Over the last twenty years, the steps-to-war thesis has become one of the dominant explanatory frameworks in the empirical literature on the causes of war. The root of this thesis is the robust finding that territorial conflicts between states are far more likely than other types of disputes to result in war, thus creating a particular problem for diplomatic resolution. While not attempting to account for all types of wars or paths to conflict, the theory contends that territorial issues tend to have unusually high salience, and states are likely to attempt to adjudicate these disputes using policies promoted by realist interpretations of the international system.Footnote 1 Presuming states typically wish to avoid war if possible given its high cost, the steps-to-war theory argues that these policies have the ultimately ironic effect of increasing the probability of war with each step taken, rather than decreasing it.Footnote 2 An understanding of this interaction between high salience issue and realpolitik tactics then, within the frame of steps-to-war, would constitute one important path to war and to our understanding of conflict processes generally.
This study takes as its starting point the fact that substantial statistical testing of the steps-to-war thesis and the relationships within it has been completed, and has largely validated its empirical claims. The next step in testing the theory is to focus directly on the underlying causal mechanism described by the framework and to assess its validity vis-à-vis other theoretical explanations for the same empirical correlations that provide support for the steps-to-war explanation. The statistical research to date, while invaluable in demonstrating the foundational relationships, has not been able to address satisfactorily whether this is the result of the steps-to-war thesis offering the best explanation for territorial conflicts as a class, or whether the different explanations represent different real paths to conflict within the class.Footnote 3 This study directly addresses those possibilities through analytical evaluation of appropriate historical cases, focusing in particular on the endogeneity question of the anticipation of war as a cause of both war and the ‘steps.’
Within the steps-to-war theory, war is ‘chosen’ step-by-step, and the probability of war is significantly decreased at any point where the next step, because of diplomatic intervention or other reasons, is not taken. This causal framework then explains how each step is individually associated with war as an outcome, and why the combination of policies would also be associated with that outcome. It is not, however, the only plausible explanation of the same empirical correlation between these variables.
Two substantial challenges related to the question of endogeneity in the theory have been raised.Footnote 4 Endogeneity problems arise when the causal arrow between independent and dependent variables may be reversed, and this problem is seen in the research into the causes of war. Levy outlines the problem in the deterrence literature as one example.Footnote 5 Studies have found that the power balance between states does not strongly impact the outcome of crises, or that deterrence does not necessarily succeed if the target state has a major ally. These findings may be explained, however, by the fact that states consider these things before challenging the status quo, and simply do not initiate crises they are not willing to take forward.Footnote 6 The power balance or the alliance may then be related to the outcome, but cannot be said to cause it because the original challenge was partially determined by the balance of power in the first place. The problem in the case of deterrence is that our conclusions can be severely skewed by the difficulty of accounting for successful deterrence – potential cases where no actual challenge was ever made. In the case of the steps-to-war thesis, it is not a problem of challenges that never existed, but rather alternative explanations that argue that the ‘steps’ are a consequence, not a cause, along the road to war.
The first argument, and the one that animates this study, is the problem of states anticipating war and preparing for it in advance of actual large-scale combat. Slantchev suggests that leaders anticipating war may ‘use territorial claims to consolidate domestic support for their war policies.’Footnote 7 This alternative suggests that the reason that arms buildups, alliances, and repeated disputes, as well as territorial disputes themselves, are associated with war is that leaders either make a decision to go to war, or assume war is highly probable given current conditions, then engage in policies to augment their capacity to fight a war in advance of the actual commencement of hostilities.Footnote 8 If states are anticipating war and acting accordingly, then the (expectation of) war is causing the ‘steps’ that occur prior to the outbreak of war, not the other way around (the very exemplar of an endogeneity problem). In that case, each policy choice is a reflection of (caused by) the anticipated war, which is in contradiction to the causal logic of the steps-to-war model.
Ripsman and Levy’s assessment of British appeasement strategies in the 1930s provides a case study of the distinction between the causal arguments.Footnote 9 History has not dealt kindly with Neville Chamberlain’s recognition of Hitler’s claim to the Sudetenland and subsequent pronouncement that he had secured ‘peace for our time.’ Ripsman and Levy argue that Chamberlain was not, as usually assumed, foolishly naїve about the likelihood of war with Germany, but rather that he was consciously and strategically buying time for Britain to rearm. Believing that an attempt to defend Czechoslovakia would result in failure with catastrophic consequences for everyone, Chamberlain deliberately sacrificed the country for the time to rebuild Britain’s armaments for the war that was likely to come. The obvious implication of this is that Britain’s arms buildup prior to the war should not be seen as a step that increased the probability that war would subsequently occur, but as conscious preparation for a war already expected (even if there was still some small uncertainty).
The distinction between the causal frames is clearly relevant in cases where leaders are trying to solidify domestic support for a war decision already taken. It is also relevant, however, if we accept that there can be unintended consequences in cases where the state assumes that war is likely to occur and chooses policies that then increase the chance of bringing about that war in a sort of final self-fulfilling prophecy.
The anticipation argument presupposes that leaders already assess the likelihood of war as very high. It is possible that their policies may then increase that probability, but it would be wrong for us to understand the war as fundamentally being caused by the intervening policies – its fundamental cause lies before those policies are taken. If this is the case, focusing diplomatic intervention on arms policies or alliance-building is largely missing the point, or the real cause, of the likely conflict. Diplomatic efforts are only likely to decrease the probability of war if they bracket those policies out and focus almost entirely on the underlying political conditions that have precipitated the move toward war, and the resolution of these issues then make policies like arms races simply irrelevant.Footnote 10
The steps-to-war thesis, in contrast, anticipates a relatively low risk of war, even with a territorial dispute (more contentious than other issues), until the intervening policies make that war more probable, effectively ‘causing’ a war that was not likely to have happened otherwise. The implication here is that diplomatic efforts to reduce arms buildups and discourage exclusive alliances should be encouraged as ways of reducing the probability that the underlying political conflict will even rise to the point that war becomes likely. Resolving the territorial dispute itself is certainly beneficial in the long run, but not absolutely required to keep the likelihood of war low in the absence of other steps being taken. The realpolitik policies themselves also must be directly addressed, such as in an arms control agreement, and considered of primary importance to preventing war.
Beyond anticipation of war, a second alternative to the explanation for the statistical relationships between the ‘steps’ and war is related to the role of interstate rivalry. Rivalry, it is argued, instead of being just another step toward war, is the cause of both war and the ‘steps.’Footnote 11 States that are rivals see themselves as embedded in a relationship of enmity that is assumed to have an indefinite future. Rivals are more likely to experience war, and they are more likely to seek allies, to engage in repeated disputes, and to build up their militaries. While it may be that each of these policies reinforces the rivalry and impacts future levels of conflict, the steps are essentially reflecting the same rivalry that the war itself reflects. Once again, one would find a correlation between the steps and war, but the causal mechanism is not fundamentally that indicated by the steps-to-war thesis: in order to understand how war comes about, our focus should be on the rivalry directly, rather than the specific impact of the steps.Footnote 12
Some attempt has been made to address the latter challenge through empirical studies controlling for rivalry in a variety of ways,Footnote 13 and while this study furthers that investigation, it focuses in particular on a systematic assessment of the anticipation of war argument. Do states anticipate war and then engage in a range of conflictual behavior in advance of actual hostilities? Or do those conflictual behaviors – arms races, alliance-building, repeated ‘testing’ disputes – actually bring about a war that might have been relatively unlikely (even over a salient issue) in their absence?
The results of the historical analysis offered here allow us to draw several conclusions. First, among the class of wars over territory, different categories of cases, or likely paths to war, can be seen. However, the results indicate that while states do ‘anticipate’ war, it is not a theoretically significant fact: anticipation of war is not ‘causing’ arming, alliance-building, etc. in a meaningful way. Rather, states anticipate war, and engage in final preparations (including the re-initiation of ‘settled’ territorial claims), after their relations with the opposing state have deteriorated over the course of time. And critically, the process of that deterioration occurs in ways predicted by the steps-to-war theory, as we see both general patterns of emerging policy as predicted by the theory, and specific individual reactions by policy makers in the cases that reflect the expected action–reaction process within the theory. The implications for practical diplomatic efforts to avert higher levels of conflict are profound.
THEORETICAL BACKGROUND
Within the steps-to-war framework, the argument is that states attempt to ‘win’ territorial disputes – either to gain sovereignty over the disputed territory or to indefinitely prevent the other state from doing so – by employing a series of realpolitik measures that are intended to demonstrate both their will to win and their capability. They make alliances to augment their power at comparatively small cost. Because there is an ongoing conflict over the territory, they are likely to engage in repeated militarized disputes short of war; they may well build up their militaries, and because of all of this, a genuine rivalry may emerge between the states. In the steps-to-war formulation, each ‘step’ increases the probability that the next will be taken. Rather than demonstrating resolve and highlighting what is intended to be a rational negotiation over the value of the particular issue at stake, each event or policy alters the context in which the next is to be taken, increasing the perception of threat and hostility and the perceived likelihood that the next step is necessary to avoid loss or the perception of weakness. Ultimately, war is seen as an essentially inevitable outcome by the involved states, particularly if hardline leaders come into power in one or both states.Footnote 14 At this point, states do begin to ‘prepare for war’, but only after the relationship has followed a fairly predictable path through the steps of conflict. War then occurs when the context has become highly contentious, and an incident between the states progressively escalates to the point of high-level violence.Footnote 15
Previous literature on the steps-to-war thesis has primarily sought to test the key proposed relationships, both individually and in combination. Because the theory contended that alliances and arms buildups, rather than deterring conflict,Footnote 16 would increase the probability of war, determining the real impact of these policy choices and the other steps on the likelihood of war has been critical to determining its validity to this point. Empirical testing has found considerable support for the relationships proposed by the theory, and these findings are robust across different units of analysis and samples.Footnote 17 Territorial conflicts are indeed more war-prone than conflict over other issues;Footnote 18 arms buildups increase the probability of conflict between states,Footnote 19 as do some kinds of alliances;Footnote 20 states that engage in multiple territorial disputes,Footnote 21 and those that fall into rivalries are more likely to find themselves experiencing war.Footnote 22
While invaluable in determining the validity of the relationships proposed by the theory, statistical modeling is not the best way of differentiating between the varying explanations for the uncovered associations. Moving beyond modeling the statistical relationships, this study takes the next logical step in testing the steps-to-war theory by addressing the endogeneity challenges, with particular emphasis on the question of what it means when states anticipate war. It does so through a close examination of a theoretically-relevant sample of historical cases, evaluating their timelines and the sequence of events that led them to war. By direct examination of historical events, we can determine if those histories fit a pattern that reflects the causal framework of the steps-to-war thesis, or clearly indicate that one or other of the arguments better explains why we see a confluence of the ‘steps’ with war.
This qualitative approach to assessment of the historical timelines of the cases is also encouraged by the fact that there is substantial flexibility built into the steps-to-war theory. There is no rigid pattern in the order of which the steps should appear. Earlier formulations of the theory, in particular, suggested that the order was not vital, only that each step individually and in combination added to the probability of war. Senese and Vasquez later argued that the logical sequence is likely to follow a rational frame, with policies assumed to have lower costs (like alliance-making) preceding higher cost policies (like arms buildups), but fundamentally, a certain amount of variance may well be seen in the order of the steps, and the causal path would still be seen as conforming to the logic of the steps-to-war.Footnote 23
Significantly, however, not every sequence or pattern of timing would provide confirmatory evidence for the theory. For the theoretical mechanism argued in the steps-to-war thesis to be valid, there must be time for each step to alter decision makers’ reading of the context, increasing their perception of hostility, etc. and thus causing the next step to be taken.Footnote 24 Senese and Vasquez build in this notion that some time must pass between step and war to consider it a causal relationship when they only include in their testing those alliances that occur at least three months before the start of a war (thus avoiding any suggestion of causality in cases like the Nazi–Soviet nonaggression pact in late August 1939).
If it is found that the steps and war occur within one short period of time, one could not easily make a strong claim that each step was causing the next, and increasing the chance of war. In such a case it would be equally plausible that the war was already seen to be a near foregone conclusion by at least one party to the conflict, and war preparation was under way.Footnote 25 In other words, we could not reject the argument that anticipation of war is the best explanation for the uncovered correlations, rather than war being seen as relatively unlikely in the beginning, but made highly probable by the policies. In contrast, if the steps are occurring across a significant stretch of time (even across changing governments or regimes) it would be difficult, if not impossible, to sustain an argument that states had anticipated a war with such certainty that all the steps can only be realistically thought of as simply preparation for an assumed future. In such situations, it is more plausible that the steps would, in fact, be having some causal impact on each other, and we could consider the pattern as providing confirming evidence for the steps-to-war model.
In addition, while considerable flexibility is assumed in the theory, one would not expect rivalry to precede other steps in the model. In the steps-to-war formulation, rivalry may well emerge as a step following repeated disputes, rapid arming, and making alliances, and thus itself increase the probability of war; however, if it precedes the other steps, then it would be difficult to argue that the steps were causing an increase in the probability of war, rather than the rivalry itself causing both steps and war.
These different theoretical explanations for the known correlations between the steps and war should, in fact, demonstrate different timing and sequential patterns among the steps, thus a careful examination of the historical timelines of theoretically appropriate cases can offer us a way of testing the validity of the claims made against the steps-to-war model. For the steps-of-war thesis to be supported, we would expect that there would be some time between the taking of the steps, reaction to them, and later escalation to war. If we find, to the contrary, that the timing of the steps and war is so compacted that the steps cannot realistically be having a causal impact on one another, or alternatively, that rivalry precedes other steps, then the theoretical explanation offered by the steps-to-war will have been largely refuted. If we find that the timing is not compacted, and that rivalry does not precede the other relevant steps, this would provide positive evidence supporting the theoretical mechanism of the steps-to-war argument, with important implications for the relevance of specific targeted diplomatic intervention into potential conflicts.
METHODS
This study employs a two-step method for testing the steps-to-war thesis against the alternative theoretical explanations for the statistical relationships among the relevant variables. In the first step, I construct and evaluate the historical timelines of the occurrence of the steps to war across a select sample of cases. I use these timelines to determine whether the cases fall into clear categories that represent one or other of the expected patterns. This step will allow me to preliminarily judge the extent to which the steps-to-war thesis is supported by the timelines, and the extent to which the logical patterns reflecting either the anticipation of war argument or rivalry argument appear in the group of cases.
In the second stage of the study, I focus explicitly on the issue of anticipation of war as the alternative causal explanation for the relationships. Developing more in-depth studies of several key cases, I examine the historical argument that the ‘steps’ simply reflected a prior decision to go to war, or a prior belief that war is likely to be unavoidable. This method will allow me to distinguish more clearly between the steps-to-war argument, on the one hand, and the anticipation-of-war argument on the other – is the causal arrow running from the decision to go to war rather than toward it? What does it mean, both in terms of historical interpretation and in terms of present political conflicts, to say that states are anticipating war?
This study uses a different unit-of-analysis than that which has been employed in prior tests of the steps-to-war thesis. Most of these have statistically assessed the probability of war in pairs of states or militarized disputes given the presence of some combination of the relevant variables. The unit-of-analysis here is pairs of states, or dyads, that have engaged in territorial claims that result in war during the time that the claim is active. I use discrete territorial claims because the steps-to-war model builds an argument for a class of wars that fundamentally emerge out of conflicts over territory – while the steps may be associated with an increase in the chance of war in disputes over other issues, the thesis is really founded in an explanation of how states deal with territorial disputes.
The frame is also narrowed to only those territorial disputes that result in war. I do that in order to apply process tracing of case studies, which allows for an empirical examination of the underlying causal mechanisms of the alternative theories for the relationships we have already observed.Footnote 26 This group is only a small proportion of all territorial claims between states, and a small proportion of all dyads that go to war. However, I am not addressing the question of whether the different ‘steps’ are associated with war; we know that they are. That question, which would necessitate evaluating all cases that went to war or not, has already been answered in statistical tests of the theory and its related elements. Because there is a clear statistical relationship between these variables and war in those empirical studies, temporal sequencing of non-war cases would not be particularly useful since most of the variables would not be there to sequence. For instance, a territorial dispute that did not end in war would also probably not have more than one or two (if any) of the variables (arms buildups, alliances, repeated territorial disputes, and rivalry) so a sequence would likely be uninformative. Rather, the question here, in cases where war occurs and the statistical relationship is essentially a given, is whether the path by which states get to war is the one that is claimed by the steps-to-war process, or is it better explained by one of the other existing arguments for the same relationships?
Huth and Allee’s territorial claims data outline all territorial claims between 1919 and1995.Footnote 27 A territorial claim begins when a state has made an official claim, and the other has responded making it clear that it will not capitulate. The claim officially ends when there is a formal settlement of the claim, typically by treaty. There are forty-one pairs of states between 1919 and 1995 that have an outstanding territorial claim that results in war, resulting in forty-two total cases employed in this study (Hungary and Yugoslavia have two discrete claims that result in war). Those forty-two cases constitute the full sample for the analysis in this study. For each of the cases, I determine the dates, within the active territorial claim, of each theoretically-relevant variable in the steps-to-war thesis. These include the history of militarized interstate disputes (MIDs) over the issue of territory;Footnote 28 the creation of politically-relevant alliances; the occurrence of unilateral and mutual rapid military buildups; and the existence and timing of interstate rivalry between the countries.
To determine when the first politically-relevant alliance was formed over the course of the territorial dispute, I begin with the Alliance Treaties and Provisions (ATOP) dataFootnote 29 and then define alliances as politically relevant or not following the method derived from Senese and Vasquez.Footnote 30 Alliances are defined as politically relevant when they are made either with another state in the same geographical region as the opposing state that is party to the territorial claim, or with a major state. Major states can be assumed to have the power, and potentially the will, to project their capacity globally, and minor states in the region may be both interested in the outcome of the dispute and have the ability to act in their own region. In this analysis, I include multilateral alliances to which both states are party, assuming that either or both could believe that the other members of the alliance would support them in the event of a crisis.
To measure rapid military buildups, I use the indicator employed by Sample, updated for this purpose.Footnote 31 The measure is calculated at both the nation–year and dyad–year level, so we can use it to examine general patterns of arming between the states involved in a territorial conflict – both times when they are mutually engaging in rapid buildups and, separately, times when one or the other state is engaging in a unilateral buildup. While the steps-to-war thesis includes arms races, not unilateral arming, it is assumed here that in an examination of timelines, this is simply more potentially useful information for evaluating the general pattern of relations between states.
I determine the existence and timing of rivalry using the Diehl–Goertz definition of an enduring rivalry.Footnote 32 An enduring rivalry exists between states if they have experienced at least six Militarized Interstate Disputes (MIDs) in the course of a twenty-year period. This measure was chosen over others in an effort to find an operational definition that would make rivalry effectively independent of the other variables in the steps-to-war formulation. While repeated territorial disputes during the course of an ongoing territorial claim are an element of the steps-to-war thesis, the Diehl−Goertz measure includes all MIDs, so the indicators are distinct.Footnote 33 It is possible that two states could be rivals at the commencement of a territorial conflict (having experienced disputes before the start of the territorial conflict), or never become rivals at all by this definition. It also avoids isolated conflicts characterized by few disputes, and thus gets at the conceptual notion of a real rivalry between states that see each other as enemies.
The timelines are then evaluated for the extent to which they support the contending explanations for the relationships among the variables. To address the question of whether we can assume that national leaders believed there was a sufficiently high probability of war to begin preparations (they ‘anticipated’ war), or they intended war, thus ‘causing’ the steps as a prelude to the expected conflict, I assume that at least one year had to pass between the activation of the territorial claim and the commencement of war to establish support for the steps-to-war argument. Senese and Vasquez build in a three-month window for alliances alone; I make it a year for the sequence. If war occurs in less than a year after a territorial dispute commences, it would be difficult to sustain an argument that each step had a causal relationship with subsequent steps. We could plausibly assume at that point that a war is being planned, and states are engaging in advance preparation for it.Footnote 34 The other timing variable here is that of rivalry in the historical frame of the territorial dispute. While the steps-to-war thesis admits substantial flexibility in the sequence of the different steps to war, rivalry at the start of a territorial conflict would be theoretically problematic. Certainly, it suggests that rivalry itself could not be dismissed as the likely cause of subsequent choices to make alliances, engage in further disputes, etc. as well as war.
FINDINGS
Systematic evaluation of the historical sequence of the cases reveals that there are categorical distinctions in their patterns. These patterns may be seen in Table 1. There are different groups of cases whose patterns are internally consistent, but distinct from one another. The makeup of those groups appears in the first instance to provide substantial support for the steps-to-war thesis, but also some support for each pattern that we would expect to see if the endogeneity concerns are valid. At the very least, they appear to represent separate conflict processes, or separate theoretically important ‘paths-to-war’ in Levy’s formulation,Footnote 35 even among the class of wars that are the result of territorial conflicts.
Table 1 Territorial Claims Divided by Classification
Notes: Rivals bolded. Cases leading to war within one year italicized.
In support of the steps-to-war thesis, there is a group of cases, nineteen of the forty-two, that follow the pattern we would anticipate if states are reacting to each step as expected by the model. We see a theoretically-predicted path: the creation of politically-relevant alliances, repeated engagement in territorial MIDs, arms buildups, and sometimes the emergence of rivalry. In each of these cases, war occurs more than a year after the commencement of the claim (typically much more than a year), suggesting that states were unlikely to be simply engaging in war preparation or presuming war was a very likely probability when they built their alliances and military resources; and rivalry, when it does appear, is late enough that it cannot plausibly be assumed to be the predominant causal factor for both the steps and war.
The remaining cases must be considered in light of the alternative arguments. For a number of the cases, rivalry occurs early enough in the course of the territorial claim that it cannot be discounted as the key explanatory variable. Twelve of the forty-two cases were designated rivals within a year of the start of the dispute. In some of those cases, war also occurred at the start of the claim,Footnote 36 but for those where war followed after some length of time, the states typically engaged in arms buildups, alliance-making, and repeated disputes in some order. This finding indicates that in some territorial claims leading to war, rivalry is happening early enough in the dispute that it may be exerting the primary causal impact on further steps taken toward war, as well as being the underlying cause of the war. While the steps may be reinforcing the rivalry, it is the rivalry that should be the focus of analysis. The argument certainly cannot be discounted given this evidence, and suggests the existence of a distinct path to war among states with territorial claims.
In seventeen (40 percent) of the forty-two cases, war occurs within a year of the start of the claim. At first glance, this might be supposed to provide confirmation of the anticipation of war argument – states intending war (perhaps if their demands are not met), broadcasting their rationale by activating a territorial claim, and then engaging in various preparatory policies before commencing the anticipated action – but further exploration of the cases indicates that the larger group of seventeen also breaks down on important lines. In many cases, anticipation of war as a realistic explanation for the war is essentially impossible. There are several cases that resulted in war almost immediately and have since experienced successive wars with continuous territorial claims and/or rivalry. These include Israeli dyads with Egypt, Syria, and Jordan, as well as the India–Pakistan dyad. Another group of seven cases follows almost immediately on the heels of the First World War, and represents battles over the boundaries of new or newly recreated states in the aftermath of empire.
These cases clearly demonstrate circumstances that require a separate explanatory framework altogether different from either the steps-to-war theory or the principal challenges related to endogeneity discussed here. In spite of the fact that war occurs within a year of the start of the territorial claim, the argument that states are anticipating war in advance and then engaging in arms buildups, alliance-making etc. cannot be considered a realistic causal framework for understanding these cases. In each case, at least one of the states did not exist for long before the territorial claim and war started – certainly not long enough to have conspired to conflict for any lengthy time, or been conspired against, or for relations to have deteriorated to the point between the states that a future war was seen as so likely that it should be driving policy. Poland’s claims against Lithuania and Russia were two of several claims by that re-emergent state at the end of the First World War, and resulted in immediate war. The breakup of the Ottoman and Austro-Hungarian Empires and the negotiation of the Versailles Treaty resulted in many new states in southeastern Europe, and new claims made there and globally. These claims involving new states included three wars involving Hungary alone in 1919 and 1920 almost immediately upon the independence of the states involved.
For the latter group of cases, it is not anticipation of war driving the claim and war preparation behavior, but rather the consequences of one war, and the death of empire, setting off the chain of events leading to a cycle of war afterwards. In some cases, the ‘steps’ to war occurred after the wars, not before, making it clear that there is no causal chain running through the creation of alliances, arms buildups, or multiple discrete disputes toward the war.Footnote 37 The immediacy of the war means that the conflict itself is the likely cause of the other steps, even in cases where there is no further war during the duration of the claim.Footnote 38
A final category where war happened very soon after the commencement of the claim is worthy of greater attention. These are the cases for which the anticipation of war argument is likely to find its clearest support. The five cases of this group include four leading up to the start of the Second World War: Italy–Greece (1940–41), Germany–Poland (1938–39), Russia–Finland (1938–41), and Hungary–Yugoslavia (1940–41), plus the Iraq–Iran (1979–[95]) claim. For these cases, alliance-making and arms buildups continued until the war itself, and historically, they are instances where the assumption is often made that war was planned in advance. If this were the case, then they would validate one important critique of the steps-to-war thesis, certainly limiting its scope substantially. The different steps are being taken, and war does occur, so the cases would provide statistical validation of the key relationships, but one could not say that the territorial claim is starting the process of conflict, instead of serving as the public justification in a case where war had already been planned.
The second part of this analysis then engages a focused assessment of this specific argument against the steps-to-war thesis. Choosing cases that most contradict the logic of the steps to war can allow us to determine the extent to which the anticipation-of-war claim truly undermines the steps to war as a theoretical explanation for war. A case-study design focusing on ‘least-likely’ cases for a theory can provide a significant test of the underlying causal mechanism, and offers an important indicator of the confidence we should have in a theory.Footnote 39 Thus, this part of the study targets the group of cases that apparently contradict the steps-to-war thesis, and best support the argument of war being anticipated in advance, and uses these to test the alternative arguments simultaneously.Footnote 40 If the anticipation-of-war framework cannot be upheld in these cases, it suggests a profound problem with the notion; and if the steps-to-war thesis is upheld with these cases, it offers substantial support to that theory’s basic causal logic.
The five cases included in this second phase of the analysis are those mentioned above: Italy’s claim against Greece (1940–41); Germany’s claim against Poland (1938–39); Russia/the Soviet Union’s claim against Finland (1938–41); Hungary’s claim against Yugoslavia (1940–41); and Iraq’s claim against Iran (1979–continuing). However, to assess the likely causal chain for individual cases more specifically, I do not limit the analysis only to the specific years of the single territorial conflict, but rather move beyond that to an examination of the recent history of the dyad more generally. By examining the longer timeline, we can address the question of what anticipation means. Slantchev suggests that states may make territorial claims because they are already anticipating war.Footnote 41 By examining the prior relationships for these claims that seem to indicate that very behavior, we can address the questions of why they were anticipating war at that point, and how they got there. Does a more detailed examination of each case really support the notion that states were choosing policies based on preparing for an already anticipated war, or that the policies themselves are fundamentally altering the underlying relationship, making a small chance of war into a large probability of one?
CASE STUDIES
Italy–Greece
The Italian territorial claim against Greece for possession of Northern Epirus was activated in 1940, and Italy attacked Greece shortly thereafter. As with several of the other cases here, the dispute and war between Italy and Greece represent the larger, multilateral conflicts of the Second World War. An examination of the history of the dyad suggests that this territorial claim was almost certainly made to justify the declaration of war against Greece, and thus, it is, in the first instance, a clear case of war anticipation from 1940. The history, however, also raises an important question about what it means to ‘anticipate’ a war, and what that really has to tell us about the validity of the steps-to-war model.
The conflicts in this case involve not two countries, but three, and the interplay of their relationships across twenty years before the official claim in 1940 also indicates a much more complex causal pattern than that inherent in the anticipation of war argument. A number of events along the path indicate both that the war could not have been realistically anticipated for a long period in advance, and at the point where it probably was planned in 1940, a great deal of theoretically-relevant negative interaction had already occurred – interaction that had shaped the dyadic relationship and increased the probability of war.
Several different territorial conflicts dating from 1919 are relevant to the interstate history of this case and understanding of the path that took these states toward war. The first was a claim made by Greece against Italy over the Dodecanese Islands.Footnote 42 Before Greece officially ceded the islands by treaty in 1928, ending the conflict, Italy and Greece experienced at least one territorial MID, the countries formed politically-relevant alliances, and engaged in unilateral and mutual military buildups over several years. Even when the two countries did not have an official territorial conflict (1928–40), they continued making relevant alliances, and persisted in patterns of intermittent arming. Greece joined international sanctions against Italy when the latter invaded Ethiopia in 1935, leading to a termination of a neutrality treaty between the two.Footnote 43 After this, Italy’s rhetoric toward Greece became increasingly hostile, particularly in relation to the strength of the relationship between Greece and the much more powerful United Kingdom.Footnote 44 Greece responded to rising Italian threats with increased military expenditures, and the armament became mutual and consistent leading up to the war after the Greek coup in 1936 and the commencement of the Fourth of August Regime under General Metaxas.Footnote 45 This pattern demonstrates continuing interactions that shaped the nature of the relationship between the two states over a considerable period, across different governments and regimes, increasing the probability of future conflict, but almost assuredly without a long-term conspiracy to launch an anticipated war in 1940.
Understanding the conflict between Italy and Greece in 1940 cannot be complete without some historical understanding of two other territorial claims in the region, however. First, following the First World War, Italy made a claim for parts of Albanian territory, and rights to hold a mandate over the whole country.Footnote 46 Not surprisingly, the Albanians objected both to the smaller territorial demands, and to the obvious challenge to their status as a sovereign nation. This dispute officially lasted only a year, until 1920, when Italy conceded to Albanian sovereignty. However, following Mussolini’s consolidation of one party rule in Italy in 1925, he reinvigorated Italian policy related to the claim. For the next decade, Italy pursued a policy of attempting to gain greater control over both the Albanian economy and policy.Footnote 47
A final historically relevant dispute among these states existed between Albania and Greece, with both countries laying claim after the First World War to the region of Epirus, the very region that Italy claimed to justify its war against Greece. The Albanian–Greek conflict was two-way, with each of them claiming territory under control of the other. In 1921, the League of Nations officially announced its support for the 1913 borders between the two countries, but it was not until 1924 that Greece agreed to comply with this determination.Footnote 48
The lead up to the Italian claim against Greece in 1940 actually began in 1939, when Italy officially reinstated its earlier claim against Albania and promptly invaded and occupied the country.Footnote 49 At this point, Italy, having captured Albania, also took on and reinstated Albania’s claim against Greece for Epirus, and subsequently attacked Greece. It was far less successful in this instance, with its military pushed well back into Albania by the Greeks, and was only able to occupy the desired territory after Germany invaded Greece as well.Footnote 50
While one may well argue that the 1940 claim (and indeed, the 1939 claim against Albania) was made in anticipation of war already in preparation, the case belies the essential relevance of the causal frame of the anticipation of war argument. Certainly any alliances made after the claim, and continued arming were in direct anticipation of the war, but it is unlikely that all of the actions before that were in preparation for a planned conflict. It is highly implausible that Italy was conspiring to war for twenty years. In 1925, Mussolini may have had ambitions, but he could not have anticipated that he would have Nazi support (since the regime did not yet exist) in his claims, and that support was critical to the unfolding situation. Further, the Greek regime itself was involved in arming throughout several years leading up to the war, arming that, given its timing, is likely to have been the result of Italian militarization of the region and threatening rhetoric. While it seems an obvious case of anticipation of war, that war was only expected after a long process that largely follows the steps-to-war logic, though with the relevant territorial claims being ‘settled’, or at least latent, for a good deal of the time.
Russia/Soviet Union–Finland
The Winter War between the Soviet Union and Finland came in the context of the Soviet Union’s 1938 claim to nickel mines in Petsamo and several Finnish islands on which it wanted to build naval bases.Footnote 51 The larger structural context was the desire of Moscow to secure Finland in the event of a German attack, presuming the Germans would use Finland as a base to march on Leningrad.Footnote 52 The war commenced within a year of the initiation of the territorial dispute, and it is likely that Russia intended from the beginning to go to war if Finland did not capitulate to the demands, especially given Finland’s awkward position between Germany and the Soviet Union. Thus, this situation almost certainly represents a case where war was anticipated at the point of the claim. Once again, however, while there is little doubt that the Soviet Union probably intended war if Finland failed to submit to Soviet demands, the case raises questions about what precisely it means to argue that the appropriate causal explanation for a correlation between the ‘steps’ and war is to be found in a prior expectation of war.
In this case, too, a territorial conflict existed earlier between Finland and the Soviet Union, with claims going in both directions, between 1919 and1920. Those claims were officially ended in 1920 with the Treaty of Dorpat in which both states made concessions. The territorial claim made by Russia to Petsamo in 1938 was a direct claim for that part of the disputed territory that Finland had been granted in the 1920 treaty.Footnote 53
In spite of the fact that the territorial dispute had been ‘settled’ by treaty in 1920, the two states continued their antagonistic behavior throughout the interwar period. During the 1920s and 1930s, while the territorial conflict was non-active, the two states continued to engage in those policies that conform to the logic of the steps-to-war thesis. Territorial MIDs took place during 1919 and 1939, only when the claims were active, but the states continued to develop a series of politically-relevant alliances (including a non-aggression pact with each other in 1932), and one or both countries were engaged in rapid military buildups throughout most of the interwar period. In the mid-1920s, the buildup was mutual, as it was from 1935 on, though by the latter period, the critical third party to the relationship was Germany, whose interactions with both the Soviet Union and Finland highly conditioned the bilateral relationship.
From these facts, covering two decades of the Soviet–Finnish relationship, it would be unrealistic to claim that the link between the ‘steps’ and war is best explained by an assertion that they were simply preparing for an expected war during all that time. Anticipating the war, and engaging in last minute additional preparations appears to be the penultimate step, but not the start of the underlying causal chain of events. Rather, this case suggests that there was a long and continual buildup, then a ‘calling of the question’ – a decision to go to war if the other did not capitulate once the territorial demand was explicitly made. Even after the territorial claim had been made, records of diplomatic activity throughout 1939 indicate the importance of both domestic politics (no politician in the democratic Finnish government wanted to commit political suicide by conceding to the Russians) and perceptions of increasing hostility (the Russians specifically interpreted diplomatic hesitation on the part of the Finns as intentional hostility) in shaping the environment in which war was finally triggered.Footnote 54
Hungary–Yugoslavia
The territorial conflict between Hungary and Yugoslavia leading into the Second World War resembles a number of similar conflicts in the region during this time. Many territorial disputes accompanied the breakup of the Austro-Hungarian and Ottoman Empires at the close of the First World War. Many resulted in small, immediate wars; some were negotiated and never re-emerged, and some continued to fester throughout much of the twentieth century.
In this case, before the 1940 claim leading to their involvement in the Second World War, the two states had a prior claim at the end of the First World War that also resulted in war. In 1919–20, Yugoslavia made a claim against Hungary, and in the Peace Treaty of 1920, Yugoslavia gained sovereignty over the disputed lands. That same territory, awarded to Yugoslavia in the initial settlement of the dispute twenty years earlier, was the target of Hungary’s reactivated claim in 1940.Footnote 55 Hungary subsequently reoccupied the territory when the Axis countries invaded Yugoslavia, causing its breakup and division, in April 1941.
Throughout the period between the two claims, there were numerous politically-relevant treaties, including treaties between Yugoslavia and its allies in which Hungary was named as a specific threat. At least one territorial MID occurred in 1922, and unilateral militarization occurred through most of the interwar era, though it was not mutual until 1940. Following Yugoslavia’s signing of the Balkan Pact (a defensive treaty) in February of 1934, Hungary began a substantial military buildup. Its expenditures in 1934 were 65 percent higher than in 1933, and it continued double digit increases each year for the remainder of the decade.Footnote 56
While the pathways of the two previous cases suggest that reinitiating the territorial claim was done to justify an intended war, this case is almost certainly not one of war anticipation. Hungary’s decision to go to war had more to do with the multilateral nature of conflicts involving Yugoslavia and Hungary’s allies than a specific decision made by Hungary to prosecute its disputes with the state.
Leading up to the war, Hitler was, in fact, trying to convince Yugoslavia to join the Axis alliance. The monarchy’s regent agreed to this alliance, but the military staged a coup, deposed the regent, and officially replaced him with the underage prince. This thwarting of his plans for the country apparently both surprised and irritated Hitler, leading him to order a German invasion of the country.Footnote 57 Italy and Hungary joined the invasion, and thus, Hungary was able to occupy the territory at the heart of its claim against Yugoslavia. If German attempts to lure Yugoslavia into the alliance had been successful, as was anticipated at the time, that country would presumably not have encouraged Hungary to attack its new ally. The relationship between the steps across the interwar period and the war, coupled with the uncertainty over Yugoslavia’s side in the war, ultimately means that the policies cannot be attributed to a clear decision to go to war in advance.
Germany–Poland
The specific territorial claim made by Germany against Poland in 1938 was a demand for the city of Danzig (Gdansk). This relatively small issue rather obviously masked Germany’s intentions with regard to the rest of Poland, and it also masks the complexity of the multiple claims around the borders of Poland, including with Germany, during the interwar period. The end of the First World War marked Poland’s re-emergence as a sovereign state, and Poland had claimed Danzig, as well as parts of Prussia and Upper Silesia at that time. Negotiations in 1921 resulted in mutual concessions by both Poland and Germany, and delineated the legal border; as part of this treaty, the Allies officially made Danzig a ‘free city.’
While the territorial conflict was officially resolved, the character of Danzig in particular put the permanency of the treaty in question from the beginning. Danzig was a largely German city, and while officially a ‘free city,’ the Polish government effectively had sovereign rights over the territory. In the intervening twenty years, the German population chafed at Polish bureaucracy, and the small Polish population chafed at discrimination by the Germans.Footnote 58 It was never a comfortable arrangement.
Following the 1921 treaty, the two states did not experience another formal territorial militarized dispute; however, the conflict over the status of Danzig and the entire Polish Corridor to the sea remained a live issue in domestic politics in both countries, their relations with each other, and their relations with other European countries. During the interwar years, Poland formed at least one alliance that named Germany as a specific threat, and multiple other politically-relevant treaties followed. During the mid-1920s, Poland engaged in unilateral military buildups, as Germany also did from 1933 forward through the war.
Relations between the countries demonstrate a fairly consistent pattern of action–reaction, with the bilateral relationship made highly complex given shifting relations with the other major states involved, and with the ongoing interest of the League of Nations over the disposition of Danzig. The rise to power of the hardline (regarding the territorial issue) Nazi party in Germany led to a direct response from the Polish government. After the 1932 elections, and escalating once Hitler was appointed Chancellor, Poland attempted to concretize its defensive relationship with France, citing evidence of increased militarization of Germany, as well as the dangers inherent in the ascension to power of the Nazis.Footnote 59 Poland responded to the Reichstag elections of 5 March 1933 (in which the Nazis were expected to dominate) by reinforcing the guard detachment at the munitions depot of Westerplatte, just north of Danzig, and parking a Polish troop transport in Danzig Harbor the day after the elections.Footnote 60 When the Danzig Senate was effectively taken over by supporters of the Nazi party in April 1933, Poland moved part of its Border Defense Force away from the Soviet frontier into the Polish Corridor and along the East Prussian border,Footnote 61 eliciting Nazi declarations in the Reichstag of defending the Reich in its ‘hour of danger.’Footnote 62 Repeated Polish attempts, citing these developments, to elicit a stronger statement or action from France and Britain failed, however, and Poland, left in a precarious position with little support, signed a non-aggression treaty with Germany at the end of 1934.Footnote 63
As the previous case demonstrated, not every case of German attack at the start of the Second World War involved a clear anticipation of the war and preparation in advance, but, in this case, Germany may well have anticipated the war with Poland by the time of its claim in late 1938. Poland’s initial diplomatic response to German claims was to invoke the 1934 non-aggression agreement,Footnote 64 but by March 1939, it had begun to call up troops,Footnote 65 and finally obtained a concrete defense treaty with Great Britain in early April. Throughout this period, Germany was carefully isolating Poland to the best of its ability through a series of treaties, including the famous non-aggression pact with the Soviet Union.
While it is evident that war was anticipated at that time, it cannot be argued that this is the best explanation for the occurrence of all the previous theoretically relevant policies taken during the interwar years. Much of this behavior crosses government and regime, as well as many years. The arrival of hardline leadership in Germany constitutes a step within the steps-to-war framework that independently increased the chance of war between Germany and Poland, and Poland’s response (and German reaction) is evidence of that dynamic. The fullness of the evidence indicates that anticipation of war is not the proper frame for understanding how this war came about.
Iran–Iraq
The final case to be examined here is that of Iran and Iraq. The discrete territorial conflict leading to war between the two countries began in 1979, when Iraq initiated a territorial claim against Iran over the placement of the international boundary within the Shatt-al-Arab waterway. Within a year, they were engaged in a bloody eight-year war known for its use of chemical weapons against enemy soldiers and waves of boys sent across mined fields to clear the way for the advance of seasoned troops.
As with each of other cases, the official activation of this claim in 1979, likely in anticipation of war at that time, obscures the important ongoing nature of the conflict. Coupled with disputes over the delineation of the land border, this conflict dates back to the imperial era when Iran made claims against British-held Iraq. The claim against Britain lasted from 1921 to 1932, at which point Iraq became independent and inherited the dispute. The dispute remained active until 1975, when it was officially settled in the Baghdad Treaty with the main navigable channel of the Shatt-al-Arab becoming the official international border (a relatively common decision in international law). Iraq’s reactivation of the claim four years later involved official abrogation of an accepted international treaty, and overturned what was assumed to be a settled boundary.Footnote 66
Throughout the period of the prior claim, from 1932 to 1975, the rival states engaged in intermittent territorial militarized disputes, engaged in unilateral and sometimes mutual military buildups, and created a number of relevant alliances. In the period between 1975 and 1979, despite their territorial claim having been settled, they were still engaged in rapid mutual military buildups through 1977.
The historical timeline of this case is also particularly interesting because it provides a good lens for preliminarily considering the importance of an element of steps-to-war that is rarely built into testing the thesis: changes in leadership and the rise of hardliners to power. For this particular territorial claim, I have supplemented the analysis with the more detailed information on the claim from the Issues Correlates of War data. In the ICOW data, it is evident that while 1975 is the termination date of the initial claim, the claim appears to have actually been left in abeyance and reactivated time after time – each time coinciding with leadership changes in either Iran or Iraq.Footnote 67
Iran temporarily abandoned its claim in 1954 when the democratic regime of Prime Minister Mossedeq was overthrown. At the time of the 1958 Iraqi coup and the end of the Hashemite kingdom in that country, Iraq reactivated the claim, and Iran followed suit the following year. The 1975 Baghdad Treaty appeared to end the conflict once and for all with a compromise solution that reflected general practice in international law, and thus would seem to be as mutually acceptable as any finale. However, 1979 marked the year of both the Iranian Revolution and the rise of Saddam Hussein to power in Iraq. In both cases, these were very hardline leaders taking control of the governments; as well, the Iraqi government saw the danger to itself of an expansion of the Iranian Shi’a revolution to its own majority Shi’a population. Iraq consequently abrogated the treaty, and relations collapsed. This information indicates that further case analysis in testing steps-to-war claims may well benefit from more in-depth assessment of domestic political change in specific cases and the impact of those changes.
The case of Iraq and Iran does add an extra wrinkle to the complexity of assessing the steps-to-war thesis because the states were also rivals from the commencement of their dispute in 1932. Since rivalry itself underpins the other endogeneity claim against the steps-to-war, it cannot be dismissed as the explanation for the historical dynamics of the case. The historical trajectory of the case, however, is particularly interesting for our understanding of these relationships and the paths taken to war involving territorial disputes. For one thing, it suggests that the adoption of realpolitik policies, as outlined in the steps-to-war, may both be caused by the rivalry, but also contribute to the maintenance and endurance of that rivalry in a more complex set of causal relationships.
The case, in addition, highlights the question of what makes states rivals in the first place. There existed an active territorial conflict here before Iraqi independence in 1932, and the inheritance of this already antagonistic relationship may well offer an explanation for the rivalry between Iraq and Iran from the start of their relationship. While Iraq and Iran may have been rivals from the start of their sovereign territorial dispute, it is unclear that this would have been the case if the claim had not already roused antagonistic behavior before Iraq’s inheritance of it.Footnote 68
The consideration of inherited conflicts may be theoretically intriguing: the steps-to-war thesis as outlined depends on the claim being between two sovereign entities whose choices regarding the claim and each other systematically affect the probability of war – there is no obvious avenue within the theory as currently configured for inherited relationships. The rivalry literature, while allowing for rivalry from the start of a relationship, does not satisfactorily deal with this situation either: it is unclear whether it would be theoretically appropriate to claim that existing dyadic rivalry is truly the cause of arming, alliance-making, and war if indeed previous sovereign claimants were engaging in those behaviors in advance of the new state taking over the claim. The new state inherits a conflict that they ‘know’ cannot easily be negotiated because it has already been active for some time, so their probability in engaging in conflict over the claim is already increased by prior ‘steps’ taken by the previous claimant. The answer to this puzzle and the implications for refining our theories, and our understanding of the interplay between rivalry and the steps-to-war, requires a more systematic analysis of cases like Iran and Iraq and others where at least one rival inherited the conflict from another state.
CONCLUSIONS
The purpose of this analysis has been to advance testing of the steps-to-war thesis by directly addressing a principal theoretical challenge to it. More specifically, the study assesses the theoretical merit of the anticipation of war argument, which stands in contrast to the underlying causal logic of the steps-to-war thesis. Substantial empirical research has demonstrated that the correlative relationships outlined by the steps-to-war theory are valid: territorial conflicts are more likely to result in war than other kinds of disputes, alliances and rapid arming increase the probability of war, as does the occurrence of repeated militarized disputes between states. Individually and collectively, these theoretically-relevant variables are related to war.
However, while the steps-to-war thesis offers one explanation for those relationships, others have argued that the occurrence of the steps is not the cause of the war, but rather that the causal arrow is moving toward the steps rather than from them. The anticipation-of-war argument asserts that states are engaging in alliance-making and arms buildups because they anticipate war and are preparing for it, and thus the ‘steps’ again are not causing war, but reflecting a state’s prior decision to wage war. In the case of rivalry, the challenge to the steps-to-war thesis is that instead of being a step on the road to war, rivalry is causing both the steps and war itself.
The results of examining the historical timelines of forty-two territorial claims that resulted in war suggest that there are different paths to war represented in this group: in addition to the post-war/post-imperial category of wars (which must be addressed by separate causal arguments altogether), rivalry itself seems to reflect a separate causal process at work in a group of these cases. The anticipation-of-war challenge to the steps-to-war thesis, however, appears to be ultimately untenable in the case of states with long-running territorial conflicts. When we do timelines of the theoretically-relevant variables leading to war, approximately half of the cases fit the pattern that we would associate with the steps-to-war causal chain. In other cases, war occurred within a year of the commencement of the territorial claim, and/or the states were rivals from the beginning of the claim. For the first group, the steps-to-war thesis is simply not applicable. However, for some of the other ‘non-steps-to-war’ cases, the logic of the steps-to-war chain actually may be highly pertinent, in cases, for instance, when a particular state dyad inherits a conflict from a previous sovereign over the territory.
The second part of the analysis focuses on the five cases that appear to support most closely the anticipation of war argument (they are most-likely cases for that theory) while challenging the causal mechanism of steps-to-war (they are least-likely cases for that argument). A closer evaluation of those cases is revealing. Of these five cases, the history of the Hungary–Yugoslavia claim (1940–41) demonstrates that war was not anticipated at all, but that the interaction pattern between the states clearly mirrors that predicted by the steps-to-war. In the other four cases, war was quite likely expected when the claim was made – it is unclear, for instance, what Poland could have done, short of relinquishing its right to remain a sovereign state, to avoid war with Germany, by the time the German Foreign Minister had his tête-à-tête with the Polish Ambassador in October of 1938.Footnote 69 Ultimately, however, the history of these cases undermines the premise of the anticipation-of-war critique of the steps-to-war thesis.
The activation of the territorial conflict in each of these cases, shortly before the war commenced, suggests that the challenging states were anticipating a high probability of war or even intending a war, and using the territorial claim to justify and/or telegraph their intent at that point. But the larger history of the dyadic relationship, and sometimes a more complex multilateral situation, suggests that the relationships between states were dynamic and developing over a considerable period, and likely that the steps being taken were affecting the context of the relationship (and the future probability of war). Evidence offered here highlights some cases of specific reaction by leaders of one country to relevant changes in the other and, in all cases, indicates a pattern of changing policies that fits the predictions of the steps-to-war thesis. It is unrealistic to assume that countries were planning war or assuming it to be highly likely and preparing across multiple (sometimes violent) leadership changes, rather than assuming that the alliance building, arms policies, and militarized disputes themselves were altering the course of the relationship and future choices toward war.
Four of the five cases clearly reflect the logic of the steps-to-war thesis. The other (Iraq–Iran) is more complex because they were also rivals at the start of the territorial claim – and, indeed, rivals from the point of the independence of at least one state in the dyad. This means that the findings here essentially lay to rest the endogeneity challenge inherent in the anticipation of war claim, but the question of the role of rivalry is relevant and important to our understanding of conflict processes. A substantial number of cases follow the logic of the steps-to-war thesis quite clearly; for others, rivals from the start, the causal framework of the rivalry literature is relevant, both as a separate path to war, and in its interplay with the steps to war. The results indicate the importance of furthering existing work on the way rivalry comes about in the first place, and how that relates to the steps to war. If rivals inherit territorial claims that are already contentious, pre-occurring policies (by other sovereign states) may be having a step-by-step impact on the likelihood of war between new pairs of states. Neither the steps-to-war thesis, nor the rivalry research, satisfactorily deals with this potential path at this point, but it indicates a direction for future research.
Ultimately, this work demonstrates that one of the endogeneity claims against the steps-to-war theory is simply not historically valid. Understanding war as anticipated and planned may explain some wars, but anticipation of war is not the best causal explanation for any of the sample of wars related to territorial conflicts once we examine their dyadic histories closely. Evaluation of the cases demonstrates that the logic of the steps-to-war thesis is supported by the historical timelines in many instances. Future understanding of the causal paths leading from territorial conflict to war must attempt to untangle the complexities of territorial claims and their histories as they relate to both rivalry and the steps to war.
In a world where national policy makers and international organizations are monitoring multiple potential conflict situations, the lessons here are significant. The steps-to-war thesis maps out a particular road to war over the issue of territory, an issue that is particularly salient and more likely to result in war than conflicts over other matters. The steps that are often taken to manage territorial claims are not neutral, but alter the political context in ways that significantly increase the chance of war being the final outcome. These steps reflect conscious choices, not structural characteristics largely beyond control. Diplomatic intervention cognizant of the measurable threats associated with common policies in the context of territorial conflicts has a real chance of significantly lowering the probability of war. When territorial disputes are active, both the participants and third parties can act on the realization that resolving these specific issues is particularly relevant to national security and international stability. When states engaged in territorial conflicts take the steps outlined here these should be seen as early warning indicators that mandate a narrowing of the focus on these specific disputes from the range of potential conflictual situations, and that the policies themselves are worthy of diplomatic intervention as well as the underlying conflict. Directing efforts in these ways could substantially lower the chance of escalation to war.
