CHAPTER ELEVEN
War, what is it good for? Absolutely nothing.
—Edwin Starr
WHAT IS THE FUNCTION OF WAR? Poets, artists, and ordinary people have long agreed with the words of the popular song that war is good for “absolutely nothing,” providing only death and destruction. And yet wars continue to occur by the deliberate choices of national leaders, encouraging the conclusion that wars must have some purpose, must be politics “by other means.”
This book has described two functions of war, two purposes that fighting is meant to serve: providing information and solving commitment problems. These answers undergird a single theoretical approach to war termination focusing on how states seek to maximize benefits and minimize costs. These answers also constitute the core of broader international relations theory about the causes of war: wars happen because states are uncertain about each others’ intentions and capabilities, and because states cannot sign binding agreements by which they agree not to attack each other. I have extended these insights about the causes of war to understand the prosecution and termination of war, exploring how war is used to solve the problems that lead to its outbreak.
This concluding chapter has two tasks. The first is summary and synthesis. The findings of the twenty-two war-termination decisions examined are summarized, grouped, and analyzed, towards providing more general conclusions as to how wars end, and how information and commitment factors interrelate in explaining war termination and international relations more broadly. The second is the discussion of the empirical finding for contemporary American foreign policy. What recommendations do these findings hold for the conduct of American foreign policy in the twenty-first century?
HOW STATES ENDED WAR
The empirical chapters in this book surveyed twenty-two war-termination decisions across six wars. Although the small and nonrandom nature of the sample precludes quantitative analysis, the data provide strong support for the model presented in chapters 2 and 3, and the general notion that both information and commitment factors are necessary to provide a complete understanding of war-termination decision-making. This section uses sets of the twenty-two cases to demonstrate several patterns predicted by the theoretical model.
Higher War-Termination Demands with Discouraging Information, Commitment Credibility Concerns, and Hope for Ultimate Victory
The conventional information-oriented wisdom (as expressed by the information proposition in chapter 2) is that combat setbacks should encourage negotiation and concessions, and combat successes should encourage increases in war-termination demands. The theoretical model, as fully developed in chapter 3, develops a set of propositions that explain the conditions under which the information proposition ought to predict accurately war-termination behavior, and conditions under which it will not accurately predict war-termination behavior. Specifically, chapter 3 develops propositions that predict the conditions under which a belligerent may maintain high war-termination demands or even raise them, even when the belligerent is experiencing combat setbacks. The theoretical argument is that a belligerent will eschew concessions in the face of bad news when it thinks the adversary will not credibly commit to a warending commitment, and if the belligerent has some hope of eventually winning the war at an acceptable cost in relation to the stakes at hand.
Seven cases fit this mold of a belligerent maintaining high war aims (or even increasing its war aims) despite battlefield setbacks, because of severe commitment concerns and a hope of eventually winning the war. They include the absence of Union concessions in summer 1862, the Confederacy’s decision not to negotiate in late 1863, the Union’s decision not to negotiate in the summer of 1864, the Confederacy’s decision not to make concessions in early 1865, Britain’s decision not to negotiate with Hitler in May 1940, the American commitment to the unconditional surrender of the Axis in early 1942, and the American decision to press for the conquest of North Korea in August–September 1950.1
In each case, the belligerent in question had suffered severe setbacks. A focus on information alone would predict that the losing belligerent should have reacted to its battlefield setbacks by reducing its wartermination demands in the hopes of ending the war sooner and avoiding the risks of fighting on. However, in each case the belligerent elected to fight on because of concerns that making concessions and/or accepting a limited war outcome would jeopardize the belligerent’s longer-term security. Jefferson Davis feared that the Radical Republicans would force a harsh reconstruction. Lincoln feared the southern states would soon demand slavery in more than just the existing slave states. Churchill feared that Hitler would demand concessions, leaving Britain at his mercy. Roosevelt feared a repetition of the Versailles debacle. Truman feared that North Korea would take asylum behind any armistice line, lying in wait for a favorable opportunity to strike again. Further, in each case the belligerent was able to maintain some hope that eventually the tide might turn, the recent setbacks notwithstanding. Britain in 1940 hoped for eventual American rescue. The U.S. in 1942 assumed that eventually mobilization of its economy would overwhelm the Axis. The CSA hoped that escalating Northern war fatigue would force the Union to end the war on terms recognizing Confederate independence. Truman hoped that American forces could hold on until reinforcements could arrive, or MacArthur’s great gamble at Inchon might turn the tide.
Beyond these seven cases, Japan from mid-1942 to mid-1945 is perhaps an eighth case. Like the other cases, Japan experienced a steady stream of combat setbacks, yet chose not to negotiate or make concessions. Also like the other cases, Japan clung to some hope of extracting a more acceptable war outcome, by persuading the United States that pursuing unconditional surrender would be prohibitively costly. This case is a bit different from others in that Japan did not face quite the same commitment problem. Its adversary, the U.S., kept to its demand for unconditional surrender, meaning the Japanese did not consider the possibility of the U.S. pushing for a limited war settlement that they would then go on to break.
These cases are the strongest evidence that a complete understanding of war termination cannot focus exclusively on information dynamics. These are all easy cases for an information-only approach, in that in each case the independent variable of battle outcomes is very clearly coded as providing discouraging information from combat outcomes, thereby predicting a reduction in war-termination demands. Yet in each case the belligerent raises or maintains high war-termination demands. They all demonstrate that commitment dynamics play an important role in understanding a belligerent’s war-termination decisions.
Lower War-Termination Demands with Discouraging Information, Commitment Credibility Concerns, and Almost No Hope for Ultimate Victory
Another group of cases is similar to the preceding group with one difference. Like the preceding group, these belligerents received discouraging information from combat outcomes and sustained credible commitment concerns about their adversaries. However, unlike the previous group the belligerents in this group held almost no hope in eventual victory. For them, the nontrivial costs of continuing to fight were enough to encourage concessions to reach even an unstable peace, rather than fight on in hopes of a victory unlikely to occur.
The first two cases are Finland in 1940 and 1944. In both periods, Finland experienced combat setbacks in its war with the Soviet Union. Finland had reason to distrust Soviet commitment credibility, if for no other reason than both the 1939 and 1941 Soviet attacks broke neutrality agreements. Yet in both cases Finland decided to make concessions and accept a limited end to the war. Both times, Finland recognized that its military prospects were collapsing rapidly, and that absolute defeat loomed as a possibility. Relatedly, Finland understood on both occasions that there was little hope for eventual victory, especially given that third parties would not be able to turn back Soviet might (in 1940, British and French aid would arrive too late, and in 1944, Germany was collapsing under the assault of the Red Army).
A third case, with thin evidence, is the Soviet Union in autumn 1941. The Soviet Union faced a much grimmer outlook in 1941 than did Britain in 1940 or the U.S. in 1942. All three Allied belligerents faced very worrisome combat setbacks at these three points in time, and all three gravely doubted the credibility of any war-ending Axis commitment to peace. However, the Soviet Union faced a much bleaker outlook in 1941 than did Britain in 1940 or the U.S. in 1942. Nazi Germany had invaded and conquered huge chunks of the Soviet Union by autumn 1941, including roughly a third to a half of the Soviet economy, and stood on the brink of seizing Moscow itself. Both Britain and the U.S. could hope that the oceans, naval power, and air power could provide enough defenses to prevent their own homelands from being invaded, although of course the U.S. felt much more secure on this score than did Britain. The theory predicts that the U.S. and Britain would reject negotiations in pursuit of an unstable limited war peace, and instead fight on, whereas the Soviet Union would in desperation offer concessions in an effort to stave off its apparently inevitable defeat. The existing evidence indicates that Stalin did decide to offer Hitler a basket of territorial concessions in October 1941, as a desperate move to end the war and save the Bolshevik state, although it is unclear whether these concessions were ever communicated to Berlin.
A final case in this group is Japan in August 1945. Up to this point in the war, Japan had continued the war despite combat setbacks, in the hopes that eventually it could threaten to impose enough costs on the U.S. to persuade it to abandon unconditional surrender. It finally agreed to make concessions and accept near-unconditional surrender once its faith in its ability to impose high costs on the U.S. collapsed, a development caused principally by Soviet entry into the war and discouraging reports about Japanese homeland defenses.
Lower War-Termination Demands with Discouraging Information, Commitment Credibility Concerns, and Prohibitively High Costs of Continued Warfare
Related to collapsing hopes of victory, another factor that can cause a belligerent facing combat setbacks and retaining credible commitment concerns to lower its war-termination aims is the prospect of prohibitively high costs of continuing to fight. The U.S. in early 1951 represents this dynamic. Its hopes of solving its credible commitment problem by overthrowing the Pyongyang government were dashed once China entered the war in autumn 1950. Chinese entry redefined the credible commitment problem: now, solving the credible commitment problem in Korea would mean escalation to war with China, and perhaps even nuclear war with the Soviet Union. As the theory predicts, the Washington decision-makers concluded that the costs of securing the luxury of solving the credible commitment problem proved too high in relation to the dangers posed by a violation of the commitment (the future conquest of South Korea). Chastened, the U.S. lowered its war-termination demands, no longer seeking the conquest of North Korea.
Japan in August 1945 also represents this dynamic. The previous section noted that Japan accepted defeat in August 1945 in part because it lost faith in its ability to impose costs on the United States. It also recognized that continuing to fight would impose exorbitant and unacceptable costs. The atomic bombings of Hiroshima and Nagasaki demonstrated the American ability to inflict several tens of thousands of casualties with a single bomb, without risking American casualties. The prospect of the literal annihilation of Japanese civilization was too much for the leadership in Tokyo to bear.
Encouraging Information Causing Higher War-Termination Demands
A few cases demonstrate the logic of war contained within the information proposition of chapter 2, when good news from the battlefront encouraged a state to demand more of its adversary. The Soviet Union in 1940 and 1944, after initial setbacks, finally reached a position of military dominance over Finland. In both instances, it took advantage of its military successes to increase its war-termination demands. Notably, its increased demands in each case were finite, as external factors (the impending Anglo-French intervention in 1940 and the remaining need to crush Germany in 1944) provided impetus to end each war quickly.
The Union in autumn 1862 is worth noting as a case that does not fit in this group. On the surface, it might seem to belong, as the Union raised its war-termination demands to seek the emancipation of the slaves after having won at Antietam. Closer examination reveals that the victory at Antietam did little to boost the Union’s confidence in its ultimate military prospects. Further, the Union raised its war aims out of concerns with its weakness, not because it was encouraged by demonstration of its strength, since Lincoln thought that emancipating the slaves would encourage Northern and Southern blacks to support the Union cause and increase its strength.
When the Nature of the Good Affects the Commitment Problem
Imposing an absolute war outcome is one solution to the credible commitment problem since it removes the adversary’s sovereignty. In some instances, the severity of the commitment problem can be affected by allocations of the good short of an absolute outcome; if the good can substantially reduce the likelihood, the adversary will break the peace agreement. Territory is the best example of this dynamic, since a belligerent’s possession of strategic territory like high ground can make defense easier, thereby making it less attractive for a defeated adversary to defect and reattack, meaning the commitment problem is reduced. The Finnish– Soviet conflicts from 1939–44 illustrate this dynamic, because the Soviets sought the acquisition of critical portions of Finnish territory to make the Soviet border more secure, to improve the defenses of Leningrad and especially naval access to the Baltic Sea, and to make Helsinki more vulnerable to attack. Although commitment factors were critical in creating Soviet preferences, in both the Winter War and Continuation War changes in Finnish and Soviet war-termination demands are well explained by a focus on information. Similarly, the cornerstone of World War I Germany’s war aims in the West was the postwar control of Belgium, to decrease the likelihood of a future British or French attack.
Other cases also demonstrate how reallocations of the disputed good can affect the commitment problem. In 1940, Churchill rejected negotiations with Hitler in part because he knew Hitler would demand reductions in British military and naval power, which would put Britain even further at Germany’s mercy. In the Korean War, negotiations over prisoners of war dragged on in part because (at least) the UN/American side recognized that the settlement of this issue would directly affect the postwar balance of power, since the UN held more Communist prisoners than vice versa, and the Chinese proposal of an all-for-all trade would provide the Communists with a relative gain in postwar manpower levels. Sending the Communist prisoners back might also make Communist soldiers less likely to surrender in future conflicts. In summer 1864 as Union fortunes looked bleak, Lincoln refused to consider retreating on emancipation, because he knew this would undermine the willingness of blacks to support the Union and thereby undercut Union military power. Raising war aims can improve the balance of power in other circumstances as well, since belligerents may broadly hope that transforming a conflict into a war of liberation may cause individuals to flock to their cause. Emancipation was motivated primarily as a tool to recruit blacks to the Union cause. In 1943, Japan hoped that offering more rights and freedoms in conquered areas would attract greater support. More generally, revolutionary regimes are likely to think that launching wars of liberation may attract oppressed peoples to their sides.2
These cases illustrate the idiosyncrasy of how divisions of the good affect the commitment problem. The good can affect the commitment problem (and, more broadly, military power) in a number of ways, including possessing important territory, affecting the motivation or ability of individuals to fight for one side or the other, and possessing specific military assets. Further, not all increments of the good may affect the commitment problem equally, as not all increments of land are equally valuable and not all military assets are equally important to the balance of power.
More generally, if the nature of the good affects the commitment problem, this mediates the effects of information and commitment dynamics. When a state captures an element of the good that reduces the commitment problem, it may feel more willing to end the war and accept a limited outcome, even if it has enjoyed a string of military victories, and especially if it fears the escalation of the costs of war. Further, even when a belligerent fears that its adversary may consider violating the terms of a warending agreement, the belligerent may be willing to abandon absolute war aims and accept a limited outcome if the limited outcome allows the belligerent to acquire an element of the good that will reduce the commitment problem. This dynamic allowed the Soviets to accept limited victories over Finland in 1940 and 1944.
Domestic Politics
Surprisingly, the role of domestic politics in this book’s cases of wartermination decision-making was limited. The possible effects of domestic politics on war termination are several, and this book considered two possible relationships. The first is the mainstream idea that the escalation of casualties increases the likelihood of concessions, especially in democracies. This hypothesis received strikingly little support. Sometimes, democratic publics were more opposed to making concessions than were democratic leaders, as was the case perhaps most clearly in Finland during the Winter War and Continuation War. This is consistent with the pattern observed elsewhere that sometimes fighting wars is more popular with the public than the leadership.3 Sometimes, the escalation of casualties makes neither the public nor the leadership significantly more interested in negotiating, as was the case for the U.S. during the Korean War from 1951 to 1953. Sometimes, public opinion is sufficiently unformed that an elected leader has the ability to push public opinion away from considering concessions, as Churchill was able to do in 1940 with his public addresses, some of the most famous and effective wartime speeches in history.4 Sometimes, rising casualties may make democratic publics more highly motivated to grant the adversary concessions in order to end the war, but leaders may be willing to swim against the tide and fight on regardless. Roosevelt shook off a few specks of public doubt in early 1942. More dangerously, in the summer of 1864, continuing the Civil War was so unpopular in the North that Lincoln accepted that it would cost him the November presidential election, but he chose to fight on and probably lose office rather than negotiate in an attempt to save his reelection bid.
Even the Vietnam War, although not examined here, has limitations as an example of a war in which mounting casualties forced a democracy to make concessions. There is strong evidence that mounting casualties did erode American public support for the war, and encouraged President Johnson not to run for reelection in 1968.5 However, while this mounting opposition eventually affected American military strategy, especially President Nixon’s policy of Vietnamization and the gradual withdrawal of American troops, it did not cause the United States to offer a stream of concessions as casualties increased and support decreased. The first major American concession did not come until 1970, some five years after the major American ground troop escalation and some two years after the Tet Offensive.6 This surprising pattern points to a puzzle: If escalating casualties do not pressure democracies to make concessions, why are wars involving democracies shorter? One answer to this puzzle may be that democracies are especially likely to initiate wars that promise to be short, but once war starts their war-termination behavior will converge with the behavior of autocratic states.7
A small set of the cases also explored the domestic politics hypothesis that semirepressive, moderately exclusionary regimes like World War I Germany may raise their demands as the war evolves into a long losing effort. Chapters 9 and 10 examined two semirepressive, moderately exclusionary belligerents fighting long losing wars. Evidence supporting the hypothesized dynamic was quite limited. The first episode that offers at best weak support for the hypothesis was the German decision not to settle in the West in 1917–18 after the defeat of Russia, and instead launch a major offensive in March 1918. The historical particulars of this case are not consistent with the hypothesis, and the evidence that Germany fought on to improve its security by acquiring Belgium is at least as strong as the evidence for the domestic politics claim. The other case, Japan in World War II, provides almost no support. As things went from bad to worse from mid-1942 through summer 1945, Japan did not raise its war aims. It also refused to make diplomatic concessions to the Allies, although not because of a desire to retain the fruits of empire to reward coalition supporters, but rather because the Japanese believed that concessions should not be made after defeat because they would show weakness, and that the Allies were pushing for unconditional surrender anyway. Further, although Japan made no formal concessions until after August 1945, as early as 1943 it did decide to retreat to a smaller defense perimeter, thereby de facto abandoning goods that could be distributed to coalition supporters. Japan also formally offered material concessions to the USSR as incentive for them to mediate. If anything, the Emperor and others in the leadership saw a greater domestic political threat from public war fatigue than from bringing home insufficient gains in the war. Their fear that continuing the war would create grave internal instability was an important factor pushing them to accept a modified form of the Potsdam Declaration in August 1945.
The absence of strong evidence favoring this latter domestic politics claim in this small number of cases is important, given that the mixed regime hypothesis applies to only a small band of cases within the universe of interstate wars. The set of losing, semirepressive, moderately exclusionary regimes that fought long wars (such regimes fighting long losing wars is the category of belligerents the theory applies to) is relatively limited. The average duration of wars from 1816–1991 was about fourteen months, and of the more than 200 belligerents that fought more than ninety wars since 1816, only nine were mixed regimes fighting losing wars that lasted longer than fourteen months.8 Beyond the cases discussed here of Germany in World War I (Germany 1916–17, as discussed in chapter 9, providing more support for the Goemans hypothesis than Germany 1917–18) and Japan in World War II, it is difficult to find supportive evidence. For example, the evidence on Russia during World War I is both thin and mixed.9 In the Mexican–American War, the war-termination behavior of Mexico (a semi-exclusionary, moderately repressive regime fighting a long losing war) is inconsistent with the predictions of the theory. Mexican president Antonio Lopez Santa Anna did not increase Mexican war aims in the face of battlefield defeats and escalating losses. Indeed, he sought to make concessions and negotiate even before Mexico City had fallen. Aside from predicting war-termination behavior, the Goemans hypothesis would forecast that, following defeat in the war, Mexico’s primary wartime leader, Santa Anna, should have lost power and suffered severe personal punishment, such as exile, death, or imprisonment. On the surface, the case does seem to offer supporting evidence, since after the war Santa Anna did flee to Venezuela in exile. However, the particulars of his fate do not neatly follow the model. Santa Anna resigned the presidency voluntarily, he did so while the war was ongoing (September 1847), and even after he resigned as president he kept his position as leader of Mexico’s army. The new president appointed by Santa Anna, Manuel Pena y Pena, did soon remove Santa Anna from his position as commander in chief, in preparation for a court martial. Santa Anna sought (and received) exile in Venezuela, although he may have been more concerned with personal threats from renegade Texas Rangers for revenge than with threats from angry Mexicans. Notably, any Mexican antipathy to Santa Anna was temporary, as Santa Anna became Mexican president again some five years after his flight to Venezuela.10 Lastly, it was Pena and not Santa Anna who signed the war-ending Treaty of Guadelupe Hidalgo the following year, and Pena himself was not thrown from power after signing the treaty.11 Pena voluntarily resigned, and resumed his seat on the Mexican Supreme Court. In short, the evidence supporting the mixed regime domestic politics hypothesis is thin.
Some suggestive evidence exists for a variant on this hypothesis, however. Various scholars have proposed that militaries have different preferences for fighting wars than do civilians—for example, that militaries prefer fighting for decisive rather than incomplete, negotiated victories.12 When there is weak civilian control of the military, then this may delay war termination, as the civilians may fear a military coup d’etat if they move too aggressively towards peace. This dynamic was demonstrated most clearly in the Japan case, where the Emperor and other civilians at times expressed fear of a military coup d’etat if the civilians moved to a peace deal that would threaten the Japanese polity. It was also seen in the Germany case, when Kaiser Wilhelm may have feared speaking out against the escalation to unrestricted submarine warfare in 1916–17 because of domestic political threats from the navy. In 1917–18 it is a bit more complicated. Since there was a military dictatorship of the Duo, it appears that both civilians and military officers supported continuing the war in the West in order to secure Belgium.
The Inefficiency of Combat
A general point from many of the cases is that combat is a relatively inefficient means of hastening war termination through information transmission. There are many reasons why combat does not efficiently provide information that in turn directly affects war-termination behavior. The fog of war makes combat outcomes often quite ambiguous, impeding the process by which expectations might converge sufficiently to permit war termination.13 Making matters worse, decision-makers are motivated to obscure true outcomes, as leaders of a state experiencing battle defeat may wish to underplay the magnitude of the loss so as not to encourage the adversary. For example, a week after the Japanese attack on Pearl Harbor, Assistant Secretary of State Breckenridge Long wrote in his diary, “The real damage at Honolulu has not been disclosed. Apparently Japan does not accurately know. To advise her would be foolish—as well as dangerous.” Reciprocally, the Japanese kept silent about the damage inflicted by the April 1942 Doolittle raid on Tokyo.14 Laboratory experiments have revealed that expectations are less likely to converge when subjects are exposed to more ambiguous information than when exposed to information with a more straightforward interpretation.15
Even when combat outcomes clearly indicate success for one side and setbacks for the other, such information still does not always encourage shifts in war-termination behavior. Belligerents such as the Confederacy in early 1865, Britain in 1940, the U.S. in 1942, and the U.S. in 1950 all refused to lower their war-termination demands even in the face of serious combat losses. The case study findings are consistent with evidence from quantitative scholarship on war duration, that wars do not become more likely to end the longer they last—that is, the accumulation of information from ongoing combat does not steadily increase the chance of war termination.16 What accounts for this inefficiency?
The cases here provide three possible explanations. First, as described in chapter 3, fears of adversarial noncompliance with war-ending agreements sometimes moot the connection between information and war-termination behavior. Although fighting on looks decreasingly attractive, the alternative of seeking a limited war outcome deal is even less attractive, as it allows the adversary to reattack in the future.
A second source of the inefficiency of the combat-information/war-termination connection is the fear decision-makers sometimes have that making concessions will send signals of weakness. Decision-makers fear communicating weakness especially during war, and may want to wait until they have demonstrated strength on the battlefield before risking any concessions or even negotiations. This dynamic reverses the prediction forecast by the information proposition, as it posits that setbacks ought to make a decision-maker less likely rather than more likely to make concessions, again disrupting the proposed combat-information/war-termination dynamic of converging expectations and eventual war termination. Among the cases described here, Japan in World War II perhaps best illustrates this negotiation-from-strength dynamic.
Third, leaders can be patient. They are often unwilling to draw conclusive inferences from individual battles about the balance of power. Even when they admit to themselves that things are going poorly they are reluctant to take the next step and offer concessions. This patience likely emerges from a number of interrelated factors. Leaders recognize there is a large variance around their estimates of the balance of power, and they are often unlikely to change their beliefs too drastically after a single battle.17 Some leaders, such as Roosevelt in 1942, have the luxury of knowing that the war will be a long one, and that the true balance of power will not be revealed until military mobilization has been completed, some months or years in the future. Part of the unwillingness to offer concessions may be confidence, if not overconfidence.18 Overconfident leaders may view battle outcomes with bias, playing down defeats and placing too much emphasis on victories. In particular, some losing leaders assume that nonmaterial factors will eventually allow their side to surmount overwhelming material odds. Examples of such (over)confidence include Churchill trusting that British moral fortitude will eventually win out, and putting the best possible spin on the aerial balance-of-power numbers; the Japanese leadership trusting that Japanese spiritual superiority will somehow trump American material power; and Davis’ belief that the fighting elan of the Southern man will triumph over the factories, mills, and foundries of the Union.
Variation in Commitment Fears
The cases provide somewhat less traction in understanding consequential variance in belligerents’ fears about commitment credibility. Regime type provides little purchase. Some belligerents were unwilling to accept a limited war outcome with autocratic adversaries (such as Britain and the U.S. against the Axis powers in World War II and the U.S. in the first portion of the Korean War), and other belligerents were more willing to accept a limited peace with autocratic adversaries (such as the Soviet Union against Germany in 1941, Finland against the Soviet Union, and the U.S. in the latter stages of the Korean War). Some belligerents were more willing to accept a limited war outcome with democratic adversaries (such as North Korea/China in the latter stages of the Korean War, and the Soviet Union in the Winter War and Continuation War), and other belligerents were less trusting of democratic adversaries to adhere to war-ending commitments (such as the CSA during the Civil War and the Soviet Union in the first stage of the Winter War).
Perhaps the single most important factor affecting variation in commitment fears is the past behavior of the adversary. Churchill and Chamberlain recalled Hitler’s treachery at Munich. Roosevelt was haunted by the dashed hopes of Versailles. Lincoln knew the Southern penchant for wielding the secession threat. In the later years of the Civil War, Davis looked back to the Emancipation Proclamation as a clear retreat by Lincoln on his most important prewar promise. Truman saw the invasion of South Korea as a repeat of the dark fascist theater of the 1930s. And yet, there are episodes in which the lines from historical experience to wartermination behavior are not as straight. Stalin knew all too well of Finnish efforts to overthrow Soviet Bolshevism in 1918 and 1941, and yet in 1940 and 1944 was willing to accept a limited outcome because of countervailing factors. In 1941, Stalin looked past the obvious lessons provided by Hitler’s abrogation of the 1939 nonaggression pact, and instead grasped for the straws of Brest-Litovsk, hoping against hope that major territorial concessions might save Soviet socialism. In 1951, Truman swallowed his fears of walking in Chamberlain’s footsteps when fears of escalation pushed him and Dean Acheson to scale back their war aims in Korea.
Commitment Credibility and International Relations
The findings have broad implications for international relations more generally. Both information and commitment dynamics are fundamental and essential dynamics driving war behavior. The conventional wisdom is correct that information does matter, but this book found that wars are not about information all the way down. Fears of commitment noncompliance have long been understood to affect how wars start, and this book has shown that they also affect how wars end.
Other areas of international relations have also emphasized the importance of credible commitments. Scholarship on international institutions has explored factors affecting the likelihood of state compliance with institutional dictates such as institutional design, domestic politics, interdependence, and others. This book has developed a new factor that can reduce the likelihood that an international commitment will be broken: absolute war outcomes. In fighting war, states can brutally solve the noncompliance problem by eliminating the other side, or at least eliminating the other side’s ability to make a choice. Note that this is a solution to the commitment noncompliance problem inapplicable to international institutions covering areas such as trade, finance, or the environment, since one state would be unlikely to consider conquering another state to prevent it from erecting illegal trade barriers, for example.
IMPLICATIONS FOR AMERICAN FOREIGN POLICY IN THE TWENTY-FIRST CENTURY
Many of the theoretical issues discussed in this book are centrally important to American foreign policy in the twenty-first century. How should the United States think about ending its wars? How can the U.S. achieve its foreign policy aims while paying as few costs in blood and treasure as possible?
One policy conclusion might be that the most effective conflict resolution policy tools are those that can reduce credible commitment fears. However, the empirical evidence on the effectiveness of these foreign policy tools is mixed. One study of all interstate wars from 1914–2001 found that tools that seem to reduce commitment fears, such as peacekeeping forces, arms control, and confidence building measures, did not significantly increase the duration of peace after war—that is, they did not reduce the chances of a belligerent breaking a peace settlement.19 Conversely, there is evidence that such tools can help resolve intrastate conflicts.20
The 9/11 attacks have increased the concerns of many American foreign-policy makers that America’s most worrisome foes—rogue states and terrorist groups—cannot be bargained with, coerced, or frightened. These foes are seen as being willing to engage in gigantically costly, even suicidal operations, because authoritarian leaders and terrorists have no concerns about civilian casualties, and they are sufficiently irrational or fundamentalist to be willing to accept even their own deaths. This makes war termination difficult, both because one might never be able to impose enough costs on such actors to persuade them to negotiate, and because one can not rely on such actors to adhere to war-ending agreements.
The George W. Bush administration framed post-9/11 conflicts in this manner, and laid out a foreign policy to address these new threats. The new foreign policy was presented in the September 2002 National Security Strategy (NSS), also known as the Bush Doctrine. It argued that deterrence would not work against leaders of rogue states and terrorist groups, and as a result the United States may need under some circumstances to strike preemptively or preventively to eliminate emerging threats.21 Iraq under Saddam Hussein served as the leading example for the Bush administration of a state undeterrable because of its leader’s callous disregard of civilian casualties, and of a state that would not abide by its international commitments, as evidenced by its apparent disregard in the 1990s of United Nations resolutions calling for dismantling its WMD and missile programs. Iran and North Korea, the other two members of the “axis of evil,” were also seen as authoritarian states willing to disregard international commitments such as those required by their signatures to the Nuclear Non-Proliferation Treaty.
The NSS is well explained by the theoretical framework developed here. The underlying assumption of the NSS was that policy toward rogue actors should not be understood as an information problem. Because rogues are willing to accept huge costs and even their own demise, the revelation of information to rogues will not work. That is, even if America could reveal enough information to convince the rogue that America has both the ability and the willingness to destroy the rogue, such revelation would be insufficient to deter the rogue, because the rogue is willing to risk suicide.
The NSS instead framed rogue actors as constituting a commitment problem. At the simplest level, rogue actors cannot be counted on to adhere to international agreements or accepted norms of behavior. War, and especially preventive war, was seen as a potential solution to the commitment problem that rogue actors present. Two categories of preventive attacks have been suggested as possible solutions to the commitment problems of rogue actors: more limited attacks aimed at destroying WMD or the scientific/industrial infrastructure used to produce such weapons, or more comprehensive attacks aimed at overthrowing the rogue leadership and imposing regime change. In the category of limited attacks, the idea is that if such a limited attack can destroy a state’s WMD capability, then the commitment problem is solved by preventing the rogue from having the military power to consider violation of any non-proliferation commitments. In some sense, it belongs to the category of wars in which acquiring an increment of the good, in this case physically destroying a state’s WMD program, helps reduce the commitment problem. In the category of regime change attacks, the commitment problem is solved more decisively through absolute war outcome.
A central question for American foreign policy is: Should preventive war be used to solve credible commitment problems? If so, under what conditions? Again, answering these questions has critical importance for American foreign policy in the twenty-first century. The U.S. will face dangerous adversaries in the years to come, including North Korea, Syria, Iran, and others, who are feared to likely renege on their international commitments. The use of preventive force is one possible solution.
A first question is whether or not the use of force works as a means of solving commitment problems. The historical record provides tentative support for the idea that at least absolute war can solve credible commitment problems. In interstate wars since 1816, when a defeated state has suffered foreign-imposed regime change or state death, it has almost never reattacked the victor.22 However, for the U.S. in the twenty-first century, absolute war will mean imposed democratization, which is a difficult strategy to execute successfully. Sometimes, the U.S. has successfully imposed democracy in the wake of absolute military defeat, such as in Japan and Germany after World War II, in Grenada after the 1983 U.S. invasion, and in Panama after the 1989 invasion. Such efforts are not always successful, however, as demonstrated by a number of other American interventions in Latin America in the twentieth century, especially before 1945. More systematic analyses have indicated that the use of force is not a reliable tool to spread democracy.23 At this writing (September 2008), it is not clear that democracy will flourish and survive in Afghanistan and Iraq following the 2001 and 2003 interventions in each of those nations. However, even if democracy fails to emerge, neither state is likely to pose an interstate threat to the United States in the short or medium term, because each war smashed the power of the Taliban and Saddam Hussein governments, and that power is not quickly or easily rebuilt. Chaos or civil war might ensue, but undesirable as these outcomes might be, they would not permit the reemergence of an interstate threat.24
That being said, pursuing absolute war as a foreign policy tool has two major problems. First, such wars can bear very high costs. Most simply, wars inflict human costs, both deaths and nonfatal injuries, on both the attacking power and on the civilians in the target state. The American military death toll from the Iraq War is in the thousands, with nonfatal injuries in the tens of thousands. At least several tens of thousands of Iraqi civilians have been killed in the violence since the outbreak of war in 2003. Sometimes human costs must be accepted in the pursuit of the protection of the nation, but that does not mean that those costs should be forgotten.
Such wars are also financially costly, potentially running tabs into the hundreds of billions or trillions of dollars. Guns versus butter concerns aside, there are many spending priorities into which those funds could be poured to improve American security against threats from WMDs and terrorism in particular, such as more aggressive acquisition of fissile materials from the former Soviet Union, improved port security at American and foreign ports, more widespread radiation detectors (especially at airports and border crossings), missile defense, foreign aid to attract support in the global war on terror and to stabilize friendly regimes, and others.
These wars also damage the military itself. The American military will most likely remain an all-volunteer force in the short and medium term since neither the Pentagon nor the American public desire a return to conscription. Because of this, very real limits exist on the total size of the American military, and fighting wars such as the Iraq War strain American military capacity, both because forces tied down in one operation cannot be used elsewhere, and because fighting such operations makes recruitment more difficult, having the effect of threatening the quality and quantity of available recruits.25
Perhaps most worrisomely, wars with the aim of regime change can have severe diplomatic costs for the United States. At the minimum, the exercise of American power to overthrow a foreign leader makes states uncomfortable, as it violates national sovereignty norms. In the application, such wars are invariably messy, drawing American military forces into behaviors that tarnish the American image abroad, including killing civilians and committing war crimes. The hope is that the world community would recognize that such wars have humanitarian ends, the overthrow of a dictator and the installation of democratic regimes, but the reality is that such actions are often seen in the worst light. America’s actions are not given fair comparison to actions taken by other regimes, and America does not get credit for its successes. The 2003 Iraq War caused anti-Americanism around the world to soar, cementing America’s image as a new imperialist power bent on killing Muslims and savaging the Muslim world. This despite the overthrow of Saddam Hussein who killed more Muslims than any other individual in history, America’s use of force in the 1990s to save Muslim Kosovars and Bosnians from marauding Serb forces, the substantial amount of economic aid given by the United States to Muslim regimes such as Turkey, Egypt, and Pakistan, the wave of disaster relief following the 2005 Pakistan earthquake, and so forth. These concerns are not new to twenty-first century American foreign policy. Indeed, one reason Britain elected not to overthrow the Turkish Sultan after World War I was because of its concerns that doing so would spark a worldwide anti-British backlash, especially among Muslims in the British colony of India.26
Anti-Americanism is not merely the loss of a popularity contest, it is a direct threat to American national security. The sources of terrorism are poorly understood, and although experts are divided on the effects of factors such as economic conditions, political institutions, and religious ideology, there is broader consensus that anti-Americanism stimulates terrorist recruitment and action. If terrorism, and WMD terrorism in particular, constitutes one of (if not the greatest) threats to American national security, the surge of global anti-Americanism must be viewed with great concern.
If the first major drawback to such wars is high costs, a second major drawback is that the benefits of such wars as compared to doing nothing and leaving the rogue state in place may be overstated. Sometimes, fears of a rogue state’s WMD capabilities may be exaggerated. During World War II, Germany was far less advanced in its nuclear weapons program then the United States feared. Iraq’s WMD capabilities were substantially less advanced at the time of the 2003 attack than was presumed by the Bush administration. The 2005 American National Intelligence Estimate (NIE) declared that Iran was pursuing a nuclear arsenal, but the 2007 NIE reversed course, proposing that Iran had abandoned its nuclear weapons program in 2003.27
Further, even when a state does possess WMD, there are other means of addressing the WMD threat than attacks intended to impose regime change. Deterrence of WMD attack has been an extremely successful policy tool. The United States has deterred a number of WMD-armed, antiAmerican dictators, such as Joseph Stalin, Nikita Khrushchev, Leonid Brezhnev, Mao Tse-tung, Saddam Hussein, and Kim Jong Il from launching WMD attacks against the U.S. or its allies. Israel likely deterred Saddam Hussein from launching biological or chemical weapons against Israeli cities during the 1991 Gulf War. Deterrence can likely succeed in the future as a means of preventing the state-to-state use of WMD. Deterrence aside, diplomacy has been able to ameliorate and even eliminate a number of WMD programs and threats. The existence of the NPT helped persuade nations such as Argentina, Brazil, and South Africa to abandon their nuclear programs. A combination of carrots and (economic) sticks pushed Libya to declare and abandon its WMD programs in 2004. Although North Korea likely violated aspects of the 1995 Agreed Framework, the agreement likely did constrain the size of the North Korean arsenal, and in 2007 it signed a follow-up agreement with the Bush Administration. And the arms control process has substantially reduced the Soviet/Russian nuclear threat since the late 1980s.
Beyond wars aimed to change a rogue’s leadership, limited wars can be used to solve credible commitment problems. However, the empirical record is less encouraging of the conclusion that limited wars can serve this function, even if one side can change the balance of power or capture an element of the good that can make future defection by the other side more costly or less likely to succeed. Among the cases in this book, the Soviet Union hoped that its territorial acquisitions following the Winter War would improve the security of its northwest region. These hopes were not met, as after the war Finland permitted Germany to base troops on Finnish soil, and Finland joined Germany’s June 1941 attack on the Soviet Union. Israel hoped that its acquisition of the West Bank, the Golan Heights, and the Sinai peninsula in the 1967 Six-Day War would dissuade its Arab neighbors from attacking again in the future, although Egypt initiated the War of Attrition in 1969, Egypt and Syria both attacked to initiate the 1973 Yom Kippur War, and in 2006 Hezbollah launched missile attacks on Israeli territory from beyond the Golan, in Lebanon.
Another limited war strategy, more pertinent to twenty-first-century American foreign policy, and which can reduce commitment credibility problems, is to degrade directly the adversary’s military capabilities, in particular the adversary’s WMD capabilities. This is often seen as an attractive approach to addressing WMD threats, because rather than launching a more general war intended to overthrow a rogue leader and climb the long steep hill to democracy, one can launch air or missile strikes against WMD facilities. Such attacks are especially attractive because they are seen to have very low human costs, either of friendly forces or of collateral damage in the target state, they impose less strain on military resources, and the international backlash would likely be less. The 1981 Israeli attack on the Iraqi reactor at Osiraq is often pointed to as a leading example of how successful such attacks can be, and American decisionmakers have often considered such attacks against Iran and North Korea in particular. The Bush Administration even considered the possible use of nuclear weapons in carrying out attacks against WMD targets.28
The principle shortcoming of such limited attacks against WMD facilities is that they are unlikely to ameliorate the WMD threat. The historical record is rather discouraging.29 The Osiraq attack destroyed a reactor unlikely to have made much contribution to a weapons program, because of the design of the reactor, the presence of international inspectors, and the Iraqi dependence on the French supply of reactor fuel. Further, the attack probably accelerated the Iraqi weapons program, by increasing Saddam’s material commitment to acquiring nuclear weapons, and by driving it underground.30
The record of other attacks is similarly unimpressive. Some limited attacks, such as airstrikes on Iraqi WMD facilities in 1991, inflict relatively little damage, because the bulk of the targets are hidden and therefore untargeted. Dispersal and concealment is a direct reaction to the threat of airstrikes. The 1981 attack caused Iraq to disperse and conceal their WMD production and storage facilities. Other states such as Iran and North Korea have learned the same lesson, and an important reason why the United States elected not to attack North Korean nuclear facilities in the mid-1990s or later was because of the fear that there were likely many secret sites that would be undamaged in any air attack. Some limited attacks were launched against programs that either likely did not exist, such as the Clinton administration’s 1998 missile attacks against alleged chemical weapons facilities in Sudan, or were ultimately unlikely to succeed anyway, such as the bombing raids against the German nuclear weapons program in World War II. Perhaps the most successful such limited attack against a WMD program were Iraqi airstrikes against the Iranian Bushehr nuclear reactor in the 1980s, which may have delayed the Iranian nuclear weapons program, although information on this episode is scant. Similarly, little is known about the context or effects of Israel’s 2007 attack on a Syrian facility thought to be part of a nuclear program.
Another important critique is that just as the success rate of launching such limited attacks is underwhelming, the success rate of not launching such attacks is quite impressive, in that no state that considered launching such an attack but elected not to ultimately regretted that decision, especially given that some preventive attacks were nuclear in nature and would have incurred massive casualties and perhaps nuclear retaliation. The United States considered launching preventive attacks against the nuclear programs of the Soviet Union in the 1940s and 1950s, against China in the 1960s, and against North Korea in the 1990s. In each case, hindsight has proven that restraint was the wiser option. The Soviet Union never used its nuclear arsenal against the United States, and never used it politically to extract substantial diplomatic gains at American expense. Further, the Soviet Union eventually crumbled, and the Soviet/Russian nuclear arsenal was substantially scaled back. China neither used its nuclear arsenal nor attacked any American ally after it conducted its first nuclear test in 1964, and it is now evolving into a peaceful commerceoriented competitor rather than a militarized major power rival. North Korea at this writing is still a militarized nuclear state ruled by an antiAmerican dictator, but it has not launched any interstate aggression since preventive attacks were first considered in the early 1990s. The 2007 nuclear agreement gives hope that diplomacy may yet yield progress.
Other states have benefited from forgoing preventive WMD attacks. Egypt (and, some argue, the Soviet Union) considered launching preventive airstrikes against the Israeli nuclear reactor at Dimona in the 1960s.31 Although Israel and Egypt fought three wars in the 1967–73 period, all were conventional, and in the late 1970s the Camp David peace process achieved for Egypt tremendous gains, including peace with Israel, the return of the Sinai peninsula, and a lucrative alliance with the United States. These gains were all eventually realized without preventive nuclear attacks, and of course Israel has not launched nuclear attacks against Egypt in the interim. The Soviet Union considered launching preventive attacks against the Chinese nuclear program in the late 1960s, and since then not only have those two countries avoided military conflict, but they also have achieved a substantial and mutually beneficial rapprochement. India and Pakistan have both considered launching preventive attacks against each other’s nuclear programs, but neither state has ever used its nuclear arsenal against the other, and the two states have enjoyed a peaceful and slowly warming relationship since 1999.32
In short, war to solve commitment problems has not been and would not be a panacea. American foreign-policy makers in the twenty-first century need to recognize the limits and costs of using wars in this manner. Other foreign policy tools such as diplomacy and deterrence may be less costly and more effective.