The history of warfare is replete with peace treaties. From the Qadesh Peace Treaty, concluded in 1269 B.C.E. between the Hittites and Egypt (which currently hangs at the headquarters of the United Nations), to the Peace of Westphalia and now-infamous Treaty of Versailles, leaders have met over battlefields and conference tables to determine, and formally agree to, terms of peace. Concluding a peace treaty following the termination of hostilities was a norm of international relations for millennia. But this is no longer the case. From approximately 1950 on, even though the absolute number of wars between states has not decreased dramatically, the number of peace treaties accompanying those wars has sharply declined. Why?
Although a growing literature on war termination and the duration of peace has addressed several important issues related to this question,Footnote 1 we lack an understanding of both why peace treaties are concluded and why belligerents have stopped using them to mark the end of interstate wars. In Iklé's famous phrase, “every war must end.”Footnote 2 But not every war ends with a peace treaty.
Whether and when a war ends with a peace treaty has practical as well as theoretical implications. Peace treaties—defined in this article as the written documents that describe contracts between belligerents to cease hostilities and resolve issues under disputeFootnote 3—serve several functions. They are primarily meant to terminate wars and restore friendly relations among belligerents, including the resumption of diplomatic relations and trade.Footnote 4 Furthermore, recent scholarship has illustrated that, problems of endogeneity notwithstanding, it is not too far a step to surmise that formal (versus informal, or even no) peace treaties also may cement peace.Footnote 5 On average, among wars that stop and resume at a later date, wars ending in peace treaties see four more years of peace compared to wars ending without peace treaties. Among the set of war-dyads that see a resumption of war at a later date, the average duration of peace for wars ending without peace treaties is eleven years; the average duration of peace for wars ending with peace treaties is twenty years. Wars that resume despite the conclusion of peace agreements also may be less lethal.Footnote 6
Concluding peace treaties serves as an implicit acknowledgment that a state of war existed previously; at least as important, peace treaties eliminate ambiguity about whether two countries are (still) at war.Footnote 7 By contrast, the absence of a peace treaty can create difficulties for belligerents seeking to normalize relations, resolve disputes, and remove troops from disputed areas, even when the original conflict appears to have ended. The absence of a Russo-Japanese peace treaty following World War II, for example, prevented the normalization of relations between these two countries for decades.Footnote 8
Recent developments in the international system—specifically, the emergence of a strong and growing canon of the law of war—have, however, created disincentives for states to conclude formal peace treaties at the termination of war. The codification of norms governing the appropriate conduct of states engaged in warfare (jus in bello) has raised the potential costs of admitting to being in a state of war. Because the conclusion of a peace treaty is tantamount to admitting that one has been in a state of war (by ending the state of war), states today may have become less inclined to conclude peace treaties.
This article contributes to at least four literatures. First, scholarship on the conclusion and duration of peace agreements has made great strides in recent years but has tended to focus almost exclusively on the post-1945 era.Footnote 9 Scholars in this area are therefore not poised to notice any long-term trends in the use of peace agreements. Second, because the decline of the use of peace treaties in interstate wars represents the decline of an international norm, this article also speaks to scholarship on the life cycle of norms.Footnote 10 With only a few exceptions, however,Footnote 11 most scholars of international norms have not yet focused on the decline (as opposed to the emergence, or strength) of international norms.Footnote 12
Third, the specific argument made here speaks to the costs and benefits of the legalization of international politics. Increasing regime complexity can lead to both intended and unintended consequences, many of which span multiple issue areas. Some view this development as a positive one, allowing gains in efficiency as well as societal well-being. For example, the creation of more trade-dispute-settlement forums may increase the reputational costs to any state considering reneging on a trade agreementFootnote 13 But there are also potential downsides to legalization and regime complexity.Footnote 14 For instance, an apparently more robust system for prosecuting war crimes on the international stage may hinder rather than help in achieving a just peace.Footnote 15 My argument falls into the latter category.
Fourth, along with the decline of territorial conquestFootnote 16 and major power war,Footnote 17 the decline of peace treaties is among a myriad of war's transformations. While it may or may not be the case that war itself is in decline,Footnote 18 it is apparent that wartime diplomacy has changed dramatically. We no longer call wars wars. Instead, they are “invasions,” “counterterrorist actions,” or “humanitarian interventions.” As such, the formalities of war—such as peace treaties—seem to have fallen away. Changes in how we label war may create the impression that war itself has gone out of vogue when it remains all too present in today's world.
A Brief History of Peace Treaties
The first known peace treaty is the Treaty of Qadesh, concluded between the Hittite Emperor Hattusilis and Ramses II of Egypt. Historian Trevor Bryce argues that Hattusilis's main motivation for concluding the peace treaty with Ramses II “was the legitimacy which the agreement conferred upon Hattusilis's occupancy of the Hittite throne.”Footnote 19 Bryce suggests that Hattusilis's usurpation of his nephew's throne placed him in an awkward position with respect to domestic and international legitimacy. Because vassals swore to specific kings and lineages and because the Great Kings of the Bronze Age concluded treaties with each other on a personal basis, Hattusilis could not be secure in his reign until and unless he received some form of recognition from key external actors. Among the four kings, Ramses was first among equals. Ramses was also sheltering Hattusilis's usurped nephew. By recognizing Hattusilis as a brother king in the Qadesh Treaty, Ramses firmly placed him as an equal. The earliest peace treaties may therefore have been used as a means of signaling and solidifying sovereignty. In the context of interstate war, it is puzzling then why peace treaties continued to be used even once a sovereign state system was well established and, further, why they have fallen out of use more recently.
Although the form of peace treaties has changed over time, they have always been characterized by highly ritualized and formal behaviors. The Ancient Greeks and Romans erected pillars to mark the conclusion of peace treaties.Footnote 20 In the Middle Ages, these agreements were often concluded through highly symbolic acts such as the “kiss of peace.”Footnote 21 Also known as the osculum pacis, the kiss of peace was similar to today's familiar handshake between heads of state upon the conclusion of a peace agreement. But the kiss of peace was even more meaningful in an era where such agreements were not likely to be written down. The kiss was considered to be a solemn oath. After England's King Henry II refused to give Thomas à Becket the kiss of peace upon the conclusion of a twelfth-century peace treaty between France and England (part of which revoked Becket's exile), Becket returned to England and was murdered shortly thereafter.Footnote 22 Thus, even before peace treaties appeared most commonly in written form, concluding such treaties was a public act surrounded by great formality. Moreover, to refuse to engage in these formalities, as Henry II did, was frowned upon; to disobey the dictates of peace risked punishment from the heavens.
Concluding peace treaties—through oaths, kisses, or written documents—has been a norm of international politics for millennia. Of late, however, this norm appears to have fallen into desuetude. Of the sixty-four wars fought in the twentieth century,Footnote 23 only twenty-three have been accompanied by formal peace treaties. The drop in peace treaty use is especially visible after World War II. The Korean War ended with an armistice, but no peace treaty. The Paris Peace Accords that terminated U.S. involvement in Vietnam serve as a formal cease-fire document between the United States and North Vietnamese without resolving any of the political issues motivating their war. The 1991 Persian Gulf war ended with a series of resolutions handed down from the UN Security Council, not negotiated between belligerent factions—although the United States obviously had a say in this case, the Iraqis played no role in determining the content of the resolutions.Footnote 24 Iran and Iraq agreed to a cease-fire and the promise of future peace treaty negotiations in 1988, but those negotiations never materialized.
On the other hand, some of the bitterest rivals of the late twentieth century have concluded peace treaties, but inconsistently so. For example, India and Pakistan signed a peace treaty after the 1971 Bangladesh War, but not after the prior (or subsequent) wars over Kashmir. Israel and Egypt signed the Camp David Accords in 1978 (Jordan also concluded a peace treaty with Israel, but decades later). And the Football War between El Salvador and Honduras was punctuated by the Treaty of Lima.Footnote 25 These few exceptions notwithstanding, the clear trend in the post–World War II era has been away from the use of formal peace treaties to conclude interstate war.
Historically, concluding peace treaties was “a standard of appropriate behavior for actors with a given identity,” that is, states at war.Footnote 26 Although we lack an account of the origins of the norm of concluding peace treaties, the endurance of this norm—both its historical trajectory and the frequency with which peace treaties have been concluded—suggests that it was fairly deeply internalized among states. In recent decades, however, we have seen a significant move away from the use of peace treaties to conclude interstate wars. This shift raises questions of theoretical as well as practical import, the most critical of which is: Why has the use of peace treaties declined?
Whither Peace Treaties?
Beginning in the mid-nineteenth century, international jurists began the process of codifying the laws of war governing armed conduct, or jus in bello, in multilateral treaties. In 1856, there was one such law—the Declaration of Paris—which governed maritime warfare. Today, there are more than forty such laws of war.
States admitting to being in a state of war during an era of codified jus in bello are unequivocally obliged to pay the costs of compliance and to bear the consequences of noncompliance with the laws of war. The costs of compliance include investment in training, as well as foregoing the use of specific weapons and military strategies. Accepting these obligations can put states at a disadvantage in the prosecution of a war. It is not the benefits, but rather the costs of concluding the peace treaties that have changed over time because the obligations attendant upon belligerent states have risen with the codification of jus in bello.
There are several sources of the costs of noncompliance. Some costs are included in the terms of several major laws of war, which dictate punishment or restitution for violations of jus in bello. These costs may be borne by states themselves, or by individuals representing states; in either case, they are costs states typically seek to avoid or limit. For example, article 3 of the Hague Convention of 1907 Respecting the Laws and Customs of War on Land states: “A belligerent party which violates the provisions of the said Regulations shall, if the case demands, be liable to pay compensation. It shall be responsible for all acts committed by persons forming part of its armed forces.”Footnote 27 Similarly, each of the 1949 Geneva Conventions requires punishment of violators of the Conventions. Further, articles 3 and 4 of the First Protocol of the 1954 Hague Convention for the Protection of Cultural Property require the return of cultural property unlawfully expropriated during war and related indemnities if these obligations are not met.Footnote 28
Belligerents also may pay a soft cost of noncompliance insofar as constituents of leaders of noncompliant states value adherence to international law, and insofar as noncompliant belligerents value their reputation with respect to the international community. For example, when India failed to live up to its commitment to repatriate, in a timely manner,Footnote 29 more than 90,000 Pakistani prisoners of war following the 1971 Bangladesh War, Pakistan filed suit against India in the International Court of Justice.Footnote 30 The Indians, knowing that being in a formal state of war effectively left them without recourse, quickly conceded and returned most prisoners. Concern for their reputation as good citizens of the international community, and proponents of international law, was critical to the Indians' political retreat. They sought, in the words of a Times of India headline at the time, to avoid a situation where world opinion would be diverted away from India's good works.Footnote 31
Peace treaties provide windows of accountability where violations of law can be tallied and punished. Especially with the creation of institutions such as the International Criminal Court, belligerents guilty of violating the law of war could pay the price for noncompliance following wars concluded by a formal peace treaty. Over the past century, international war crimes tribunals have become increasingly common.Footnote 32 With this new era of accountability comes a reluctance of state leaders to expose themselves and their soldiers to liability.
As the body of codified jus in bello has grown, states might therefore prefer to create some ambiguity regarding their status as belligerents in order to make arguments that they may be exempt from adhering to the laws of war. Prior to the codification of these laws, their applicability in any given conflict could be subject to debate. Today, however, once a state of war has been acknowledged by the belligerents—and such acknowledgments typically occur via a declaration of war or a peace treaty—the applicability of jus in bello is indisputable. Thus, the absence of a peace treaty at the conclusion of hostilities leaves certain doors open with respect to restitution, the effect of war on contracts, neutrality rights, and perhaps even war crimes.
Fundamentally, this argument is about how actors behave strategically with respect to international law. To be clear, I am not arguing that jus in bello does not apply if a war is unaccompanied by a peace treaty and/or a state of war is not acknowledged by one or more belligerents. Modern jus in bello treaties such as the 1949 Geneva Conventions attempt to plug exactly this loophole by stating that the law applies in times of “armed conflict,” and not just “declared war.” But international lawyers have lacked definitions of “undeclared war” and “armed conflict.”Footnote 33 Indeed, delegates to the 1907 Hague Convention were expressly unwilling to define war, apparently wanting to retain the flexibility of naming their armed conflicts “expeditions” or otherwise.Footnote 34 In pre-negotiations for the 1949 Geneva Conventions, an International Committee of the Red Cross (ICRC) representative demonstrated a similar reluctance, stating that the committee “was not in favour of too much precision”Footnote 35 when it came to a strict delineation of the applicability of the conventions. States, too, should prefer to remain in the vaguer areas of the law, and concluding a peace treaty would eliminate this vagueness.Footnote 36
Although we should not expect policymakers to admit to this state of affairs, evidence from at least one late-twentieth-century war illustrates this logic. Reflecting on the British government's query regarding whether the UK should declare war on Argentina during the 1982 Falklands War, the British Foreign Ministry's legal advisor wrote: “It was perhaps the most momentous piece of legal advice which I ever had to give during 34 years of service in the Legal Branch of the Diplomatic Service; but it was also one of the easiest.”Footnote 37 Declaring war “would attract considerable baggage in international law,”Footnote 38 and was quickly rejected. Not admitting to a state of war affected how the British described their behavior, particularly with respect to jus in bello. For example, they created maritime “exclusion zones” rather than “blockades” around the Falklands, and memoirs of the war indicate that these vocabulary choices were dictated by the Foreign Ministry legal department. According to Britain's official history of the war, “the FCO's Legal Advisor strongly advised against describing this as a ‘blockade’ in any public statements, as in international law this term was closely connected with a declared or acknowledged war.”Footnote 39 Although “the military would have been content to call it a blockade, this created problems in international law.”Footnote 40 By retaining ambiguity, they were able to engage in behavior that was probably not legal under international law.Footnote 41
Growing opportunities for accountability for violations of jus in bello could, understandably, make decision makers nervous about exposing themselves and their soldiers to prosecution. One way to limit exposure is to avoid admitting to a state of war.Footnote 42 Peace treaties constitute a clear signal that a belligerent has been in a state of war, whereas avoiding a peace treaty maintains some ambiguity. Thus, as codified jus in bello has proliferated over the past century and a half, states may prefer to avoid concluding formal peace treaties as a legal strategy to decrease the likelihood of paying the costs of compliance and the price of noncompliance.
H1: As codified jus in bello proliferates over time, states will be less likely to conclude peace treaties.
H1 suggests a systemic cause for the declining use of peace treaties, one that reflects a change in the existence and number of international laws governing the conduct of war. But how do systemic factors affect unit-level behavior? It might be the case that the general proliferation of codified jus in bello produces a blanket change at the unit level, but it also seems plausible that those belligerents that have ratified more relevant treaties will be especially sensitive to the possibility that concluding a formal peace treaty unequivocally activates jus in bello. This distinction parallels the difference between customary and positive (that is, treaty) law. Indeed, it may be that we should expect to observe a stronger effect of the proliferation of codified jus in bello at the unit/dyad level prior to the elevation of the status of this body of law to customary law, which would be observed at the systemic level:
H1a: Dyads with greater levels of ratifications of jus in bello treaties will be less likely to conclude wars with peace treaties compared to dyads with lower levels of ratifications of jus in bello treaties.
Consistent with the general logic that the proliferation of codified jus in bello should lead to a decline in peace treaties is the notion that formal declarations of war—which also clearly activate the laws of war and have also been on the decline—should be observed (or not observed) in tandem with peace treaties. By my logic, wars that begin with declarations of war should be especially likely to conclude with formal peace treaties because the declaration would have already placed belligerents over the bright line of having to comply with jus in bello.
H1b: Wars that begin with declarations of war are especially likely to conclude with formal peace treaties.
It is important to note, though, that there should also be some exceptions to this rule. States that are eliminated from the international system will not be able to conclude a peace treaty. It is also possible that belligerents begin a war expecting to be compliant with jus in bello, but see those expectations dashed during the course of the war; thus their incentive to conclude a formal peace treaty is diminished. Indeed, states that violate jus in bello should be particularly averse to concluding peace treaties that may leave their leaders and soldiers vulnerable to demands for restitution or punishment. Moreover, violations of laws of war that are most specific regarding punishment of violation may be especially likely to undermine the conclusion of a peace treaty. For example, the 1954 Hague Convention on the Protection of Cultural Property contains clear provisions for violation of laws of war regarding the protection of cultural property. Although appropriating cultural treasures as the spoils of war was a norm of international politics since at least the Roman Empire, this norm changed dramatically beginning in the nineteenth century. Following the Napoleonic Wars, both art appreciation and international lawyers enjoyed ascendance. This combination led to growing norms in favor of protecting cultural property during warfare; these norms were first written down in a series of draft treaties and, most important, in the 1954 Hague Convention.Footnote 43 The 1954 convention states: “The High Contracting Parties undertake to take, within the framework of their ordinary criminal jurisdiction, all necessary steps to prosecute and impose penal or disciplinary sanctions upon those persons, of whatever nationality, who commit or order to be committed a breach of the present Convention.”Footnote 44 Many postwar trials have included charges of plundering. Several high-level Nazis were convicted of looting and plunder of cultural property as a result of the Nuremberg trials in 1945–46.Footnote 45 Since its creation in 1993, the International Criminal Tribunal for the former Yugoslavia (ICTY) has prosecuted several Serbian and Croat military officers for such actions as bombing and plundering the Old Town of Dubrovnik and bombing the famous Stari Most bridge in Mostar.Footnote 46
Even when criminal charges are not on the table, the practice of restitution and return for plundered cultural treasures has become accepted and common after war. Restitution is often required as part of peace terms, especially for the losing party. For example, the Allies' peace treaty with Austria after World War II contains extensive provisions for the return and restitution of various types of property, including cultural property. By contrast, Germany and Russia—two countries that did not conclude a peace treaty with each other—continue to debate the fate of thousands of cultural objects taken from Germany by Russian soldiers during World War II. In the absence of a peace treaty, the Russians are unable to make legal arguments about compensatory restitution that could allow them to keep these objects.Footnote 47 Thus, we should expect that belligerents that violate the laws of war regarding protection of cultural property should be unlikely to conclude formal peace treaties, compared to belligerents that comply with the laws of war regarding protection of cultural property.
H1c: States violating jus in bello that are very specific regarding punishment should be less likely to conclude formal peace treaties compared to states that comply with jus in bello.
Given the nature of this argument, it is much more likely that indirect rather than direct evidence will be available. Policymakers should be reluctant to state explicitly their intent to create ambiguity about the applicability of the laws of war; we should not expect to find prolonged discussions of the calculations made to avoid the formalities of war as a means of reducing the costs and risks of compliance and noncompliance. In this respect, the evidence from the Falklands War should be viewed as strong support from an unexpected quarter.
Additional and Alternative Explanations
Two main classes of alternative explanations also could account for the decline in the use of peace treaties. The first refers to a secular trend. It is possible that something other than the proliferation of codified jus in bello or the control variables discussed shortly is driving the conclusion of formal peace treaties, and that this variable is best captured as a time trend. Even though time, on its own, is a relatively atheoretical variable, it could capture other changing factors that could be affecting the declining use of peace treaties. Thus, it is included in the analysis as a sort of omitted variable check.
The second type of alternative explanation relies on a general understanding of why states sign peace treaties under some conditions, but not others. These explanations refer to the particular circumstances of wars that could lead to an increased likelihood of concluding a peace treaty. All of these conditions also fall under the rubric of the transformation of war: war has become less likely to end in victory and with territorial exchange, more likely to be fought under the aegis of the UN and by democracies, and less likely to be fought between major powers.
War outcomes
Military victory is often accompanied by spoils, the ability to rewrite history, and an increased likelihood of durable peace. When one side has clearly won a military conflict, conference tables may become more efficient than battlefields as locations for resolving political issues. This is not to say that the winning side necessarily dictates the terms of peace unilaterally. While there are certainly cases where this has occurred (with the Treaty of Versailles being an obvious example), as long as the losing side has not been eliminated completely, negotiations can linger on even in the wake of a decisive military defeat.Footnote 48 Nonetheless, military victory may be more likely to produce a peace treaty than a war that ends in a draw because the alternative—a return to the battlefield—is not a very credible threat for the losing side. This logic suggests the following hypothesis:
H2a: Wars ending in military victory will be more likely to be accompanied by peace treaties compared to wars that end in draws.
Separately, Fortna and Toft have noted a decline in victory in wars.Footnote 49 Both interstate and civil wars are much more likely to end in draws today. If one of the key purposes of a peace treaty is to codify a settlement of issues over which war is fought,Footnote 50 we should expect that peace treaties will decline as draws increase.
Scholarship on enduring rivalries, by contrast, suggests a different relationship between military victory and peace treaties.Footnote 51 To the extent that wars—such as those between India and Pakistan over Kashmir or those between various Arab states and Israel—are related, war that ends in a military victory does not necessarily mean that political issues have been resolved.
Many recent wars suggest another, related, change in war: today, wars are more likely to end in regime change than in territorial conquest or more traditionally limited military victories. Insofar as regime change is an increasingly common mode of war termination, it could be the case that peace treaties have become unnecessary. Instead of forging an agreement between foes, what may be more useful in these types of cases is a treaty of alliance or friendship. This observation suggests the following hypothesis:
H2b: Wars concluding with foreign-imposed regime changes (FIRCs) will be less likely to be accompanied by a peace treaty compared to wars that do not conclude with FIRCs.
A related possibility refers to wars that conclude with the destruction of a state, in which case there is no one on the other side of the table with whom a peace treaty could be signed. This logic suggests that wars ending in the death of states will be less likely to be accompanied by peace treaties than wars ending with the retention of sovereignty for the belligerents.Footnote 52
H2c: Wars concluding in state deaths will be less likely to be accompanied by peace treaties compared to wars that do not conclude with state deaths.
Territorial conquest
Another circumstance that could increase the probability of a peace treaty being concluded is territorial exchange as a result of the war. Territory, as is well known, is a major cause of war.Footnote 53 If territory is ceded or taken in war, a peace treaty provides an important opportunity to codify that change. In effect, the peace treaty can serve as a type of deed—a public document certifying ownership of a particular piece of territory. Indeed, this may be one of the most important purposes of peace treaties, and suggests another hypothesis:
H3: Wars in which one side receives territory from another are more likely to be accompanied by peace treaties compared to wars in which territory does not change hands.
Recent scholarship on territorial conflict and, in particular, on the emergence of a norm against territorial conquest in the mid- to late twentieth century, suggests that territorial exchange as a result of warfare is likely to be frowned upon today.Footnote 54 Interestingly, this prohibition does not seem to prevent the emergence of territorial wars in the same era; territory remains an extremely salient cause of war. Two of the most important post-1945 clusters of wars—those between India and Pakistan over Kashmir, as well as those in the Middle East over Israel's borders—are indisputably territorial conflicts. Nonetheless, if a war is being fought over territory, but a strong international norm exists against territorial aggression, the public transfer of territory as a result of interstate war may become unacceptable.
The decline of major power war
The clear decline in major power war since World War II provides another possible explanation for the decline in peace treaties. Insofar as major powers care about the status of their treaty partners, we might expect that they will be particularly likely to conclude peace treaties with each other, but not with nongreat powers.
H4: Major powers are more likely to conclude peace treaties with each other than with nonmajor powers.
Democracies and peace treaties
There are good reasons to think that democracies will be more likely than nondemocracies to conclude peace treaties. For example, peace treaties are somewhat analogous to contracts in domestic politics and are consistent with notions of conflict resolution being accompanied by codification. Thus, we might also expect to observe a relationship between the regime type of belligerents and the conclusion of a peace treaty.
H5: More democratic dyads will be more likely to conclude peace treaties compared to less democratic dyads.
War duration
Bargaining theories of war suggest that conflicts where the bargaining space is narrow and even shrinking will be particularly long and difficult to end.Footnote 55 By this logic, we should expect a negative association between war duration and peace treaties; those wars that present the hardest challenges of termination may also present extreme difficulties of codifying the termination of the war and settling the issues at dispute.
H6: Longer wars will be less likely to be concluded with peace treaties compared to shorter wars.
Peace treaties and jus ad bellum
International legal scholars attribute the decline in the use of formal peace treaties to the rise of the UN, for two reasons. First, they suggest that UN Security Council (UNSC) resolutions may act as substitutes for peace treaties.Footnote 56 The second reason to link the rise of the UN with the decline of peace treaties is as follows: if war, and especially aggression, are considered to be illegal acts by international law, then few states may wish to admit complicity through a formal peace treaty.Footnote 57
Although the UN Charter does not prohibit war itself—indeed, it permits defensive war—it does prohibit wars of aggression. Aggressor states can and, more often than not, do portray themselves as victims. Even so, with the rise of an institution as powerful (relative to its predecessors) as the UN, it may seem safer simply to avoid any behaviors that would clearly put aggressor states in the realm of official warfare. One way of achieving this goal is to avoid concluding peace treaties.
H7: UN member states are less likely to conclude peace treaties compared to non-UN member states.
Defining and Coding Peace Treaties
I define a peace treaty as a written document that describes a contract between belligerents to cease hostilities and resolve issues under dispute. It must be signed and ratified by parties to a conflict. Similarly, Black's Law Dictionary defines a peace treaty as “a treaty signed by heads of state to end a war.”Footnote 58 Typically, peace treaties include several articles meant to codify the settlement of disputes, especially: a preamble; political and territorial clauses; financial, economic, and juridical clauses; and safeguards and guarantees.Footnote 59 Although wars can end—even formally—via unilateral declarations, such declarations do not constitute peace treaties precisely because they are unilateral. Cease-fires, armistices, and truces also do not in themselves constitute peace treaties because they are meant to create temporary cessations of hostilities; they are not necessarily meant to resolve issues under dispute.
The Treaty of Paris that concluded the Spanish-American War is a good baseline example of a peace treaty. Subsequent to the cessation of hostilities in early August 1898, the U.S. and Spanish representatives negotiated a temporary peace protocol, signed on 12 August. The protocol left several key issues, notably the future status of the Philippines, unresolved; these issues were to be deferred to the official peace conference, which opened in Paris on 1 October 1898. Although at a key moment in the negotiations there was a very real possibility of both sides walking away from the negotiating table, a treaty was, ultimately, concluded. The final treaty included provisions regarding: the U.S. occupation of Cuba; transfer of Puerto Rico and the Philippines to the United States; repatriation of prisoners of war; and indemnities.
It is worth considering some more ambiguous cases, most of which I do not code as peace treaties. When a group of Western states, plus Japan, invaded China to relieve their besieged legations in Beijing during the Boxer Rebellion of 1900, their path was one of rape and pillage.Footnote 60 The war ended with the Final Protocol, which contained provisions for indemnities, reparations, and trade policy, but it was clearly not intended to be a peace treaty. Grob writes that concern for the domestic legitimacy of the agreement precluded its being treated as a formal peace treaty, and that its being called a “protocol,” rather than a peace treaty “served the purpose of denying the legal existence of war.”Footnote 61
Several additional ambiguous cases refer to instances of wars that preceded—and then essentially bled into—World War II, making distinguishing the conclusion of these wars from the conclusion of World War II exceptionally difficult. These cases include: the Sino-Soviet War of 1929, where a planned peace conference was preempted by Japan's invasion of Manchuria; the 1937 Sino-Japanese War, whose conclusion was complicated by the subsequent civil war in China; and the 1939 Nomonhan War (between the Soviet Union and Japan), whose conclusion seems to have been determined as much by a border commission as by a formal peace agreement.
Also challenging are cases of formal agreements that clearly do not cover all the political issues under dispute. The Tashkent Declaration that concluded the Second Kashmir War between India and Pakistan, for example, provided for the military withdrawal of both sides and resumption of diplomatic relations, but deferred resolution of political issues to another series of (unsuccessful) talks. Thus, while Tashkent includes minimal political content, it does not really address the issues at stake in the war and is therefore not coded as a peace treaty in this study. This agreement contrasts with the Simla Agreement that concluded the 1971 Bangladesh War. The Simla Agreement and the Tashkent Declaration read very similarly, but for the fact that the main issue under dispute in the Bangladesh War—the independence of East Pakistan/Bangladesh—is resolved in the Simla Agreement, which is coded as a peace treaty in this study.Footnote 62 Similar to the Tashkent Agreement, a 1991 agreement associated with the 1977 Vietnamese-Cambodian war set the stage for a UN peacekeeping mission to prepare for free elections in Cambodia, but did not itself resolve the main issues at stake.
The UN's role in peace negotiations complicates our understanding of post-1945 peace treaties. Both the Iran-Iraq War and the 1990 Persian Gulf War were accompanied by UN Security Council resolutions that laid out the terms of peace, and to which the relevant parties agreed. Although these resolutions serve similar purposes as peace treaties, their form diverges so dramatically from what we would conceive of as a traditional peace treaty that it is difficult to code them as such. Rather than two sides negotiating terms across a conference table, these resolutions were more like ultimatums handed down from a third party (especially in the case of the Iran-Iraq War).
The main distinctions between peace treaties and other peace-related agreements and instruments speak to both form and substance. On substance, agreements that address procedural but not substantive issues are not coded as peace treaties; for example, a cease-fire agreement that demarcates a buffer zone and establishes a timeline for future talks would not constitute a peace treaty, but an agreement that solidified a border and resolved other outstanding issues could. On procedure, a peace treaty must be written, negotiated among the belligerents, and considered a peace treaty by the belligerents. Thus, an oral agreement, unilateral withdrawal, UNSC resolution, or written agreement that is expressly stated not to be a peace treaty would not be considered as such. In the empirical analysis that follows, I include many of these ambiguous cases as peace treaties in robustness checks to ensure that my findings are not an artifact of my definition.
Approximately one-half of the ninety-four interstate wars from 1816 to 2007 were accompanied by peace treaties, but this percentage looks quite different when we examine different time periods.Footnote 63 Of the fifty-six wars that ended before 1950, forty (more than 70 percent) were accompanied by peace treaties. By contrast, only six of thirty-eight (approximately 15 percent) of wars ending after 1950 were concluded with formal peace treaties. This trend is illustrated in Figure 1.
FIGURE 1. Percentage of wars ended with peace treaties, 1820–2007
Empirical Analysis
The data set used in this article is composed of nondirected war dyads (as an example, the United States and Germany in World War II yield only one observation in the data set). Because the dependent variable—whether a peace treaty is concluded—is binary, logistic regression is used to analyze the data. Given the large number of multilateral wars in the data set, the models are clustered by war to produce robust standard errors.
Description of Independent Variables
I use the War Initiation and Termination (WIT) data setFootnote 64 to analyze the decline in peace treaties in interstate war. The universe of cases that bounds the data set is the Correlates of War (COW) list of interstate wars from 1816 to 2007.Footnote 65 For each belligerent in each war, extensive secondary research was conducted to determine the military (as distinguished from the political) outcome of the war, whether territory changed hands between dyad members fighting each other, and what type of agreement (if any) was reached between the two states at the conclusion of the war.
The variable jus in bello is based on the ICRC's list of codified international humanitarian laws since the nineteenth century.Footnote 66 The codified international laws that make up this list begin with the 1856 Declaration of Paris and continue with the 1899 and 1907 Hague Conventions as well as the 1949 Geneva Conventions. The list also includes such laws as the 1980 Conventional Weapons Convention and the 1998 Rome Statute, which created the International Criminal Court. The list is meant to include all laws that pertain to belligerent conduct during the course of warfare. Treaties regarding the conditions under which states may go to war—such as the Kellogg-Briand Pact or UN Charter—or those regarding acceptable treatment of citizens by governments that are meant to apply primarily during peacetime—such as the International Covenant on Economic, Social, and Cultural Rights—are not included. Some treaties that cover both peacetime and wartime issues, such as the Convention on the Rights of the Child, which includes provisions regarding the use of child soldiers, however, are included, although these comprise a very small number of treaties in the list. For each year, I counted the number of laws of war governing the use of force.
The variable mean ratifications, defined as the mean number of jus in bello treaties each dyad had ratified at the start of the conflict, is used to test H1a. Unlike the systemic jus in bello variable, mean ratifications sees a fair amount of cross-dyad variation. Moreover, mean ratifications is much less correlated with the year in which a war ends than the systemic jus in bello variable.Footnote 67 States do not automatically ratify laws of war; indeed, they often wait a great deal of time to do so. For example, the United States did not ratify the 1980 Conventional Weapons Convention until 1995, the UK did not ratify the 1856 Declaration of Paris until 1956, and China did not ratify the 1954 Hague Convention Regarding the Protection of Cultural Property until 2000.Footnote 68
A technical reason to treat the proliferation of jus in bello as a systemic variable is that peace treaties require at least two signatories. If the belligerents have ratified different numbers of (different) jus in bello treaties at the time a war ends it is difficult, theoretically, to know which belligerent's level of ratification should be most predictive of a peace treaty's conclusion. This issue becomes especially complicated in multilateral wars. That said, the fact that ratification data provide a dyad-specific independent variable that does not necessarily track directly with time is an important advantage to bear in mind.
declaration of war was also taken from the WIT data set, and is defined as a public proclamation of a state's intention to engage in hostilities with another state that is issued according to the laws of the issuing state. Thus, while the Gulf of Tonkin Resolution authorized the U.S. president to commit troops to Vietnam, it (famously) did not constitute a declaration of war. By contrast, the U.S. declaration of war against Japan in World War II, which was requested by the president and resolved by Congress, does qualify as a declaration of war.Footnote 69
To test H1c, which addresses compliance with jus in bello, I use Morrow and Jo's data on compliance with the laws of war regarding protection of cultural property in all interstate wars in the twentieth century.Footnote 70 My analysis employs Morrow's ordinal compliance index, which is a composite compliance score based on the magnitude, frequency, degree of centralization, and clarity of violations.Footnote 71
Although the Morrow and Jo data cover nine issue areas (aerial bombardment, armistice/ceasefire, chemical and biological weapons, treatment of civilians, protection of cultural property, conduct on the high seas, prisoners of war, declarations of war, and treatment of the wounded), I focus exclusively on noncompliance with cultural property laws for two reasons. First, it is one of the issue areas among the laws of war that contains specific provisions for punishment. Second, unlike most of the other issue areas considered by Morrow and Jo, all the documents used to generate their coding rules for protection of cultural property are actual treaties, rather than declarations or draft conventions. Given that my key causal mechanism rests precisely on the proliferation of codified jus in bello treaties, incorporating the additional element of uncodified laws of war into a measure of compliance would potentially add a great deal of error to the analysis. Because my unit of analysis is nondirected dyads, I use the greater of the two possible values of noncompliance for the dyad members. The number of observations for regressions that include this variable will be significantly lower than in other regressions because of missing data on this variable, because the Morrow and Jo data do not include wars added to the latest list of interstate wars from COW, and because their data set covers only the twentieth century.Footnote 72
The variable military victory was taken from the WIT data set, which includes a variable for military outcome as a tie or draw, a slight victory for one side, a clear victory for one side, or an extreme victory. I transform this variable into a binary variable, where “military victory” = 1, if the war ended in a clear or extreme victory.Footnote 73 The variable firc codes for whether a foreign-imposed regime change has occurred as a result of the war and is used to test H2b. The coding for this variable is taken from Downes and Monten's list of such regime changes.Footnote 74 The Downes and Monten list was then cross-referenced with the COW list of interstate wars to determine which regime changes were, in fact, associated with wars. Fazal's coding of state death as “the formal loss of control over foreign policy to another state” is used in this study to determine whether any dyad member was eliminated at the conclusion (or during the course) of hostilities.Footnote 75
territorial change simply describes whether territory changed hands as a result of the war. Taken from the WIT data set, it is coded relative to the territorial status quo ante bellum, on a five-point scale from no territory changing hands; through colonial possessions or protectorates that change hands or become independent; ownership changes over small or unimportant bits of territory; large or important changes in territorial control of sovereign territory; and entire states being taken over. As with the military victory variable, I transform this variable into a binary variable, where “territorial change” = 1, if any territory exchanged hands during the war.
To test H5, I use data from the Polity IV project. Democracy is measured on a scale from −10 to 10, with −10 being the most autocratic and 10 being the most democratic.Footnote 76 The joint polity measure used in the peace treaties regressions was created by taking the lower of the polity scores in the dyad.Footnote 77
TABLE 1. Logistic regressions on peace treaties, by war-dyad
* p < .10;
** p < .05;
*** p < .01.
Three dummy variables distinguish great power/minor power, minor power/minor power, and great power/great power dyads, with great power/great power dyads being the reference category. Both the list of major powers and start and end dates to measure war duration (in days) were taken from COW.
Annual membership in the UN was obtained from the UN.Footnote 78 This variable codes for whether neither, one, or both belligerents in the dyad were members of the UN.
Finally, I also control for war end year—again, using the COW start and end dates—to address the possibility that a secular trend accounts for the decline in peace treaties. If this variable is negative and significant, and overwhelms the effect of the jus in bello variable(s), one would have to conclude that another cause is driving the decline in the use of formal peace treaties. If, however, the effect of codified jus in bello is robust once time is controlled for, it would be difficult to argue that some other effect of time is at play.
Results
Logistic regressions on peace treaties were conducted on nondirected war-dyads from 1816–2007. The statistical results lend support to the hypotheses regarding the proliferation of the law of war. Model 1 employs the systemic jus in bello variable, and finds a significant effect of the proliferation of codified jus in bello. This is so even when I control for the end year of the war (see Model 5); the coefficient on war end year is positive and not significant, and therefore time cannot account for the declining use of peace treaties in interstate war in this model. By contrast, wars such as the Korean War, which ended in the 1950s, just after the creation of the 1949 Geneva Conventions, were approximately 50 percent less likely to conclude with a peace treaty compared to wars that ended in the 1850s, such as the Crimean War, just as the first law of war was codified.
Model 2 employs the mean ratifications variable as an alternative to the systemic measure of jus in bello, and yields similar results. Belligerents in the 75th percentile for ratification are 60 percent less likely to conclude a peace treaty compared to belligerents in the 25th percentile for ratification. This contrast represents the difference between the Mexican-American War (which ended with a peace treaty) and the first Gulf War, which did not.
The results also indicate that war-dyads in which at least one party has issued a declaration of war are more than 35 percent more likely to conclude peace treaties compared to dyads in which neither party has issued a declaration of war. This result lends support to H1b, and suggests that wars that begin formally are more likely to conclude formally.
Model 4 begins to assess the claim that noncompliance with jus in bello will lessen the likelihood of concluding a peace treaty.Footnote 79 The coefficient on noncompliance with cultural property laws is both negative and statistically significant, suggesting that belligerents that are less compliant with cultural property laws of war are also less likely to conclude peace treaties. This result supports H1c, which suggests violation of those laws of war most subject to punishment appears to be associated with a lower likelihood of concluding a formal peace treaty.
Results for variables associated with alternative and additional hypotheses are more mixed. Perhaps surprisingly, neither military victory nor foreign-imposed regime change appear to be related significantly to peace treaties. Also, contrary to the expectation of bargaining theorists of war, war duration does not appear to be related to the conclusion of a peace treaty. joint polity also fails to attain statistical significance, a result that ought not be particularly surprising, for at least two reasons. First, democratic peace theorists suggest that democracies will be unlikely to fight each other.Footnote 80 Because these theorists do not suggest that the democratic peace is restricted to a particular time period, it is difficult to see how joint democracy could explain the decline in the use of formal peace treaties. Second, if the democratic peace theorists are incorrect, the increased number of democracies in the post-1950 world would suggest that we could see more democracy versus democracy wars, which would lead to an increase in formal peace treaties.
The most important inconsistent finding refers to the un membership variable. In Model 3, the coefficient on UN membership is negative and statistically significant, consistent with H7. But the coefficient on un membership becomes positive and loses statistical significance in Model 4—which covers only the twentieth century—casting doubt on the role of the UN as a depressor of peace treaty use. Because the emergence of the UN is concurrent with a series of other important shifts in international relations that could depress the use of peace treaties, such as the Cold War and the emergence of nuclear weapons, this measure effectively controls for several other variables as well.Footnote 81
Among the control variables, three tend to generate statistically significant coefficients. Great power dyads are significantly more likely to conclude peace treaties compared to minor power/minor power dyads or minor power/great power dyads. Given the decline of great power war that has accompanied the decline of peace treaties, it does seem that this particular change in war is one driver of the decreased use of peace treaties. Not surprisingly, wars that are accompanied by state deaths are more than 50 percent less likely to conclude in formal peace treaties compared to wars that do not end in state deaths. And conflicts that end with territorial exchange are 40 percent more likely to conclude in a peace treaty compared to conflicts without territorial exchange, lending support to the notion that peace treaties can serve as deeds of a kind. Moreover, insofar as a norm against territorial conquest operates in today's international system, it is consistent that we would be less likely to see belligerents codifying their conquests in peace treaties, even if conquest does occur.
Tests for robustness, such as exclusion of the world wars, using the log of duration, and including the highest (rather than the average) level of ratification for a dyad yielded stable results with respect to the coefficients on the jus in bello and mean ratifications variables; the one great power and state death coefficients sometimes lose statistical significance in these specifications, while the coefficients on territorial exchange and military victory are likely to become more statistically significant. Similarly, as reported in Model 6, coding UN Security Council resolutions such as those following the Iran-Iraq War and the 1991 Gulf War as peace treaties also generates stable results. The wars of the 1970s, which see an unusually high number of peace treaties, were also dropped from the analysis as a robustness check; the results were stable.
As an additional set of robustness checks, I changed the unit of analysis to the war rather than the war-dyad.Footnote 82 Given that nearly half of the wars under consideration are multilateral, this approach reduces the likelihood that the largest multilateral wars in particular are driving the results. This alteration in the unit of analysis reduced dramatically the number of observations, but nonetheless yielded a significantly negative coefficient for both the jus in bello and mean ratifications variables. Likewise, territorial change remained a significantly positive predictor of peace treaties. Although declaration of war, noncompliance with cultural property laws, and un membership all took the correct sign, none of these variables generated statistically significant coefficients in this set of tests. To some extent, the disparity in results when the war-dyad versus the war itself is the unit of observation should not be surprising. Many variables (such as whether a declaration of war is issued) are dyad specific and therefore difficult to aggregate to the war level for multilateral wars. Although Morrow and Jo do tend to aggregate their coding of violations of jus in bello for multilateral wars, restricting the data to the twentieth century yields a very small number (forty-six) of observations. For regressions that include this variable, it is surprising to see any significant results. un membership, on the other hand, tends not to vary much across dyads in the same war, casting further doubt on the effect of UN membership on the conclusion of peace treaties.
Among all the explanations for the demise of peace treaties considered here, the claim that the proliferation of codified jus in bello accounts for this trend appears to be most supported by the evidence. Figure 2 illustrates the marginal effect of the proliferation of codified jus in bello on the likelihood that a peace treaty will be concluded. The steeper slope beginning at around five laws of war may be because of the fact that the nature of codified jus in bello sees a dramatic change as this canon of the law of war develops. The earliest codified laws of war were arguably designed to make the prosecution of war easier for belligerents. For example, the 1856 Declaration of Paris governs contraband on neutral ships but offers belligerents great leeway in defining what counted as contraband.Footnote 83 Similarly, the 1864 Geneva Convention on the Treatment of the Wounded helped all sides to a conflict—at little cost—by preventing maltreatment of the wounded and affording protection to medical personnel on the battlefield.
FIGURE 2. Probability of peace treaty as jus in bello increases (includes confidence intervals)
The steepening slope in Figure 2 at around five laws of war reflects a sea change in the character of international humanitarian law. The 1899 and 1907 Hague Conventions on the Laws and Customs of War on Land were the early harbingers of this change. They prohibited attacking undefended areas and contained provisions for treatment of occupied territories, but certainly did not devote much time to considering the role of the civilian in war. This balance shifted more dramatically with the creation of the Fourth 1949 Geneva Convention “Relative to the Protection of Civilian Persons in Time of War.”Footnote 84 Today's belligerents face a much more complicated and costly legal landscape than their nineteenth-century predecessors.Footnote 85
Conclusion
Interstate war has changed in important ways. Among these changes is the demise of peace treaties in interstate war, a trend that is both puzzling and troubling. That a major cause of this trend could be an unintended consequence of the development of the modern law of war should be reason for concern. As legal scholar Blum states, “If we don't admit to being at war, we don't have to consider the position of others regarding how we should be fighting it, or winning it.”Footnote 86 That said, if future research lends additional support to the argument made in this article, the policy implication is not necessarily that the law of war should be rolled back. Even though peace treaties are on the decline in interstate wars, they appear to be increasingly common in the conclusion of civil wars. Of 151 post-1945 civil wars identified by Doyle and Sambanis,Footnote 87 for example, approximately one-third were accompanied by peace treaties. Doyle and Sambanis identify two 1950s peace treaties, one in the 1960s, six in the 1970s, four in the 1980s, and thirty in the 1990s. Their data exhibit both a general upward trend over the entire period and a dramatic increase of peace treaties in civil wars in the 1990s. While the number of civil wars begun in the 1990s is also much higher than in previous decades (and the number of wars ended in a given decade roughly tracks with the number of wars begun in the decade), there is a fifteen-fold increase in the use of civil war peace treaties from 1950 to 2000, but only a three-fold increase in the outbreak of civil war in the 1990s compared to the 1950s. To some degree, this gap may be because of the fact that, historically, most civil wars have been total wars, with one side eliminated entirely, thereby also eliminating any need for a peace treaty. More recently, as civil wars have become increasingly secessionist, peace treaties have become valuable for rebel groups seeking recognition and legitimacy. These same rebel groups may wish to convey the impression that they will be good citizens of the international community by broadcasting their adherence to jus ad bellum in their declarations of independence,Footnote 88 and to jus in bello in their invitations to the ICRC to monitor their conduct during warfare.Footnote 89
The contrast between the declining use of peace treaties in interstate war and their increasingly frequent use in civil wars has important implications for those who argue that war itself has undergone a fundamental transformation. For those who believe war has simply shifted from the interstate to the intrastate arena, the declining use of peace treaties in interstate war is still puzzling but should not be of too much concern. For those who believe all war is on the decline, the declining use of peace treaties in any realm of warfare should be concerning, as peace treaties may decrease the likelihood of war's recurrence. But for those who are skeptical that war has changed fundamentally, the decline of peace treaties points to a different, perhaps more insidious, finding: war may have changed in name only. Ironically, efforts to regulate the horrors of war may have made war's end more difficult to identify and, in so doing, undermined the work of those trying to moderate its effects and prevent its occurrence.
