The causes of war

The essence of the bargaining model as applied to the causes of war is that states experience disputes over the settlement of some issue, and war is one means by which states can achieve (or maintain) a better settlement. I structure my discussion around Fearon's seminal paper, in which he shows how a bargaining (or “rationalist”) model explains the outbreak of war. Fearon uses a bargaining model to argue that if two states in dispute know the outcome of a possible war, they should in general prefer to reach a deal that would reflect the hypothetical postwar political settlement, rather than fight, reach that same settlement, and also suffer the costs of war. This view assumes that fighting itself is costly—that the belligerent always suffers some negative utility, no matter how the issue at stake is settled. In fact, this assumption becomes necessary when one asserts that two states would always prefer to reach a bargain without fighting rather than fight and then reach the same bargain.¹⁷

Within his bargaining model, Fearon develops three conditions under which war is possible. First, there may be disagreement between the two sides as to the likely outcome of a war. If at least one side overestimates its chances of winning or underestimates the opponent's resolve, it may not raise its settlement offer enough to make the other side choose the offer over war. This lack of bargaining space stops a war-avoiding compromise.¹⁸

This proposition immediately presents a puzzle. If complete information can help states avoid fighting, and states prefer to avoid fighting, then why don't states completely reveal their capabilities and resolve to one another? States have an incentive to exaggerate their capabilities and resolve in an attempt to extract better offers. Costly, more-credible signals of resolve, such as the mobilization of troops, may help states better reveal capabilities to one another, albeit at the expense of raising the risk of accidental war.²¹

Fearon's second condition: war may occur because of an inability to commit not to fight in the future. If there are advantages to be had from striking first, then the outcome of the war differs depending on who attacks first, and the inability in anarchy for two sides to credibly commit not to attack may preclude a settlement. Additionally, if one side grows in power, then the outcome of war may change in the future, meaning that a bargain that is sufficiently war-avoiding now may not be when the balance changes. The inability of the growing side to commit not to attack in the future may undermine the ability of the two states to reach an accommodation.²³

Fearon's third condition: bargaining may not avoid war if the item under dispute is indivisible. The all-or-nothing status²⁴

The conduct and termination of war

Here, I lay out how the bargaining model views the conduct and termination of war. A preliminary task to understanding the conduct of war and its relation to bargaining is to define combat, the actual activity of war. Combat is a violent clash between at least two politically distinct groups organized to wield force. War consists of sustained and substantial episodes of combat. Clausewitz saw episodes of combat as the building blocks of war, pointing out that “[c]ountless duels go to make up a war,” and even more tersely that “[e]ssentially war is fighting.”²⁶

An armed force engaged in combat tries to accomplish one or more of three immediate tasks: the destruction of military forces, the occupation of territory, and the destruction of civilian assets. Destruction of military forces means just that: the killing, disabling, or capture of military personnel, and the destruction, disabling, or capture of military equipment. Territory is occupied when one can control movement into, out of, and through it. Destruction of civilian assets includes killing and injuring noncombatants and/or destroying and damaging material items, such as the means of production. Importantly, capturing territory and destroying forces can be both means and ends. An army might destroy a defending force in order to capture a piece of territory. Alternatively, capturing territory might allow the destruction of enemy forces—for example, if seizing high ground provides fire control of an area.

The bargaining model proposes that military means are used as part of the bargaining process, to advance political ends. There are a variety of ideas about exactly what role combat plays within war (specifically, how it relates to bargaining). A common theme within the newer bargaining-model scholarship is acceptance of Clausewitz's general distinction between limited (or “real”) and absolute war.²⁷

I identify here two separate means by which combat can accomplish political goals within the bargaining model of war. First, combat can seek total conquest, or the utter destruction of the enemy's means to resist, in order to achieve victory in a Clausewitzian absolute war. Second, combat can reduce uncertainty about the capabilities or resolve of the combatants.

The utter destruction of the enemy's means to resist is the outcome of absolute war, and it is of course accomplished through the bloody reality of combat. After obliterating the adversary's military capabilities, the victor has the luxury of establishing whatever political settlement it likes. Alastair Smith and Allan Stam offer a straightforward model of this perspective, in which states seek to occupy forts and the war ends in absolute victory when one side captures all of the forts. Darren Filson and Suzanne Werner describe a somewhat similar dynamic, in which losers suffer a greater loss of resources following a battle than do winners—a relevant imbalance, because a state must stop fighting and lose decisively if its resources reach a certain minimum. Hence, accumulating a series of battlefield successes can accomplish decisive victory by depleting the opponent's resources. Both the Smith/Stam and Filson/Werner models also allow that as a decisive outcome approaches—that is, as one side controls fewer and fewer forts (Smith/Stam) or as one side's resources dip close to their minimum (Filson/Werner)—the losing side may be motivated to offer a negotiated deal acceptable to the winning side, to avoid a decisive defeat. In Robert Powell's costly-processes model, in contrast, each side faces some risk of collapse after each battle, and battles that do not end in collapse do not provide information to the two sides about what the chances of collapse are.³⁰

How exactly could combat hasten a decisive defeat in this manner? If it aims to occupy territory, then the accumulation of battle victories means the progressive conquest of more and more territory. While the conquest of this territory may not affect the likelihood of the outcome of future battles, it may bring the invader closer to decisive territorial capture and result in absolute victory. The most prominent means of decisive territorial capture is seizure of the enemy's capital city, which in some cases (including many in World War II) has meant the total collapse of an enemy's ability to resist, because the capital is the seat of so much administrative and economic power. However, capture of the capital does not always mean decisive defeat. During the War of 1812, for example, British forces sacked Washington without forcing American capitulation; and during World War II, the Soviet Union might have kept fighting after the capture of Moscow, since in the fall of 1941 the civil government was evacuated from Moscow to the more secure city of Kuibyshev.³¹

Destruction of military forces and capture of economic capacity might achieve the same goal as capturing the capital. A series of battles could conceivably grind down a belligerent's armed forces to essentially nothing, allowing a decisive victory by literally destroying the enemy's capacity to resist. And attacks against a nation's economic capacity might eventually do enough damage to undermine its war production completely. This is sometimes the ultimate aim of counterindustrial bombardment, although it has never proven successful to this degree. The United States came close to wrecking Japan's industrial capacity in World War II through aerial bombardment and the destruction of Japan's merchant marine.³²

In short, this is an apolitical notion of combat's role in advancing national aims: brute force is used to crush the enemy's means of resistance. Some have argued that this conception of combat as brute force can also be applied to the accomplishment of limited aims. According to this perspective, war occurs when one side has completely lost faith in the chances of striking a bargain to settle the dispute and thus resorts to violence to impose a change. Brute force is used to destroy part of the adversary's military assets or to seize a piece of territory. Schelling discussed the possibility of accomplishing limited aims through physical force in his distinction between using violence as brute force and using it as coercion.³³

A more central role of combat in the bargaining model of war is the reduction of uncertainty. As discussed, a common bargainingmodel explanation of the outbreak of war is uncertainty about the outcome of a hypothetical war. Combat can reduce uncertainty by providing information about the actual balance of power. Or as some have put it, combat in a limited (or “real”) war reveals information about what the outcome of an absolute war would be.³⁴

Combat might also reduce uncertainty about resolve and about military effectiveness. A defender that takes casualties in a battle may convey resolve to the other side. For example, Japanese troops inflicted heavy losses on invading American forces during the 1945 Iwo Jima and Okinawa campaigns. This caused the United States to update its beliefs and increase its casualty estimates for an invasion of Japan, which in turn strengthened the case for dropping the atomic bomb as a means of making such an invasion unnecessary.³⁸

Some bargaining models allow for states to update their beliefs about each side's ability to absorb and/or inflict costs.³⁹

How are costs actually inflicted in combat? Note first that the focus here is on pain for its own sake—that is, costs unrelated to the balance of power. Combat can inflict costs in any of the ways described above. Capturing territory inflicts costs, both because territory has intrinsic value to nations and because it may provide economic benefits. Destroying military forces can inflict pain even aside from the question of the balance of power if it undermines the ability of the leadership to stay in power, for example, or if the leadership desires to use its military forces in another theater. Destroying civilians or civilian assets also means inflicting costs, although research on aerial bombardment indicates that bombing civilian assets is generally unsuccessful as a tool of coercion, despite the fact that bombing military assets can work.⁴²

Notably, there are differences among the newer, formal wartime-bargaining models. For example, Wagner sees combat as reducing uncertainty over who would win the eventual absolute war, whereas Filson/Werner and Smith/Stam see combat as reducing disagreement over who would win forthcoming individual battles. Both Powell and Filson/Werner allow a belligerent to update its beliefs on the basis of negotiations, because if an adversary keeps fighting without making concessions, the belligerent updates its beliefs about the adversary; Wagner and Smith/Stam, however, do not. Powell's model permits a focus on resolve that many of the other models do not. Smith/Stam relaxes the standard game theory assumption employed by the other models that actors share common prior beliefs.⁴³

The consequences of war

Lastly, the bargaining model helps explain the consequences of war, specifically the stability of a postwar peace. Two important hypotheses have emerged from different versions of the bargaining model. Werner lays out the bargaining-model claim that wars are about the revelation of information about power and capabilities, and that the end of war creates a readjustment of goods consistent with a new understanding of the distribution of power. Her proposition is that the more the power balance between two former belligerents changes in the years following war, the more likely war becomes, as the two states will resort to war to realign the distribution of goods to match the changing balance of power.⁴⁴

Smith and Stam produce a second argument. They propose that the more battles are fought, the more information is revealed, causing the expectations of the two sides to converge. Additionally, they posit that the greater the convergence of expectations about capabilities, the more stable will be the postwar peace. This argument provides a stronger theoretical basis for the long-standing war-weariness hypothesis, the speculation that peace lasts longer after long wars than after short wars.⁴⁵

The empirical record

There are islands of empirical support for the bargaining model, with studies finding, for example, that (1) dyadic power parity produces uncertainty that makes war more likely, (2) economic interdependence reduces uncertainty, which reduces conflict, (3) changes in the postwar balance of power can make the recurrence of war more likely (though this is a disputed finding), (4) anticipated conflict costs affect prewar bargaining demands, (5) war settlements reflect belligerents' expectations about future costs and probabilities of winning, (6) more intense wars are likely to be followed by longer periods of peace, and (7) semirepressive, moderately exclusionary regimes fight longer wars.⁴⁶

The existing empirical scholarship does raise questions for the bargaining approach, however. The finding that wars are not duration dependent (meaning that they are neither more nor less likely to end the longer they last) presents a puzzle. If war is caused by uncertainty and combat reduces uncertainty by revealing the true capabilities of both sides, then wars ought to become increasingly likely to end as they endure, as fighting battles continually reduces uncertainty until there is sufficient agreement on capabilities to create bargaining space and to permit war termination.⁴⁹

More centrally, however, the core of the bargaining model remains to be tested. This will call for examination of three factors: estimates of capabilities, estimates of resolve, and the exchange of offers between two sides. A critical task is collecting data on battles fought during war. Testing intrawar hypotheses of the bargaining model requires linking changes in negotiations with changes in each side's estimates of the war's likely outcome, which in turn are generally determined by battlefield events. The quantitative approach requires collecting data on individual battles within wars and keeping track of wins and losses. There exists one data set for battles, devised by the Historical Evaluation and Research Organization, but its inconsistent coverage of battles across wars makes it inappropriate for this application.⁵¹

In addition to quantitative analysis, case studies of individual wars will be necessary to test the central ideas of the bargaining model. Case studies are likely to do better than quantitative analyses at following the exchange of offers, noting in particular how one side might increase or decrease its settlement offer to the other side in reaction to battle outcomes and changing domestic political conditions. Hein Goemans provides one such case study in his examination of bargaining during the First World War, with a focus on the German offense in March 1918 as a “gamble for resurrection.”⁵²

Old wine in new bottles?

Does the newest formal work offer important advances over the nonformal work of the 1950s through the 1970s? One criticism might be that nearly all of the new hypotheses can be found somewhere in works by Blainey, Schelling, Kecskemeti, Iklé, or Pillar. This is a general criticism of formal work, that it (especially scholarship on war) amounts to old wine in fancier bottles.⁵⁴

Second, and relatedly, formalization helps the theoretical development of the model. Even though the very basic insights have been developed by earlier, nonformal scholars, formal approaches have explored and expanded the model, searched for potential contradictions or inconsistencies, opened new channels, and developed new hypotheses. To name a few contributions produced thus far: Formal analysis helped to develop—and to provide a solution for—the puzzle that fighting might not lead to the creation of bargaining space because the winning state may simply increase its aims.⁵⁵

Third, the new formal approaches offer, for the first time, the possibility of linking the causes, prosecution, termination, and consequences of war into a single theoretically consistent framework. The older work limits itself to one or two of these phases, as Schelling focuses mostly on causes; Kecskemeti, Iklé, and Pillar on termination; and Blainey on causes and a bit on consequences. Some of the newer models build frameworks that would permit seeing all phases as part of a bargaining process, an ambition that falls short of providing a grand unified theory of war but still offers a more comprehensive vision than previously existed.

Deterrence, the spiral model, and cognitive psychological biases

The bargaining model overlaps with deterrence theory. In deterrence theory, an actor (a defender) attempts to deter another state (a potential attacker) by promising to make a specific response if the second state takes some action. Deterrence succeeds when the potential attacker does not attack. Though there are many variations on deterrence theory, a basic proposition is that “[d]eterrence is predicted to succeed when the expected utility of using force is less than the expected utility of not using force.”⁵⁹

The bargaining model does make one important, different prediction. Classic deterrence theory predicts that as an imbalance of power between two states grows, war becomes more probable. The bargaining model, on the other hand, argues that it is not an imbalance of power that causes war, but rather disagreement over the balance of power (and over the likely outcome of war). From this emerges the prediction that balances of power might be more significantly associated with war than are imbalances of power, because there is more uncertainty about the outcome of an eventual war between two evenly matched states than about two unevenly matched states.⁶⁰

Another major theory of war is the spiral model. In contrast to the deterrence model, which argues that war occurs out of greed, the spiral model predicts that war occurs out of fear. Anarchy makes the state prepare military forces, which threaten its neighbors and push them to arm; the resultant arms race makes all sides less secure, a dynamic known as the security dilemma. The spiral model forecasts two principal paths to war. Preventive war occurs when one state perceives the balance of power shifting against it and attacks sooner rather than later, when conditions would be less favorable. Preemptive war occurs when one state attacks in order to forestall a perceived imminent attack.⁶¹

Both types of war fit into the bargaining model. A changing balance of power may introduce commitment problems that make a war-avoiding bargain difficult to attain, leading to preventive war. Or incomplete information about preferences in the context of a shifting balance of power may make a preventive war more likely.⁶²

The versions of the deterrence and spiral models laid out thus far have retained rationalist assumptions. Many scholars have built versions of these theories that incorporate nonrationalist assumptions and look instead to cognitive psychology. Some have argued that cognitive biases affect leaders' perceptions, so that leaders make attribution errors or maintain images of a hostile enemy even in the face of disconfirming evidence. Such biases may reinforce spirals of hostility by making defensively oriented moves look offensive and conflict more likely. Attribution errors cause states not to update their beliefs about the intentions of other states in patterns predicted by rationalist deterrence theory. Others have argued that motivated biases affect leaders' perceptions. For example, a leader may feel he faces domestic political pressures that demand an international victory in order for him or her to maintain power at home. The leader is then motivated to perceive that an international adversary can be defeated, and may then believe that victory is possible, even as the adversary becomes more powerful. In other words, the standard deterrence proposition that a more powerful defender makes deterrence success more likely does not hold.⁶⁴

The cognitive-psychological school would level at least one general critique of the bargaining model: that revelation of information about the opponent during peacetime ought not significantly reduce the likelihood of war breaking out, and that revelation of information about the opponent during wartime ought not significantly reduce the likelihood of war ending. In the prewar case, the bargaining model would propose that, ceteris paribus, as two sides reveal information about their capabilities and intentions, bargaining space will open up to permit a waravoiding bargain. Such revelations might come through increased transparency of societies (through democratization, for example), costly mobilization during international crises, the performance of armies in wars with third parties, and so forth. The cognitivepsychological school would doubt the ability of such revelations to preserve peace. Leaders' images of other countries as hostile or weak will likely persevere even in the face of credible evidence to the contrary. Further, if external factors such as domestic developments create an environment conducive to motivated bias, then leaders may develop beliefs that make war seem more appealing despite unfavorable military conditions. During wartime, similar dynamics may operate, precluding battle outcomes from causing leaders to update their beliefs about the resolve or capacity of the enemy. This is the view of some of the older war-termination literature, that cognitive and motivated biases are likely to swamp clear-headed thinking during war.⁶⁵

Conducting decisive, competitive tests of these two perspectives may be difficult, as the difference is one of degree rather than kind. That is, the bargaining model says that the revelations that emerge from mobilization or battles are more likely to cause war avoidance and war termination, whereas the psychological perspective argues that such revelations will probably be less effective, though not completely useless.⁶⁶

Organization theory and the conduct of war

Like cognitive psychology, organization theory is built on a set of assumptions that might be inconsistent with the unitary, rationalist assumptions of some versions of the bargaining model. Beliefs about preferences and capabilities in particular are filtered through organizations, frequently militaries. This might introduce systematic biases into the formation and evolution of beliefs. Some have argued, for example, that militaries tend to view adversaries as intractably hostile, implying perhaps that military intelligence might not recognize actions intended to signal benign intent.⁶⁸

The question of organizational rigidity is something of a double-edged sword for the bargaining model, however. Toorigid militaries may not recognize that they have overestimated their own capabilities and underestimated those of their adversary, so again, information may not be taken from battle outcomes to provide war-terminating bargaining space. Yet moreflexible militaries that recognize the underperformance of troops in the field may not be content to accept it; such militaries may be entrepreneurial, changing strategy and the application of technology so as to minimize disadvantages and maximize advantages as they appear.⁷¹

Formal bargaining-model frameworks tend not to account for entrepreneurship.⁷²

Entrepreneurship can also lead to rapid changes in military strategy, slowing the creation of bargaining space. The rapid German conquest of Poland should have encouraged the convergence of expectations between France and Germany about the likelihood of a future Franco-German war, perhaps enabling a war-avoiding settlement of some kind. However, the French drew lessons from the Polish campaign, hoping to do better than the Poles. France inferred that a German attack in the West would come through the Belgian plains. French General Maurice Gamelin commented: “The chief lesson to be learnt from the Polish campaign was the penetrative power of the speedy and hard-hitting German armoured formations …. Belgium was the more likely target because the Ardennes was not a good tank country and there were good defences in Luxembourg.” France confidently prepared for a decisive battle in Belgium. This strategic choice might have caused Germany to reconsider its attack plans and strike a war-avoiding deal with France, but Germany also did not sit on its hands. Germany devised a strategy to exploit French expectations: a feint into Belgium to lure Allied forces forward, coupled with a decisive thrust through the Ardennes to cut off the bulk of Allied forces and separate them from Paris.⁷⁷

How might variation in military entrepreneurship affect war outcomes? The straightforward argument is that entrepreneurship is positively correlated with victory. More adaptive militaries are more effective, maximizing their own strengths and their enemies' weaknesses. The bargaining model, however, would make two different predictions. First, militaries with higher levels of entrepreneurship should fight longer wars. If the capability of militaries changes from battle to battle, then this will slow the convergence of expectations necessary for war termination. Second, states confident in the high levels of entrepreneurship of their militaries may find it more difficult to avoid war through bargaining. Entrepreneurship makes militaries more effective, but this is relatively difficult to convey to adversaries (as opposed to the obvious advantage of possessing fleets of combat aircraft, for example). More generally, the bargaining model might simply incorporate entrepreneurship into its assumptions, accepting the simple premise that if states update their beliefs and settlement offers they ought also to update their military strategies— since, after all, strategy is strategic.

Organization theory might offer other critiques of the bargaining model beyond issues of the rigidity/entrepreneurship of militaries. Organizations use quantitative indicators to demonstrate success or failure: firms use profit statements, graduate programs use job placement rates, and so on. Existing versions of the bargaining model assume that two opposing militaries use the same metric to determine success and failure. For example, success might be measured by the acquisition of forts (as in the Smith/Stam model), so as one side gains forts from the other, the former experiences success and the latter experiences failure. However, in some conflicts, militaries may have different measures of success; two opposing sides could conceivably observe the same battle outcome with both concluding that they were successful, coming no closer to agreement on the eventual outcome of the war. This is one interpretation of the Vietnam War, as the United States sought to inflict North Vietnamese casualties (the infamous body count) and North Vietnam sought to inflict American casualties. The occurrence of combat and casualties on both sides caused each to conclude that it was doing well, perhaps delaying the termination of the war. Similarly, in 1812 Napoleon viewed the French capture of Moscow as a step toward victory, whereas Russian General Mikhail Kutuzov saw it as hastening French overextension and defeat, commenting when he ordered the retreat from Moscow that “Napoleon is a torrent which we are as yet unable to stem. Moscow will be the sponge that will suck him dry.” Note that the bargaining model is not completely incapable of accounting for organizations drawing inferences. The Smith/Stam relaxation of the noncommon-priors assumption permits two sides to draw different inferences from the same event.⁷⁸

A more common problem is when all actors within a single side do not agree on the actual metric of success—that is, no indicator is dominant. The army might focus on territory captured, the navy on enemy tonnage sunk, the air force on bombing sorties carried out, the foreign service on allies lined up, and so on.⁷⁹

Domestic politics and the formation of preferences

The role of domestic politics in world affairs has dominated the past decade of international-relations research. How can domestic political factors fit into the bargaining model? Most applications of the bargaining model have treated states as unitary actors. One model has incorporated domestic politics by distinguishing among regime types, generating an array of theoretical predictions that account for a variety of apparently contradictory empirical findings—for instance, that democracies win wars but are especially prone to being challenged militarily.⁸⁰

The bargaining model lends itself well to other applications of domestic politics. Within war, variations in regime type may encourage different bargaining strategies. If a belligerent perceives its democratic opponent as particularly vulnerable to costs, it may be encouraged to initiate a battle as a means of demonstrating its capacity to inflict costs—as was perhaps Iraq's intention when it launched a ground attack against coalition forces at Khafji in late January 1991, or as was probably North Vietnam's intention when it launched the Tet offensive in January 1968.⁸¹

Another cut on domestic politics and the bargaining model concerns the “gamble for resurrection.” A state losing a war may adopt a new strategy—a gamble for resurrection—that might increase the chances of winning at the expense of also increasing the chances of decisive defeat. In and of itself, the gamble is a useful conceptual step forward for the bargaining model, as it to some extent endogenizes military strategy by allowing states to choose between a strategy with higher mean and lower variance of outcomes and one with lower mean and higher variance. In terms of domestic politics, one might expect that leaders more fearful of suffering moderate defeat—such as those in semirepressive regimes, who might be executed or imprisoned as a result—to be more likely to gamble for resurrection than are other kinds of leaders, such as democratic or dictatorial leaders. Goemans describes the German offensives on the Western Front in the spring of 1918 as a gamble by the German leadership staking the captured conquests in the East to defeat France in the West and maintain its hold on power.⁸⁴

Domestic-politics theories can produce a deeper critique of the bargaining model. A fundamental assumption of the bargaining model is that fighting itself, divorced from the political issues at stake, is intrinsically costly. That is why the model sees war as a puzzle: both sides are better off striking a political settlement instead of fighting to reach that same political settlement, as each side's net utility must reflect the political benefit of the deal minus the costs of fighting.

But some domestic-politics theories have posited that leaders may prefer fighting over reaching a settlement peacefully. They may engage in conflict to divert public concern from internal problems, rallying citizens around the flag to increase support of the leadership. The evidence of such dynamics in the United States is generally weak.⁸⁵

Constructivism

Constructivism, a theory that emphasizes the importance of intersubjective social factors in determining critical phenomena such as identity and interests, claims that war is best understood as a social convention determined and shaped by norms and culture, not as a rationalist choice reflecting costs and benefits. The fundamental causes of war, then, are best explored in light of factors like the nature of international culture (i.e., whether it fosters Hobbesian anarchy or Rousseauvian cooperation) and more specific cultural factors, such as taboos on the use of nuclear weapons.⁸⁷

Constructivism's hypothesis that practice generates interests may also be outside the bargaining model. The bargaining model proposes that a state's preferences, especially regarding what constitutes acceptable terms of peace in relation to expectations about the course of future costs and battle outcomes, are stable across the war. Constructivism might propose that the practice of violence could polarize these preferences, pushing states to make more extreme demands as the war proceeds and an image of the enemy as an intractably hostile barbarian becomes solidified. One hypothesis might be that particularly bloody wars are more likely to have extreme terms of settlement, because states hold out longer for better terms. Arguably, the intensity of combat during World War II in the Pacific may have pushed the United States to demand unconditional surrender.⁸⁸

Constructivism also poses a deeper challenge to the bargaining model, again returning to the assumption that fighting is costly. Some strains of constructivism posit that fighting serves important social functions, principally the formation of group identity. The definition of self requires the definition of other, and, some argue, the essential nature of the relationship between self and other is competitive and ultimately conflictual.⁸⁹

Interestingly, some constructivist propositions may fit into the bargaining-model framework. The bargaining model predicts that war may become more likely if the issue at stake is indivisible. Some items in international politics may be physically divisible, but discourse and culture may make them practically indivisible. By describing the city of Jerusalem as practically if not literally indivisible, constructivism may help explain why the Arab-Israeli conflict seems intractable. It may similarly shed light on the Indo-Pakistani conflict, describing Kashmir as having characteristics of indivisibility.⁹²

Summary

The relationships between the bargaining model and existing theories of war and international relations are varied and potentially quite fruitful. The conflict with cognitive psychology is one of degree rather than kind, as the psychological perspective would predict a slower, less efficient process of updating beliefs. Organization theory offers a challenge to current versions of the bargaining model, although it should help the future theoretical development of the bargaining model to become more nuanced. Some versions of the bargaining model are beginning to capture domestic political factors, and future theoretical development will continue in this trend.

Probably the most direct challenges to the bargaining model come from theoretical perspectives that propose that fighting may take on a net positive utility, even when divorced from the issues at stake. Such fighting may serve the purpose of bolstering identity or a government's hold on domestic political power. While the broadest forms of this hypothesis—predicting omnipresent war as a means of reinforcing national or gender identity—may be overdrawn, the possibility that specific conditions, such as internal political change, may make war for its own sake desirable should not be ignored; indeed, this view is beginning to acquire some empirical support. In the broadest sense, this critique is part of the older point that because formal theory cannot describe where preferences come from, its contribution to our understanding of world politics will always be limited.⁹³