Non-aggression pacts: context and explanation

Abstract

The existing literature offers contrasting views on the causes and effects of non-aggression pacts. Some scholars contend that these agreements impose audience costs that prevent an ongoing rivalry from escalating to war. Others claim that states use non-aggression pacts to signal to others that their rivalry is over and that their future relations will be peaceful. Scholars disagree as to the impact non-aggression pacts have on violent conflict. I demonstrate that various definitional and coding issues beset the literature, resulting in the incorporation of many agreements that should not be considered as non-aggression pacts. I then make a threefold argument about non-aggression pacts. First, non-aggression pacts came into being in the 1920s amid emerging norms proscribing interstate warfare. Second, they saw frequent use in interstate Europe. Nazi Germany and the Soviet Union used them to manipulate those norms so as to make themselves appear more acceptable despite their revisionism. Finally, many friendship treaties, which have been miscast as non-aggression pacts, are a separate type of agreement that became common among those post-colonial states that acquired independence during and immediately after the Cold War. Timeless arguments regarding non-aggression pacts thus reify these agreements and overlook key motives behind their use.

What are non-aggression pacts and what impact do they have on violent conflict? This twofold question is deceptively simple. On the face of it, non-aggression pacts are international agreements that states conclude to declare to one another that they will not use military force against each other. And indeed, states have signed many agreements that themselves are explicitly called non-aggression pacts. However, many international treaties contain clauses relating to mutual non-aggression – not least the United Nations Charter – but are not labelled as such. Complicating matters is that the academic literature has yielded contradictory arguments as to the causes and effects of these agreements. Leading arguments commonly emphasize the public nature of non-aggression pacts, how these agreements have their roots in interstate rivalry, and how they engage the reputation of their signatories. Nevertheless, one theory postulates that, amid an ongoing rivalry, states use non-aggression pacts to raise audience costs and so reduce the likelihood of war. Another theory advances the notion that states use non-aggression pacts to signal to third parties that their rivalry is over, thus raising confidence in the prospects of their future cooperation. Some scholars find that they have no impact on whether violent conflict breaks out, whereas others find that these agreements make it less likely.

I argue that these contradictions are the result of analysts miscoding agreements as non-aggression pacts and thus lumping them together with other treaties – in particular, friendship treaties – that should be distinguished from one another. The disagreements that empirical scholars have had regarding the uses and consequences of non-aggression pacts may be due to these definitional challenges. More critically, however, the underlying problem in the international relations literature with regard to non-aggression pacts is that the arguments given have a timeless quality to them. Yet, as I show below, the non-aggression pact as a species of international agreement is largely bound to a particular historical context.

The purpose of this article is to elaborate on the conceptual and historical issues involved with non-aggression pacts. I first review the standard definitions that international security scholars have invoked to conceptualize non-aggression pacts, as well as the data used for analysing them. I push back against the common perspective that non-aggression pacts constitute a type of alliance. I also show that scholars have overstated the frequency of non-aggression pacts in the historical record. I then develop three strands of the main argument that this article advances: first, that non-aggression pacts appeared in the 1920s amid emerging norms proscribing interstate warfare; second, that they saw frequent use in interwar Europe due to a combination of post-1918 territorial disputes and revisionist great powers; and third, that friendship treaties, which have been mistakenly identified as non-aggression pacts, in fact, are a separate type of agreement that became especially common among those post-colonial states that acquired independence during and immediately after the Cold War.

In elaborating these arguments, I show that general statements regarding the causes and effects of non-aggression pacts are problematic. They reify non-aggression pacts by ignoring their historicity and by neglecting those key patterns in the data. For example, arguments about finished rivalries and audience costs must reckon with how non-aggression agreements were especially popular among Nazi Germany and the Soviet Union in the 1920s and 1930s – states not known for concluding rivalries in good faith and, for that matter, bearing audience costs. I contend that these otherwise revisionist great powers used non-aggression pacts to manipulate emerging norms about peaceful conflict resolution so as to justify their actions and to prevent enemy coalitions from forming. Although themselves a product of a specific historical period as well, many of those friendship treaties miscast as non-aggression pacts have served more benign purposes – specifically, for newly independent countries to build diplomatic relations among themselves and to pursue developmental objectives in an international legal order that had allowed their subordination in the first place. Put together, the scholarship on non-aggression pacts has shown a fundamental misunderstanding of the many international agreements that researchers have tried to explain.

Defining and theorizing non-aggression pacts

The literature on non-aggression pacts has developed haphazardly over the decades, with one result being how this type of agreement receives much less attention than other forms of security cooperation. Below, I examine how scholars to date have defined non-aggression pacts and review the main findings of this literature. Thereupon, I delve into the definitional and coding issues that characterize this literature.

Current understandings of non-aggression pacts

According to Brett Ashley Leeds, non-aggression pacts are agreements between two or more states that obligate them ‘to refrain from military conflict with one another’.¹ Scholars have argued that they are a type of alliance because, in the words of Bruce Russett, they count as ‘a formal agreement among a limited number of countries concerning the conditions under which they will or will not employ military force’.² Leeds and her co-authors, thus, incorporate non-aggression pacts in their Alliance Treaty Obligations and Provisions dataset.³ They find that, of all the 212 cooperative security arrangements that they code from the period between 1815 and 1944, 49% obligate defensive military support, 26% pledge offensive military support, and 24% promise non-aggression.⁴ Non-aggression pacts are treaties that usually contain provisions for ‘non-aggression’ but have no clauses that pertain to consultations in the event of an attack or the undertaking of offensive or defensive military operations that involve third parties. As these treaties are a ‘type of conflict management agreement’, scholars tend to distinguish non-aggression pacts from peace treaties or ceasefire agreements.⁵ Non-aggression pacts often lack the institutionalization and specificity that peace treaties have. Unlike ceasefires, states do not sign non-aggression pacts to terminate a military conflict.

Two major theoretically informed empirical arguments about non-aggression pacts are worth mentioning. Both are rationalist in that they are premised on how states use institutions to signal their intentions, either to each other or to third parties, as well as to shape the behaviour of others via the information that these agreements help convey.⁶ Both arguments assume that the signatory states are acting in good faith and so see non-aggression pacts as legalistic arrangements that, at least in the short term, provide mutual benefit.⁷ Both also share the perspective that rivalry matters for the formation of non-aggression pacts. A rivalry refers to a situation where at least two states perceive one another, rightly or wrongly, as a competitor for the same scarce resources in the shadow of possible military escalation.⁸ Rivalries tend to endure and can lead to outright military conflict, especially if the rivals compete for territory and influence.⁹

The first argument under consideration is from Mattes and Vonnahme. They argue that a rivalry can be a persistent problem for the states involved, leading them to consider sometimes how to manage their relationship so that the rivalry does not spiral out of control. The non-aggression pact offers one useful solution. What is relevant for Mattes and Vonnahme is the public nature of these agreements. Mattes and Vonnahme observe that ‘[b]y formalizing and publicizing their pledges not to threaten or use force, the signatories impose audience costs on whoever violates the agreement’ in times of major tensions.¹⁰ As what may be the case with other international agreements, writing down and formalizing non-aggression pacts raise the ex-post costs of noncompliance and thus help states signal to each other (and to third parties) the significance that they attach to the agreement. With the state breaking a non-aggression pact risks appearing deceitful, if not frivolous, about its international commitments, audience costs become manifest following the misdeed. That is, when reneging on a public pledge, a state damages its reputation among international observers. This outcome is undesirable because states will subsequently find eliciting future cooperation from others to be more difficult.¹¹ Audience costs can also be domestic if a government’s supporters believe that national honour is at stake when public pledges are made.¹² Wishing to avoid these costs, states abide by the non-aggression pact, thereby reducing the potential for military conflict between them, at least for the short term.¹³ Their rivalry might yet endure but it becomes manageable.

Lupu and Poast offer a contrasting argument for why states sign non-aggression pacts. They agree explicitly with Mattes and Vonnahme that rivalry offers an important basis for the non-aggression pact. However, Lupu and Poast contend that non-aggression pacts arise when the rivalry is over. The problem that states signing non-aggression pacts wish to solve is that because rivalries often last long, outside observers may be unsure whether a particular rivalry is truly over. Those third parties may wrongly rely on past behaviour to make inferences about the future relations between those states that have had a rivalry, thus creating an information asymmetry. Accordingly, non-aggression pacts signal to third parties that the probability of conflict between former rivals is much lower than what is perhaps commonly understood. Broadcasting such information may be helpful for attracting investment, commercial interest, or security cooperation from third parties, especially those that had been aligned with the former rival when the competition had lasted.¹⁴ Citing two studies, including the one by Mattes and Vonnahme, Lupu and Poast add that what makes this signal more credible is the alleged record of non-aggression pacts being associated with reduced violence. When describing their theoretical mechanism, they mention neither audience costs nor, for that matter, reputation. In sum, Mattes and Vonnahme’s explanation presumes the rivalry persists but becomes a management problem, whereas Lupu and Poast’s explanation turns on the rivalry being over and needing to be clarified as such.

Do non-aggression pacts really live up to their name? To date, scholars have uncovered mixed evidence about whether non-aggression pacts achieve what they are supposedly intended to do. Andrew Long and his co-authors find that non-aggression pacts increase peace in their study of alliance treaties.¹⁵ Mattes and Vonnahme determine that they dampen the likelihood of militarized conflict, but they note that their influence weakens over time. This finding is consistent with other work that emphasizes audience costs. Others assess that non-aggression pacts do not at all decrease conflict.¹⁶ Some scholars even find that non-aggression pacts are associated with increased conflict.¹⁷

Definitional and coding issues

This literature has produced important insights on why and how states would use non-aggression pacts. However, definitional and coding problems mark this body of scholarship. These problems illuminate why scholars have uncovered mixed evidence about their purported effectiveness.

One fundamental definition problem is the notion that non-aggression pacts should be considered as alliances. The ultimate purpose of military alliances is to help two or more states coordinate defence and foreign policies in view of their shared national security goals.¹⁸ Non-aggression pacts are consistent with this definition. However, states could use an alliance to deter, and defend against, a commonly perceived threat;¹⁹ to avoid isolation and to prevent counterbalancing coalitions;²⁰ to coordinate offensive or preemptive military operations;²¹ or even to resolve disputes among themselves.²² Crucially, many, if not most, treaty-based military alliances that come to mind for scholars – the 1879 Dual Alliance between Austria–Hungary and Germany, the 1892 Franco-Russian alliance, the 1902 Anglo-Japanese alliance, the North Atlantic Treaty Organization, US-led bilateral alliances in East Asia, and the Warsaw Pact – contain promises of at least one signatory coming to the defence of the other. Military alliances often do imply promises of non-aggression between their signatories, but the modal alliance involves states using these security arrangements primarily to boost their security vis-à-vis some adversarial third party, whether explicitly defined or not.²³ In contrast, the essence of a non-aggression pact is that states primarily use them to foreswear military action against one another. Even those military alliances that serve to overcome internal disagreements – what Patricia Weitsman calls ‘tethering alliances’ – have featured mutual defence clauses that non-aggression pacts do not have.²⁴

If non-aggression pacts are really alliances, then the resulting imprecision could lead to odd interpretations of the empirical record. Poland, for example, had so-named non-aggression pacts with both Nazi Germany and the Soviet Union in the 1930s, but, from the perspective of contemporary Polish leaders, Poland only had an alliance with France for much of that decade before concluding another one with the United Kingdom. Claiming that Poland simultaneously had an alliance with Nazi Germany and the Soviet Union would obscure critical differences between these security arrangements and distort the historiographical debate about why Warsaw was reluctant to conclude defence agreements with Moscow. To invoke a more contemporary example to illustrate the absurdity, the non-aggression pact signed by the two Koreas in 1991 suggests that Seoul has two alliances, one with the United States and another with North Korea, its main adversary.

Another problem with how scholars have defined and theorized non-aggression pacts is that they are much more contextually bound than how scholars have talked about them. Simply put, scholars have reified non-aggression pacts. This problem is not unique to non-aggression pacts. John Gerard Ruggie questioned on similar grounds those theoretical assertions about international anarchy and its supposedly enduring, unchanging quality throughout the centuries. Thinking of change strictly in terms of the distribution of capabilities distorts the sociological shift from medievalism to modernity in world history.²⁵ Similarly, John Hobson and George Lawson highlight the tendency in some international relations scholarship of what they call ‘tempocentrism’, which takes a ‘reified present and extrapolates this back in time to render all history amenable to transhistorical, universalist analysis’. One implication of this historiographical approach is that ‘all actors and, indeed, the international system itself are presented as homologous or isomorphic’.²⁶ Instead of treating non-aggression pacts as universal and timeless, one must ask why they emerged at a particular time and saw their most extensive use during one historical period.

The definition that Leeds offers thus seems straightforward enough, but its timeless quality – its reification – lends itself to measurement problems that can undermine empirical analysis. A key issue is that the meaning of non-aggression has changed over time, and with it, the very concept of a non-aggression pact. Few agreements bear the name ‘non-aggression pact’. Those that do appeared most frequently in the interwar period. Europe saw the largest portion of such agreements in that limited time frame. Of the 137 agreements examined by Yonatan Lupu and Paul Poast in their study, only 29 contain the word non-aggression pact in their name. Of these 29 so-named non-aggression pacts, 22 were negotiated between 1919 and 1941. Nazi Germany signed six of these non-aggression pacts.²⁷ The Soviet Union signed 12 of them. Included in these numbers is the 1939 Nazi-Soviet Pact. Of course, the content of the treaty is what matters for coding an agreement as a non-aggression pact or something other. Nevertheless, coding decisions on this basis can still be problematic.

Consider the recent leading studies described above that examine non-aggression pacts. Yonatan Lupu and Paul Poast examine 123 bilateral non-aggression pacts and 14 multilateral ones, whereas Michaela Mattes and Greg Vonnahme examine 75 (politically relevant) non-aggression pacts. Using the latter’s data, Figure 1 plots the number of their observations per year between 1900 and 2001. The observations themselves are very diverse. Lupu and Poast include ‘any treaty that includes a non-aggression provision, but does not also include any defensive, offensive, or consultative alliance provision’.²⁸ The multilateral non-aggression pacts include, problematically, the 1929 Litvinov Protocol (ATOP#2240), the Association of Southeast Asian Nations (ATOP#3755), and the Economic Community of West African States (ATOP#3810). Fifteen peace treaties or attempted ceasefires are miscoded as non-aggression pacts despite how Leeds’s definition carefully distinguishes them from non-aggression pacts.²⁹ Ulterior motives underpinned two coded non-aggression pacts – the Nazi-Soviet Pact (ATOP#2470) and the 1960 China-Burma border treaty (ATOP#3380) – in that the signatories used the agreements to direct aggressive action against third parties. At least six agreements are too ambiguous to code reliably as non-aggression pacts, whether due to insufficient data or because they appear to address specific border issues. At least 45 agreements are non-aggression pacts without any economic provisions, the vast majority of which are bilateral, with some exceptions being the Pan-American Treaty (ATOP#2105) and the Protocol on Politics, Defence, and Security Cooperation (ATOP#4985).

Figure 1. Lupu and Poast (2016) data on non-aggression pacts.

Another problem with the data is the inclusion of many friendship treaties. These agreements allow states to establish diplomatic relations with one another and contain pledges to pursue non-military cooperation. As Heather Devere, Jane Verbitsky, and Simon Mark note, ‘while treaties have been interchangeably labelled as peace or friendship treaties in some contexts, in the main, peace treaties have signified the end of hostilities between the parties, whereas friendship treaties have been used for agreements about use and development of resources, territorial integrity, access to harbours, trading lanes and fisheries, and promises of cooperation’.³⁰ One can add educational and cultural exchanges. Even so, making non-aggression pledges to manage rivalries or to demonstrate their conclusion does not figure in this list. Friendship treaties might include non-aggression provisions, but so do some military alliances. If friendship treaties are excluded from Lupu and Poast’s analysis, as in Figure 2, the frequency of actual non-aggression pacts falls considerably.

Figure 2. Lupu and Poast (2016) data without friendship treaties. See Supplementary files for more information on the observations.

Why do these coding issues matter? One is that the resulting measurement error will impact hypothesis testing, especially if scholars have overstated the frequency of non-aggression pacts in the historical record. Another is that coding decisions may illuminate why scholars have yielded contradictory insights on what drives states to conclude non-aggression or what effects these agreements might have on violent conflict. As mentioned, Mattes and Vonnahme examine fewer non-aggression pacts than Lupu and Poast, the latter of whom include many friendship treaties in their analysis. Of interest to Mattes and Vonnahme are those non-aggression pacts that are ‘politically relevant’ – that is, they are concerned only with those ‘states that have the potential to interact with each other, and thus to fight, are more likely to be motivated by the fear that conflict between the signatories is possible’.³¹ Hypothetically, the audience cost argument may be most applicable in those pure non-aggression pacts that states such as those in interwar Europe used amid a highly threatening international environment. In contrast, arguments about peaceful former rivals may be most germane with respect to friendship treaties. The discussion below suggests that this twofold hypothesis may not be valid, however.

The conceptual evolution of non-aggression

The definitional and coding difficulties that characterize this literature cannot find a resolution simply by developing a more reliable measurement of non-aggression pacts. Deeper issues relating to international history are at play, and these issues make case comparisons across different historical periods too problematic from a research design perspective. The non-aggression pact may very well be a form of signal, but it is one that cannot be reduced to a rationalist framework since it is embedded in a larger set of intersubjective understandings about appropriate and legitimate behaviour.

Three observations deserve mention. The first is that a large majority of formal non-aggression pacts did not appear until after the League of Nations became established. Any argument about non-aggression pacts must explain why they emerged following the First World War. The second is that, in terms of periodization, formal non-aggression pacts appeared most frequently in Europe between the two world wars, with Nazi Germany and the Soviet Union being their most common users. The third observation concerns the post-1945 prevalence of friendship treaties, which Lupu and Poast problematically cast as non-aggression pacts. Decolonized or newly independent states used these treaties to help establish and to define their diplomatic relations with others. Incidentally, friendship treaties signed during the first half of the Cold War often involved the People’s Republic of China. Superficially at least, some association appears to exist between ideologically revisionist great powers and the use of non-aggression pledges. I unpack each of these observations in turn.

Why no formal non-aggression pacts before the 1920s?

In the leading data sets on non-aggression pacts, the first three of such coded agreements are the Central American Conference (ATOPID #1450), the Treaty of Amity between Bolivia and Ecuador (ATOPID #1467), and the Pan-American Treaty to Avoid or Prevent Conflicts between the American States (ATOP #2105), concluded in 1907, 1911, and 1923, respectively. The first so-named non-aggression pact is that which the Soviet Union and Turkey had concluded in 1925 (ATOPID #2145). Why did no formal non-aggression pacts appear before the 1920s? This question has significance for their empirical study. Although scholars have proposed timeless, universal mechanisms to explain the drivers and consequences of non-aggression pacts, these agreements are instead contextually bounded and so should not be reified.

The straightforward answer to the question raised above relates to the international legal discourse that developed in the early 20th century. Definitional controversies about the word ‘aggression’ intensified in the aftermath of the First World War amid a broader effort aimed at weakening nationalism and crafting a new international order centred on the rule of law.³² Prior to that cataclysmic event, aggression was largely seen as a matter of private criminal law – something carried out by individuals against others – with some early international thinking focused either on problematic distinctions between legitimate and illegitimate aggression or on Christian notions of what was just and unjust. Writing in 1935, American political scientist Quincy Wright observed that ‘the words [‘aggression’ and ‘aggressor’] appear very little in treatises on international law until after the World War’.³³ The 1919 Paris Peace Conference and the subsequent formation of the League of Nations precipitated a major debate on the concept of aggression. Experts could not agree on how it could be defined, let alone on whether agreeing upon a definition was itself worth doing. Having a clear definition could serve propagandistic purposes that states could exploit to brand others negatively. It could also produce a sense of compulsion among international organizations and their members that they must respond to actions that approximate that definition. Crucially, malign actors could design their hostile actions in such a way as to skirt international law.³⁴ Whatever the merits of these arguments, the need for consensus within the League of Nations, or the United Nations decades later, meant that the concept of aggression remained unclear.

Ambiguity over the term did not, of course, prevent its invocation. The war guilt clause – inserted into the Treaty of Versailles to justify Germany’s payment of significant reparations – ‘crystallized’ the idea that aggression denoted the initial military attack.³⁵ Article 10 of the Covenant of the League provided that ‘[t]he Members of the League undertake to respect and preserve as against external aggression the territorial integrity and existing political independence of all Members of the League’, adding that ‘[i]n case of any such aggression or in case of any threat or danger of such aggression the Council shall advise upon the means by which this obligation shall be fulfilled’.³⁶ Although the Covenant left the term ‘aggression’ undefined, it did distinguish, with Article 12, lawful wars from unlawful ones based on whether belligerents first submitted their dispute to arbitration, judicial settlement, or an inquiry by the Council. Hence, the League of Nations sought to promote collective security.

Subsequent legal developments made only tentative steps towards clarifying the concept of aggression. A draft Treaty of Mutual Assistance, put forward by the Third Commission of the Assembly of the League of Nations in 1923, attempted to criminalize wars of aggression. However, the effort stumbled on definitional problems. The failed 1924 Geneva Protocol tried to strengthen the linkage between aggression and compliance with the formal legal procedures set forth by the Covenant. In subsequent years, other legal efforts focused on prohibiting war. The 1925 Treaty of Locarno made exceptions for self-defence for lawful wars as defined by the Covenant, as well as for violations of Articles 42 and 43 of the Treaty of Versailles that relate to the demilitarization of the left bank of the Rhine. In 1928, the Sixth Pan-American Conference resolved that ‘war of aggression constitutes a crime against the human species … all aggression is illicit and … prohibited’.³⁷ It did not provide a definition of aggression, nevertheless. More significant still was the Kellogg–Briand Pact. Although its brief text makes no reference to the word ‘aggression’, its signatories agreed to renounce war as a tool for settling ‘all disputes or conflicts of whatever nature or of whatever origin they may be’.³⁸ Arguably, the most serious effort to make precise the meaning of the word ‘aggression’ came with two treaties signed in London in 1933 that involved the Soviet Union and its neighbours. Article 2 defines an aggressor as the first state that declares war against another state; invades the territory of another state even without a war declaration; involves its ground, naval, or air forces against those of another state even without a war declaration; imposes a naval blockade on the coasts or ports of another state; and supports armed groups formed on its own territory that invades that of another state.³⁹

A discourse thus emerged that reflected new and evolving norms and understandings regarding the desirability of military force in resolving disputes. To relate to Australian legal scholar Julius Stone’s observation, ‘[a]s a term of legal characterization, however, the term [‘aggression’] could scarcely be meaningful before the attempts, beginning with the League, to limit the customary licence to go to war’.⁴⁰ Oona Hathaway and Scott Shapiro argue that, before the First World War, rulers and international legal theorists saw war not only as a legitimate foreign policy instrument but also as the basis of international law itself. The experience of the First World War itself was what produced this normative shift in international law. Military conflict obviously remained a feature of international politics, but states increasingly had to justify their own actions in accordance with legal principles and ideas that put greater emphasis on pacificism.⁴¹ This normative shift has had perverse consequences. According to Tanisha Fazal, the strengthening of international humanitarian law over the course of the 20th century has discouraged states from declaring formal war on each other or to negotiate peace treaties with one another so that they would not be held internationally liable.⁴²

This post-1919 climate provided the context for when states began to develop non-aggression pacts as a distinct instrument of foreign policy. An agreement explicitly labelled as a non-aggression pact was an appealing innovation precisely because it aligned with the emerging legal and normative architecture that had come to mark international relations at this historical juncture. Therefore, states can use a so-called non-aggression pact in trying to certify their intent to abide by the pacifism that imbues much of the international legal discourse. Non-aggression pacts became a rational tool to ensure consistency and regularity in international politics in a way that was not possible before the 20th century and, more specifically, the First World War.

Why were formal non-aggression pacts most commonly used in interwar Europe?

Formal non-aggression pacts appeared most frequently in interwar Europe, with the vast majority of those concluded in Europe involving the Soviet Union, Nazi Germany, or some state that had not existed before the First World War. Of the 29 agreements to be explicitly labelled as non-aggression pacts before 2002, 22 of them were concluded in the 17 years between 1925 and 1941.⁴³ As Table 1 shows, Nazi Germany signed 6 non-aggression pacts and the Soviet Union signed 12 of them. Of those 22 signed during this 16-year period, 17 involved at least one signatory that did not exist before 1918, whether because it was a new state that was the primary inheritor of a now defunct empire (e.g., the Soviet Union and Turkey for the Russian and Ottoman empires, respectively) or because it was a state that obtained independence partly as a result of an empire no longer existing (e.g., Latvia vis-à-vis the Russian empire). Thus, any explanation of non-aggression pacts must consider why certain great powers – specifically, Nazi Germany and the Soviet Union – were wont to use non-aggression pacts and not others (e.g., the United Kingdom and the United States). Such an explanation must also contend with how relatively new countries, typically located in East Central Europe, tended to be signatories.

Table 1. Non-aggression pacts concluded between 1925 and 1941. Some cases are ambiguous and are marked with an asterisk. The Salonika Agreement contained clauses that went beyond promises of non-aggression, including the lifting of arms restrictions on Bulgaria that had been first raised in the 1919 Treaty of Neuilly-sur-Seine. The Treaty of Non-Aggression between Germany and the Soviet Union is also dubious, considering its commercial provisions and, crucially, division of East Central Europe into ‘spheres of influence.’

The cases themselves offer clues. As military conflict and the physical exhaustion of the Central and Eastern European land empires provided the crucible for new countries to come into existence, territorial disputes were rife. That the peacemakers sought to settle national disagreements with the post-war Treaties of Versailles (1919), St. Germain (1919), and Trianon (1920) only set the stage for more disputes. As Annemarie Sammartino observes, ‘the ethnographic realities of Central and Eastern Europe meant that each state had a highly ambivalent relationship to national borders, and a series of overlapping territorial claims destabilized these settlements as soon as they were drawn up’.⁴⁴ Territorial disputes in other regions were less frequent at this time. Africa, Asia, and the Middle East remained largely under European colonial control, with many of the territorial disputes that took place outside of Europe involving a European power challenging the claims of another European state. In total, 92 disputes saw European participation in some form. Of the 118 new territorial disputes around the world that broke out between 1919 and 1939, 50 of them were in Europe.⁴⁵ After all, states were most numerous in Europe during the interwar period and the physical space there is relatively compact (see Figure 3).

Figure 3. Percentage of states that are European (Data drawn from Correlates of War 2017).

The majority of the non-aggression pacts signed in interwar Europe involved either Nazi Germany or the Soviet Union. Their over-representation in the data requires explanation, because non-aggression pacts were prima facie at variance with their own intentions for undermining the territorial status quo with force. Both ideologically revisionist great powers had a fraught relationship with the League of Nations and international law. Partly because of how both Berlin and Moscow lost territorial control in East Central Europe amid domestic upheaval and battlefield defeats, they perceived the international order, defined by the League of Nations and its associated post-war border agreements, as exclusionary, unfriendly, and, in the Soviet case, anti-socialist. Germany joined the League of Nations in 1926 but withdrew from it in 1933 after Adolf Hitler gained power. Upon seizing power, the Nazi regime purged those government representatives, legal jurists, and university faculty members it saw as undesirable because of their racial make-up and commitment to internationalism.⁴⁶ The Bolsheviks regarded international law as a tool of the bourgeoisie that could threaten the socialist revolution and imperil efforts to achieve communism.⁴⁷ The Soviet Union did join the League in 1934 but was expelled in 1939 following its invasion of Finland. In sum, Germany and the Soviet Union were revisionist, whether out of a desire to take back at a minimum those lands lost to newly created states in the post-war settlements, a profound disagreement about the structure of the global economy, or both.

Yet, in a normative environment that was placing heavier emphasis on the peaceful resolution of conflict than ever before, embracing the discourse of non-aggression had strategic value. From the perspective of signalling theory, a revisionist state can act contrary to its ‘type’ by signing a non-aggression pact. Any state can hypothetically sign a non-aggression pact whatever its attitude towards the status quo, but those groups of states that see themselves and understand each other as status quo-oriented should not feel the need to sign one at all. As Randall Schweller notes, states publicly known as genuinely favouring the status quo should not face security dilemmas with one another.⁴⁸ Indeed, status quo states can still have territorial disputes but do so with the understanding that war between them is highly unlikely, as was the case of Denmark and Canada with respect to Hans Island.

Revisionist states, however, might find non-aggression pacts useful precisely because these agreements allow them to portray themselves as more favourable to the status quo than they really are, at least within the timeframe of the pact. They can show themselves as being in keeping with the emerging international legal climate, whatever their own feelings about it. The post-1919 international legal community within the German academy itself was sceptical of international law, largely seeing it as a plaything of the great powers.⁴⁹ Nevertheless, notwithstanding the expansionist political programme he had sketched in Mein Kampf, Adolf Hitler often made speeches as Chancellor proclaiming his peaceful intentions. He latched onto mainstream ideas of collective security, equality, and national self-determination to legitimate Germany’s resurgence.⁵⁰ Joseph Stalin apparently disavowed expansionism with his intent to cultivate ‘socialism in one country’. Although far removed from any position of authority in the interwar Stalinist system, the early Soviet international legal theorist Eugene A. Korovin wrote that certain principles and rules in international law were useful for the Soviet Union while it went about the transition from socialism to communism.⁵¹ As one study of interwar Soviet foreign policy put it, ‘[t]he Soviet Government considered that the non-aggression pact was a new type of diplomatic instrument to be differentiated sharply from the imperialist instruments such as traditional alliances which merely fomented wars and the collective security concept which lay at the centre of the League of Nations, to the Soviets an organisation of capitalist states bent on aggression, not peace’.⁵² Ironically, in abjuring traditional diplomatic instruments, Soviet leaders adopted the same language of peace that was becoming more common in capitalist societies.

Signing non-aggression pacts thus could have short-term tactical benefits, but not necessarily those identified by the international relations literature cited earlier. Specifically, revisionist states might be afraid of a counter-balancing coalition coming into formation against them before they are adequately prepared to wage war. Non-aggression pacts are useful because they can involve strategically exploiting the normative discourse to confuse and to divide potential adversaries. By using the language of non-aggression, the two revisionist powers might have wanted to make assurances that they were not as revisionist as perhaps feared and that they were more willing than generally believed to abide by emerging norms about peaceful conflict resolution, especially as these two countries did assert territorial claims against relatively young neighbourly states.

Non-aggression pacts are thus often part of a broader wedge strategy, one used by a revisionist great power to manipulate the rhetoric and purported values of an international order so as to justify its own actions while preventing counter coalitions from forming or gathering strength against it.⁵³ Indeed, rising states that disagree with at least some fundamental aspects of an international order can resort to these kinds of non-military methods.⁵⁴ Such power politics take place not only against the backdrop of international anarchy, they can also take place in the context of international organizations, international law, norms, and rules.⁵⁵ Accordingly, non-aggression pacts might impose audience costs,⁵⁶ but their real purpose may have been to create the conditions that make a prospective challenger appear less hostile than it is. Much of the existing scholarship about non-aggression pacts should not assume that non-aggression pacts are ultimately a Pareto-efficient and legalistic arrangement that must be observed in good faith when in fact revisionist governments can use these agreements to empower themselves.⁵⁷

This alternative interpretation about non-aggression pacts aligns with the historical record given their relatively frequent use during the interwar period. Hitler feared being the target of a preventive European war (Europäische Präventivkrieg).⁵⁸ After all, Germany’s military in 1934 remained ill-prepared to fight. Between 1932 and 1934, Germany endured a serious economic crisis, while its gold and currency reserves steeply declined.⁵⁹ Only in December 1933 did its military settle on an expanded armaments programme that envisioned having enough divisions – with the conclusion of its first phase planned for 1937 – to ‘mount an effective defence against a combined attack by Poland and France’.⁶⁰ Nazi diplomacy focused on overturning the Treaty of Versailles and expanding territorial control, but Hitler contrived to prevent an anti-Nazi coalition forming. For his part, alarmed by the Japanese invasion of Manchuria in 1931 and the threat it represented to the Soviet territory, Stalin advocated ‘serious preventive measures of a military and non-military character’ – measures that encompassed troop deployments to the Soviet Far East, a major expansion of the Red Army, and, not least, non-aggression pacts with the Soviet Union’s neighbours.⁶¹ Hence, the series of non-aggression pacts Moscow signed and attempted to sign in the early 1930s.

The question remains: why would countries agree to having non-aggression pacts with such outwardly revisionist countries? Most of those countries that signed non-aggression pacts with either Nazi Germany or the Soviet Union were those that did not exist as independent states before 1918. Their international situation was precarious. Territorial disagreements were rife in Central and Eastern Europe. The three Baltic countries enjoyed no alliance with a major power that would guarantee their security and territorial integrity. Poland had a bilateral military convention with France that dated back to 1921, but Polish leaders increasingly became sceptical about the robustness of this arrangement due to the inability and unwillingness of French political leaders to prepare for a major power war that would involve launching (counter-)attacks against Germany at the outset.⁶² By the 1930s, French leaders came to believe that any military action, on their part, would require British support, which London was reluctant to provide due to public war weariness and the acceptance on the part of the British political elite that some sort of territorial revisionism in East Central Europe was desirable, if not inevitable. Putting the weakness of alliance ties in sharp relief was that potential aggressors on the European continent – Nazi Germany and the Soviet Union – had revisionist intentions towards the territorial status quo.

For the allied or non-allied countries thus trying to survive in East Central Europe, non-aggression pacts were a low-cost way to manage diplomatic relations with those potential aggressors, at least in the short term. Having this flavour were London agreements that involved the Soviet Union and provided for a definition of aggression.⁶³ Although the 1934 German–Polish declaration of non-aggression was to be valid for 10 years, Polish leader Józef Piłsudski doubted that peaceful Polish–German relations would last 5 years. Nor did he believe that the new agreement could be leveraged at the direct expense of Moscow. Shortly after signing the agreement with Germany, he renewed (and extended) the non-aggression pledge signed earlier with the Soviet Union to allay concerns that Poland and Germany would collude against it. He pessimistically remarked that ‘[h]aving those two pacts we are sitting on two stools – that can’t last long. We must know… which one we will fall off first and when’.⁶⁴ Distrust of Germany lingered, but Piłsudski naively thought that Poland will not be the first target of attack because Hitler and other Nazi leaders lacked Prussian ties.⁶⁵ In the view of his minister in Berlin, a belief that Piłsudski purportedly shared, what constrained Hitler was German domestic politics, for Hitler ‘is afraid of power which will confront him with the problem of the fulfilment of the promises which in fact are not possible of realization’.⁶⁶ The non-aggression pact bought time, particularly amid concerns about the military and political reliability of France.⁶⁷ Similarly, according to one contemporary account, Estonia and Latvia signed non-aggression pacts with the Soviet Union when they did out of a perception that developments in East Asia made their great power neighbour weak, which in turn would permit a normalization of their diplomatic relations.⁶⁸

To be sure, during the interwar period itself, states had contemplated but did not sign various non-aggression pacts. Early Soviet overtures for concluding non-aggression pacts were unsuccessful with Estonia, France, Latvia, and Poland.⁶⁹ According to Małgorzata Gmurczyk-Wrońska, in the 1920s, ‘the idea to conclude a non-aggression pact in the form of a common regional agreement would become one of the most important elements in the Polish diplomacy towards the Baltic States and the USSR’.⁷⁰ A system of non-aggression pacts would emerge, but they remained bilateral, invariably featured Moscow, and were never multi-lateralized. Lithuania negotiated one with Germany in the mid-1930s, but none would be signed.⁷¹ Hungary could have negotiated a non-aggression pact with members of the Little Entente – Czechoslovakia, Romania, and Yugoslavia – but those talks broke down in part due to Hungary’s proclaimed right to rearmament and its refusal to reaffirm the territorial status quo.⁷² As such, quantitative studies of these agreements suffer from a selection bias by focusing simply on those that were signed. Moreover, many of these observations are linked to one another, whether during their negotiation or in the minds of the decision-makers, thus violating the statistical assumption that they are independent of one another. Nevertheless, those agreements that were mooted but did not come into fruition still took place in a specific historical milieu.

Post-colonial states, friendship treaties, and non-aggression pacts

Some studies lump together non-aggression pacts with friendship treaties. Yet, the data drawn from those studies suggest that friendship treaties themselves may also be borne in a particular historical context since, as Figure 4 shows, they flourished after 1945. To some extent, their apparent post-1945 frequency is misleading. Friendship treaties have existed for almost as long as societies have engaged in treaty-making. In the modern period, settler countries concluded friendship treaties with indigenous groups, as in the case with those negotiated between the British Crown and the First Nations in Canada.⁷³ After 1945, many treaties feature the word ‘friendship’ in their nomenclature but may not necessarily be a friendship treaty per se. The treaty alliances that the Soviet Union established early in the Cold War are indicative. The Warsaw Pact was predicated on the Treaty of Friendship, Cooperation, and Mutual Assistance between the Soviet Union and seven of its satellites in 1955. The Sino-Soviet Treaty of Friendship, Alliance, and Mutual Assistance framed the defence pact between the People’s Republic of China and the Soviet Union.

Figure 4. Friendship treaties (only) pulled from the data used by Lupu and Poast 2016. It is noteworthy that these treaties are only those previously coded as non-aggression pacts and so do not represent the universe of all friendship treaties concluded between states. See the Supplementary Files for information on these observations.

The available data show that, in the post-1945 period, those countries that appear to have signed friendship treaties – specifically, those friendship treaties without defence clauses – were often post-colonial. Devere, Verbitsky, and Mark also observe that ‘[f]riendship treaties have primarily been signed between the larger powers and smaller nations, between communist states, Asian nations, and South American countries’.⁷⁴ The data from the Cold War period support this observation. Post-colonial states often were the ones concluding friendship treaties with each other during this period. The reason for doing so may also be a result of the unique historical context in which those states had become independent. After all, these states at once needed to address their own fragility and integrate themselves into a legal international order that colonial states had previously devised to help justify their own imperialism. The perceived solution back then was ‘development’, which was, as legal scholar Antony Anghie observes, ‘understood by the new states principally in terms of the furtherance of industrialization and modernization’ and was thus ‘expected to marginalize ethnic identity’ that could otherwise prove destabilizing for them.⁷⁵ Hence, as Devere and her co-authors note, why many friendship treaties may have featured clauses that relate to resource management, maritime access, and educational exchanges.⁷⁶ Moreover, the consolidation of the sovereignty norm strengthened borders, however arbitrary and imposed by European powers they may have been, thereby avoiding a repeat of the situation that vexed East Central Europe between the two world wars.⁷⁷ The reason for concluding those friendship treaties thus may have had much less to do with rivalry – since they had little recent history of it – and more to do with latching onto international legal norms and emerging ideas about socio-economic development. These friendship treaties were not non-aggression pacts by another name.

The biggest wave of friendship treaties, however, came when the Cold War ended and Moscow lost control of East Central Europe. Of the 33 friendship treaties captured in the data that are signed in 1991 and thereafter, 25 involve a country that had been part of the Soviet Union.⁷⁸ A typical example is the friendship treaty between Belarus and Kazakhstan (ATOPID#4765). Another two friendship treaties signed between 1991 and 2001 involved a country that had been part of the communist bloc as a nominally independent country (e.g., Poland). Accordingly, a large majority of friendship treaties were those between the newly formed sovereign countries that had to establish diplomatic relations with one another. Some of these agreements took place in the context of forging new alliance ties. NATO’s 1995 study on enlargement affirmed that ‘[s]tates which have ethnic disputes or external territorial disputes, including irredentist claims, or internal jurisdictional disputes must settle those disputes by peaceful means in accordance with OSCE [Organization for Security and Co-operation in Europe] principles. Resolution of such disputes would be a factor in determining whether to invite a state to join the Alliance’.⁷⁹ The logic underpinning their use of these treaties thus resembles the theoretical argument made by Lupu and Poast. That is, post-communist states wanted to use the friendship treaty to demonstrate to others that relations among themselves and their neighbours will not be troublesome, thereby helping them to embed themselves in liberal international order.⁸⁰ These post-communist states were not necessarily former rivals. Many of them had been formally allied with one another under the aegis of the Soviet Union. They would soon become treaty allies again, albeit under the US leadership via NATO. To be sure, not all agreements signed between post-communist states reflected this logic. One agreement between Hungary and Russia that was mistakenly cast as a non-aggression pact concerned Soviet debt and the withdrawal of Red Army troops (#ATOP4115).

For their part, non-aggression pacts became less frequent after the Second World War, relative to the interwar period as well as to post-war friendship treaties. However, the People’s Republic of China signed many non-aggression pacts during the first half of the Cold War. Its co-signatories were generally decolonized countries that had recently obtained their independence from Western European countries. China had already articulated the Five Principles of Peaceful Coexistence in its 1954 agreement with India that served to recalibrate trade relations between India and Tibet following the latter’s annexation into China. Reflecting basic Westphalian norms of state sovereignty, these principles included mutual respect for territorial integrity and sovereignty, mutual non-aggression, non-interference in each other’s internal affairs, equality and cooperation for mutual benefit, and peaceful co-existence. The Five Principles of Peaceful Coexistence became entrenched in the 1982 Chinese constitution and had guided its foreign policy throughout the Cold War, particularly with respect to countries in sub-Saharan Africa.⁸¹ Abutu Lawrence Okolo claims that this policy ‘formed part and parcel of its wider policy thrust to boost its political influence globally and expand the “One China” principle which was to diplomatically isolate Taiwan’.⁸² China’s co-signatories would go on to provide China with such diplomatic support at the United Nations. In some respects, the logic of China’s use of non-aggression pledges resembles how Nazi Germany and the Soviet Union used them. China was a revolutionary country out of step with the prevailing international order. Although it would pursue an aggressive foreign policy in East Asia early in the Cold War, with much hostility directed towards the Nationalist government in Taiwan, it had to demonstrate its pacific credentials more broadly so as to elicit a much-desired international support. It found a receptive audience in those decolonized states that also wished to affirm Westphalian principles, not least to safeguard their newfound sovereignty.⁸³

Only a handful of non-aggression pacts exist in the record after 1991. The clearest case is the 1991 non-aggression pact struck between North Korea and South Korea, although it too contained provisions calling for enhanced cultural contact and trade.⁸⁴ Superficially, this non-aggression pact appears to demonstrate how formal rivals have entered a new phase in their relationship that they wish to convey to other countries around the world. Yet, the context once again matters. That North Korea’s treaty ally – the Soviet Union – had lost control of its holdings in East Central Europe and was seeking rapprochement with South Korea provided some impetus for North Korea to seek a diplomatic change on the peninsula, albeit one that would give cover to its fledgling nuclear weapons programme.⁸⁵ As much as North Korea would accept international safeguards on its nuclear activities in 1992, 1 year after concluding its non-aggression pact with South Korea, the International Atomic Energy Agency deemed the country to be in non-compliance with its non-proliferation commitments in 1993, thereby precipitating a new peninsular crisis that saw North Korea announce its intent to withdraw from the Nuclear Nonproliferation Treaty.⁸⁶ Pyongyang acted in bad faith and arguably used the language of non-aggression to conceal its own threatening activities.

Conclusion

At first glance, non-aggression pacts seem self-evident: they are agreements, not unlike alliances, that states conclude to demonstrate to each other and to others their lack of intent to go about military hostilities. To explain why states negotiate these agreements, scholars have offered at least two explanations, both of which share a common emphasis on the public nature of these agreements and how rivalry matters for their formation. One explanation is that states sign these agreements to create audience costs that would raise the reputational and diplomatic fall-out of launching an attack, thereby making the rivalry more stable and predictable. Another argument is that these agreements convey to third parties that a rivalry between the signatory states has ceased, thereby building global confidence regarding the prospects for their future cooperation. Scholars have also uncovered mixed evidence about the effectiveness of these agreements, with some uncovering no effect and others claiming otherwise. These arguments have a timeless quality: they are general enough to work across history and regions.

Alas, these agreements are highly contextual in nature and so what these foregoing explanations do is to reify the non-aggression pact. Indeed, the non-aggression pact is not an alliance and, more importantly, it did not truly emerge as a species of international agreement until after the First World War. It was an instrument most often used by Nazi Germany, the Soviet Union, and those European countries that had obtained their independence following the treaties that concluded that global conflagration. Accordingly, the non-aggression pact fit an emerging normative and international legal discourse that was critical of war as an instrument of statecraft. The revisionist states that signed them made use of their discursive potential: agreeing to a non-aggression pact went against their type and appealed to the pacifism that began to permeate international law. Non-aggression pacts allowed these states to demonstrate their compliance to emerging norms about peaceful conflict resolution even when their intentions towards international order were revisionist. In so doing, they could legitimate their own actions so as to prevent a counter-coalition from forming against them. Still, the East Central European states that signed non-aggression pacts with those revisionist states did so because their existence was tenuous, in part because of their many territorial disputes. In the post-1945 period, the non-aggression pact explicitly named as such became rarer. Friendship treaties were much more commonplace, often involving, at least during the Cold War, post-colonial states that were seeking to open diplomatic, economic, and cultural exchanges as newly sovereign states seeking to pursue development.

Scholars should thus be wary of reifying non-aggression pacts, let alone generalizing about them across time and space. Arguments that non-aggression pacts generate audience costs overlook that these agreements were most prevalent in the interwar period among European states, especially ideologically revisionist great powers and newly formed states. The underlying logic may well be plausible, but the empirical gap remains in view of the wide set of territorial disputes in other historical periods and places that did not precipitate non-aggression pacts. Even so, non-aggression pacts during the interwar period were arguably less a tool for imposing audience costs and more an instrument that served as a wedge strategy against enemy coalitions. Indeed, arguments that non-aggression pacts convey the peacefulness of a once troubled rivalry need to consider that the modal great power signatories of these agreements were Nazi Germany and the Soviet Union – revisionist and ideological challengers to the international order that had tactical reasons to appear congruent with greater legalization of world politics. Non-aggression pacts and friendship appear in greater frequency during different historical periods, and so should be analytically distinct from one another. A failure to heed their unique historical context helps account for the contradictory findings that scholars have reported as to the causes and effects of these agreements.