CHAPTER THREE

Credible Commitments and War Termination

Promises like pie crusts are leaven to be broken.

—Vladimir Lenin

THERE IS NO WORLD GOVERNMENT to enforce laws, treaties, or promises. Conspiracy theories notwithstanding, there is no fleet of black helicopters or blue-helmeted stormtroopers at the ready to carry out the rulings of institutions like the United Nations (UN) or the World Trade Organization (WTO). Without such support, state-to-state promises can be as fragile as Lenin’s pie crusts. This absence of government, literally anarchy, has severe and often tragic consequences, as it makes the maintenance of international order and the prevention of crimes against humanity such as genocide much more difficult. More generally, anarchy is perhaps the single most important factor that distinguishes global politics from national politics, as only in the latter are there governments that use monopolies of force to maintain law and order.

Scholars in the realist theoretical tradition have used the anarchy insight to propose that the inability of states to make enforceable commitments is an important cause of war. Peace and international order rest, according to realists, on a stable balance of power. If the balance of power is changing, declining states may fear that rising states will eventually demand changes in the international order reflecting the rising states’ emerging strength. Declining states would like some commitment from the rising states not to demand concessions or attack as they grow in power, but because of anarchy such guarantees would not be enforceable and hence not credible. Declining states may become motivated to attack, thinking that war now would be preferable to concessions or war later. This logic was behind the initiation of several wars, including the German attack on Russia in 1914, the Japanese attack on Pearl Harbor in 1941, the Israeli attack on Egypt in 1956, and the American attack on Iraq in 2003.

Although the credible commitment insight has long been applied to international relations theory and to the causes of conflict, it has almost never been applied to war termination, and to interstate war termination in particular.¹ Specifically, although classics in the post-1945 realist canon meticulously draw out the implications of the information and commitment insights for the causes of war, they say almost nothing about applying these insights to the termination of wars.² This chapter takes this very conventional realist insight about commitment credibility and applies it to the area of war termination.³ Doing so completes the development of the war-termination theory begun in the previous chapter, demonstrating that a comprehensive understanding of war termination requires incorporating the dimensions of both information and commitments. Information and commitment dynamics sometimes work in contrast, but can also work in complement. Chapter 2 demonstrated that information dynamics often shape a belligerent’s expectations about future costs, specifically, the costs of continuing to fight and pursue victory. This chapter will demonstrate that commitment dynamics are often about shaping a belligerent’s expectations regarding future benefits, specifically that commitment problems can serve to make some war-termination settlements more valuable than others.

ENFORCING WAR-ENDING BARGAINS

The theory employed in the bargaining model of war was imported from economics. Economists are most interested in understanding phenomena such as how buyers and sellers set prices and how labor and management settle on long-term wage contracts. However, economic actors (especially those interacting within a mature national economy) enjoy one important advantage over states: they know their contracts will be enforced. If a party violates the terms of a business or labor contract, the other party can resort to the legal system to enforce the terms of the contract, receive financial compensation, and even have the violator jailed. Most of the bargaining literature in economics assumes compliance with bargains reached.⁴ Much of the bargaining model of war literature imports this enforcement assumption, that the terms of any war-ending bargain will enjoy automatic compliance.⁵

This assumption of automatic compliance, as applied to international relations, deserves further examination. A war-ending settlement is of course an international agreement, as the belligerents agree to stop fighting and perhaps to reallocate the disputed good. However, there is no world government to enforce the war-ending contract. Different international relations theories offer varying perspectives on the likelihood of state compliance with international agreements, but critically all theories agree that the degree of compliance varies, and must not be assumed. Realism famously proposes that compliance with international agreements is likely to be rare, as lack of enforcement, fear of adversarial defection, and relative gains concerns all push states to ignore such agreements. Realism proposes that the only agreements that enjoy compliance are those that are trivial, either calling for cooperation on peripheral issues, or demanding behavior that states would have engaged in even absent an agreement.⁶ Institutionalism is more optimistic, forecasting that the presence of appropriately designed international institutions can facilitate interstate cooperation.⁷ Constructivism proposes that cooperation is most likely when the states share a collective identity, and/or possess otherinterest as well as self-interest.⁸ But centrally, realism, institutionalism, and constructivism all agree that state compliance with international agreements is not automatic.

Falsely assuming that compliance with international agreements is automatic can be dangerous, as the possibility of non-compliance affects both the existence and character of international agreements. States are less likely to sign an agreement that is unlikely to enjoy satisfactory compliance. For example, American suspicion about possible Soviet cheating on arms control agreements kept progress in that area slow, and was the basis of President Ronald Reagan’s famous declaration, “Trust, but verify.” Neglecting how decision makers think about the risks of non-compliance warps our understanding of how treaties are crafted.⁹

Assuming automatic compliance with war-ending agreements is especially suspect. Clausewitz recognized that states break peace treaties: “Lastly, even the ultimate outcome of a war is not always to be regarded as final. The defeated state often considers the outcome merely as a transitory evil, for which a remedy may still be found in political conditions at some later date.”¹⁰ It is not difficult to imagine why states would break ceasefire agreements.¹¹ The benefits to breaking such agreements can be great since restarting a war can offer the possibility of capturing even more of the valuable disputed good, or recapturing whatever was lost in the last war. Egypt and Syria attacked Israel in 1973 in the hopes of recapturing territories they had each lost in the 1967 Six-Day War. World War I gave France the chance to recapture the provinces of Alsace and Lorraine, lost to Germany in the Franco-Prussian War, and the 1940 German attack on France gave Germany the chance to get Alsace and Lorraine back after France reacquired these lands in the Versailles agreement that ended World War I. Conversely, there are dangers if one side fears the other may unilaterally defect to capture more of the disputed good. Israel preemptively attacked Egypt in 1967, fearing Egyptian defection on the cease fire agreements ending the 1948–49 and 1956 wars. Perhaps not surprisingly, the rate of compliance with war-ending agreements is far from perfect. Over the 1914–2001 period, nearly one third of all interstate war ceasefires (56 out of 188) eventually broke down into renewed war.¹²

This is of course a grim picture, that the international order is a jungle, and states can never fully trust each other. Some might reply that trust can emerge in international relations, that states can come to trust each other and cooperate.¹³ However, belligerents consumed by the heat of war are especially unlikely to trust each other. Two states cannot build much trust while fighting. Making matters worse, there is likely a history of broken trust between two belligerents, as an attacker often breaks a neutrality or border agreement when it decides to launch a war.

TOTAL VICTORY, TOTAL PEACE: ABSOLUTE WAR AS A SOLUTION TO CREDIBLE COMMITMENT PROBLEMS

If a wartime belligerent fears its adversary might break a war-ending agreement and reattack after war ends, one possible solution to this noncompliance problem is the imposition of an absolute war outcome. Rivals sometimes continue to break ceasefire agreements and attack each other. This cycle can end when one side suffers total military defeat, enabling the victor to impose terms that would directly prevent the adversary from breaking the peace treaty and reattacking in the future.¹⁴ Clausewitz summarized this logic at its most brutal: “So long as I have not overthrown my opponent, I am bound to fear he may overthrow me.”¹⁵

An absolute war outcome can, perhaps paradoxically, take many forms. The most severe peace terms would be the annihilation of the adversary’s population. This outcome is thankfully mostly theoretical, with the possible exception of Athens’ slaughter of all of the Melian men and enslavement of its women and children during the Peloponnesian Wars.¹⁶ Lesser versions of the annihilationist strategy have been considered and sometimes put into action. Half of Paraguay’s total population and most of its adult male population were killed in the War of the Triple Alliance in the 1860s, although within less than a century Paraguay had recovered sufficiently to fight another war, although with different adversaries.¹⁷ World War II saw several examples of annihilationist behavior, including Hitler’s elimination of twelve million Jews and other minorities as imagined threats to the Nazi state, and the 1940 Soviet execution of nearly 22,000 Polish military officers and policemen in the infamous Katyn massacre to eliminate possible future threats to Soviet power.¹⁸ More generally, combatants sometimes engage in annihilationist strategies during wartime to defeat guerrilla forces.¹⁹

Extinguishing the adversary’s sovereignty and annexing its territory is a more common though still severe solution to the commitment problem. This approach means dismantling the adversary’s government and military, thereby eliminating its ability to launch a new war. Some have referred to such an outcome as “state death.”²⁰ After such an action, adversarial defection would be possible only after the recovery of sovereignty, which in turn might occur through an internal regime change causing the conquering government to disgorge its conquests (such as the Baltic states exiting the Soviet Union following the collapse of Communism), secession by the conquered state (such as Ireland’s successful twentieth-century fight for independence, throwing off more than eight centuries of British rule), or rescue by the military action of a third party (such as the Allied liberation of France in 1944 or the UN liberation of Kuwait in 1991).

A third means by which a war outcome can prevent defection is through foreign imposed regime change. A belligerent may impose a puppet regime, install democratic institutions, and/or hardwire pacifism into a nation’s laws and constitution as a means of preventing postwar defection.²¹ Foreign imposed regime change cripples a state’s foreign policy sovereignty and renders compliance with war-ending agreements far more likely by changing the target state’s preferences for war and its political institutions. Regarding preferences, the architect of a foreign-imposed regime change executes, exiles, imprisons, or at least permanently removes from power militarist leaders and their subordinates, dismantles pro-war industrial complexes, and empowers or imports leaders with more compliant and pacific preferences. The Soviet Union eliminated nonCommunist Hungarian leaderships in 1945 and 1956, replacing them with more pliable Communists each time. The Allies held war crimes trials in Germany and Japan after World War II, purging these countries of militarists.²² Saddam Hussein was thrown from power and captured in 2003, and then executed in 2006.

Foreign-imposed regime change also means overhauling the political institutions of the target country. Some democratic belligerents may see the installation of democratic governance as a promising means of preventing an adversary’s future aggression. The proposition that democracy is a cure for war and militarism has recurred in American foreign policy over the past century. In his April 2, 1917 speech asking for a congressional declaration of war on the Central Powers, Woodrow Wilson proposed that the taproot cause of the war in Europe was Prussian militarism, and that peace required the democratic transformation of Germany. He famously declared that, “The world must be made safe for democracy. Its peace must be planted upon the tested foundations of political liberty.”²³ During World War II, Roosevelt and then Truman saw Japanese, Italian, and German authoritarianism as intertwined with militarism, and that democratization of these societies would cure them of their warlike tendencies. The 2001 war in Afghanistan was intended to rid that country of the anti-American Taliban, and democracy was seen as a long-term inoculation against the Taliban’s return. The goal of the 2003 Iraq War was the overthrow of Saddam Hussein, and the Bush administration saw the installation of democracy there as the best way to prevent Iraq from reemerging as a threat. As Bush remarked in a critical October 2002 speech, “regime change in Iraq is the only certain means of removing a great danger to our nation.”²⁴ Of course, regime change need not culminate in democratization, as governments sometimes install nondemocratic puppet regimes. After conquering France in 1940, Nazi Germany installed the undemocratic Vichy regime. After World War II ended, the Soviet Union helped put in place a number of friendly Communist regimes, both in former foes (such as [East] Germany, Hungary, and Romania) and former victims of Nazi aggression (such as Poland and Czechoslovakia).

As part of a foreign-imposed regime change, a target is also often forced to hardwire pacifism—resulting in restrictions on the size, deployment, and/or weaponry of the national military—into its new political institutions. One severe (although unimplemented) proposal was Secretary of the Treasury Henry Morgenthau’s 1944 plan for Germany after World War II. Morgenthau recommended that all Germany be deindustrialized as well as demilitarized, with the intent of removing the very foundations of German military power.²⁵ Morgenthau’s proposal was rejected, but Germany was forced to accept constraints on the use of its armed forces, as expressed in the 1949 German Basic Law.²⁶ American occupation forces also wrote the 1946 Japanese constitution, which declares in Article Nine: “Aspiring sincerely to an international peace based on justice and order, the Japanese people forever renounce war as a sovereign right of the nation and the threat or use of force as means of settling international disputes. . . . In order to accomplish the aim of the preceding paragraph, land, sea, and air forces, as well as other war potential, will never be maintained. The right of belligerency of the state will not be recognized.”²⁷ Article 11 of the 1947 Italian constitution “repudiates war as an instrument offending the liberty of the peoples and as a means for settling international disputes.”²⁸ Afghanistan’s post-Taliban 2004 constitution explicitly declares that, “The state prevents all types of terrorist activities.”²⁹ Iraq’s 2005 constitution also commits it to fighting terrorism, and not acquiring weapons of mass destruction.³⁰ These institutional changes can affect societal preferences by eradicating militarist culture, through means such as altering the educational system.³¹

Each of these outcomes—annihilation, annexation, and foreign imposed regime change—solves or at least substantially ameliorates the peace settlement compliance problem by de jure or de facto revoking of the adversary’s sovereignty, thereby directly preventing the defeated state from violating a war-ending agreement. I classify these war outcomes as absolute. As to whether this definition overlaps exactly with Clausewitz’s definition of absolute war, Clausewitz himself is a bit imprecise on this point. He talks about forcing the defeated state “to do your bidding” as descriptive of an absolute outcome, and that absolute war means to render the defeated state “politically helpless or militarily impotent, thus forcing him to sign whatever peace we please.” He also describes absolute war as “completely governed and saturated by the urge for a decision.”³²

Some may prefer a narrower definition of an absolute war outcome, that it represents zero concessions by the victorious side, and/or the utter annihilation of the defeated state. The narrow definition would arguably offer some advantages. Most generally, it would accurately demonstrate that essentially every war outcome involves choice by both the victor and the defeated. On the victor’s side, in nearly all wars the victor makes at least some concessions. For example, even as part of the “unconditional” surrender of Japan in 1945, the Allies allowed that the Emperor could remain as the spiritual head of the Japanese government. On the side of the defeated state, in nearly all wars the loser must decide to stop fighting. Even those suffering state death or foreign-imposed regime change have the opportunity to continue resistance at some level, perhaps in post defeat guerrilla warfare. Such insurgencies have occurred. Violent acts of resistance arose in Nazi-occupied Europe after official government surrenders to German forces. Although Nazi Germany surrendered formally and unconditionally to the Allies in May 1945, there persisted after the surrender some guerrilla activity of the Nazi Werwolf underground against Allied occupation forces. The underground was not a completely renegade movement, and had received support and direction from the Nazi government prior to the surrender.³³ The Japanese government in August 1945 made deliberate efforts to ensure that members of the military complied with the terms of the surrender. The government required high-level military leaders to sign an agreement of compliance with the surrender terms, and to disarm and defuel aircraft to prevent their use for kamikaze attacks. These efforts were not completely successful, as there was a coup attempt before the surrender terms were announced to the Japanese public, and a navy captain did independently organize a postsurrender insurrection lasting several days calling for defiance of the surrender terms.³⁴ More recently, although American conventional forces achieved foreign-imposed regime change in Afghanistan in 2001 and Iraq in 2003, in both states enduring and extensive armed resistance against American forces in the form of irregular insurgencies has continued.

Again, some might claim that the presence either of any victor’s concessions or of any post-surrender violence is sufficient to disqualify a war outcome from being absolute. Under these conditions, absolute outcomes have essentially never occurred, and become purely theoretical. However, the utter non-existence of absolute wars is a position to which even Clausewitz, who proposed that absolute wars are uncommon, would disagree. Clausewitz thought of Napoleon’s wars as absolute. He also allowed that conditions sometimes demand absolute war, such that a leader pursues absolute war “when he can or when he must.”³⁵

In this book, absolute war means removing the defeated state’s ability to organize resistance. As noted, in practice this can take the form of annihilation, but more commonly means state death, occupation, or foreign-imposed regime change. This definition is consistent with Clausewitz’s conception of absolute war, summarized earlier. This definition matches the heart of the commitment component of the theory described later in this chapter, as a war outcome that essentially removes the possibility of the defeated state reneging on a war-ending settlement.

Some may object that because this broad definition would mean that some war outcomes would be inaccurately described as including no victors’ concessions, it would underestimate the role played by information dynamics in determining war termination and war outcomes. However, it is not necessarily the case that the termination of a particular war was dominated by information dynamics if the victor enjoyed a decisive battlefield victory and still offered tiny concessions. Often, if a victor achieves decisive victory on the battlefield and intends to annex its adversary or impose a new regime, it may make concessions in relatively trivial areas in order to facilitate postwar stability. That is, the concessions do not detract from the victor’s ability to accomplish its aims of permanently removing the (interstate) threat posed by its adversary, and in fact may further this goal by providing for greater political stability within the nation of the defeated. For example, under the terms of surrender ending the American Civil War, the Confederacy as a political entity was extinguished but Confederate soldiers and officers were allowed to keep their horses, permitting them to plant crops more easily and feed their families, thereby reducing the likelihood of widespread famine in the southern states in the coming year. Lincoln and his leading commanders recognized that allowing Confederate soldiers to return to their farms with their horses and even their guns would help move the country towards economic and social normalcy, and would improve the chances that these exrebels would accept the Union and its laws as their own.³⁶

The case of World War II Japan also demonstrates this point. The historical argument made by some is that because the Allies conceded to Japan’s summer 1945 demand that the Emperor not be tried as a war criminal, the terms of Japan’s August 1945 surrender constitute a limited rather than absolute outcome. However, the concession over the Emperor was allowed in part because the United States (correctly) decided that doing so would increase the likelihood of its acceptance by the Japanese military.³⁷ Further, the Emperor concession itself was extremely limited, and merely implied by the language that “the ultimate form of government in Japan shall . . . be established by the freely expressed will of the Japanese people.” It was severely curtailed by the additional statement that “From the moment of surrender the authority of the emperor and the Japanese Government to rule the state shall be subject to the Supreme Commander of the Allied Powers.”³⁸ Certainly, the concession did not in intention or effect prevent the United States from accomplishing its central aims of demilitarizing Japan and completely refashioning the Japanese political landscape. In implementation, the Emperor became a politically irrelevant and powerless figure in the postwar period.

A limitation of this definition is that it may code wars as absolute even when violent insurgency emerged after the formal surrender. This is especially important as a policy issue since from a belligerent’s perspective sometimes the benefit of breaking an adversary’s ability to launch state-to-state violence is outweighed by the cost of creating an enduring insurgency. Many have made this point in the context of the 2003 Iraq War, that the interstate threat from Saddam Hussein’s Iraq could have been contained without war through deterrence, but the postwar insurgency has been gigantically costly in human, financial, and geopolitical terms.³⁹

Assessing the causes of postwar insurgency is an important question, but such a task is beyond the scope of this book. More relevant to this work are the causes of postwar interstate violence, which are mostly different from the causes of postwar insurgency. Regarding postwar interstate violence, the assumption is that annihilation, state death, or foreign imposed regime change virtually eliminate the possibility that the defeated state will launch interstate violence against the victorious state. The empirical basis of this assumption is discussed in further detail later in these pages. Regarding postwar insurgency, several factors affect the likelihood of insurgency. A general point is that the “actor” that would decide to launch an insurgency—the population at large or some subgroup(s) of the population—is often different from the actor that would decide to launch an interstate war: the government. So, the victorious state is often not bargaining directly with the potential insurgent, meaning there is no opportunity for the potential insurgent to commit to not launching a postwar insurgency. The victorious state can attempt to offer concessions to the defeated state in an attempt to encourage the defeated state to discourage postwar insurgency, as was the case with postwar Japan. However, several other factors are perhaps more important in understanding the causes of postwar insurgencies. Sometimes leaders, such as Confederate General Robert E. Lee at the close of the American Civil War, reject postwar insurgency not because of some concession offered by the victory, but simply because insurgency would be pointlessly destructive.⁴⁰ Structural factors, such as the geographic terrain in the country, and postwar occupation policy choices are likely to be critical. Studies have found that “work or starve” threats are effective in managing populations in occupied areas. The provision of domestic security is critical. Perhaps the single biggest mistake of the post-2003 Iraq War, for example, was the early dismissal of the Iraqi Army.⁴¹ That being said, the prospect of fighting a costly counterinsurgency campaign in the wake of the absolute defeat of a state’s conventional forces may be enough to dissuade a state from pursuing absolute victory (see the following).

In sum, the ability of an absolute war outcome such as annihilation, occupation, annexation, or foreign-imposed regime change to eliminate the compliance problem offers real temptations to belligerents who fear the other side will not adhere to its war-ending commitments. A belligerent may worry that any war-ending agreement that leaves intact the adversary’s sovereignty may allow the adversary to reinitiate war, and that this problem can be solved by achieving absolute victory and eliminating the other side’s sovereignty. As Immanuel Kant noted, “For some confidence in the character of the enemy must remain even in the midst of war, as otherwise no peace could be concluded and the hostilities would degenerate into a war of extermination.”⁴² The central hypothesis emerging from the credible commitment logic is that the more a belligerent fears its adversary will violate war-ending commitments, the more likely that belligerent will be to pursue absolute victory. This is the “commitment proposition.” When compliance fears dominate, belligerents may doggedly pursue an absolute war outcome, not seeking a limited war outcome to divide the disputed good and leave the other side’s sovereignty intact. The logic is that a belligerent is relatively uninterested in a limited settlement if the adversary is unlikely to abide by its terms. In comparison, achieving an absolute war outcome would solve the compliance problem by eliminating the adversary’s sovereignty, so the belligerent becomes motivated to ignore the possibility of a limited settlement and pursue an absolute outcome.

One way to frame the commitment proposition is to understand that it affects the expectations about the benefits of continuing to fight: How much utility or welfare does a belligerent gain for each added increment of the good it might obtain in a possible war-ending settlement? The relationship of increments of the good to increments of utility can be illustrated graphically, in which the x axis represents increments of the good obtained in the postwar settlement, and the y axis represents increments of overall utility. A common assumption is that the shape of the benefits curve is linear since each increment of the disputed good is as valuable as every other increment. However, if a belligerent perceives that the adversary will defect on a war-ending agreement and reattack, then this means the belligerent will get much less welfare out of any outcome short of absolute victory, because the belligerent will not be able to enjoy the subabsolute allocation of the good for long before war resumes. However, if the belligerent is able to achieve absolute victory, it will be able to consume the good without fear of losing its gains in a new war. Hence, under such conditions the shape of the benefits curve should be relatively flat for most increments of the good, meaning that acquiring additional increments of the good provides little extra utility, but that the curve should jump up for the possession of all of the good, as possession of all of the good reflects absolute victory and the solution of the postwar commitment problem. This step function relationship of good increments to utility under conditions of fears about postwar non-compliance is represented in figure 3.1.

Copyrighted image removed by Publisher

The step function in figure 3.1 introduces the possibility of divergence between the predictions of the information and commitment propositions. A belligerent that is both losing battles and fears postwar commitment problems may elect not to make concessions to the adversary in the hopes of achieving a limited war outcome, as the information proposition forecasts. A limited outcome deal in which a belligerent gets part of the good may hold relative value to the belligerent, because if the adversary breaks the deal and reattacks, the belligerent may not be able to consume the portion of the good it received under the limited outcome settlement. Hence, the belligerent gathers few benefits from stopping the fighting to reach a limited outcome settlement. The belligerent instead may fight on in the hopes of achieving an absolute victory, which would solve the postwar commitment problem, as the commitment proposition would predict. Achieving an absolute victory would permit the unthreatened consumption of the entire good because of the solution of the commitment problem, a prize that may encourage belligerents to fight on, sometimes even in the face of discouraging combat outcomes. Formal theoretical analysis confirms that the more severe the credible commitment fears, the less likely that combat outcomes are to affect war-termination behavior and discourage an adversary from pursuing absolute victory. Under some conditions of severe credible commitment fears, a belligerent may make an initial, moderate offer, and if the offer is rejected and the belligerent loses a battle, the belligerent may actually raise its demands (contra the information proposition), viewing the rejection of the first offer as evidence of the adversary’s true power, and so encouraging the pursuit of absolute victory.⁴³ In sum, when commitment concerns are severe, discouraging information coming from the battlefield may not cause a belligerent to abandon its pursuit of absolute victory.

This process of commitment fears attenuating the effects of information revelation is illustrated by French war-termination behavior during the 1870–71 Franco-Prussian War. The war began in July 1870, after Otto von Bismarck of Prussia maneuvered France into war over a diplomatic slight. The war proceeded well for Prussia, with a string of important victories at Metz, Sedan, and elsewhere. As the information proposition forecasts, Prussia began to raise its war aims as it achieved early victories, especially towards seeking territorial gains from France.⁴⁴By early December, a Prussian victory in arms seemed at hand. However, contrary to the expectations of the information hypothesis, the French refused to budge on making the territorial concessions that Prussia demanded. The historian Michael Howard commented in his classic study of the war, “By all the normal customs of warfare observed by the regular armies and the traditional statesmen of Europe, the defeats suffered by the French armies between 30^th November and 5^th December should have made the [French] Government of National Defence sue for peace; and it was assumed by the Germans and by Europe that they would do so. . . . In military logic there now seemed no prospect of defeating the Germans, nor was there any reason to suppose that a prolongation of the war would secure a more favorable peace. The struggle was kept alive only by the will and the energy of a few men at the centre of power, who inflexibly refused to admit defeat.”⁴⁵

Part of the French motivation in rejecting a negotiated peace was the desire to inflict absolute defeat on Prussia in order to solve the enduring problem of incredible Prussian commitments, and thereby achieve a solution to the Prussian threat that did not rely on an incredible Prussian commitment not to attack. French Minister of War Leon Gambetta wrote in a private letter in early January 1871, “The whole country understands and wants a war to the end, without mercy, even after the fall of Paris. . . . The simplest clearly understand that since the war has become a war of extermination covertly prepared by Prussia for thirty years past, we must, for the honour of France and for our security in the future, finish for good this odious power. . . . We shall prolong the struggle to extermination.”⁴⁶

THE EMPIRICAL RECORD: THE FREQUENCY AND EFFECTIVENESS OF ABSOLUTE WAR OUTCOMES

Some scholars declare the frequency of absolute war outcomes to be very low, interpreting this as evidence of the centrality of information in the process of war termination, as combat and intrawar diplomacy usually eventually provide enough information to permit war termination short of an absolute outcome.⁴⁷ This assumption deserves closer examination. To discern the frequency of absolute war outcomes, I first combine two datasets on war outcomes.⁴⁸ The first dataset is a list of all state deaths from 1816–1992.⁴⁹ A total of thirty-five violent state deaths occurred during this time period, some (but not all) of which occurred in the context of interstate wars. A second dataset is of all foreign-imposed regime changes occurring at the close of interstate wars from 1816–1980. A total of twenty-six such events occurred during this period, although some wars (such as World War II) experienced more than one foreign-imposed regime change (note that most foreign-imposed regime changes are not violent state deaths). As noted, for the purposes of this book, both state death and foreign-imposed regime change constitute absolute war outcomes.

There is no universally accepted list of wars or war participants, although most recognize that wars are generally violent conflicts between states that result in at least 1,000 combat deaths. Scholars differ over, for example, whether and how complex multilateral wars like World War II should be broken up into several wars. The Correlates of War (COW) project provides a commonly used dataset on interstate wars.⁵⁰ It lists seventy-nine interstate wars from 1816–1997, and within this set there are eighteen wars (23 percent) in which at least one belligerent experienced state death and/or foreign-imposed regime change. Allan Stam and I made some modest changes to the COW dataset, breaking up some large multilateral wars into component conflicts, and updating the list to 2001.⁵¹ In our dataset of ninety-eight wars, there are twenty-five absolute war outcomes (26 percent). Beyond these groupings, in some wars that ended in limited outcomes at least one belligerent sought at least at one point during the war the absolute defeat of its adversary. Such wars include the Korean War, the Iran–Iraq War, the Winter War, the Finnish– Soviet Continuation War, the War of Israeli Independence, and others.

Overall, absolute war outcomes have occurred in roughly one quarter of interstate wars. This is certainly a higher frequency than the extreme rarity predicted by the view that the information provided by combat is sufficient for efficient war termination. It better fits the perspective of the theory presented here, that sometimes information dynamics dominate, allowing wars to end prior to an absolute outcome, and sometimes states pursue absolute victory, perhaps seeking lasting solutions to credible commitment problems.

Knowing that absolute war outcomes occur with minor to moderate frequency, the next question is, do such outcomes in fact prevent states from violating their war-ending commitments not to reinitiate war? Here, the historical record is quite clear, that absolute war outcomes very significantly decrease the likelihood that war will reerupt between two states. Since 1816, when a war has ended in state death, the defeated state almost never itself comes back to life to attack the victor, except in the circumstance when it is “rescued” by a third party, as Britain and the United States rescued European states “killed” by Nazi Germany in World War II, and the UN Coalition rescued Kuwait from Iraqi annexation in 1991. But, these episodes are not examples of noncompliance by the dead state, but rather of third-party action.

Foreign-imposed regime change is also a very effective means of reducing the chances of war reerupting between two former belligerents. Systematic studies examining conflicts from 1816 to 2001 have produced strong support to this effect, as there are virtually no examples of a state suffering foreign-imposed regime change and then after war’s end attacking its former adversary. Generally, war is very significantly less likely to reerupt between former belligerents when it ends with foreign-imposed regime change compared to when war does not end in foreign-imposed regime change.⁵² Indeed, perhaps the greatest foreign policy success of the twentieth century, the transformation of the Nazi German, Fascist Italian, and Imperial Japanese war machines into prosperous, pacific, internationally engaged democracies is largely attributable to foreign-imposed regime change.

INFORMATION AND COMMITMENT FACTORS IN INTERACTION: WHEN DO WARS REMAIN LIMITED?

If former belligerents at least sometimes break war-ending agreements, states cannot effectively bind themselves to comply with war-ending agreements, and absolute war outcomes promise to reduce very substantially the likelihood of noncompliance, then why don’t states always pursue absolute war outcomes? The information proposition offers one answer to this question. Recall from the previous chapter that the information approach sees war as breaking out because of disagreement over the balance of power. War ends when disagreement is reduced sufficiently such that bargaining space between the two sides opens, and the warending settlement reflects the new understanding the two sides have about the true balance of power between them. That is, if the battles fought during the war reveal that one side is stronger than was thought before war began, then the war-ending settlement will provide more of the disputed good to that stronger side then it possessed before the war began.⁵³ For example, the U.S.–Mexico War of 1846–48 clearly demonstrated American military superiority, and common knowledge of this superiority dissuaded Mexico from reattacking after war’s end to reacquire the substantial territories it had lost in the war-ending Treaty of Guadalupe Hidalgo. When both sides agree on the military balance and that the distribution of the good reflects the military balance, the warending agreement is a self-enforcing equilibrium. Both sides agree on the balance of power, and that if the war restarted, it would end with the same outcome. As a result, neither side has an incentive to break the commitment, pay the sunk costs of fighting, and not end up receiving more of the disputed good.

However, this self-enforcing peace equilibrium may break down if after the war one side comes to believe it can reinitiate the war with better chances of acquiring more of the disputed good. Three sets of factors might cause a belligerent to think it might enjoy better chances of winning.⁵⁴ First, there might be a postwar shift in the aggregate balance of power. Most simply, there might be a change in the balance of military power between the two former belligerents, if one belligerent enjoys faster demographic or economic growth than the other, decides to devote a greater fraction of its economy to defense spending than the other, or acquires some new military technology or strategy that enhances its military power. The core theoretical insight is that a rising state cannot commit to adhere to a goods-dividing agreement with the declining state, and such circumstances can make war more likely, even when all agree on the balance of power and its rate of change.⁵⁵ As noted at the outset of this chapter, when the balance of power is changing, the state growing stronger may go to war to expand its share of the disputed good. The acquisition of advanced Soviet SAMs encouraged Egypt to attack Israel in 1973 in the hopes of recapturing the Sinai desert, breaking its 1967 ceasefire agreement with Israel. Or, the state facing relative decline may launch a war while conditions are more favorable, fearing that as the adversary grows stronger it will demand more or even launch a new attack. As noted, a number of wars have been started by fearful states in relative decline. Systematic studies exploring whether postwar changes in the balance of power make the reeruption of war more likely have produced mixed results.⁵⁶

Second, even if the aggregate balance of power is stable, one side may conclude at some point that it will enjoy advantages from attacking first, temporarily tipping the balance in its favor, enough to allow it to make important military gains. The primary source of such a faith in surprise attack would be that after peace is concluded, the adversary might demobilize or redeploy its forces, creating a window of opportunity allowing an attacker to remobilize its own forces secretly and enjoy a passing military advantage as the target slowly remobilized. International crises are more likely to escalate to war when one side mobilizes its forces secretly, perhaps because the secret mobilizer is genuinely bent on war, and is trying to create a temporary military advantage.⁵⁷ Differences in mobilization speeds may create windows of opportunity, tempting one side to strike first. However, the evidence that (perceived) first strike advantages make war more likely is mixed.⁵⁸

Third, war might reerupt if a new leadership came to power in one country, with either a higher or lower resolve for fighting. If a higher resolve leadership came to power, that country might be willing to restart the war, hoping to make greater gains because of its greater willingness to bear costs. If a lower resolve leadership came to power, that country might be attacked, as the other side might seek to exploit the new low resolve leadership and extract greater concessions.⁵⁹ Perhaps most famously, Germany became much less willing to accept the settlement ending World War I, even at the risk of war with Britain and France, after Adolf Hitler and the Nazi party took power in 1933. If a regime change in a former belligerent occurs through revolution, renewed conflict may be especially likely, as revolutionary regimes such as fundamentalist Iran in the late 1970s and France in the 1790s are especially likely to launch wars of liberation and to attract attacks from fearful neighbors.⁶⁰Notably, though, postwar regime change can have the opposite effect and improve relations if a new, less hawkish leader takes power. Anglo-Argentinian relations improved dramatically when the ruling Argentinian junta was thrown from power after the 1982 Falklands War. More generally, enduring rivalries between states are significantly more likely to end when there is internal political change within one of the adversaries.⁶¹

Any of these three developments, change in the aggregate balance of power, the appearance of a first-strike advantage, or change in leadership, might be sufficient to cause a postwar peace to break down. The key issue is whether states fighting wars forecast that any of these developments might occur after war’s end. Unfortunately, states can never be certain that such future changes in power or leadership will not happen. Adverse shifts in the balance of power are always possible, the enemy may develop new and supremely effective weaponry first, and/or the enemy may discover some way to carry out an effective surprise attack. An adversary’s leadership may be replaced by more hawkish individuals, or one’s own successors may be more dovish, inviting challenge. Therefore, states are to some degree always fearful that adversaries may violate war-ending agreements. Further, the very fact that war has broken out between these states helps create suspicions between them. Given these problems, the question then becomes, why do not belligerents always pursue absolute war outcomes as a means of solving the compliance problem? Why do we ever observe limited war outcomes? Addressing these questions is critical since the answers to them in turn constitute hypotheses as to when we should observe states accepting limited outcomes, and when states pursue absolute outcomes. These questions have four answers.⁶²

Vanished Hopes for Victory

Rejecting negotiations for a limited war outcome to fight on for better terms and perhaps an absolute war outcome presumes there is some possibility, however distant, of eventually prevailing. A belligerent will ignore bad news from the front and fight on only if it has some hope of turning the tide in the future. It may be willing to fight on even with only a slim chance of eventual victory, rather than accept war termination now and an unstable peace settlement. It may be encouraged to fight on if it thinks its probability of victory may rise in the future, since perhaps in the future the balance of power will shift when the belligerent fully mobilizes its economy, if it can deploy a new wonder-weapon against the adversary, or if a third party intervenes on its behalf.

Conversely, a state with essentially no hope of turning the tide faces a different calculus. Assuming that the costs of continuing the war are nontrivial (see the following), a belligerent may be willing to accept defeat now, even if it means a limited war settlement susceptible to being violated or even an absolute defeat, rather than fight on, suffer costs, and likely suffer the same fate. Therefore, a belligerent, even if it doubts the credibility of an adversary’s commitment to a war-ending settlement, becomes more likely to lower its war aims and seek war termination as its chances for improving its military prospects in the future approach zero. This hypothesis has been derived formally.⁶³

The information environment may or may not affect whether a belligerent reaches this conclusion. A belligerent will resist the grim conclusion that things are utterly hopeless, even in the face of combat setbacks, if its long-term strategy for victory has not been discredited. Combat setbacks may not discredit a belligerent’s long-term strategy for victory if the combat setbacks do not cast doubt on factors the belligerent is counting on. For example, if a belligerent is expecting a powerful ally to eventually enter the war and turn the tide, the belligerent will not become excessively discouraged if its own forces perform poorly. Or, if a belligerent is counting on a future strategic or technological innovation to change the balance of power on the battlefield, it will not be discouraged by battlefield losses. Indeed, some have postulated that bad combat performance may cause a belligerent to change military strategy rather than reduce its war-termination offer.⁶⁴ However, if combat or international events cause a belligerent to lose faith in its strategy for victory, if a potential ally joins the other side, if the belligerent was counting on victory through economic mobilization and the adversary captures wide swaths of its economy, if no strategic or technological innovations present themselves, or if a belligerent was counting on exacting enemy concessions through coercion and combat outcomes cause it to lose faith in its ability to inflict cost on the enemy, the belligerent may become sufficiently discouraged to accept defeat, even an absolute defeat.

Escalating Costs

Belligerents may be unwilling to reject absolute war outcomes and pursue limited war settlements if the costs of continued fighting promise to escalate significantly (this dynamic parallels the discussion surrounding figure 2.1 in chapter 2). Absolute war outcome as a solution to credible commitment problems is in some sense a luxury that a state can purchase, and states may elect not to purchase the luxury if the cost becomes too high. Small states facing more powerful adversaries, such as Finland confronting the Soviet Union in 1940, Japan facing the United States during World War II, and Iraq facing the United States in 1991, would certainly like to pursue absolute military victory to eliminate the large threat of the adversary, but doing so is seen to be prohibitively costly because of the unfavorable military imbalance.

High costs dissuade states from pursuing absolute war outcomes in other circumstances as well. A large belligerent with the military capacity to crush a small foe at acceptable human and financial costs may decide not to do so for other reasons. The belligerent may fear that the small foe will escalate to the use of weapons of mass destruction (WMD), as Iraq considered doing in 1991 if the UN Coalition had marched on Baghdad.⁶⁵ Alternatively, although the larger belligerent may feel confident in its ability to defeat its adversary at reasonable cost, if the larger belligerent is worried about the possibility of a third power soon intervening on behalf of its smaller adversary, the prospective escalation in the costs of fighting may be sufficient to push the more powerful country to accept a limited outcome.⁶⁶ In the 1973 Yom Kippur War, Israel eventually achieved decisive military superiority over Egypt and Syria. The combined external pressures of the United States and the Soviet Union, including the threatened withdrawal of military support by the former and the possibility of military intervention on the Arab states’ behalf by the latter, pushed Israel to accept a limited war outcome, and in particular spare the Egyptian army from complete annihilation.⁶⁷

Some belligerents may view the post-victory costs of administering an absolutely defeated foe to be prohibitive. At the outset of the 1846–48 U.S.–Mexico War, the United States hoped to acquire California, New Mexico, and perhaps other territories, although the unexpected success of American military forces, including the capture of Mexico City, opened the door for increased American war aims.⁶⁸ The war-ending Treaty of Guadelupe Hidalgo provided the U.S. with more substantial territorial gains than it had originally hoped for, but it stopped short of demanding the annexation of the entire country. This is puzzling from the perspective of the commitment proposition, as a “common expansionist objection [to the Treaty] was that it would be foolish to expect the treacherous Mexican government to carry out the terms of any treaty.”⁶⁹ However, the majority in the U.S. Congress viewed the ingestion of all of Mexico to engender too many costs and dangers, including the difficulties of annexing an alien and hostile people, the risks that maintaining a large American army would undermine American democracy, and concern that annexation would incite European opposition. Underlying these issues were even more complicated debates about the effects of slavery, as advocates took different positions as to whether annexation would support or undermine slavery in America.⁷⁰ A century later, President Dwight Eisenhower worried about the costs of absolute victory in discussions about the possibility of nuclear war with the Soviet Union. He argued that occupying the Soviet Union “would be far beyond the resources of the United States,” and fretted about the hollowness of a nuclear victory: “Here would be a great area from the Elbe to the Vladivostok . . . torn up and destroyed, without government, without its communications, just an area of starvation and disaster. I ask you what would the civilized world do about it? I repeat there is no victory except through our imaginations.”⁷¹ The George H. W. Bush administration was also dissuaded from marching on Baghdad in 1991 partly because of the costs of such a total victory, especially the escalation of anti-American insurgency warfare and international diplomatic backlash.⁷² Conversely, the George W. Bush administration decided in 2002–3 to overthrow Saddam’s regime under the flawed assumptions that the financial costs would be low and that Iraq would quickly enjoy stability, prosperity, and democracy.⁷³

Costs of fighting aside, the perceived consequences of a broken agreement may push a belligerent to accept a limited outcome. Would a new war following a breach of the war-ending agreement threaten the conquest of the belligerent? Such an outcome might be more likely if the adversary is at least as powerful as the belligerent or contiguous to the belligerent. Conversely, if the enemy is far away and/or less powerful, a breach of the war-ending agreement may be seen as being less costly, making a limited (if unstable) outcome more acceptable.

In sum, a belligerent, even if it is fearful that its adversary might break a war-ending agreement, will be more likely to accept a limited war outcome if it believes the costs of continuing the war promise to escalate significantly. Note that accepting a limited war outcome may mean making concessions, or it may mean accepting the adversary’s standing wartermination offer.

Postwar Peacekeeping Tools

Postwar peacekeeping tools might also reduce commitment fears and thereby affect the interplay of commitment and information dynamics. One possibility would be to create international institutions as a means of helping peace endure, such as inviting third parties to monitor and keep the peace, establishing demilitarized zones, signing arms control agreements, pursuing confidence building measures, and so forth.⁷⁴ However, such institutions are unlikely to provide comprehensive solutions to the postwar commitment problem. Not all of these tools, such as thirdparty peacekeepers, are always available. Usually, the former belligerents need to agree to allow the entry of peacekeepers, and of course a third party needs to be willing to supply them. Peacekeepers have been more commonly available after 1945 than before, and even in the post-1945 period are not always available or employed when available.⁷⁵ Further, the evidence that these tools help keep the peace between states is mixed, with some studies casting doubt on the ability of such tools to help peace endure.⁷⁶ Lastly, many of these tools are designed most directly to help prevent the accidental escalation of conflict. They are not necessarily intended to prevent the deliberate reignition of conflict. For example, UN peacekeepers were sent to the Sinai peninsula after the 1956 Suez War. As tensions mounted between Egypt and Israel in late spring 1967, Egypt simply ordered the peacekeepers to leave, and war broke out in June.⁷⁷ Nevertheless, this existing scholarship implies that a belligerent, even one fearful that its adversary will break a war-ending commitment, will be more likely to accept a limited war outcome if a third party promises to deploy peacekeeping troops or related measures.

International institutions aside, another way to reduce the commitment problem, thereby permitting war termination short of an absolute outcome, is to pre-deploy forces to reduce the possible mobilization advantages an adversary might try to gather. A belligerent might be defending a geographically far-flung ally, and it might fear that following a limited war outcome, if the belligerent brought its forces home from the ally’s territory, the adversary might reattack sometime in the future, hoping to conquer the ally before the belligerent could remobilize and redeploy its forces to the ally’s territory. One possible solution to this is the permanent forward deployment of forces by the belligerent, which would reduce the mobilization speed advantage, and also possibly serve as a tripwire automatically engaging the belligerent.⁷⁸ This was part of the American defense strategy in Western Europe especially after Eisenhower adopted the New Look defense policy, by which pre-deployment of American conventional forces guaranteed American involvement in the event of a Soviet invasion of Western Europe.⁷⁹

However, the pre-deployment strategy is designed to address only a very specific kind of commitment problem, a possible short-term shift in the balance of power gained by mobilization advantages. It also assumes that the belligerent is willing to deploy standing forces to the ally in question for years if not decades on end, a policy that may not be economically or politically sustainable.

Capturing Goods that Reduce the Commitment Problem

Under some conditions, a belligerent may be willing to forgo an absolute war outcome because a limited war outcome may offer at least a partial solution to a credible commitment problem. Most centrally, seizure of at least part of the disputed good may decrease the likelihood that the adversary will defect on a war-ending agreement by making the prospect of a future attack by the adversary prohibitively costly or unlikely to succeed. Even a state fearful of an adversary’s incredible commitment may be willing to accept a limited outcome, if that outcome provides a partial solution to the credible commitment problem and pursuing absolute war as a complete solution of the credible commitment problem promises to be prohibitively costly.

How exactly might acquisition of part of the disputed good decrease the chances the war-ending commitment would be broken? Most straightforwardly, if acquiring a specific component of the good makes a future attack less attractive to an adversary, then this may help alleviate the commitment problem. Specifically, possessing a specific increment of the good may make it easier to repel a future attack by the adversary, thereby decreasing (though not eliminating) the adversary’s incentive to renege on the war-ending commitment and launch a future attack. Most commonly, this occurs when the attacker can seize a strategically critical piece of territory that makes national defense easier. Examples include mountain ranges, all the territory up to a body of water such that the national boundary becomes the shore of the water body, thick forests, and so forth. Empirical studies have shown the importance of geographic terrain on war outcomes, and other work has found that a state is less likely to attack if territorial features such as mountains or water bodies promise to make attack difficult.⁸⁰

The Franco-Prussian War of 1870–71 demonstrates this dynamic. Prussia’s desire for territorial gains was driven by credible commitment concerns. It perceived France as an enduring threat, and saw territorial acquisition as one means of making the French commitment not to attack in the future more credible by making the execution of such an attack more difficult. As Prussian Prime Minister Otto von Bismarck told a French representative, “Over the past 200 years, France has declared war on Prussia thirty times and . . . you will do so again; for that we must prepared, with . . . a territorial glacis between you and us.”⁸¹ Elsewhere, Bismarck stated that the French border city of Strasbourg would become Prussia’s “Gibraltar,” and that Strasbourg was “the key to our house.”⁸² The outcome of the war reflected Prussia’s needs, since in the peace settlement France handed over to Prussia the provinces of Alsace and Lorraine, and the city of Strasbourg.

An increment of the good can make future defense easier if the increment contributes to military-industrial power. Two belligerents may be in dispute over a large territorial area, but key economic resources (such as coal mines, oil fields, or factories) may be concentrated in one small portion of the area, meaning that possession of that small portion may increase a belligerent’s military power and make it less likely that its adversary will renege on a war-ending agreement and attack in the future. The idea that military conquest can mean the capture of resources, which in turn fuels military power and makes future victory easier, has been called by some the “cumulativity of resources.”⁸³ In the twentieth century, military victors have been able to convert captured economic resources into future military power, as a conquered foe yields roughly half the economic resources that an ally of similar size provides.⁸⁴ Germany certainly reaped tremendous gains from its March 1939 seizure of Czechoslovakia. Czechoslovakia possessed tremendous supplies of arms and world-class munitions factories. In April 1939, twenty-three trains filled with ammunition and weapons were sent from Czechoslovakia to Germany per day.⁸⁵

The theoretical point that the capture of increments of the disputed good can affect the balance of power has been explored in some bargaining models.⁸⁶ Notably, the existing work generally maintains the assumption that each increment of the good is equally valuable—that is, acquiring each increment offers a similar change to the balance of power. This work has produced several results, including that as each increment of the good is captured and the balance of power shifts, the side gaining the increment may have an incentive to demand even more of the opponent, reflecting its recognition that the balance of power is shifting, and that when the good contributes to the balance of power, war is more likely.⁸⁷

Copyrighted image removed by Publisher

Here, I relax some important assumptions of this past work. I first relax the assumption that all increments of the good are equally valuable. Some increments of the good, such as some portions of territory, may be more valuable than others. Put another way, oil fields, coal mines, mountains, and munitions factories may not be equally distributed throughout the entirety of a territorial area.⁸⁸ I term an increment as “supervaluable” if it contains more value than other increments of the good. For example, a portion of territory might be supervaluable if it contains the only oil reserves in the entire country. Note that if the disputed good is territory, then the supervaluable increment could be the first, last, or any intervening increment of territory. If it lies in the middle of the disputed area, then the relationship between possession of the good and utility is illustrated in figure 3.2.

Like the curve in figure 3.1, the curve in figure 3.2 is not linear, but rather at one point spikes up (a “step function”). In figure 3.1, the last increment of the good is supervaluable since it represents the absolute defeat of the adversary’s military. In figure 3.2, the middle increment of the good is supervaluable, representing the possibility that any increment of the good can be supervaluable in its contribution to the military balance since, for example, the oil fields or a critical mountain range may rest in the center of the country.

If the utility return on the good is nonlinear as described in figure 3.2, some important implications follow. If the marginal return to fighting drops off after the belligerent seizes the supervaluable increment of the good, that belligerent may be more likely to accept war termination on limited terms, especially if the costs of war promise to escalate, even if its armed forces were performing well in combat. Therefore, after a belligerent has captured an increment of the good that alleviates its postwar commitment problem, it will become more likely to accept war termination on limited terms, even if it is enjoying success on the battlefield.

One critique might be that if the increment of the good is so valuable, then a war-ending settlement that allows for the exchange of the good from loser to winner may be difficult to reach, because the loser will be especially averse to abandoning that part of the good. However, the supervaluable increment of the good may make defense easier, but not make attack easier. This is most likely the case for geographic features such as mountain ranges and bodies of water, which facilitate defense but do not necessarily facilitate the attack. So, the capture of a supervaluable increment may not necessarily increase the offensive power of the victor, pushing it to continue fighting and raise its demands.

Two belligerents are especially likely to settle on a war-termination agreement in which one side gives up a supervaluable increment of a good short of suffering absolute military defeat when one side has proven its decisive military superiority, but the costs of pursuing future military operations threaten to escalate for that side. The winning side has the military advantage such that it could push further, but it does not because the costs of going further threaten to escalate the conflict, and it is less motivated to keep going because additional increments of the good are not as valuable. The losing side is willing to accept the loss of the supervaluable increment, because it recognizes that continuing to fight is unlikely to allow it to keep the supervaluable increment, and also continuing to fight introduces the dangers of suffering greater losses, including absolute defeat.

An example of these dynamics is the termination of the 1967 Six-Day War between Israel and its Arab adversaries. Israel achieved tremendous success on the battlefield, capturing in short order the Gaza Strip and Sinai from Egypt, the Golan Heights from Syria, and East Jerusalem and the West Bank from Jordan. A historical puzzle is why Israel decided to stop when it did. After achieving such great and unexpected success on the battlefield, perhaps the growth in Israeli confidence should have encouraged Israel to consider raising its war aims, in pursuit of overthrowing the Arab regimes. Indeed, Israel did have contingency plans for going farther, including a 1964 plan called Hatchet that envisioned two Israeli divisions swiftly conquering the Syrian capital city of Damascus. During the war, one Israeli newspaper had as a huge headline on its front page, “Mission: Damascus!” However, Israel accepted war termination on limited terms even in the wake of battlefield victory, in part because its limited territorial gains had substantially increased its security and helped solve its credible commitment problem by making any future Arab attack on Israel more difficult, and in part because continued military operations would engender growing diplomatic costs, especially regarding relations with the United States. The acquisition of a substantial territorial buffer in Sinai and the high ground up to and including Mount Hermon in the Golan were especially important in reassuring Israeli security concerns.⁸⁹ Arab states accepted the loss of these territories because the shocking wave of Israeli military successes left them unconfident they could recapture their lost lands. Syria and Egypt did eventually reattack in 1973, their self-confidence restored by the acquisition of advanced Soviet military technologies.

The 1991 Iraq War provides a further example of a winning military from a country concerned with postwar agreement compliance deciding to accept a limited outcome that shifted the balance of power. The George H. W. Bush administration was deeply concerned about the long-term military threat posed by Saddam Hussein’s Iraq, based on Iraq’s allegedly battle-tested conventional forces, touted as the third biggest in the world, and its WMD programs, the details of which emerged after Iraq’s August 1990 invasion of Kuwait. The United Nations coalition stopped its advance after liberating Kuwait, foregoing the opportunity to conquer Baghdad and impose an absolute outcome. Some have speculated that the decision to do so reflected both domestic and international political constraints, specifically that although the public strongly supported the use of military force to liberate Kuwait, it would not support a longer war that promised higher casualties, and that the supporting coalition would break apart if the United States exceeded its UN mandate and marched on Baghdad.⁹⁰ For its part, Iraq accepted the loss of Kuwait as a limited outcome of the war, clearly recognizing that the UN coalition had demonstrated its military superiority.

The United States and UN accepted the limited outcome of the 1991 Gulf War because they believed this outcome substantially ameliorated the commitment problem, by significantly reducing Iraqi conventional and unconventional military power. Regarding Iraqi conventional forces, a central aim of the war was to smash the Iraqi army, beyond the liberation of Kuwait. This was expressed most clearly in a January 1991 meeting with Bush and his advisers. Bush and his national security adviser Brent Scowcroft argued against the acceptability of a peaceful Iraqi withdrawal from Kuwait. The problem, they pointed out, is that it would have been politically impossible for American forces to remain in the region indefinitely as a deterrent to further Iraqi aggression, and when the American forces eventually left, Iraq could exploit the shift in the immediate balance of forces and reinvade Kuwait.⁹¹ As a result, the good in dispute was not just the sovereignty of Kuwait, but also the strength of the Iraqi army, and achieving the American preference on the latter (destroying Iraqi conventional military power) would be necessary to make the Iraqi commitment to respect the war-ending settlement on the former (recognize Kuwait’s sovereignty) more credible. At the January 1991 meeting, Bush explicitly argued that the U.S. desired both Kuwaiti liberation and the destruction of the Iraqi army, remarking, “If they crack under force, it is better than withdrawal.”⁹² In February, Soviet General Secretary Mikhail Gorbachev floated the possibility of an Iraqi withdrawal from Kuwait over a period of weeks. The U.S. viewed this proposal as flawed in part because it would have allowed the Iraqi military to withdraw intact, giving Saddam the ability to reattack Kuwait in the future.⁹³ The Bush administration’s counterproposal (rejected by Iraq) of a seven-day withdrawal would have forced Iraq to abandon a sizable fraction of its heavy forces in Kuwait, because many units were disabled or dug in.⁹⁴

The UN had similar views about Iraqi WMD. Absent the overthrow of Saddam’s regime, the concern remained that Iraq might in the future use WMD against its neighbors, and like the degradation of Iraq’s conventional military forces, the destruction of Iraq’s WMD infrastructure was part of the good that constituted coalition war aims. The U.S. perceived that even with a limited outcome, it could reduce the WMD credible commitment problem. The UN coalition conducted extensive airstrikes against known Iraqi WMD facilities during the war, and the operational success of these strikes gave the coalition confidence that Iraq’s WMD capacity had been gutted. Further, the deal ending the war required Iraq to comply with UN Security Council resolutions on WMD disarmament, and to allow UN inspectors to visit suspected WMD sites. At war’s end, all were confident that these two measures crippled the Iraqi WMD threat, and that the inspectors would be able to wrap up their work within six months at the most.⁹⁵ The reality was different, as the world quickly learned that there were wide secret swaths of the Iraqi WMD program left untargeted during the war. Iraq’s ongoing resistance to UN inspectors in the 1990s further raised fears that Iraq was defying the terms of the 1991 settlement. Perhaps ironically, it became evident in 2004 that Iraq had abandoned its WMD programs in the late 1990s.⁹⁶

ISSUE INDIVISIBILITY

The previous discussion of commitments has framed commitment effects in the context of the relationship between increments of the good and utility—that is, the benefits curve of continued fighting. Another factor that could shape the benefits curve is so-called “issue indivisibility,” the claim that some issues cannot be divided in a bargain between two disputants, and that the presence of issue indivisibility can make war more likely even between two rational, fully informed states.⁹⁷ Indivisibility is easiest to think of as physical indivisibility, a good that cannot materially be divided, or would be destroyed if division is attempted. The biblical story of King Solomon provides the classic example, in which two women are in dispute over a child, and Solomon’s offer to cut the child in two as a solution to the dispute is rejected by one of the women because doing so would of course kill the child. International relations are commonly much more about easily physically divisible goods like territory and natural resources. Even settling on political control within a national government is subject to divisibility, through institutionalized power-sharing in the parliament or even the office of the president, the latter exemplified by Bosnia’s three-headed presidency since the middle 1990s. Probably the best example of a literally physically indivisible good in international relations might be a state’s possession of a first nuclear weapon. However, even the possession of a nuclear weapon can be divided somewhat. The dual key system in NATO during the Cold War allowed the United States to own nuclear weapons based abroad, while the host nation controlled the missile used to deliver the nuclear weapon. Further, nuclear possession is functionally divisible through side payments, as American promises of extended nuclear deterrence have persuaded states like (West) Germany, Japan, South Korea, and Taiwan to forgo national nuclear programs.

A number of scholars have maintained that states treat some goods in international relations as if they were indivisible, even if they can be physically divided. Some scholars have posited that cultural and social forces sometimes push a belligerent to frame a good, especially a territorial area, as having value only if its entirety is possessed, perhaps because the territory is seen as “sacred space.”⁹⁸ Another, more rationalist theory proposes that reputation effects may create indivisibility effects. An actor may perceive that making concessions to split a good may encourage the other actor to make more challenges in the future (also known as “salami tactics”), or encourage other actors to make challenges, a dynamic related to what economists call the “chain store paradox.”⁹⁹

Issue indivisibility fits into the theoretical framework envisioned here. Specifically, because the issue pertains to the benefits side of the calculus, issue indivisibility is a factor that shapes the relationship between increments of the disputed good and a belligerent’s utility. The curves in figures 3.1 and 3.2 describe two possible relationships between these factors. If a belligerent genuinely thinks an issue is indivisible, this would mean that that side would receive no welfare from any increment of the good except when all of the good is acquired, meaning the benefits curve ought to look something like figure 3.1, a line with a near zero slope, but then a step function at the right-hand side jumping up if the belligerent can capture the entire good. The war-termination prediction might be that if a war is fought over an indivisible issue, then belligerents are less likely to negotiate for a limited outcome to split the good, and are more likely to fight to the finish.

In practice, issue indivisibility is unlikely to play a central role in war initiation or war termination. Some have proposed that issue indivisibility can at least in theory be overcome through side payments or lotteries, although such mechanisms may be infeasible because states cannot credibly commit to carrying out the terms of such a deal—that is, giving up claim on an indivisible issue on the basis of the outcome of a random number draw.¹⁰⁰ A more aggressive claim is that no issue is ever literally indivisible, either physically or perceptually. Even when an actor places a very high value on possession of a good such that it appears to treat the issue as indivisible, the actor will almost never be completely indifferent between possessing none of the good and some of the good (true indivisibility requires this indifference). Pakistan would not be indifferent to acquiring an extra 10 percent of Kashmir, and the Palestinians would not be indifferent to reacquiring the 50 percent of Jerusalem they controlled before the Six-Day War.

As an empirical matter, scholars have been unable to point to many episodes of war-causing indivisibility. Note that even with the two leading examples, Kashmir and Jerusalem, the parties possessing less than 100 percent of the good go years without attacking. Pakistan and India have fought some wars over Kashmir, although the vast majority of their fifty years since independence have been without conflict. Regarding Jerusalem, the Arab states have not attacked with the aim of recapturing Jerusalem since Israel took control of all of it in 1967. Interpretation problems arise with other alleged episodes of issue indivisibility.¹⁰¹ More generally, public statements alone are insufficient to prove indivisibility, as actors are motivated to claim that an issue under dispute is indivisible in order to provide increased bargaining leverage.¹⁰² For example, in the 1944 war-termination negotiations between Finland and the Soviet Union, the former claimed that the pre–Winter War 1939 borders must be restored, because of the indivisibility of all of Finnish territory. One Finnish delegate argued that the portion of Finland lost at the end of the Winter War in 1940 “constitutes in the economical as well as in the national sense so to speak an organic part of our nation’s body, and it affects our total existence in a most decisive way. . . . I repeat once more that the ceded part of Karelia belongs organically to Finland, and that our people cannot conceive that this frontier will be a final one.”¹⁰³ Yet the Finns gave up this territory as part of the 1944 peace treaty with the Soviet Union, accepting a division of this allegedly indivisible territory peacefully for the more than six decades that have elapsed since. Similarly, regarding whether Germany would give up the Alsace-Lorraine territories acquired in the 1870–71 Franco-Prussian War in any peace deal to end World War I, the German foreign minister Richard von Kuhlmann declared to the German Reichstag in 1917, “We have but one answer to the question of whether Germany will make any concessions in the matter of Alsace-Lorraine. That answer is no, no, never! So long as a single German can still lift a gun, the integrity of the Reich territory, handed down to us by our fathers as a glorious heritage, shall never be the subject of any negotiations or concessions. Alsace-Lorraine is Germany’s shield, the symbol of German unity.” Kuhlmann admitted in his memoirs that the speech was intended to restore patriotic unity in the Reichstag, and Germany of course returned the territories to France as part of the Versailles Treaty.¹⁰⁴

AN INTEGRATED RATIONALIST THEORY OF WAR TERMINATION

The past two chapters have laid out ideas about war termination that share some common assumptions: the state is the primary actor, states are rational, anarchy is unavoidable, and uncertainty is quite common. The two chapters together have presented a rationalist theory of war termination that contains as its two principal strands information and commitment dynamics. These two dynamics sometimes work in complement, and they sometimes pull in opposite directions. This chapter has laid out some of these tensions as well as conditions under which we should expect information or commitment dynamics to predominate. The following several chapters address the next task, assessing the power of this theory by applying it to wars that have actually occurred.