with value in industrial cities is not readily exchangeable with military power, but it does offer great potential benefits in the long term, as investments of capital, labor, and technology yield increasingly greater economic growth, which can be eventually converted into military power. Although the acquisition of strategically important territory also stimulates a shift in the military balance between adversaries, seizure of industrial cities is key to a long-term shift in power balance. Thus, more densely populated territories comprising industrial cities are expected to be more detrimental to ceasefires when acquired in the preceding war. Hypothesis 2a (Potential Utility of Acquired Territories I):Ceteris paribus, the greater the amount of territory acquired from the opponent during a preceding war, the more short-lived the ceasefires. Hypothesis 2b (Potential Utility of Acquired Territories II):Ceteris paribus, the more densely populated the acquired territory, the more short-lived the ceasefires. The following sections test these hypotheses against data on interstate ceasefires that have occurred since World War II and illustrate their underlying logic in relation to the Sino-Vietnamese conflicts. 5 Research design The hypotheses mentioned above are tested against data on ceasefires following interstate wars between 1946 and 1997, originally compiled by Fortna (2004).2 The analysis involves two parts. The first part tests the theoretical claim that territorial acquisition is one of the barriers to achieving durable ceasefires (Hypothesis 1). Next, it examines the link between ceasefire duration and characteristics of acquired territories (Hypotheses 2a and 2b). 5.1 Data Fortna’s dataset covers 48 ceasefires following 26 full-scale interstate wars between 1946 and 1997. In the dataset, a ceasefire is defined as ‘an end to or break in the fighting, whether or not it represents the final end of the war’ (Fortna, 2004, p. 45). Each case was based on the pair of countries involved in a conflict, dividing multilateral wars into several cases of principal belligerents. Each case rendered a series of observations representing each year or less after a ceasefire, so that time-varying covariates could be included in the analysis. In total, the 48 ceasefires generated 876 observations. Although new wars occurred and some ceasefires broke down after 1997, the analysis period was not extended beyond this point because some key variables were unavailable for the later wars. 5.2 Dependent variable The dependent variable of the analysis is the ceasefire duration, ‘measured from the date of the ceasefire to the start of another COW [Correlates of War] war between the same two belligerents, if there was one’ (Fortna, 2004, p. 48). Among the 48 ceasefires, war resumed in 22 cases, including the First Kashmir War and the First Turco-Cypriot War. In the other 26 cases, peace lasted beyond the time horizon of the analysis, and thus it was right-censored as of 1998. 5.3 Explanatory variables The first part of the analysis examines how territorial acquisition affects ceasefire duration. Each ceasefire was thus checked using the territorial change dataset of the COW Project to determine whether its preceding war resulted in any territorial acquisition (Tir et al., 1998). A binary variable, territorial change, was then created and coded as 1 when any violent territorial acquisition had been associated with a ceasefire, and as 0 if otherwise. In addition, two variables were constructed for capturing the varying degrees to which the seized territory is expected to empower the victor. The first factor considered is the size of the seized territory. The area of exchanged territory was included in the analysis based on the ‘Area of Unit Exchanged in Square Kilometers’ listed in Tir et al. (1998). Area was simply defined as the natural log of the land area obtained by the new owner through war. Observations from cases in which no territorial transfer occurred were assigned a value of 0. The second factor is the population density of the acquired territory. The density of exchanged territory was calculated by dividing the population residing therein by the area of the corresponding territory using the ‘Area of Unit Exchanged in Square Kilometers’ and the ‘Population of Unit Exchanged’ as listed in Tir et al. (1998). The natural log of the calculated density was used to create the variable density, thereby adjusting its diminishing marginal effect. Observations from cases in which no territorial change occurred were assigned a value of 0. 5.4 Other covariates The analyses also considered the following covariates. Related to territory, variables that represent the geographic proximity of adversaries and the issue at stake were included. The binary variable, contiguity, is coded 1 if adversaries are contiguous by land or separated by less than 150 miles of water. To measure whether the war was over contentious issues, a binary variable denoting whether adversaries’ existence was the major stake in the war was included. The variable was created by Fortna (2004) based on the ‘gravity of value threatened’ in the International Crisis Behavior (ICB) dataset (Brecher and Wilkenfeld, 2000). Because territory is one of the essential elements of a sovereign state, conflicts over existence necessarily involve territorial disagreements. The analysis also incorporated territorial integrity at stake, which is coded as 1 when one or both adversaries sought to seize a portion, but not the entirety, of the opponent’s territory. This variable was also derived by Fortna (2004) from the ICB dataset. In addition to these factors, the strength of ceasefire agreements was included, following Fortna (2004), who argued that a strong formal agreement contributes to durable peace. Fortna represented the strength of agreements using the agreement strength index, a composite measurement adding the following 10 factors: formal acceptance of a ceasefire proposal; withdrawal of military forces to status ante; formation of demilitarized zones; agreement on arms control procedure; involvement of peacekeeping forces; strength of enforcement measures by third-party involvement; strength of internal enforcement effort; number of paragraphs in the agreement text; agreement on confidence-building measures; and the level of dispute-resolution efforts. Following Werner and Yuen (2005), battle consistency and interrupted war were also incorporated into the analysis. The variable of battle consistency is measured by the duration over which one side is consistently winning battles toward the end of the war (the ‘last tide’ of the war) as a proportion of the total duration of the war. Interrupted war is a binary variable that takes 1 if ‘(1) there is clear evidence of significant third-party pressure to ceasefire, and (2) the ceasefire is invoked before the parties negotiate the final settlement’ (Werner and Yuen, 2005, p. 273), and 0 if otherwise. The indicator of actual shifts in power is represented by the post-war change in relative capabilities as created by Werner (1999) from the Composite Index of National Capability (CINC) score of the COW Project (Singer et al., 1972). This variable is measured by the absolute difference in adversaries’ growth rates. A country’s growth rate is obtained by the annual growth of its capability as a proportion of its capability the previous year. The variable is then obtained by taking the difference between the growth rates of the adversaries. Therefore, the larger the value of the change in relative capabilities, the more significant is the shift in the adversaries’ balance of power. Furthermore, a military tie was coded as 1 when a war ended in a tie and as 0 if otherwise. The cost of war was measured by the natural log of the total number of battle deaths. Last, adversaries’ history of conflict was measured by the number of militarized interstate disputes that occurred between the dyad’s members divided by the number of years that the dyad had existed in the interstate system when war started. These three variables were retrieved from the dataset provided by Fortna (2004). Descriptive statistics of all the variables are reported in Table 1. Table 1 Descriptive statistics Variables  Mean  Min  Max  Std. Dev.  Obs.  Territorial acquisition  0.15  0.00  1.00  0.36  891  Area  1.36  0.00  12.07  3.24  891  Density  0.68  0.00  7.36  1.69  891  Contiguous  0.66  0.00  1.00  0.48  876  Existence at stake  0.26  0.00  1.00  0.44  876  Territorial integrity at stake  0.17  0.00  1.00  0.38  876  Agreement strength  4.36  0.00  9.50  2.39  876  Battle consistency  0.42  0.00  1.00  0.43  876  Interrupted war  0.28  0.00  1.00  0.45  876  Change in relative capabilitiesa  0.12  0.00  5.43  0.28  770  Military tie  0.45  0.00  1.00  0.50  876  Cost of war  9.30  5.60  14.04  2.35  876  History of conflict  0.75  0.00  3.41  0.83  876  Variables  Mean  Min  Max  Std. Dev.  Obs.  Territorial acquisition  0.15  0.00  1.00  0.36  891  Area  1.36  0.00  12.07  3.24  891  Density  0.68  0.00  7.36  1.69  891  Contiguous  0.66  0.00  1.00  0.48  876  Existence at stake  0.26  0.00  1.00  0.44  876  Territorial integrity at stake  0.17  0.00  1.00  0.38  876  Agreement strength  4.36  0.00  9.50  2.39  876  Battle consistency  0.42  0.00  1.00  0.43  876  Interrupted war  0.28  0.00  1.00  0.45  876  Change in relative capabilitiesa  0.12  0.00  5.43  0.28  770  Military tie  0.45  0.00  1.00  0.50  876  Cost of war  9.30  5.60  14.04  2.35  876  History of conflict  0.75  0.00  3.41  0.83  876  aTime-varying covariate. Agreement strength might change over time if follow-up agreements are reached. Table 1 Descriptive statistics Variables  Mean  Min  Max  Std. Dev.  Obs.  Territorial acquisition  0.15  0.00  1.00  0.36  891  Area  1.36  0.00  12.07  3.24  891  Density  0.68  0.00  7.36  1.69  891  Contiguous  0.66  0.00  1.00  0.48  876  Existence at stake  0.26  0.00  1.00  0.44  876  Territorial integrity at stake  0.17  0.00  1.00  0.38  876  Agreement strength  4.36  0.00  9.50  2.39  876  Battle consistency  0.42  0.00  1.00  0.43  876  Interrupted war  0.28  0.00  1.00  0.45  876  Change in relative capabilitiesa  0.12  0.00  5.43  0.28  770  Military tie  0.45  0.00  1.00  0.50  876  Cost of war  9.30  5.60  14.04  2.35  876  History of conflict  0.75  0.00  3.41  0.83  876  Variables  Mean  Min  Max  Std. Dev.  Obs.  Territorial acquisition  0.15  0.00  1.00  0.36  891  Area  1.36  0.00  12.07  3.24  891  Density  0.68  0.00  7.36  1.69  891  Contiguous  0.66  0.00  1.00  0.48  876  Existence at stake  0.26  0.00  1.00  0.44  876  Territorial integrity at stake  0.17  0.00  1.00  0.38  876  Agreement strength  4.36  0.00  9.50  2.39  876  Battle consistency  0.42  0.00  1.00  0.43  876  Interrupted war  0.28  0.00  1.00  0.45  876  Change in relative capabilitiesa  0.12  0.00  5.43  0.28  770  Military tie  0.45  0.00  1.00  0.50  876  Cost of war  9.30  5.60  14.04  2.35  876  History of conflict  0.75  0.00  3.41  0.83  876  aTime-varying covariate. Agreement strength might change over time if follow-up agreements are reached. 5.5 Statistical model Ceasefire duration was examined through duration analysis to enable proper treatment of the right-censored processes. The current study employed the Weibull model, assuming that the baseline probability of war resumption monotonically changes over time. The same analyses using the Cox model were run here to reduce the risk of arriving at a conclusion sensitive to assumptions about the shape of the baseline hazard, although a residual analysis suggests that the Weibull model better fits the data. Corresponding results are reported in the Supplementary Material. 6 Results The results of the analysis are reported in the form of coefficients that show the marginal effects of variables on the hazard function, that is, the instantaneous probability of failure at a certain point in time conditional on survival up to that point. A factor with a negative coefficient decreases the instantaneous probability of ceasefire breakdown and thus extends its duration. For substantive interpretation, the hazard ratio is also reported: the instantaneous probability of ceasefire failure increases by the factor of the ratio, given survival up to that point. Predictors of the hazard ratio exceeding a value of 1 thus increase the likelihood of war resumption. 6.1 The effect of territorial acquisition The main results of the duration analysis on Hypothesis 1 are reported in Table 2. Model 1 examines the effect of territorial acquisition by controlling for factors such as the characteristics of war, the form of ceasefire agreements, and post-war shifts in the balance of power. Territorial acquisition is found to have a disturbing effect on ceasefire duration. When one or both adversaries acquire new territory through war, a succeeding ceasefire, when achieved, faces a higher risk of failure. The effect of territorial change is robust across different models and model specifications, as is discussed later and in the Supplementary Material. In all models, territorial acquisition decreases ceasefire duration. Table 2 Ceasefire duration and territorial acquisition (Weibull model)   (1)  (2)  (3)  (4)    coef.  coef.  coef.  coef.  Territorial acquisition  2.16***  2.57***      (0.23)  (0.70)      Area      0.29***        (0.05)    Density        0.38*        (0.15)  Contiguous    1.29  1.10  1.60    (1.10)  (0.98)  (1.10)  Existence at stake    2.50*  2.45*  2.60*    (1.22)  (1.11)  (1.30)  Territorial integrity at stake    −3.53***  −3.84***  −2.96***    (0.66)  (0.54)  (0.86)  Agreement strength  −0.23  −0.35  −0.32  −0.37  (0.13)  (0.23)  (0.22)  (0.29)  Battle consistency  −1.85  −7.31*  −7.69**  −6.78*  (1.03)  (2.85)  (2.89)  (3.09)  Interrupted war  2.39***  6.94***  7.13***  6.02***  (0.49)  (0.95)  (1.04)  (1.30)  Change in relative capabilities  0.88***  1.29***  1.24***  1.33***  (0.15)  (0.31)  (0.29)  (0.38)  Military tie  2.25*  4.52  4.17  4.40  (0.92)  (2.81)  (2.36)  (2.69)  Cost of war  −0.36*  −0.87***  −0.87***  −0.94***  (0.16)  (0.13)  (0.14)  (0.15)  History of conflict  0.78***  0.60  0.68  0.73  (0.17)  (0.62)  (0.54)  (0.61)  P  0.18  0.33**  0.32**  0.26*  (0.13)  (0.12)  (0.13)  (0.12)  N  770  770  770  770  BIC  137.9  130.0  130.2  141.0  Log-likelihood  −35.7  −31.8  −31.9  −34.0    (1)  (2)  (3)  (4)    coef.  coef.  coef.  coef.  Territorial acquisition  2.16***  2.57***      (0.23)  (0.70)      Area      0.29***        (0.05)    Density        0.38*        (0.15)  Contiguous    1.29  1.10  1.60    (1.10)  (0.98)  (1.10)  Existence at stake    2.50*  2.45*  2.60*    (1.22)  (1.11)  (1.30)  Territorial integrity at stake    −3.53***  −3.84***  −2.96***    (0.66)  (0.54)  (0.86)  Agreement strength  −0.23  −0.35  −0.32  −0.37  (0.13)  (0.23)  (0.22)  (0.29)  Battle consistency  −1.85  −7.31*  −7.69**  −6.78*  (1.03)  (2.85)  (2.89)  (3.09)  Interrupted war  2.39***  6.94***  7.13***  6.02***  (0.49)  (0.95)  (1.04)  (1.30)  Change in relative capabilities  0.88***  1.29***  1.24***  1.33***  (0.15)  (0.31)  (0.29)  (0.38)  Military tie  2.25*  4.52  4.17  4.40  (0.92)  (2.81)  (2.36)  (2.69)  Cost of war  −0.36*  −0.87***  −0.87***  −0.94***  (0.16)  (0.13)  (0.14)  (0.15)  History of conflict  0.78***  0.60  0.68  0.73  (0.17)  (0.62)  (0.54)  (0.61)  P  0.18  0.33**  0.32**  0.26*  (0.13)  (0.12)  (0.13)  (0.12)  N  770  770  770  770  BIC  137.9  130.0  130.2  141.0  Log-likelihood  −35.7  −31.8  −31.9  −34.0  Note: Standard errors are clustered by conflict and reported in parentheses. * P < 0.05, **P < 0.01, ***P < 0.001. Table 2 Ceasefire duration and territorial acquisition (Weibull model)   (1)  (2)  (3)  (4)    coef.  coef.  coef.  coef.  Territorial acquisition  2.16***  2.57***      (0.23)  (0.70)      Area      0.29***        (0.05)    Density        0.38*        (0.15)  Contiguous    1.29  1.10  1.60    (1.10)  (0.98)  (1.10)  Existence at stake    2.50*  2.45*  2.60*    (1.22)  (1.11)  (1.30)  Territorial integrity at stake    −3.53***  −3.84***  −2.96***    (0.66)  (0.54)  (0.86)  Agreement strength  −0.23  −0.35  −0.32  −0.37  (0.13)  (0.23)  (0.22)  (0.29)  Battle consistency  −1.85  −7.31*  −7.69**  −6.78*  (1.03)  (2.85)  (2.89)  (3.09)  Interrupted war  2.39***  6.94***  7.13***  6.02***  (0.49)  (0.95)  (1.04)  (1.30)  Change in relative capabilities  0.88***  1.29***  1.24***  1.33***  (0.15)  (0.31)  (0.29)  (0.38)  Military tie  2.25*  4.52  4.17  4.40  (0.92)  (2.81)  (2.36)  (2.69)  Cost of war  −0.36*  −0.87***  −0.87***  −0.94***  (0.16)  (0.13)  (0.14)  (0.15)  History of conflict  0.78***  0.60  0.68  0.73  (0.17)  (0.62)  (0.54)  (0.61)  P  0.18  0.33**  0.32**  0.26*  (0.13)  (0.12)  (0.13)  (0.12)  N  770  770  770  770  BIC  137.9  130.0  130.2  141.0  Log-likelihood  −35.7  −31.8  −31.9  −34.0    (1)  (2)  (3)  (4)    coef.  coef.  coef.  coef.  Territorial acquisition  2.16***  2.57***      (0.23)  (0.70)      Area      0.29***        (0.05)    Density        0.38*        (0.15)  Contiguous    1.29  1.10  1.60    (1.10)  (0.98)  (1.10)  Existence at stake    2.50*  2.45*  2.60*    (1.22)  (1.11)  (1.30)  Territorial integrity at stake    −3.53***  −3.84***  −2.96***    (0.66)  (0.54)  (0.86)  Agreement strength  −0.23  −0.35  −0.32  −0.37  (0.13)  (0.23)  (0.22)  (0.29)  Battle consistency  −1.85  −7.31*  −7.69**  −6.78*  (1.03)  (2.85)  (2.89)  (3.09)  Interrupted war  2.39***  6.94***  7.13***  6.02***  (0.49)  (0.95)  (1.04)  (1.30)  Change in relative capabilities  0.88***  1.29***  1.24***  1.33***  (0.15)  (0.31)  (0.29)  (0.38)  Military tie  2.25*  4.52  4.17  4.40  (0.92)  (2.81)  (2.36)  (2.69)  Cost of war  −0.36*  −0.87***  −0.87***  −0.94***  (0.16)  (0.13)  (0.14)  (0.15)  History of conflict  0.78***  0.60  0.68  0.73  (0.17)  (0.62)  (0.54)  (0.61)  P  0.18  0.33**  0.32**  0.26*  (0.13)  (0.12)  (0.13)  (0.12)  N  770  770  770  770  BIC  137.9  130.0  130.2  141.0  Log-likelihood  −35.7  −31.8  −31.9  −34.0  Note: Standard errors are clustered by conflict and reported in parentheses. * P < 0.05, **P < 0.01, ***P < 0.001. Model 2 in Table 2 incorporates the geographic proximity of adversaries and the issues at stake into Model 1. The effect of territorial acquisition remains consistent even when these factors are controlled for, that is, territorial change results in shorter ceasefires regardless of adversaries being contiguous or of existential and territorial issues at stake. Moreover, in Models 1 and 2, the effect of territorial acquisition is observed with controlling for post-war shifts in relative capabilities. According to the results of Model 2, among variables relating to territories, the geographical contiguity of adversaries does not affect ceasefire durability in a significant manner. In contrast, it was found that ceasefires are less likely to endure when an adversary’s existence as a sovereign state is threatened, although the magnitude of the variable’s substantive effect might be moderate, as is explained later. On the other hand, ceasefires are maintained for a longer period if their territorial integrity, but not existence, is challenged. The contrasting effects of these variables suggest that the more contentious the issue at stake is, the more likely the adversaries are to return to war. Additionally, interrupted wars are found to be associated with shorter ceasefires and wars with consistent battle outcomes tend to be followed by a more durable peace, as Werner and Yuen (2005) demonstrated. However, the analysis presented here is unable to determine whether third parties caused unstable ceasefires or whether particularly unstable ceasefires attracted third-party intervention. Moreover, consistent with Werner (1999), the post-war change in relative capabilities increases the risk of ceasefire breakdown. Furthermore, the greater the cost of the preceding war, the more durable the subsequent ceasefire, which is consistent across all model specifications. Figure 1 visually presents the hazard ratio of each variable estimated from Model 2. It shows that the occurrence of territorial acquisition considerably increases the probability of war resumption. The instantaneous probability of war resumption following territorial acquisition is estimated to be 13 times higher (95% confidence interval [CI] [3.29, 51.66]) than observed in cases involving no territorial change, ceteris paribus. Moreover, the resumption of war appears more likely when existence is at stake, showing the estimated hazard ratio of 12.2 (95% CI [1.12, 131.77]). However, from a substantive point of view, the existential issue might have only a moderate effect on ceasefire duration, taking into account its large standard error. The influence of a shift in relative capabilities appears moderate, considering that a one-unit increase of the variable signifies a very large shift in the balance of power such that the power of one side is doubled in a span of one year. Figure 1 View largeDownload slide Effects on the likelihood of war resumption. The circles indicate the hazard ratio of each predictor and the line around them shows the 95% confidence interval. The scale of the horizontal axis is adjusted so that the behavior of all variables is substantively and intuitively understandable. Curtailed parts are denoted by vertical double lines. Created from Model 2 in Table 2. Figure 1 View largeDownload slide Effects on the likelihood of war resumption. The circles indicate the hazard ratio of each predictor and the line around them shows the 95% confidence interval. The scale of the horizontal axis is adjusted so that the behavior of all variables is substantively and intuitively understandable. Curtailed parts are denoted by vertical double lines. Created from Model 2 in Table 2. Looking at the left side of the figure, territory at stake and battle consistency greatly contribute to stable ceasefire outcomes. In other words, war is less likely to resume if it was fought over issues of territorial integration, annexation, and separatism, but not the existence of country. Adversaries are also less likely to return to war if they are able to infer their balance of power accurately from consistent battle outcomes of the preceding wars. The cost of war also has a non-negligible effect. To give an example of substantive interpretation, the figure shows the effect of the variable when the total number of battle deaths increases from 2,980 to 8,100. 6.2 The potential utility of exchanged territory Hypotheses 2a and 2b are then examined to analyze how the effects of territorial acquisition vary depending on the characteristics of territorial change. Model 3 in Table 2 provides the estimation results of the duration analysis incorporating the size of the acquired territories in occurrences of territorial acquisition. Consistent with the hypothesis, ceasefire breakdown is positively related to the size of the area exchanged between adversaries. The hazard ratio for area is 1.33, with 95% CI [1.20, 1.47]. To give an example, a one-unit increase in area is roughly equivalent to the difference between the Golan Heights (7.4) and West Bank (8.7) acquired by Israel from Syria and Jordan, respectively, through the Six-Day War. Considering that the observed value of area ranges from 5.6 to 12.1 with a mean value of 8.8 for cases involving territorial change, the result suggests that the effect of the variable is substantively important. Model 4 in Table 2 incorporates the density of exchanged territory while dropping area because of their high collinearity (the correlation is 0.91). The result shows that ceasefires are less durable when there is an exchange of more densely populated territory. In cases involving territorial change, the observed value of density varies from 2.1 to 7.4, with a mean value of 4.4. For instance, the density of the Golan Heights was 3.6 and that of the West Bank was 4.7. Hence, the hazard ratio of 1.47, with 95% CI [1.09, 1.98], indicates the variable’s non-negligible effect on ceasefire duration. Next, a graphical representation in Fig. 2 depicts how the population density of the acquired territory affects the likelihood of war resumption over time. The plots report the estimated hazard function (the left column) and the estimated survivor function (the right column) for three scenarios: ceasefires following (i) no territorial acquisition, (ii) acquisition of territories with average density, and (iii) acquisition of territories with the highest density recorded in the dataset. The 95% CIs are not reported for the sake of readability. According to the illustrations, supposing that 10 years (3,650 days) have passed since the war terminated without any territorial change, the likelihood of the ceasefire enduring until then is slightly higher than one following the acquisition of most densely populated territories. After 20 years (7,300 days), the possibility of a ceasefire preceded by the transfer of most densely populated territories remaining intact is much smaller than one involving no territorial change. Figure 2 View largeDownload slide Estimated hazard function and survivor function for density. All other variables are held at their mean values. The 95% CIs are not shown in the plots for the sake of readability. Created from Model 4 in Table 2. Figure 2 View largeDownload slide Estimated hazard function and survivor function for density. All other variables are held at their mean values. The 95% CIs are not shown in the plots for the sake of readability. Created from Model 4 in Table 2. 6.3 Discussion The first part of the analysis demonstrates that territorial acquisition increases the likelihood of war resumption and that this effect is not attributed to the contiguity of adversaries, the importance of territorial issues, and the actual shift in relative capabilities. Therefore, the findings suggest that territorial acquisition increases the likelihood of war resumption because it generates a commitment problem by shaping adversaries’ expectations for future power shifts. Potential concerns with the above analysis would be that the results depend on methodological assumptions and influential observations in the dataset. To address the first concern, the same analyses are run using the Cox model. A series of robustness checks are also conducted to control for the effect of possible outliers and influential observations, such as the Sinai War between Israel and Egypt and the First Turco-Cypriot War. The results remain unchanged, and are reported in the Supplementary Material. The second half of the analysis shows that the effect of territorial acquisition varies by territories’ potential utility to empower the victor. In other words, the seizure of large and densely populated territories through war causes less durable ceasefires. An alternative interpretation of the results may be that governing the local population inhabiting a newly acquired territory is more difficult than occupying an underpopulated land. The local population may organize rebellion(s) or work as operatives when the territory’s former ruler wages a counter-offense. More heavily populated territories are thus more difficult to secure control over and are more susceptible to war resumption. Although this study alone cannot reject this possibility, this alternative argument serves as a limited account of failed ceasefires, considering that they are not durable when democratic countries are victorious.3 When lack of effective governance over the occupied territory is the major barrier to a stable ceasefire, democratic countries should be able to promote post-war peace because they tend to have greater state capacity and to provide better governance. However, the supplemental analysis shows the opposite, implying that the main barrier is the victor’s potential for power growth rather than lack of governance. This statistical analysis illuminates the links between territorial acquisition, a commitment problem, and war recurrence. The next section applies this argument to cases in the Asia and Pacific region: the Sino-Vietnamese conflicts over the land border and in the South China Sea. The analysis also complements the statistical analysis, demonstrating that the patterns of their conflicts were affected by the utility of the seized territories in each region. 7 Illustration: Sino-Vietnamese conflicts The People’s Republic of China (hereafter, China) and the Socialist Republic of Vietnam (hereafter, the SRV or Vietnam) are known for their long-time rivalry. Although both countries uphold a communist ideology, this commonality has not implied shared interests even during the Cold War. Their disputes are most intensely militarized along their land border, over which they fought two full-scale wars in 1979 and 1987. They have also engaged in less intensified and sporadic militarized disputes over the South China Sea, where armed clashes have happened in 1979 and 1988, and then frequently since 2010. Without doubt, the ceasefires between 1979 and the late 1980s at land and sea greatly differ in terms of the intensity of military actions (wars on land and militarized disputes at sea) and are not readily comparable. However, contrasting them will shed light on the above-mentioned dynamics while holding other factors, such as the presence of the Soviet Union and the ongoing civil war in Cambodia, constant. The First Sino-Vietnamese War broke out on 14 February 1979, when China invaded Vietnam beyond the disputed land border. After China’s extensive invasion and firm resistance by the Vietnamese troops, war was terminated when China announced its withdrawal from the occupied territory in March 1979. Historical accounts disagree on who won the war. Some suggest that ‘[i]t was a total defeat for China’ (Bercovitch and Jackson, 1997, p. 189), whereas others claim that Vietnam lost in terms of war damage (Hood, 1992, p. 59). The war imposed considerable costs on both sides; more than 50,000 soldiers lost their lives and 500,000 people were killed in the war, although contradictory figures are provided by various sources (Hood, 1992, p. xv; Bercovitch and Jackson, 1997, p. 189). Although Beijing and Hanoi initiated peace negotiations immediately after the war, fighting along the border resumed soon after. By early 1981, the number of violent incidents committed by the Chinese allegedly exceeded 4,000 cases, and China accused Vietnam of instigating more than 2,000 armed provocations and the killing of 240 Chinese (Chang, 1986, pp. 79–80). The fighting culminated in January 1987 through a conflict referred to as the Second Sino-Vietnamese War. Confrontations lasted until March 1987, resulting in as many as 3,000, mostly Chinese, fatalities (Bercovitch and Jackson, 1997, p. 216). The dispute over the South China Sea contrasts with the failed ceasefire over the land border. The first confrontation at sea occurred between China and South Vietnam in 1974 when China seized the entire Paracel Islands, and South Vietnam in turn occupied part of the Spratly Islands. After reunification in 1975, the SRV inherited territorial claims and control over the Spratlys from South Vietnam. Then, the first incident between China and the SRV occurred in April 1979 when ‘China seized a Vietnamese reconnaissance boat allegedly conducting sabotage activities in the Paracels’ (Chang, 1986, p. 79). Fighting was limited in intensity and no significant military confrontation was reported in the South China Sea until 1988. Stability broke down in 1988, when a Chinese naval vessel attacked a Vietnamese gunboat near the Spratly Islands, resulting in more than 70 casualties. Thus, the ceasefire terminating the First Sino-Vietnamese War failed quickly, whereas stability was preserved after the armed clash over the South China Sea. As noted above, the two incidents in 1979 significantly differ in intensity and there might be room for argument regarding whether a ‘ceasefire’ occurred in the South China Sea. Nevertheless, it is remarkable that the South China Sea incident did not escalate or recur as opposed to that over the land border. This contrast is even more intriguing considering that a costly and uninterrupted war is commonly considered less prone to recurrence. This study focuses on the utility of seized territories at land and sea. During confrontations along the land border, no large territorial change was reported4; however, China and Vietnam exchanged small but important territories: strategically situated hills along the border. These hills served as strategically important locations from which the occupier shelled the opponent in conflicts that ‘often took the form of artillery duels’. For instance, in mid-1981, ‘the two sides engaged several thousand troops and fought … over a few hills in the Fakashan area of China’s Guangxi region’ (Chang, 1986, pp. 79–80). Similarly, in 1984, ‘Chinese troops proceeded to seize a couple of hilltops that China claimed to be inside Chinese territory but which had been occupied by Vietnam since the war and repeatedly used to shell Chinese border villages for five straight years.’ The offense invoked unprecedented levels of counterattacks by Vietnam ‘in an attempt to recapture the strategic high grounds that Hanoi claimed were inside Vietnamese territory’ (Chang, 1986, pp. 80–81). The main stakes in the South China Sea were underwater oil and mineral reserves. Although the islands also had strategic value, the archipelagos ‘have attracted new attention since the late 1960s when these areas became known as rich in undersea oil deposits’ (Chang, 1986, pp. 14–16). This ambition for oil and minerals was fueled by the global trend of setting exclusive economic zones (EEZs) to include maritime resources. Indeed, when the SRV took over Spratly Islands, ‘Hanoi reportedly had also started to conduct talks with foreign oil companies on the resumption of oil exploration and exploitation in areas adjacent to this archipelago’ (Chang, 1986, p. 28). Therefore, China and Vietnam exchanged two vital territories: strategically located hills and islands containing maritime resources. Although they did not exchange industrial cities as posed in the theoretical argument, the two territories had different implications for future power shifts during the period between 1979 and 1988 as described in the theory: strategically located hills had greater utility in terms of stimulating power shifts than maritime resources, which both China and Vietnam lacked the capacity to exploit. The value of these hills was readily consumable by both countries that could improve their military position by simply consolidating their control over the hills (Carter, 2010). The vitality of these hills was compounded by the underdeveloped economic and military capabilities of the adversaries. Vietnam’s economic performance was unsatisfactory during this period due to the long Vietnam War, poor administrative capacities, and corruption among the political elite. Its economic and military capacities mostly depended on assistance from the Soviet Union (Hood, 1992, p. 66). Although the new leadership that took office in 1986 decided to turn away from a centrally planned economy and toward a market economy through a process of ‘Doi Moi (economic renovation)’, its economic growth was limited during the 1980s. In China, similarly, the economy and society, at large, were devastated by the Cultural Revolution. Although conditions began to change in 1978 when Deng Xiaoping took leadership and applied economic reforms, more than half the population still lived below the poverty line as reported in 1981 (Ravallion and Chen, 2007). China’s gross domestic product (GDP) per capita remained at less than $2,000 (in 1990, in US dollars) until 1992.5 For adversaries with limited capacity, the hills were conceived of as an essential source of future power shifts. In contrast to the hills, both China and Vietnam lacked capacities to exploit offshore oil fields in the deep waters of the South China Sea. China was only able to exploit offshore oil from the shallow waters of the Gulf of Tonkin (Lo, 1989, pp. 60–61). To improve its capacity, under the new leadership of Deng Xiaoping, China established the China National Offshore Oil Corporation (CNOON) in 1982, and signed a number of contracts with foreign companies throughout the 1980s (Chang, 1986, p. 72). However, by the early 1980s, their activities were still limited to shallow sea areas in the Tonkin Gulf and off the coast of Hainan Island (Lo, 1989, p. 61). By the end of 1982, China had extended its focus to the South China Sea and installed oil rigs there. Nevertheless, it was not able to initiate actual oil production, even in the Gulf, until 1986 (Hood, 1992, p. 132). Similarly, Vietnam suffered from lack of technological capabilities to extract maritime resources. Although Vietnam and the Soviet Union allegedly installed three or four oil rigs in the Gulf as early as in 1975 and 1976, ‘it was learned that probably only two were actually working’ in 1984. Even in the shallow waters of the Gulf, it was not until April 1986 that ‘the Vietnamese-Soviet drill crews discovered their first major producer … in the Bach Ho (White Tiger) field’. Disappointed by the result, Vietnam considered forging a contract with Western companies, but the USSR did not allow Hanoi to do so (Hood, 1992, pp. 131–133). Interestingly, in contrast to the stability of the South China Sea, a couple of incidents occurred in the Gulf of Tonkin during this period. Quarrels in the Gulf were clearly related to oil exploration in the area: Hanoi warned that ‘if China continued to search for oil in the disputed waters in the Gulf of Tonkin, companies … would have to bear the consequences’ (Hood, 1992, p. 132). In fact, 14 armed Vietnamese boats clashed with Chinese fishing trawlers in the Gulf in 1982. Similarly, in 1983, two Chinese fishing boats operating in the Gulf were attacked by Vietnamese forces (Chang, 1986, pp. 76–77). Also interesting is the fact that the armed clash of 1988 over the South China Sea was preceded by China’s discovery of oil in the Spratlys. Although ‘[u]ntil late 1987, the Spratly potential was largely unknown and unexplored’, the situation drastically changed in November 1987 when ‘China announced it had discovered rich oil and gas deposits under the Spratlys after a survey of 180,000 square kilometers.’ Thus, the successful discovery of the oil reserve appears to have disrupted stability in the area. Hence, the above illustrations suggest that territorial acquisition and the utility of seized territories explain the dynamics of the Sino-Vietnamese conflicts occurring between 1979 and the late 1980s. 8 Conclusions This study demonstrated that territorial acquisition increases the risk of war resumption because it shapes expectations for future shifts in the balance of adversaries’ power and consequently makes peaceful renegotiation more likely to fail because of a commitment problem. The duration analysis of ceasefires following wars occurring between 1946 and 1997 shows that territorial acquisition renders likely resumption of war. Moreover, some types of territorial acquisition pose a more serious commitment problem than others based on the characteristics of the transferred territory: the size and the population density. In short, territorial acquisition that would likely produce a larger future shift in power was demonstrated to be significantly more detrimental to lasting ceasefire. The findings presented here thus contribute to the literature on recurrent war in several respects. First, although disputes over territory have been especially conflictual, the effect of territorial acquisition on recurrent war has been insufficiently studied in the literature. This study filled the gap, demonstrating that territorial acquisition has detrimental effects on post-war peace. Moreover, this study also found that the negative effect of territorial acquisition is not completely attributed to the contiguity of adversaries or the vitality of territorial issues. Another explanation, therefore, was offered. Territorial acquisition generates a commitment problem, which obstructs peaceful renegotiation of ceasefires and causes recurrent war. The study also showed that the effect of territorial acquisition varies depending on the utility of the acquired territory. These findings are somewhat contradictory to intuition-based results. The common belief is that a decisive victory contributes to post-war peace. However, if the size of a victory is defined by the volume of the seized territories as opposed to military outcomes, a significant victory may not lead to post-war peace. The implications of this research extend to current incidents between China and Vietnam over the South China Sea. After several years of peace since 1988, controversy reemerged in 2010 when Chinese forces arrested nine fishermen accused of carrying explosives close to the Paracel Islands (The Associated Press, 2010). This event has been followed by several militarized incidents that were apparently motivated by a desire for maritime resources. Although these incidents are similar to those occurring during the 1970s and 1988s, they have become more frequent than before. Viewed in the framework of this study, this increase in the number of armed clashes over the South China Sea is possibly affected by the economic and technological development of both countries in exploiting maritime resources in deep water areas. It is not argued that the mechanism identified in the current study serves as the sole and complete explanation of the Sino-Vietnamese rivalry to date. Rather, it sheds light on the dynamics of their rivalry revolving around issues of territorial acquisition and their economic development, hoping to advance our understanding of their ongoing dispute and to find a way to prevent it from further escalating. Acknowledgements For helpful comments on this project, the author is grateful to Miao-ling Lin Hasenkamp, David Leblang, Jeffrey Legro, Kazuto Ohtsuki, Anna Pechenkina, Todd Sechser, Allan Stam, Changhe Su, Atsushi Tago, three anonymous reviewers, the editors of International Relations of the Asia-Pacific, and all participants of the 2013 Joint Conference of International Studies Association and Peace Science Society, the 2013 Kobe Sakura Meeting, the 2017 Annual Conference of the Midwest Political Science Association, and the Workshop at Fudan University. The author also thanks Virginia Page Fortna for making her dataset available online. Special acknowledgement goes to Daichi Yamada for his research assistance. This research was supported by the JSPS KAKENHI 15K1699805. Supplementary material Supplementary material is available at International Relations of the Asia-Pacific online. Footnotes 1 The list of cases provided by Fortna (2004) is supplemented by the author in reference to the Expanded War Data (v.2.1) listed in Gleditsch (2004). 2 The dataset can be downloaded from http://www.columbia.edu/~vpf4/peacetime.htm (24 February 2018, date last accessed). 3 Corresponding results are reported in the Supplementary Material. 4 No territorial exchange was identified in the dataset. However, this omission does not seriously undermine the validity of the statistical analysis, because at worst it would only cause an underestimation of the effect of territorial change. 5 The Maddison-Project, available at http://www.ggdc.net/maddison/oriindex.htm, 2013 version (24 February 2018, date last accessed). References Beardsley K. ( 2008) ‘ Agreement without peace? International mediation and time inconsistency problems’, American Journal of Political Science , 52( 4), 723– 740. Google Scholar CrossRef Search ADS   Bercovitch J., Jackson R. ( 1997) International Conflict: A Chronological Encyclopedia of Conflicts and Their Management, 1945-1995 . Washington, DC: Congressional Quarterly. Blainey G. ( 1988) Causes of War , 3rd edn. New York: Free Press. Google Scholar CrossRef Search ADS   Brecher M., Wilkenfeld J. ( 2000) A Study of Crisis. Ann Arbor, MI: University of Michigan Press. Bremer S.A. ( 1992) ‘ Dangerous dyads: conditions affecting the likelihood of interstate war, 1816-1965’, Journal of Conflict Resolution , 36( 2), 309– 341. Google Scholar CrossRef Search ADS   Carter D.B. ( 2010) ‘ The strategy of territorial conflict’, American Journal of Political Science , 54( 4), 969– 987. Google Scholar CrossRef Search ADS   Carter D.B., Goemans H.S. ( 2011) ‘ The making of the territorial order: new borders and the emergence of interstate conflict’, International Organization , 65( 2), 275– 310. Google Scholar CrossRef Search ADS   Carter D.B., Goemans H.S. ( 2014) ‘ The temporal dynamics of new international borders’, Conflict Management and Peace Science , 31( 3), 285– 302. Google Scholar CrossRef Search ADS   Chang P. ( 1986) The Sino-Vietnamese Territorial Dispute . New York: Praeger. Diehl P.F., Goertz G. ( 2000) War and Peace in International Rivalry . Ann Arbor, MI: University of Michigan Press. Google Scholar CrossRef Search ADS   Fearon J.D. ( 1995) ‘ Rationalist explanations for war’, International Organization , 49( 3), 379– 414. Google Scholar CrossRef Search ADS   Fortna V.P. ( 2004) Peace Time: Cease-fire Agreements and the Durability of Peace . Princeton, NJ: Princeton University Press. Gleditsch K.S. ( 2004). ‘ A revised list of wars between and within independent states, 1816-2002’, International Interactions , 30( 3), 231– 262. Google Scholar CrossRef Search ADS   Goertz G., Diehl P.F. ( 1992) Territorial Changes and International Conflict . London: Routledge. Google Scholar CrossRef Search ADS   Grieco J.M. ( 2001) ‘ Repetitive military challenges and recurrent international conflicts, 1918–1994’, International Studies Quarterly , 45( 2), 295– 316. Google Scholar CrossRef Search ADS   Hensel P.R. ( 1994) ‘One thing leads to another: recurrent militarized disputes in Latin America, 1816-1986’, Journal of Peace Research , 31( 3), 281. Google Scholar CrossRef Search ADS   Hensel P.R. ( 1996) ‘ Charting a course to conflict: territorial issues and interstate conflict, 1816-1992’, Conflict Management and Peace Science , 15( 1), 43– 73. Google Scholar CrossRef Search ADS   Hensel P.R. ( 2001) ‘Contentious issues and world politics: the management of territorial claims in the Americas, 1816–1992’, International Studies Quarterly , 45, 81– 109. Google Scholar CrossRef Search ADS   Hood S.J. ( 1992) Dragons Entangled: Indochina and the China-Vietnam War . Armonk, MY: ME Sharpe. Lo C. ( 1989) China’s Policy Towards Territorial Disputes: The Case of the South China Sea Islands . London: Routledge. Lo N., Hashimoto B., Reiter D ( 2008) ‘ Ensuring peace: foreign-imposed regime change and postwar peace duration, 1914-2001’, International Organization , 62, 717– 736. Google Scholar CrossRef Search ADS   Powell R. ( 2006) ‘ War as a commitment problem’, International Organization , 60( 1), 169– 203. Google Scholar CrossRef Search ADS   Quackenbush S.L. ( 2006) ‘ Identifying opportunity for conflict: politically active dyads’, Conflict Management and Peace Science , 23( 1), 37– 51. Google Scholar CrossRef Search ADS   Quackenbush S.L. ( 2010) ‘ Territorial issues and recurrent conflict’, Conflict Management and Peace Science , 27( 3), 239– 252. Google Scholar CrossRef Search ADS   Quackenbush S.L., Venteicher J.F. ( 2008) ‘ Settlements, outcomes, and the recurrence of conflict’, Journal of Peace Research , 45( 6), 723– 742. Google Scholar CrossRef Search ADS   Ravallion M., Chen S. ( 2007) ‘ China’s (uneven) progress against poverty’, Journal of Development Economics , 82, 1– 42. Google Scholar CrossRef Search ADS   Reed W., Chiba D. ( 2010) ‘ Decomposing the relationship between contiguity and militarized conflict’, American Journal of Political Science , 54( 1), 61– 73. Google Scholar CrossRef Search ADS   Senese P.D. ( 2005), ‘ Territory, contiguity, and international conflict: assessing a new joint explanation’, American Journal of Political Science , 49( 4), 769– 779. Google Scholar CrossRef Search ADS   Senese P.D., Quackenbush S.L. ( 2003) ‘ Sowing the seeds of conflict: the effect of dispute settlements on durations of peace’, The Journal of Politics , 65( 3), 696– 717. Google Scholar CrossRef Search ADS   Senese P.D., Vasquez J.A. ( 2003) ‘ A unified explanation of territorial conflict: testing the impact of sampling bias, 1919–1992’, International Studies Quarterly , 47, 275– 298. Google Scholar CrossRef Search ADS   Senese P.D., Vasquez J.A. ( 2005) ‘ Assessing the steps to war’, British Journal of Political Science , 35( 4), 607– 633. Google Scholar CrossRef Search ADS   Singer J.D., Bremer S., Stuckey J. ( 1972) ‘Capability distribution, uncertainty, and major power war, 1820-1965’, in Russett B. (ed.), Peace, War, and Numbers , pp. 19– 48. Beverly Hills, CA: Sage Publication. The Associated Press ( 2010) Vietnam: seeking fishermen’s release (2010, October 7). The New York Times, Late Edition. http://www.nytimes.com/2010/10/07/world/asia/07briefs-Vietnam.html (24 February 2018, date last accessed). Tir J. ( 2003) ‘ Averting armed international conflicts through state-to-state territorial transfers’, Journal of Politics , 65( 4), 1235– 1257. Google Scholar CrossRef Search ADS   Tir J., Schafer P., Diehl P.F. et al.   ( 1998) ‘ Territorial changes, 1816–1996: procedures and data’, Conflict Management and Peace Science , 16( 1), 89– 97. Google Scholar CrossRef Search ADS   Vasquez J.A. ( 1993) The War Puzzle . London: Cambridge University Press. Google Scholar CrossRef Search ADS   Vasquez J.A. ( 1995) ‘ Why do neighbors fight? Proximity, interaction, or territoriality’, Journal of Peace Research , 32( 3), 277– 293. Google Scholar CrossRef Search ADS   Walter B.F. ( 2003) ‘ Explaining the intractability of territorial conflict’, International Studies Review , 5( 4), 137– 153. Google Scholar CrossRef Search ADS   Werner S. ( 1999) ‘ The precarious nature of peace: resolving the issues, enforcing the settlement, and renegotiating the terms’, American Journal of Political Science , 43( 3), 912– 934. Google Scholar CrossRef Search ADS   Werner S., Yuen A. ( 2005) ‘ Making and keeping peace’, International Organization , 59( 2), 261– 292. Google Scholar CrossRef Search ADS   © The Author(s) 2018. Published by Oxford University Press in association with the Japan Association of International Relations. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com

Journal

International Relations of the Asia-PacificOxford University Press

Published: Mar 22, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off