Precrisis military hostility and escalation in international crises

Abstract

Why do some international crises escalate into violence while others do not? I argue that an understanding of crisis behavior necessitates awareness of the processes preceding a crisis and propose two mechanisms that link precrisis hostility with crisis violence. The population fatigue mechanism suggests that precrisis hostility interrupts the everyday life of a population, strengthening its demand for a harsher response. If leaders behave with restraint during a crisis and after the crisis hostilities resume, leaders risk facing greater dissatisfaction from their now even more fatigued population. The baseline for resolve mechanism suggests that if states are actively hostile before a crisis, they must maintain hostility during a crisis to credibly demonstrate resolve. Findings on international crises between 1918 and 2010 support the interphase relationship: a crisis escalation is significantly more likely after a hostile precrisis period. The behavior of opponents can indicate how crises might unfold even before they begin.

Keywords

Crisis escalation international conflict international crisis recurrent conflict respite period rivalry violence-begets-violence waiting time

Introduction

In October 2003, for the first time in three decades, the Israeli air force carried out an airstrike on Syrian territory. Retaliating for a suicide bombing in which 21 people were killed in the Israeli city of Haifa, Israeli jets raided a suspected training camp of the Palestinian Islamic Jihad near Damascus (Myre, 2003). These events coincided with the thirtieth anniversary of the October 1973 Yom Kippur War, the last war fought between Israel and Syria over the disputed Golan Heights.¹ In such a tense atmosphere, the prospects of a new severe clash were increasing. Syrian president Bashar al-Assad blamed Israel for “trying to terrorize Syria” and drag it into a new regional war. Addressing a special United Nations Security Council meeting to discuss the crisis, the Syrian United Nations ambassador, Fayssal Mekdad, twice threatened that “Syria is not incapable of establishing a resistance and deterrence balance that would force Israel to reconsider its calculations.”² Israeli prime minister Ariel Sharon also warned that Israel “will strike its enemies in every place and with any means,” and urged Israelis to prepare “as if the next war was around the corner” (Adams, 2003). And yet, amid the heated rhetoric, the longtime rivals refrained from escalating their crisis into a war. After nearly two months of acute diplomatic tensions, the Israeli–Syrian crisis faded. Was the averting of violence related to the scarcity of military incidents between the states before their crisis? Does military hostility during periods preceding crises make crisis escalation more likely? The exploration of such an interphase relationship—particularly the impact of military hostility prior to a crisis on violence in the subsequent crisis—is at the center of this study.

The linkage between past and future interactions has been a central theme in scholarship on rivalries and crises (Brecher, 2008; Colaresi et al., 2008; Gibler and Owsiak, 2018; Goertz and Diehl, 1993; Leng, 1983; Maoz and Mor, 2002; McLaughlin Mitchell and Thies, 2011; Senese and Vasquez, 2008). What is still largely missing in the literature, however, is a discussion of respite periods, or intervals, between conflict events. How violence and the outcome of a previous crisis affect the onset and violence of a subsequent crisis is frequently analyzed (Brecher and Wilkenfeld, 2000: 168; Fortna, 2003: 351–54; Hensel, 1994: 291–92), but relatively little is known about the effects of interactions during the periods preceding crises. In contrast to previous studies that stress the attributes of crises, I emphasize the influences that the behavior of states in the absence of a crisis might have on the probability of crisis escalation. I argue that the presence of chronic military hostility before a crisis creates domestic and strategic pressures that influence the calculations of states about the necessity of violent escalation.

Building on the literature on rivalry and crises, I suggest two potential explanations for why a precrisis hostility might lead states to escalate the crisis. Both mechanisms suggest that precrisis hostility makes crisis escalation more likely. Yet they diverge on their predictions of which states—the targets or the perpetrators of the precrisis hostility—are more likely to escalate in the crisis phase.

The first mechanism, population fatigue, provides the perspective of a target state. In this explanation, precrisis hostility pushes leaders to employ greater violence during a crisis to better manage the expectations of their exhausted population. When a population is constantly annoyed by military incidents, the domestic quest to decisively halt the harassing activities of the adversary increases. If, instead, leaders hesitate to use violence during the crisis, and after the crisis military hostility against the population resumes, the leaders risk facing greater audience costs from their now even more tired and dissatisfied population.

The second mechanism, the baseline for resolve, suggests that precrisis hostility increases crisis violence by setting a higher threshold for demonstrating credible resolve during a crisis. Committing hostile acts in the absence of a crisis reveals a state’s willingness to use military force. A reluctance to use force during a crisis, despite committing precrisis hostility, might appear as a sign of weakness. This might lead the adversary to expect greater concessions, making it more difficult for the state to end the crisis in a favorable way. According to this explanation, resolve is relative to the past behavior of a state, linking precrisis hostility with crisis violence.

I examine the link between precrisis interactions and crisis violence using the International Crisis Behavior (ICB) dataset. Findings on dyadic international crises from 1918 to 2010 provide strong evidence for the interphase relationship in international conflicts. Results from binomial logistic regression models show that military hostility before crises increased the odds of severe crisis violence by more than half. In fact, two out of every three international crises that escalated into serious clashes or war occurred in the aftermath of a hostile precrisis period.

Empirical assessment of the theoretical mechanisms shows that the two mechanisms work in conjunction. It also lends stronger support for the baseline for resolve mechanism, and somewhat weaker support for the population fatigue mechanism. This suggests that at times of crisis, states may be more concerned about the credible demonstration of resolve to their opponent. Results are robust for different specifications of precrisis hostility and crisis violence and remain valid when controlling for potentially confounding factors, such as the characteristics of crisis actors, their dyadic history, and attributes of their past and ongoing crises, as well as when taking selection effect into account. As such, this study proves that precrisis hostility is an important element for understanding crisis behavior. While other factors can have a strong effect on crisis escalation, precrisis hostility—an overlooked element in crisis research—can provide a good indication of how the next international crisis is likely to unfold.

Research on crisis escalation and precrisis hostilities

Crisis escalation is a central topic in international relations research (Asal and Beardsley, 2007; Brecher and Wilkenfeld, 2000; Colaresi and Thompson, 2002: 274; James, 2019; Levin-Banchik, 2018; Reed, 2000). This research can be classified into three general approaches that emphasize different crisis and crisis actor attributes when explaining violent escalation.

The first approach focuses on the impact of the attributes of crisis actors, such as their political regime, military and economic capabilities, or possession of nuclear weapons (Asal and Beardsley, 2007; Ben-Yehuda and Levin-Banchik, 2010, 2014; Lektzian and Souva, 2009; Mishali-Ram, 2006). Some actor attributes, such as political regime, change rarely and slowly. In such cases, understanding variation in crisis violence within conflicts requires considering additional, more dynamic factors.

The second approach to crisis escalation addresses precisely this point. It focuses on the characteristics of an ongoing crisis, such as the impact of a crisis trigger, the gravity of the threat, or the number of actors perceiving this threat (Andersen-Rodgers, 2015; Brecher and Wilkenfeld, 2000; Trumbore and Boyer, 2000). In line with conflict-begets-conflict theory, for instance, Wilkenfeld (1991) finds that a crisis that begins with a violent act is considerably more likely to involve severe violence than crises with a non-violent trigger. While a focus on immediate action–reaction addresses the variability of cases, crises in this approach are often studied as separate events.

In contrast, the third approach to crisis escalation considers the broader history between adversaries and emphasizes that different crises might be related. The attributes of a previous crisis, such as its level of violence or how it ended, might affect the risk of violent escalation in the next crisis. Such a focus on the past relationships between adversaries is at the heart of the rivalry approach (Colaresi and Thompson, 2002; Diehl, 2006). We know that one confrontation leads to another, and that once states become involved in a conflictual event, they are more likely to experience conflict in the future (Colaresi and Thompson, 2002; Hensel, 1994; Senese and Vasquez, 2008). Such linkage is explained by the central role of memories about the past and the expectation of future conflict (Goertz and Diehl, 2000: 226). This means that conflict is a process. Yet empirical analyses of conflicts focus mainly on tense and hostile episodes in that process. With a few exceptions (Goertz et al., 2016; Leng, 1983: 414–415), what enemies do in the absence of a crisis remains largely overlooked.

This study builds on the rivalry approach. It considers dyadic history and past relationships as a key to the explanation of crisis behavior. Unlike previous research, however, it looks beyond analysis of links between subsequent crises and addresses precrisis periods.³ The focus on the variation in precrisis hostility, rather than on the role of the attributes of crises, past or present, or of crisis actors, is what makes this study different. This study builds on the rich knowledge in previous literature on rivalry and international crisis escalation, and aims to contribute an additional, precrisis angle to this literature.

We know that longer periods between conflicts tend to make conflict onset and escalation less likely (Braithwaite and Lemke, 2011: 117; Lektzian and Souva, 2009: 28). However, do the interactions during the periods between conflicts matter? And if so, how does precrisis hostility affect the likelihood of crisis escalation?

The lack of focus on precrisis interactions in previous crisis research is surprising given that many of the quantitative analyses of conflict include peace years as a common variable in their models. One of the problems with this practice, from a rivalry perspective, is that these analyses consider the years without military confrontations or crises as years of peace. The intervals between confrontations are usually measured in terms of their duration (Lektzian and Souva, 2009: 26; Mettler and Reiter, 2012). They also are usually implied to be homogeneous in terms of states’ interactions. The presence of the peaceful activities, or the absence of activities at all during peace years, is often assumed rather than evaluated empirically.

This study challenges this approach. It demonstrates that periods before recurring crises might have differing levels of interstate hostility. This means that the episodes we usually label peace years may not be periods of peace. We need to look beyond the duration of peace years and also study the interactions during these periods. It is important to address what the adversaries are doing before crises, explore variation in their precrisis hostilities, and analyze their impact on crisis escalation.

Focusing on the hostile dimensions of international conflicts, I theorize that the presence of military hostilities prior to a crisis makes crisis escalation more likely. I suggest two mechanisms through which a precrisis hostility can increase violence in a subsequent crisis. These mechanisms should be considered as supplementary and can be contextualized as a two-level game (Putnam, 1988; Trumbore and Boyer, 2000). In particular, leaders in a conflict must pay attention to domestic audiences and pressures, as the first mechanism suggests, and must also deal with the international consequences of their behavior, as the second mechanism stresses.

Domestic pressure: population fatigue

The first mechanism stresses the importance of domestic pressures. It focuses on population fatigue as a driving force behind public preference for military action in a crisis. This mechanism suggests that experiencing continued military hostilities disrupts the normal life of a population, tires it, and facilitates its expectation of harsh measures against the harassing state. This mechanism builds on audience costs and rivalry research as well as on psychological literature.

The rivalry literature has given considerable attention to studying the impact of domestic audiences (Vasquez, 2009; Wright and Diehl, 2016). Colaresi (2004: 567) suggests that “domestic politics may trap leaders into perpetuating a rivalry” owing to anticipation of an electoral penalty for overcooperation. In his empirical analysis of leaders’ tenures, Colaresi (2004) finds that dovish leaders are more likely to lose power than their hawkish counterparts. Supporting this logic, Davies and Johns (2016: 343) use a survey experiment and find that “a hawkish failure wins more public sympathy than a failed inducement.” The population in rival states might request a more coercive crisis behavior to restore a sense of national pride that has been challenged by repeated violence (Crescenzi et al., 2007: 657). Alternatively, military action might be more popular because using force might be considered as a simpler solution to ongoing violence (Davies and Johns, 2016: 347). What is common to these explanations is that a history of persistent violence increases public expectation and pressure for violent escalation during crises.

Recent psychological literature also demonstrates that exposure to violence makes civilian populations less supportive of political compromise and more favorable to militant policies (Canetti, 2017: 942; Peffley et al., 2015: 829). According to Canetti et al. (2017: 102) persistent violence radicalizes populations because it increases psychological distress and magnifies the collective threat perception.

The fatigue mechanism is therefore activated as a result of chronic violence between crises. The population expects its leaders to punish a lingering aggressor and deter it from continuing attacks (Vasquez, 2009: 217–218). Such a demand increases during times of crisis when disruptions of ordinary life are at their peak and in the public spotlight (Chong and Hall, 2017: 26; Fearon, 1994: 577). When a population has been under attack prior to a crisis, it may consider crisis violence a necessary bad, a “last push,” to bring about a change. During the Second Israel–Lebanon War in July 2006, for example, two of every three Israelis, or 67%, believed that fighting in Lebanon must continue despite the rocket fire and casualties, until Hezbollah’s threat was removed (Peace Index, July 2006). During times of crisis, the population might be more willing to mobilize and accept the burdens of war if it previously has been attacked. In such circumstances, the population can anticipate that sacrifice during a crisis might pay off by preventing the resumption of hostile actions by an opponent in the aftermath of a crisis.

The expectation of violence by a population cannot alone explain why leaders would take population fatigue into account and choose to escalate the crisis. Here, the consideration of domestic audience costs comes into play. For a population, tired from a military hostility before a crisis, the crisis epitomizes a hope for change. If leaders behave mildly and miss the opportunity for change, that is, they fail to provide the expected policy outcome, they risk declining domestic support (Nomikos and Sambanis, 2019). Yet, if they utilize violence during a crisis, they can claim that they acted decisively, used all available options, and thereby behaved appropriately and proved their commitment to the security of their population (Debs and Weiss, 2014; Leng, 2004: 56).

After a crisis, to limit potential criticism, the leaders can recall their firm stand. If hostility resumes in the aftermath of a crisis, they can more confidently claim that others—for instance, the rival state or the international community more generally—should be held responsible for the absence of change (Fortna, 2003: 343). Such a shift of blame can be perceived as a more credible excuse by the population if leaders prove that they have used harsh measures against the perceived aggressor during a crisis. In contrast, hesitation to escalate a crisis, despite the population fatigue from chronic hostility before the crisis, can cause greater popular dissatisfaction, and thereby greater potential audience costs, with the leaders, especially if the hostility resumes (Davies and Johns, 2016; Debs and Weiss, 2014; Nomikos and Sambanis, 2019). This leads to a general hypothesis:

Hypothesis 1: crisis escalation into severe violence will be more likely following a hostile precrisis period.

Strategic calculations: baseline for resolve

The second mechanism leads to a similar general hypothesis but emphasizes the strategic calculations of states, particularly the demonstration of resolve. The importance of signaling a credible resolve in international affairs has been widely recognized in the conflict literature, most notably in crisis bargaining theory (Fearon, 1994; Leng, 2004; Lupton, 2018a, b; Schelling, 2008). According to Fearon (1994), adversaries possess private information about their capabilities and resolve, contributing to uncertainty and a war of nerves when interests collide. Convincing an opponent that you have high resolve and are capable and willing to fight hard over a foreign policy interest can reduce the risk of escalation by helping to compel your opponent to make concessions and back down in a crisis.

Not every action in the bargaining process is equally compelling (Lupton, 2018b). How to assess the credibility of threats is a debatable topic. Press (2005) distinguishes between two approaches. The past actions theory determines credibility by considering the previous commitments and behavior of states. In contrast, he develops the current calculus theory that suggests that credibility is best judged by looking at the opponent’s preferences now, such as at the balance of power and interests, rather than at their past actions. Press finds that during military crises, state leaders often worry about the historical consequences of their own actions, and yet assess the credibility of their opponents based on current, rather than past, contingences.

Joining the conversation between scholars who assess credibility of resolve based on present (Press, 2005; Wiegand, 2011) or past crisis behavior (Lupton, 2018b), this study suggests addressing an additional layer—the analysis of state behavior in the intervals between crises. In this explanation, precrisis hostility, rather than behavior in current or past crises, serves as a reference baseline for stronger signals of resolve in a crisis. To be sure, past and current interactions can also matter. Past major crises, for example, are entrenched within the national memory (Vasquez, 2009: 226), and leaders often recall a state’s past experiences to justify their course of action domestically and internationally (Abramson and Carter, 2016). While recalling previous traumatic events during times of crisis can be convenient for mobilization and rhetoric, it might be less suitable for assessing the state’s resolve in a current crisis. First, a long time may pass between major crises, making precrisis activities a more accurate indicator of a state’s current intentions. On average, it takes more than six years for the same states to become enmeshed in a repeated international crisis.⁴ Even in the most heated conflicts, such as those between India and Pakistan or between Arab states and Israel, some crises reoccur after more than a decade of relative calm. The activities shortly before a crisis are simply closer in time to the crisis trigger, and, as such, they are likely to be increasingly relevant for a state’s decisions during that crisis.

If so, why consider state behavior before, rather than during, a crisis? The answer is because states’ incentives to misrepresent resolve are greatest during a crisis (Wolford, 2014: 146). During a precrisis period states follow a more routine course of action. Misrepresentation of resolve becomes a less urgent matter. The precrisis activities represent a more typical behavior of states, and thus can serve as a reliable, although not the only, reference point for evaluating the coherence of a state’s resolve.

So, if a state often resorts to low-level military force before crisis, but refrains from violence during a crisis, its opponent can interpret such a change as a reduced resolve and a sign of weakness. Expecting rather low resistance from the irresolute state, the opponent might harden its position and request greater concessions, making it more difficult for a state to end the crisis in a favorable way (McManus, 2014; Wolford, 2014: 146).

According to this explanation, the precrisis hostility and crisis violence are linked because the credibility of the resolve in one phase of a conflict can be judged by the coherence of resolve across the consecutive phases. The precrisis hostility locks states into their positions, creates a precedent, and sets the stage for crisis violence (Fearon, 1994, 577; Leng, 2004: 56–57; Wiegand, 2011).

In the absence of hostility, even a relatively restrained act, such as a military threat, a low level of mobilization, or a limited airstrike, may be enough to credibly signal resolve (Lupton, 2018b; Schelling, 2008: 141). The scarcity of direct military incidents between the United States and North Korea prior to the 2017 nuclear crisis, for example, can indicate why during a crisis the White House considered a limited surgical operation as sufficient to demonstrate American resolve against North Korea (Kahl, 2017). Similarly, the first direct American military operation against Syria during the civil war, the single-site cruise missile strike against a Shayrat airbase on 7 April 2017, was portrayed as a strong enough signal of American resolve despite its limited scope. Secretary of Defense Jim Mattis defined the strike as a sufficient measure to deter the Syrian government, which “would be ill-advised ever again to use chemical weapons” (US Department of Defense, 2017).⁵

Comparing the mechanisms

While leading to a similar expectation of severe crisis violence in the aftermath of a hostile precrisis period, the two mechanisms differ on which states—the targets or the perpetrators of the hostility—will be more likely to use severe crisis violence.

According to the population fatigue mechanism, crisis escalation is driven by the sense of desperation among the population within the target state and the desire of their political leaders to reduce the domestic audience costs. The mechanism presents the perspective of the target state. The population is tired because it is subject to the chronic hostile acts and expects that a harsh retaliation will stop it from being a target.

According to the baseline for resolve mechanism, states escalate their crisis to demonstrate their resolve in a consistent manner. This mechanism presents the perspective of a state using hostile acts before crisis. Resolve is relative: the strength and credibility of the perpetrator’s resolve is judged by its past behavior. If a state was actively hostile during a precrisis period, it must maintain hostility during the following crisis to credibly convince an opponent of its resolve. This leads to two specific expectations about crisis escalation by the target and by the perpetrator of hostile acts in the absence of a crisis:

Hypothesis 2 (population fatigue): targets of precrisis hostility will be more likely to escalate their subsequent crisis into severe violence compared with states that were not targeted before a crisis.

Hypothesis 3 (the baseline for resolve): perpetrators of precrisis hostility will be more likely to escalate their subsequent crisis into severe violence compared with states that did not commit military hostility before a crisis.

Research design

I estimate the odds of crisis escalation in the aftermath of a hostile precrisis period with binomial logistic regression. I use two units of analysis. A non-directed crisis-dyad is appropriate to test the first hypothesis because of the interest in the precrisis–crisis link, regardless of which of the states engaged in precrisis hostility or escalated the crisis. I use directed crisis-dyad to examine the second and third hypotheses, which focus on assessing the proposed mechanisms. The change in unit of analysis reflects the interest in the foreign policy behavior of specific actors in a crisis—targets or perpetrators of precrisis hostility.

The ICB dataset (version 12, 1918–2015; https://sites.duke.edu/icbdata/) serves as a primary source of information in this study. Following Brecher and Wilkenfeld (2000: 3), an international crisis is a situation in which national leaders perceive an increased threat to their country’s basic values, feel they have a finite time to react to this threat, and believe that there is a heightened risk of military escalation with another state. I use an online dyadic version of ICB (Hewitt, 2003) that has been updated to 2015 to identify recurrent crises between the same pairs of states.⁶ I then use an actor-level version of ICB for information on the foreign policy behavior of crisis actors during international crises.

Crisis escalation is the dependent variable in this study. Derived from the ICB, the variable identifies the intensity of violence employed by each crisis actor during its foreign policy crisis. In line with previous studies on crisis escalation (e.g. Mettler and Reiter, 2012), I dichotomize this variable and code 0 when a state employed no or minor violence during a crisis, and 1 for serious clashes and full-scale war. For non-directed analysis, a crisis is considered to escalate if at least one of the crisis actors in the dyad employed serious clashes or war. For the directed analysis, the focus is on the violence employed by state A in a dyad.

Precrisis hostility, the key independent variable, denotes conflictual military activities by states during a six-month period preceding their crisis. I limit the temporal span of the precrisis period to establish a consistent chronological length among dyads. This is important because the duration of intervals between crises varies greatly. The choice of a six-month window is an arbitrary limit and is supplemented in robustness checks with both shorter and longer time windows.⁷

The information on military hostilities is derived from the dyadic Militarized Interstate Dispute (MID) dataset (Maoz et al., 2019). With details on the start date of dyadic disputes, I identify whether the threat to use force, display of force, or use of force by and between states took place during the six-month period preceding their crisis.⁸ Since military disputes and security crises may overlap, I exclude military disputes that started less than two weeks before a crisis. This is done to diminish the risk that the differences between disputes and crises are driven entirely by differences in coding rules. Excluding military disputes that started less than two weeks before a crisis helps to ensure that hostile military behavior began before a crisis, even if at some point it evolved into a crisis, continued during the crisis, or was even sustained after the crisis.

The MID and ICB projects both focus on conflictual security events between states. Nevertheless, there are meaningful conceptual differences between militarized interstate disputes and international crises (Hewitt, 2003; Hewitt and Goertz, 2005; Leng, 2004). First, military disputes are conflictual events that denote hostile behavior, regardless of how this behavior is perceived by opponents in the conflict. Second, militarized disputes, as the name suggests, must involve a military dimension with at least one of the states making a military threat, displaying military force, or using force against other states. Third, committing such military hostility must be intentional to qualify as a military dispute. In contrast, international crises are a perceptual phenomenon. A crisis may take place without any of the states committing a hostile act or even having an intention to do so. An expectation of massive retaliation for a terror attack by another actor, for instance, may trigger a crisis for both actors even if there was not yet a threat, display, or use of force by either of the opponents in a dyad. While many of the crises may involve an intentional threat, display, or use of force by one state against the other, none of these is a necessary or a sufficient condition of an international crisis. Similarly, while some military disputes may involve a perception of threat, time pressure, and the risk of war—the three conditions of a crisis—none of these is required for an event to be considered a military dispute. This means that military disputes and international crises are conceptually different types of interstate conflictual security events.⁹

The empirical overlap between some military disputes and international crises makes MID less than an ideal choice for the analysis of the precrisis–crisis link. And yet, among available datasets, the MID serves as the appropriate source of information on precrisis hostility for several reasons. First, unlike event datasets that collect information from specific open sources, MID records all instances of military hostility that states explicitly direct toward one another whether these hostilities were reported in the media or not. This diminishes the risk of a reporting bias that might affect results on the impact of precrisis hostility. Second, the temporal scope of the MID dataset is broader than that of other available event datasets, and it provides consistent information on the use of military force for most of the period covered by the ICB. The latest version of the dyadic MID ends in 2010, while the ICB continues to 2015, so overlap between the projects allows the study of the precrisis–crisis link for the period of more than 90 years. Thus, in the absence of a designated dataset on the precrisis behavior, the dyadic MID serves as a useful and reliable source of information.

For non-directed analysis, to probe hypothesis 1, I operationalize two variables on precrisis hostility. The first is a binary measure indicating the presence or absence of military hostility between crisis actors. The presence of precrisis hostility is coded 1 when at least one of the states in a dyad commits a hostile military act—that is, threatens, displays, or uses force—during the six-month period before their crisis. The second variable addresses the extent of continuation of precrisis hostility, operationalized as the number of days during which states in a dyad were involved in a military dispute before their crisis. This is a count variable that ranges from 0 to 180, namely from situations when states did not threaten, display, or use military force before their crisis to situations when the entire period was fully militarized. Values between the extremes indicate various degrees of hostility continuation, with an average of 61 hostile days during a six-month period before a crisis.¹⁰

For directed analysis, to test hypotheses 2 and 3, I operationalize two binary variables. A state is coded as the perpetrator of precrisis hostility when it threatens, displays, or uses military force against its opponent during a six-month period before a crisis and as a target of precrisis hostility when the opponent commits these hostile military acts. The information is derived from MID variables hihosta and hihostb that indicate if, rather than in which order, each state in a dyad threatened, displayed, or used military force during a dispute.

Control variables

To confront the possibility of a spurious relationship between precrisis hostility and crisis violence, I control for additional factors that might affect crisis escalation.

The immediate violence and acute threat that adversaries face at the beginning of a crisis may eliminate the impact of the preceding hostility. The results on the precrisis–crisis link may simply reflect the violence-begets-violence process within a crisis (Wilkenfeld, 1991). I control for how a crisis begins and for what threat it involves. Based on ICB, crisis trigger is dichotomized to indicate whether an international crisis starts with a violent act, such as a border clash or a large-scale invasion. The variable is coded 1 when any state in a dyad (non-directed measure) or state A in a dyad (directed measure) experiences a violent crisis trigger. Gravity of threat is a dichotomous variable coded 1 when any state in a dyad (non-directed measure) or state A in a dyad (directed measure) perceives a highest level of threat in their crisis, namely the threat of grave damage or a threat to their existence.

A past crisis, as a traumatic major event, can also affect states’ decision to escalate (Colaresi and Thompson, 2002; Leng, 1983). It could be that precrisis hostility echoes the continued effect of a previous crisis, rather than playing a distinct role. I control for violence and the outcome of a previous crisis between the states using ICB. Past crisis violence is a dichotomous variable coded 1 when any state in a dyad (non-directed measure) or state A in a dyad (directed measure) utilized serious clashes or war in their previous crisis. Past crisis outcome indicates whether a previous crisis ended with an agreement. The dichotomous variable is coded 1 when any state in a dyad (non-directed measure) or state A in a dyad (directed measure) agrees to a compromise at the termination point of their past crisis.

Past interactions may have a cumulative impact stemming from a conflict history, beyond a single crisis episode. Many of the international crises take place in the setting of protracted conflict or rivalry, a context that makes crisis escalation more likely (Colaresi and Thompson, 2002). The theory offered in this study does not assume the presence of such a context, and yet two-thirds of the dyadic crises in the analysis took place in the setting of a protracted conflict. I therefore control for whether a crisis is an isolated event or part of a longer chain of repetitive hostilities over the same issue. Based on ICB, protracted conflict is a dichotomous variable coded 1 when a crisis is part of a protracted conflict setting for any state in a dyad (non-directed measure) or for state A in a dyad (directed measure). I also control for the number of past crises, a count variable of previous dyadic crises, and conflict duration, which indicates the number of years from the first crisis to a given crisis.

Past research indicates that democratic states are less inclined to use violence in crises (Brecher and Wilkenfeld, 2000: 149; Maoz, 1998). I control for democratic regime using the Polity IV dataset. Joint democracy is coded 1 when both states in a crisis dyad (non-directed measure) or state A in a dyad (directed measure) score at least 6 on the Polity measure. Lastly, I control for the power of adversaries in a crisis. For non-directed analysis, capability asymmetry is calculated by dividing the CINC (Composite Index of National Capability) score of a stronger state by a weaker state in a crisis dyad. A higher value indicates a greater power gap, namely greater power asymmetry. For directed analysis, relative capability is calculated by dividing the CINC score of state A by that of state B, so that a higher value indicates the greater power of state A in a crisis dyad.

Results: effects of precrisis hostility on crisis violence

Findings on the effects of precrisis hostility on crisis violence in Tables 1 and 2 provide empirical support for the first hypothesis. As expected, the use of military force by states prior to a crisis makes crisis escalation more likely.

Table 1.

Precrisis hostility and crisis escalation.

Military hostility before crisis	Crisis escalation
Military hostility before crisis	Absent: crisis involvednone or minor violence	Present: crisis escalated into serious clashes or war
Absent	55%, N = 240	45%, N = 194
Present	37%, N = 152	63%, N = 261
Total	46%, N = 392	54%, N = 455

Note: Pearson chi-square = 29.12, p < 0.000. The table shows the relationship between crisis violence and military hostility during a six-month period preceding an interstate dyadic crisis. The unit of analysis is a non-directed crisis-dyad (N = 847, 1918–2010).

Table 2.

Logistic regression: effect of precrisis military hostility on escalation into severe violence in a subsequent crisis.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Precrisis hostility
Presence of military hostilitybefore crisis	2.124*** (0.338)		1.621**(0.282)	2.353***(0. 496)	2.330***(0.366)	2.164***(0.396)	1.492(0.400)
Extent of continuity of militaryhostility before crisis		1.007***(0.001)						1.006**(0.002)
Crisis attributes
Violent trigger			5.602***(0.882)				7.560***(2.048)	7.521***(2.031)
Threat of grave damage orto existence			6.264***(1.202)				4.799***(1.178)	4.604***(1.132)
Previous crisis_t−1
Severe violence inpast crisis				2.795***(0. 603)			1.960**(0.473)	1.877*(0.468)
Agreement asan outcomeof past crisis				0.747(0.166)			0.873(0.220)	0.773 (0.189)
Dyadic attributes
Joint democracy					0.203*(0.130)		0.732 (0.586)	0.593 (0.480)
Capability asymmetry					0.999*(0.001)		1.000 (0.003)	0.999 (0.003)
Dyadic history
Number of past crises						0.993(0.067)	0.895 (0.054)	0.898 (0.055)
Protracted conflict						4.109***(1.000)	4.724***(2.002)	3.009*(1.467)
Protracted conflict × precrisishostility						0.852 (0.347)	1.343 (0.805)	1.537 (0.558)
Years since first crisis						0.971**(0.010)	0.987 (0.009)	0.988 (0.010)
Observations	847	847	847	512	803	847	500	500
Wald χ²	22.47***	37.89***	164.65***	37.20***	38.95***	101.55***	113.38***	125.39***
Pseudo R²	0.025	0.046	0.215	0.077	0.051	0.115	0.301	0.312

Note: *p < 0.05; **p < 0.01, ***p < 0.001. Robust standard errors are presented in parentheses, clustered by non-directed crisis-dyads. Observations refer to non-directed dyadic international crises, 1918–2010. The values listed above parentheses are odds ratios, which indicate a change in the odds of crisis escalation to serious clashes or war by change in one unit for independent variables. A value lower than 1 indicates a negative relationship; a value higher than 1 indicates a positive relationship.

Table 1 presents a first look at the interphase relationship. It shows that, overall, 54% of dyadic international crises escalated into serious clashes or war. When crises were preceded by military hostility, the escalation rate was higher, 63%. This means that states that had committed military hostilities against one another before a crisis significantly increased their risk of being involved in a violent crisis (p < 0.000). In fact, two of every three dyadic international crises that took place in the aftermath of a hostile precrisis period became violent.

Table 2 provides further evidence of the risks of precrisis hostility, supporting the first hypothesis. The bivariate analysis in Models 1 and 2 shows that both the presence and the dominance of military hostility, respectively, are statistically significant in increasing the odds of crisis escalation into severe violence. Model 1 shows that the odds of violence were more than twice as high after a hostile precrisis period. Model 2 shows that for every additional day of military hostility in the six-month period preceding a crisis, the odds of serious clashes or war increased by 0.7%. If, for example, during a precrisis period the opponents used military force against one another for a total of 30 days, the odds of their crisis escalation increased by more than 20% compared with a crisis that took place in the aftermath of a non-hostile period. This means that the more prevalent military hostility was before crisis onset, the more likely the states in a dyad were to experience severe violence in a subsequent crisis, supporting the first hypothesis. Illustrative of such a precrisis–crisis link is the explanation of the Russian decision to escalate the Russian–Georgia crisis in 2008 into a war. In an interview, Dmitry Medvedev, president of Russia during the crisis, emphasized that the “conflict had been simmering for years.” When “it became clear that those were not just minor provocations but a full-blown attack … I had to give the order to open fire on Georgian troops,” Medvedev explained (Medvedev, 2013). Precrisis violence, even if it is minor and insufficient to trigger a crisis, can still have an impact on the decisions of states to escalate during the subsequent crisis.

Multivariate models in Table 2 demonstrate that the effect of precrisis hostility holds when controlling for potentially confounding factors, such as a violent trigger or a protracted conflict setting. The results on control variables are generally consistent with previous research on crisis escalation. The attributes of an ongoing crisis and the dyadic history are powerful predictors of crisis violence.¹¹ A crisis that began with a violent trigger, involved a high gravity of threat, or was part of a protracted conflict had the greatest odds of escalating into serious clashes or war, as Models 3 and 6 suggest.¹² Crisis violence was also more likely if a previous dyadic crisis involved violence, as Model 4 shows. The outcome of the previous crisis, however, had no effect on crisis escalation. There was also a pacifying effect of a joint democracy and power asymmetry on violence in ongoing crises, although it became statistically not significant in the general model.

The substantive effect of precrisis hostility on crisis violence remains consistent across all models examined. This effect is also statistically significant across all the models, with and without control variables, except for Model 7. Compared with bivariate results, the effect of precrisis hostility is weaker when control variables are included, but it is still meaningful as even in this case precrisis hostility increases the odds of severe crisis violence by half. The comparison of findings in Models 7 and 8 indicates that the extent of continuity of military hostility, rather than its mere presence, has a more profound effect on crisis escalation. Taken together, the findings in Table 2 show that precrisis hostility is an important element for understanding crisis behavior, although not the only or the primary predictor of crisis violence. While other factors, such as a violent trigger, can have a strong effect on crisis escalation, precrisis hostility, especially when it becomes chronic, can provide a good early indication of the risks of potential escalation in the next dyadic crisis.

Assessing the mechanisms

So far, the analysis shows that international crises are increasingly likely to escalate into severe violence in the aftermath of a hostile precrisis period. What explains such an interphase relationship? Do states escalate their crises because their domestic audience is tired of the precrisis hostility and expect leaders to act decisively, as the population fatigue mechanism suggests? Or do states escalate crises because their use of militarily hostile acts in the absence of a crisis sets a higher threshold for a credible demonstration of resolve during subsequent crisis, as the baseline for resolve mechanism suggests? In other words, who is more likely to escalate a crisis to severe violence: the targets or the perpetrators of precrisis military hostility?

To probe the mechanisms, I use a directed crisis-dyad unit of analysis. The motivations and behavior of state A toward state B may differ from that of state B toward state A. Table 3 presents the results from logistic regression models estimating the crisis escalation of targets and perpetrators of precrisis hostility.

Table 3.

Logistic regression: effect of being targeted and using precrisis military hostility on escalation into severe violence in a subsequent crisis (directed).

	Model 1	Model 2	Model 3
Precrisis hostility
Actor is a target	1.545 (0.326)*		0.518 (0.650)
Actor is a perpetrator		1.667 (0.346)**	1.910 (1.593)
Actor is both a target and a perpetrator			4.315 (2.063)**
Crisis attributes
Violent trigger	4.929 (0.965)***	4.964 (0.964)***	4.912 (0.968)***
Threat of grave damage or to existence	4.571 (0.916)***	4.570 (0.914)***	5.002 (1.021)***
Previous crisis_t−1
Severe violence in past crisis	1.990 (0.414)**	1.986 (0.414)**	2.089 (0.439)***
Agreement as an outcome of past crisis	1.243 (0.254)	1.230 (0.253)	1.181 (0.242)
Actor attributes
Democracy	0.759 (0.161)	0.763 (0.162)	0.729 (0.155)
Relative capability	1.003 (0.006)	1.002 (0.007)	1.004 (0.004)
Dyadic history
Number of past crises	0.922 (0.043)	0.917 (0.043)	0.923 (0.043)
Protracted conflict (PC)	2.538 (0.668)***	2.433 (0.641)**	3.941 (1.459)***
PC × both a target and a perpetrator			0.291 (0.153)*
Years since first crisis	0.988 (0.008)	0.988 (0.008)	0.988 (0.008)
Observations	681	681	677
Wald χ²	162.41***	162.98***	158.78***
Pseudo R²	0.247	0.249	0.256

Notes: *p < 0.05; **p < 0.01, ***p < 0.001. Robust standard errors are in parentheses, clustered by directed crisis-dyad. Observations refer to directed dyadic international crises, 1918–2010. Precrisis hostility, the key independent variable, is measured for the six-month period preceding a crisis.

Before assessing each mechanism separately, it is important to note that the complex relations between states might blur the line between user and target type of states. First, states often reciprocate: a state may be both using and being targeted by military force. Second, multiple military acts can occur during the months preceding a crisis. A state may use military force in one event and become a target in another.¹³ Findings in Table 3 show that consideration of such complexity is crucial. In particular, Model 3 shows that precrisis hostility has a strong effect on crisis escalation when an actor has both committed and experienced violence before a crisis. In such situations, the odds of crisis escalation to serious clashes or war are roughly four times higher, compared with the situations when a state is only a user, a target, or neither a user nor a target of precrisis hostility. This suggests that the two mechanisms, the population fatigue and the baseline for resolve, work in conjunction with each other. This conclusion is in line with previous research, specifically with Putnam’s (1988) two-level game theory that suggests that national leaders must consider both domestic pressures and the international consequences of their actions. “Neither of the two games can be ignored by central decision-makers,”Putnam (1988: 434) explains.

When considering the mechanisms separately, the empirical results lend stronger support for the baseline of resolve mechanism, and somewhat weaker support for the population fatigue mechanism. Models 1 and 2 demonstrate that both targets and perpetrators of precrisis hostility are significantly likely to escalate their crises, as expected by the second and third hypotheses, respectively. Yet states that commit military hostility before a crisis (Model 2) have somewhat higher odds of employing severe crisis violence than states that are the targets of precrisis hostility (Model 1). This conclusion is strengthened by findings in Model 3, addressed earlier, which includes targets and perpetrators of hostility in the same equation. The effect of being only a target is negative and not statistically significant, while the substantive effect of only being a user is positive. This suggests that states do not necessarily need to be attacked by their opponent before their crisis to employ severe violence during the crisis. At the same time, committing hostility before a crisis appears to increase the risk of crisis escalation. The use of force against an opponent before a crisis can increase the threshold for what can be considered as resolute behavior during the crisis, locking the state into its position and urging it to use higher violence to demonstrate its firm stand over the stakes. This is consistent with crisis bargaining theory. The enemy is “the most important audience” in judging a state’s resolve, Schelling (2008: 141) explains.

To be sure, the analysis of mechanisms as a stand-alone should be considered as suggestive, given the already mentioned complexity of state relations and the necessity to consider both the domestic and international spheres. Yet, while it is clear across all models that both mechanisms may be at work, the finding that the baseline for resolve has a somewhat more profound role is an important step in understanding the role of precrisis behavior in increasing the risk of crisis escalation.

While further and in-depth analysis of the mechanisms, separately and together, is needed, this study demonstrates that precrisis hostilities matter. In choosing crisis behavior, states often take into account both the domestic and international implications but seem to give some primacy to demonstrating their resolve through a consistent use of force across conflict phases. This means that the behavior of states before their crisis can help us better understand their behavior during a crisis. The exploration of interphase relationships should therefore be taken seriously in conflict and security scholarship, and especially in international crises as major acute episodes in conflict processes.

Robustness checks

Numerous robustness checks were performed to verify that the results are not driven by the choice of operationalization for the key variables. Re-examination of the precrisis–crisis model with alternative measures of crisis violence and precrisis hostility supported the main results, in terms of both substantial findings and their statistical significance. The main conclusion of this study remains intact: international crises, when preceded by hostile precrisis periods, were considerably more likely to escalate into severe violence compared with other international crises where hostility during the precrisis period was absent or more limited.

The results from the robustness checks are presented in the Supplemental Material. Here, I summarize the main points. In the original model, I used severity of crisis violence as the dependent variable. This variable implies that a crisis actor has employed violence as a primary crisis management technique, but the severity, or intensity, of this use differed. For the robustness checks I use two alternative, broader specifications of crisis violence in the ICB. The first indicates whether the violence was a primary management technique in the first place, while the second focuses on how central, or important, state leaders believed that this violence was to achieving their goals during the crisis. With the use of alternative measures for crisis violence, the results of this study remain intact.

With regard to the key independent variables, two main changes were implemented for the robustness checks. First, precrisis hostility was refined to include only the use of force, rather than all expressions of military hostility. Second, I altered and relaxed the restrictions on the length of the precrisis period. The main analysis reported in this study focused on a six-month period preceding a crisis. For the robustness checks, I looked at hostility for one month, two months, and one year preceding a crisis, as well as at military hostility during the entire interval between recurrent crises. As before, the results remained largely unchanged, leading to the same conclusion of the existence of the interphase relationship between crisis behavior and precrisis hostility.

Lastly, to account for potential selection bias, I used a Heckman probit model. The crisis onset was analyzed at the first stage and crisis escalation at the second stage in the model (Reed, 2000). Results in Table 4 indicate that even when accounting for crisis onset, the effect of the main independent variable remains positive and statistically significant. This means that precrisis military hostility still increased the likelihood of crisis escalation, demonstrating the robustness of the findings on the precrisis–crisis link in international conflicts.

Table 4.

Robustness checks: Heckman selection models (dyad-year, 1918–2010).

	Robustness for Model 7in Table 2	Robustness for Model 8 in Table 2
Crisis escalation
Presence of military hostility before crisis	0.321***(0.084)
Extent of continuity of military hostility before crisis		0.003***(0.001)
Violent trigger	1.267***(0.104)	1.259***(0.101)
Threat of grave damage or to existence	0.924***(0.098)	0.899***(0.097)
Severe violence in past crisis	0.780***(0.101)	0.759***(0.010)
Agreement as an outcome of past crisis	−0.067(0.092)	−0.101(0.092)
Joint democracy	0.087(0.312)	0.010(0.314)
Capability asymmetry	−5.79(0.000)	−0.0001(0.000)
Number of past crises	−0.069(0.036)	−0.062(0.036)
Protracted conflict	0.703***(0.101)	0.636***(0.099)
Protracted conflict × precrisis hostility	−0.007(0.004)	−0.008(0.004)
Years since first crisis	−1.716**(0.514)	−1.719**(0.496)
Crisis onset
Severe violence in past crisis	0.154***(0.032)	0.154***(0.032)
Agreement as an outcome of past crisis	−0.095**(0.036)	−0.095**(0.037)
Joint democracy	−0.380***(0.089)	−0.380***(0.089)
Capability asymmetry	−0.0002**(0.0001)	−0.0002**(0.0001)
Number of past crises	0.087***(0.010)	0.087***(0.010)
Peace years	−0.047***(0.004)	−0.047***(0.004)
Peace years²	0.001(0.0001)	0.001***(0.0001)
Peace years³	−3.36***(5.98)	−3.35***(5.99)
Observations	12,376	12,376
Wald χ²	355.90***	397.62***

Note: *p < 0.05; **p < 0.01, ***p < 0.001. Robust standard errors are presented in parentheses, clustered by directed crisis-dyads. Observations refer to dyad-years.

Conclusions

Motivated by the need for greater attention to the intervals between recurrent confrontations in conflict research, this study focused on the link between precrisis hostility and crisis escalation. It has demonstrated that a threat, display, or use of military force before a crisis can increase the risk of severe violence during a subsequent crisis. Results provide consistent evidence that international crises that occur in the aftermath of a militarily hostile period are significantly more likely to escalate into serious clashes and full-scale war. This means that periods preceding crises, rather than being merely the intervals of peace on the sidelines of a conflict, are important elements for understanding crisis behavior. Periods between recurrent confrontations matter. In studying why some international crises escalate into violence while others do not, we must look not only at past and ongoing crises but also outside the confines of a major confrontation and address the role of states’ behavior during the routine period of their relationship.

Two theoretical possibilities were suggested for connecting hostility before crises and violence in crises: domestic population fatigue and consistent demonstration of resolve. Findings have provided support for both explanations but showed that it is mainly the resolve mechanism that is at work. Committing military hostility in the absence of a crisis sets a higher threshold for the crisis management options that states can use to demonstrate their firm stand. Resolve is relative. The past routine behavior of states that involves a relatively low risk of misrepresentation can serve as a credible reference point for evaluating the resolve. A self-imposed restraint during a crisis, despite the active use of military hostility beforehand, signals a state’s weakness and creates a barrier for it to reach a more favorable crisis outcome. Thus, a state using military force before a crisis has an increased propensity to escalate in a subsequent crisis.

The next step in assessing the effects of precrisis hostility requires a thorough qualitative examination of state motives. Collection of new data on interactions during the intervals between crises will advance future work on the empirical evaluation of the theoretical mechanisms that operate behind the links of crisis escalation and precrisis hostility.¹⁴ The theoretical framework and empirical evidence from this study will hopefully stimulate a greater interest in precrisis interactions and their role in the evolution of international conflicts. For its part, this study demonstrates that precrisis hostilities matter and can add to our understanding of international crisis behavior.

These findings have important implications for research on international conflicts, crises, and wars. Adding to the extensive research on the tense spikes in world politics, they demonstrate how integrating a precrisis perspective can contribute to the better understanding of crisis escalation. They emphasize that crisis and precrisis periods are both essential components of a conflict process—a conventional reflection of the real world that has yet to be fully integrated into conflict studies. It was more than three decades ago when Leng (1983, 415), in one of the pioneering studies on recurrent crises, suggested that “we cannot entirely rule out the possibility that some of the bargaining that we observed [during international crises] may be caused by events occurring between the crises.” And yet, while different approaches in the conflict literature agree that international conflicts consist of interlinked phases, the intervals between recurrent confrontations are usually set aside in favor of studying more acute conflict situations. This study challenges such practice and joins a few recent efforts that call greater attention to various segments of a conflict process (Diehl, 2006; Iakhnis and James, 2019; Senese and Quackenbush, 2003). We still know relatively little about respite periods between recurrent confrontations, and even less so about the role they play in conflict escalation. This study should encourage future research to seriously consider the role of the precrisis interactions in their theoretical and empirical analyses of international crises and recurrent conflicts.

The findings also have important policy implications. First, they make it clear that history matters. Interactions during precrisis periods can serve as early signals for crisis behavior. The behavior of states today, even during relatively tranquil and routine periods in their relationship, shapes what happens tomorrow. It might be tempting for a state, especially for a weaker or dissatisfied one, to behave more boldly in the absence of a crisis as the risk of immediate escalation is relatively low (e.g. Maoz and San-Akca, 2012). Yet such behavior can have undesirable longer-term consequences. While chronic use of limited force may not be sufficient for massive retaliation and immediate escalation, it should not be considered as a safe policy for conflict management. Eventually, many international crises recur, especially when the issues under contention remain unresolved. When this happens, states that behaved aggressively in the absence of a crisis might find themselves trapped in a violent maze. Crises do not arise from “nowhere”; they are a part of broader conflict processes, which include both the previous crises, as past studies have extensively discussed, and the precrisis periods, as highlighted in this study.

There is also a more positive implication of the findings on the precrisis–crisis link. The stronger support for the baseline for resolve mechanism suggests that states can, in advance, by manipulating their own precrisis behavior, have at least some control over crisis escalation. While a target state cannot choose whether to be a target, the perpetrator of hostility can decide whether or not to commit military hostility during a precrisis period. By limiting the use of hostile military activities in the absence of a crisis, leaders not only can manage their conflicts in a more peaceful way during precrisis periods but can also reduce the risk of violence during their next international crises. This policy will not solve international conflicts, but at least it might contribute to a more positive climate between states and greater regional and international security.

Supplemental Material

Appendix_Table_1___Precrisis_Military_Hostility_and_Escalation_in_International_Crises – Supplemental material for Precrisis military hostility and escalation in international crises

Supplemental material, Appendix_Table_1___Precrisis_Military_Hostility_and_Escalation_in_International_Crises for Precrisis military hostility and escalation in international crises by Luba Levin-Banchik in Conflict Management and Peace Science

Supplemental Material

Appendix_Table_2___Precrisis_Military_Hostility_and_Escalation_in_International_Crises – Supplemental material for Precrisis military hostility and escalation in international crises

Supplemental material, Appendix_Table_2___Precrisis_Military_Hostility_and_Escalation_in_International_Crises for Precrisis military hostility and escalation in international crises by Luba Levin-Banchik in Conflict Management and Peace Science

Research Data

precrisis – for Precrisis military hostility and escalation in international crises

precrisis for Precrisis military hostility and escalation in international crises by Luba Levin-Banchik in Conflict Management and Peace Science

This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https://creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage).

Research Data

precrisis_Do_file – for Precrisis military hostility and escalation in international crises

precrisis_Do_file for Precrisis military hostility and escalation in international crises by Luba Levin-Banchik in Conflict Management and Peace Science

Research Data

precrisis_dyadyear – for Precrisis military hostility and escalation in international crises

precrisis_dyadyear for Precrisis military hostility and escalation in international crises by Luba Levin-Banchik in Conflict Management and Peace Science

Footnotes

Acknowledgements

I would like to thank Zeev Maoz, Michael Brecher, Fiona Ogunkoya, Patrick James, David Lake, Laura Sjoberg, Matthew Hoffman, John Raymond, Hemda Ben-Yehuda and the anonymous reviewers for their helpful feedback at different stages of the project.

Author’s note

Earlier versions of this study were presented as part of the Emerging Scholar Forum at the Annual Meetings of International Studies Association West in Pasadena, California in 2017 and 2018. An earlier version of this paper was awarded the Best Faculty Paper Award by ISA-West in September 2019.

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the Israel Institute Postdoctoral Research Grant (Grant Number GL#20163).

ORCID iD

Luba Levin-Banchik

Supplemental material

The Online Appendix is available online through the SAGE CMPS website:

Notes

References

Abramson

Carter

(2016) The historical origins of territorial disputes. American Political Science Review 110(4): 675–698.

Adams

(2003) War rhetoric escalates across tense Mideast lines: Sharon vows to hit enemies in “any place,” as Syria warns that it is “not a weak state”. Globe and Mail (Toronto), A1.

Andersen-Rodgers

(2015) No table necessary? Foreign policy crisis management techniques in non-state actor-triggered crises. Conflict Management and Peace Science 32(2): 200–221.

Asal

Beardsley

(2007) Proliferation and international crisis behavior. Journal of Peace Research 44(2): 139–155.

Ben-Yehuda

Levin-Banchik

(2010) The dangers of terror in world politics: International terror crises, 1918–2006. Studies in Conflict & Terrorism 34(1): 31–46.

Ben-Yehuda

Levin-Banchik

(2014) Regime and power in international terror crises: Strong democracies fight back hard. Terrorism and Political Violence 26(3): 504–522.

Braithwaite

Lemke

(2011) Unpacking escalation. Conflict Management and Peace Science 28(2): 111–123.

Brecher

(2008) International Political Earthquakes. Ann Arbor, MI: University of Michigan Press.

Brecher

Wilkenfeld

(2000) A Study of Crisis. Ann Arbor, MI: University of Michigan Press.

10.

Canetti

(2017) Emotional distress, conflict ideology, and radicalization. PS: Political Science & Politics 50(4): 940–943.

11.

Canetti

Elad-Strenger

Lavi

Guy

Bar-Tal

(2017) Exposure to violence, ethos of conflict, and support for compromise: Surveys in Israel, East Jerusalem, West Bank, and Gaza. Journal of Conflict Resolution 61(1): 84–113.

12.

Chong

Hall

(2017) One thing leads to another: Making sense of East Asia’s repeated tensions. Asian Security 13(1): 20–40.

13.

Colaresi

(2004) When doves cry: International rivalry, unreciprocated cooperation, and leadership turnover. American Journal of Political Science 48(3): 555–570.

14.

Colaresi

Thompson

(2002) Strategic rivalries, protracted conflict, and crisis escalation. Journal of Peace Research 39(3): 263–287.

15.

Colaresi

Rasler

Thompson

(2008) Strategic Rivalries in World Politics: Position, Space and Conflict Escalation. Cambridge: Cambridge University Press.

16.

Crescenzi

MJC

Kathman

Long

(2007) Reputation, history, and war. Journal of Peace Research 44(6): 651–667.

17.

Davies

GAM

Johns

(2016) The domestic consequences of international over-cooperation: An experimental study of microfoundations. Conflict Management and Peace Science 33(4): 343–360.

18.

Debs

Weiss

(2014) Circumstances, domestic audiences, and reputational incentives in international crisis bargaining. Journal of Conflict Resolution 60(3): 403–433.

19.

Diehl

(2006) Just a phase? Integrating conflict dynamics over time. Conflict Management and Peace Science 23(3): 199–210.

20.

Diehl

Goertz

(2000) War and Peace in International Rivalry. Ann Arbor, MI: University of Michigan Press.

21.

Fearon

(1994) Domestic political audiences and the escalation of international disputes. American Political Science Review 88(3): 577–592.

22.

Fortna

(2003) Scraps of paper? Agreements and the durability of peace. International Organization 57(2): 337–372.

23.

Gibler

Owsiak

(2018) Democracy and the settlement of international borders, 1919 to 2001. Journal of Conflict Resolution 62(9): 1847–1875.

24.

Goertz

Diehl

(1993) Enduring rivalries: Theoretical constructs and empirical patterns. International Studies Quarterly 37(2): 147–171.

25.

Goertz

Diehl

(2000) (Enduring) rivalry. In: Midlarsky

(ed.) Handbook of War Studies II. Ann Arbor, MI: University of Michigan Press (pp. 129–163).

26.

Goertz

Diehl

Balas

(2016) The Puzzle of Peace. Oxford: Oxford University Press.

27.

Hensel

(1994) One thing leads to another: Recurrent militarized disputes in Latin America, 1816–1986. Journal of Peace Research 31(3): 281–297.

28.

Hewitt

(2003) Dyadic processes and international crises. Journal of Conflict Resolution 47(5): 669–692.

29.

Hewitt

Goertz

(2005) Conceptualizing interstate conflict: How do concept-building strategies relate to selection effects? International Interactions 31(2): 163–182.

30.

Iakhnis

James

(2019) Near crises in world politics: A new data set. Conflict Management and Peace Science. First published online: doi.org/0738894219855610.

31.

James

(2019) What do we know about crisis, escalation, and war? A visual assessment of the International Crisis Behavior Project. Conflict Management and Peace Science 36(1): 3–19.

32.

Kahl

(2017) The United States should resolve to avoid war with North Korea in 2018. Foreign Policy, 27 December. Available from: http://foreignpolicy.com/2017/12/27/the-united-states-should-resolve-to-avoid-war-with-north-korea-in-2018/.

33.

Lektzian

Souva

(2009) A comparative theory test of democratic peace arguments, 1946–2000. Journal of Peace Research 46(1): 17–37.

34.

Leng

(1983) When will they ever learn? Coercive bargaining in recurrent crises. Journal of Conflict Resolution 27(3): 379–419.

35.

Leng

(2004) Escalation: Competing perspectives and empirical evidence. International Studies Review 6(4): 51–64.

36.

Levin-Banchik

(2018) Assessing knowledge retention, with and without simulations. Journal of Political Science Education 14(3): 341–359.

37.

Lupton

(2018a) Signaling resolve: Leaders, reputations, and the importance of early interactions. International Interactions 44(1): 59–87.

38.

Lupton

(2018b) Reexamining reputation for resolve: Leaders, states, and the onset of international crises. Journal of Global Security Studies 3(2): 198–216.

39.

Maoz

(1998) Realist and cultural critiques of the democratic peace: A theoretical and empirical re-assessment. International Interactions 24(1): 3–89.

40.

Maoz

Mor

(2002) Bound by Struggle: The Strategic Evolution of Enduring International Rivalries. Ann Arbor, MI: University of Michigan Press.

41.

Maoz

San-Akca

(2012) Rivalry and state support of non-state armed groups (NAGs), 1946–2001. International Studies Quarterly 56(4): 720–734.

42.

Maoz

Johnson

Kaplan

Ogunkoya

Shreve

(2019) The Dyadic Militarized Interstate Disputes (MIDs) Dataset version 3.0: Logic, characteristics, and comparisons to alternative datasets. Journal of Conflict Resolution 63(3): 811–835.

43.

McLaughlin Mitchell

Thies

(2011) Issue rivalries. Conflict Management and Peace Science 28(3): 230–260.

44.

McManus

(2014) Fighting words: The effectiveness of statements of resolve in international conflict. Journal of Peace Research 51(6): 726–740.

45.

Medvedev

(2013) Medvedev: Russia’s top priority in S. Ossetia War was to defend our citizens’ interests. https://www.rt.com/news/georgia-south-ossetia-medvedev-interview-012/.

46.

Mettler

Reiter

(2012) Ballistic missiles and international conflict. Journal of Conflict Resolution 57(5): 854–880.

47.

Mishali-Ram

(2006) Ethnic diversity, issues, and international crisis dynamics, 1918–2002. Journal of Peace Research 43(5): 583–600.

48.

Myre

(2003) The Mideast turmoil: Airstrike; Israel attacks what it calls a terrorist camp in Syria. New York Times, 6 October. Available from:https://www.nytimes.com/2003/10/06/world/mideast-turmoil-airstrike-israel-attacks-what-it-calls-terrorist-camp-syria.html.

49.

Nomikos

Sambanis

(2019) What is the mechanism underlying audience costs? Incompetence, belligerence, and inconsistency. Journal of Peace Research, published online 11 June. https://doi.org/10.1177/0022343319839456.

50.

Peace Index (2006) July. Peace Index Project, directed by Ephraim Yaar and Tamar Hermann. Tel Aviv: Tami Steinmetz Center for Peace Research and Evens Program in Mediation and Conflict Resolution of Tel Aviv University. Available from: http://www.peaceindex.org/indexMonthEng.aspx?num=33&monthname=July.

51.

Peffley

Hutchison

Shamir

(2015) The impact of persistent terrorism on political tolerance: Israel, 1980 to 2011. American Political Science Review 109(4): 817–832.

52.

Press

(2005) Calculating Credibility: How Leaders Assess Military Threats. Ithaca, NY: Cornell University Press.

53.

Putnam

(1988) Diplomacy and domestic politics: The logic of two-level games. International Organization 42(3): 427–460.

54.

Reed

(2000) A unified statistical model of conflict onset and escalation. American Journal of Political Science 44(1): 84–93.

55.

Schelling

(2008) Arms and Influence: With a New Preface and Afterword. New Haven, CT: Yale University Press.

56.

Senese

Quackenbush

(2003) Sowing the seeds of conflict: The effect of dispute settlements on durations of peace. The Journal of Politics 65(3): 696–717.

57.

Senese

Vasquez

(2008) The Steps to War: An Empirical Study. Princeton, NJ: Princeton University Press.

58.

“Special U.N. Security Council Meeting to Discuss Israel’s Attack in Syria” (2003) CNN, 5 October. Available from: http://transcripts.cnn.com/TRANSCRIPTS/0310/05/se.03.html.

59.

Trumbore

Boyer

(2000) International crisis decision-making as a two-level process. Journal of Peace Research 37(6): 679–697.

60.

US Department of Defense (2017) Statement by Secretary of Defense Jim Mattis on the U.S. military response to the Syrian government’s use of chemical weapons. Press release no. NR-129-17. Available from: https://www.defense.gov/News/News-Releases/News-Release-View/Article/1146758/statement-by-secretary-of-defense-jim-mattis-on-the-us-military-response-to-the/.

61.

Vasquez

(2009) The War Puzzle Revisited. Cambridge: Cambridge University Press.

62.

Wiegand

(2011) Militarized territorial disputes: States’ attempts to transfer reputation for resolve. Journal of Peace Research 48(1): 101–113.

63.

Wilkenfeld

(1991) Trigger–response transitions in foreign policy crises, 1929–1985. Journal of Conflict Resolution 35(1): 143–169.

64.

Wolford

(2007) The turnover trap: New leaders, reputation, and international conflict. American Journal of Political Science 51(4): 772–788.

65.

Wolford

(2012) Incumbents, successors, and crisis bargaining: Leadership turnover as a commitment problem. Journal of Peace Research 49(4): 517–530.

66.

Wolford

(2014) Power, preferences, and balancing: The durability of coalitions and the expansion of conflict. International Studies Quarterly 58(1): 146–157.

67.

Wright

Diehl

(2016) Unpacking territorial disputes: Domestic political influences and war. Journal of Conflict Resolution 60(4): 645–669.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.08 MB

0.20 MB

0.00 MB

133.11 MB