Review Of Empirical Studies Of Conventional Deterrence

CIAO DATE: 7/99

Review Of Empirical Studies Of Conventional Deterrence

1999

Center for Global Security and Democracy
Rutgers University

Since the early 1960s, a number of major empirical investigations have explored the effectiveness of conventional military threats in deterring aggression. [See, inter alia, Arquilla and Davis (1992); George and Smoke (1974); Huth (1988); Huth and Russett (1984); Jervis, Lebow, and Stein (1985); Karsten, Howell, and Allen (1984); Lebow (1981); Lebow and Stein (1990); Maoz (1983); Mearsheimer (1983); Quester (1986); Russett (1967); Shimshoni (1988); Wolf (1991); and Zinnes, North, and Koch (1961)] These studies employ a variety of statistical, case-study, and focused comparison methodologies and examine a diverse set of historical cases, ranging over time from 431 BC to 1991. Unfortunately, because the studies apply different definitions of key terms, pose different questions, and test different hypotheses, they are not directly comparable with each other.

Nonetheless, a careful examination of this literature reveals a striking consensus on certain general propositions about conventional deterrence. This consensus is remarkable not only because of the methodological differences between the studies but because of substantial underlying theoretical disagreements, between researchers who assume that interstate relations are inherently conflictual and states are rational, calculating actors and those assuming interstate relations are potentially cooperative and states are potentially irrational actors.

On the basis of this accumulated research, it is possible to reach three critical conclusions about the use of threats of conventional force to deter adversaries seriously considering military aggression or compel such adversaries to halt or reverse their actions. These three findings, along with much of the evidence that supports them and a number of the policy implications that flow from them, are somewhat counterintuitive, particularly given the theorizing about nuclear deterrence popularized during the Cold War. For this reason, as well as because of the important implications for the design and conduct of American security policy that follow from them, these findings about conventional deterrence are worth examination.

The first conclusion is both simple and highly cautionary: many potential adversaries are, at least at times, undeterrable. The nature of conventional deterrence is that it will regularly fail, even in cases where commitments to respond are “clearly defined, repeatedly publicized, and defensible, and the committed state [gives] every indication of its intention to defend them by force if necessary.” [Lebow (1985), 211] The possession of decisive conventional military superiority and the visible demonstration of resolve will not necessarily permit the United States to deter attacks on friends and interests. Three policy implications of this finding deserve highlighting:
- Unlike nuclear deterrence, conventional deterrence does not result in a robust, stable stalemate but in a fluid and competitive strategic interaction that, at best, buys time during which underlying disputes or antagonisms can be resolved
- Over the long run, in the absence of political resolution of conflicts, upholding international commitments is likely to require war, either because conventional deterrence has failed or because war is necessary to establish or re-establish the conditions required for conventional deterrence to operate.
- Because of the insensitivity of potential aggressors to information, efforts to “signal” resolve through diplomatic or military measures, such as redeploying forces into region during crisis, have limited political effectiveness. Force movements may in some cases be important because they change the military balance in the region and thereby manifestly foreclose particular military options, but in general they do not appear to have a significant impact on potential aggressors’ political calculations about the credibility or relevance of deterrent commitments.
The second conclusion regarding conventional deterrence is the importance of denying potential aggressors the ability to achieve a quick victory. Potential aggressors are typically not attracted by the promise of victory in a long war—too much is uncertain and potential costs are too uncontrollable. A decision to undertake military aggression usually requires a perception that an option for rapid success exists. A necessary, though not sufficient, condition for conventional deterrence to hold, therefore, is that all options for a military fait accompli, either through a decisive “blitzkrieg” or a successful war of limited aims, be foreclosed. Two key policy implications, supported by the empirical evidence, follow:
- The overall military balance is less relevant to deterrence than the balance of combat-capable forces in theater or immediately deployable to theater. Because potential aggressors are unlikely to attack unless they believe they have found a way to win quickly, threats to defeat them in a long war, however credible, are unlikely to affect their calculations.
- Threats to deny the potential adversary its objectives are more likely to be effective than threats to punish or retaliate.
The third conclusion is that in the event that deterrence fails and aggression occurs, threats to use conventional military force to impose societal pain and destruction are less likely to compel the aggressor to make peace than are threats to defeat its military forces and to deny it its politico-military objectives.

This report is organized in three sections. The first reviews the findings of empirical investigations of conventional deterrence regarding the factors leading potential adversaries to be undeterrable. The second explores the evidence suggesting the importance of the capacity to deny quick victory. The third briefly considers the literature regarding the effectiveness of conventional military force in compelling an aggressor to halt its actions once begun. [George and Simons (1994); Pape (1996)]

I. The Undeterrability of Adversaries

Potential aggressors have regularly proven undeterrable, despite the best efforts of intelligent, committed deterrers possessing superior military capabilities. Historically, this rather surprising proclivity of the weak to ignore conventional deterrent threats and attack the interests and friends of the strong, or even the strong themselves, has three causes [Wolf (1991), 5]:

The potential aggressor may be strongly motivated to challenge the status quo and, given the less-than-total costs of military defeat in a conventional conflict, may conclude that the costs of continuing to accept the status quo outweigh the expected costs of even losing a conventional war.
The potential aggressor may believe, correctly or incorrectly, that it has developed a military option that politically or militarily “designs around” the deterrent threat.
The potential aggressor may be unable to understand or respond rationally to the deterrent threats that it faces.

Highly Motivated Challengers

The most obvious reason why a potential aggressor may be undeterrable is that the magnitude of pain associated with conventional threats is simply insufficient to deter given the pressure to take action. In some cases, even the virtual certainty of losing a conventional war may be more attractive than failing to act. Potential aggressors may conclude that military defeat is not necessarily incompatible with achieving some political goal or advantage: the ultimate Egyptian military defeat in 1973, for example, nonetheless forced Israel and its backers to resume a stalled diplomatic process. Short-term military losses may permit a government to mobilize popular support and liquidate domestic problems, and may result in major diplomatic breakthroughs. Especially—but not only—for states motivated by “a strong commitment to particular values, a psychopathological leader, or a ‘crazy state’ mentality,” risking a losing war may be entirely consistent with achieving domestic or international political objectives. [Wolf (1991), 5]

What this suggests, of course, is the Clausewitzian observation that conventional war is a political activity, and its utility will be judged in political terms, not purely military ones. Based on a detailed historical examination of Israeli efforts to deter its Arab neighbors, one scholar concludes that “With the specter of Armageddon missing, conventional force is, very simply, useful. As Bernard Brodie described war in the Renaissance, so it remains useful for conventional powers today: ‘War was something that had to be risked not only occasionally, and strictly for matters of great moment but constantly, on all sorts of issues.’ Deterrence cannot hope to be a robust phenomenon in such an environment. There should be (as we know there are) many situations in which, regardless of a defender’s threats, his challenger benefits by resorting to violence.” [Shimshoni (1988), 15] So long as the costs of “defeat” are not too high, and so long as the costs associated with the dishonor of nonaction and the benefits of “doing the right thing” are seen as significant, war will be a useful political tool for dissatisfied states.

Potential aggressors may calculate that the costs of military defeat will be tolerable in two situations. First, when opponents are relatively evenly matched, potential aggressors may assume that the deterrer will be unable easily to score a decisive victory and will find it difficult or expensive to achieve a position from which it can impose intolerable pain on the potential aggressor’s people or force a change in the potential aggressor’s government. (A cautionary note is necessary here, however: while statistical studies support the commonsense finding that “when all other things are equal, threats tended to succeed when the target perceived the threatener’s ability to inflict serious damage on it to be substantial,” there is no clear relationship between the actuality of inferiority and the perception of it and, as a result, “we find no evidence that weaker or smaller states are more likely to yield to threats than are stronger, larger states... Weak targets appear to be intransigent as often as strong ones.” [Karsten, Howell, and Allen (1984), 69, 55])

But, second, it should be equally clear that even a much weaker aggressor may sometimes reasonably anticipate that the probable costs of beginning a losing war will be limited and acceptable: a potential aggressor may rationally calculate that even though its opponent possesses the conventional military—or even nuclear—capacity to impose extraordinary or unacceptable hardship, it will not in fact do so. The international political costs of imposing draconian punishment on an aggressor state, its government, or its society may be substantial, and a highly motivated challenger may take this into account in estimating the costs of a losing war. As Robert Jervis has noted in considering the lessons of the Gulf War for understanding post-Cold War political-military realities, “Perhaps the main inhibitor of the U.S. threat to punish others—especially with nuclear weapons—is not retaliation by the adversary but rather the expected adverse response on the part of world and U.S. public opinion.... [U]nless the provocation were viewed as extreme, many would be outraged at massive death and destruction visited on an offending third-world country. Again, much would depend upon the circumstances and the way in which the punishment was exacted (blockades may produce as much suffering as bombing, but they are much less dramatic), but the fact that the United States is relatively immune to direct counterpunishment does not mean that this form of violence can easily be used.” [Jervis (1994), 128] A noteworthy implication follows from this: presuming that international and public outrage are reduced to the extent that the target of deterrent violence is military rather than “civilian,” threats to deny an adversary its objectives by eliminating its military capabilities or to punish a regime by destroying its military assets are likely to be more credible than threats to impose societal pain.

One additional point is worth highlighting. The historical record suggests that the willingness of potential aggressors to launch attacks that fly in the face of the military odds may well increase if the odds are lengthened. When underlying political differences remain unresolved, measures by a deterrer to increase its military capacity may stimulate, rather than prevent, a deterrence failure. “If in the estimate of attackers, the long-term status quo presents an intolerable condition, they may choose to go to war regardless of the current balance of power. The impetus to go to war... may be the result of an assessment that the correlation of forces is shifting even more dramatically toward the defender and, therefore, opportunities for success will only decline.” [Allan (1994), 219; see also Lebow (1985), 215]

In sum, even an accurate appreciation of the fact that they will lose may not be sufficient to deter potential aggressors. Even when confronted by a deterrer with clearly superior military capability, and especially when the balance of power is moving against them, potential aggressors may elect war as the only, or best, means of ending an unacceptable international political stalemate or resolving unacceptable domestic political difficulties.

Challengers Able to “Design Around” Deterrent Threats

If the first problem for conventional deterrence is that potential aggressors may be willing to accept the consequences of fighting and losing a war, the second is that potential aggressors may be able to convince themselves, correctly or incorrectly, that they will be able to avoid the threatened outcome. In contrast to nuclear deterrent threats, conventional deterrent threats involve “contestable costs.” [Harknett (1994), 86] In conventional deterrence, the credibility that the deterrent threat can be implemented and will have the anticipated impact depends critically on details of the deterrer’s capacity and strategy and, more to the point, on the countermeasures the aggressor is able to take. Further, because the deterrer must communicate its threats and convincingly demonstrate its means of carrying them out, the deterrer must provide the potential aggressor with the information most useful to “design around” these threats either politically or militarily. [Harknett (1994); Shimshoni (1988); Lebow (1985), 214-15] In other words, when used against an intelligent and motivated opponent, every conventional deterrent threat contains within it the seeds of its own demise. To tell a potential aggressor what fate will befall it if it misbehaves is to give it the critical information it needs to prevent that fate from befalling. In the ensuing strategic interaction, the potential aggressor possesses substantial structural advantages. As Harknett summarizes, conventional deterrence’s “fundamental problem is not easily solved by strategy or reputation because it lies within the inherent dynamic process produced by reliance on non-nuclear forces. Conventional deterrence is hampered by the nature of the weapons upon which it is based. The ability of conventional forces to inflict pain on an opponent is highly dependent on the skill shown in their application, on the capabilities possessed, and on the counter-capabilities employed by an opponent. The destructiveness of conventional weapons, and thus the cost that can be threatened, will only be felt over time after achieving military victory or dominance over some sector of the battlefield. In the mind of a potential challenger, therefore, conventional weapons and their attendant costs hold out the prospect of technical, tactical, or operational solution, and it is this prospect that makes them less effective deterrents to war.” [Harknett (1994), 88-89]

The contestable nature of conventional deterrent threats has important implications.

There is an important informational asymmetry inherent in conventional deterrence. For conventional deterrence to operate, the deterrer must credibly communicate its conditional intentions (what it will do and under what precise circumstances) and its abilities. The potential aggressor, however, is free to keep its evolving political and military capabilities secret until such time as it challenges the status quo. In other words, the deterrer must show all—or at least most of—its cards, while the potential aggressor is free to use this information and give nothing back.

Of course an informational asymmetry exists in all deterrence relationships. But it should be clear that there are two substantial differences between conventional deterrence and nuclear deterrence. First, more information needs to be conveyed in conventional deterrence. Because of the magnitude of the damage nuclear weapons can impose and the near-certainty with which this damage can be imposed, regardless of the relative skill or tactical advantages of the adversaries, the informational burden of nuclear deterrence is relatively light. The precise technical details of the weapons and targeting strategy are simply unlikely to matter much to decisionmakers during crises: so long as at least minimal second-strike capabilities exist, a nuclear exchange is unlikely to be appealing regardless of how well or badly the other side is prepared to execute its blow. Knowledge that the weapons exist and could, even in only moderate numbers, be delivered may well be sufficient to deter. “The mutual possession of nuclear weapons greatly simplifies these [knowledge] requirements, obviating the need for detailed mutual understanding of each other’s capabilities, desires, and rationalities. But our consideration of conventional weapons and deterrence suggests an acute problem for common knowledge. The nonnuclear world is messy and detail intensive; yet sharing the knowledge of this detail of the tangible and intangible, physical and political, constant and dynamic aspects of relative will and skill is critical for both the generation and comprehension of relative and credible deterrent threats.” [Shimshoni (1988), 16] Again, the point here is that capability is an issue in conventional deterrence in a way that it is not in nuclear deterrence: “the nature of conventional forces invites skepticism at a level that few deterrence theorists have emphasized—that of capability. Due to the contestable nature of conventional forces, it is a state’s capability to inflict costs that is most likely to be questioned by a challenger.... This is not to say that the uncertainties around a state’s commitments and vital interests are unimportant, but rather to stress that the most problematic area for conventional deterrence is in establishing a credible capability.” [Harknett (1994), 89]

Second, the cost of conveying information is higher in conventional deterrence. In nuclear deterrence, even if the deterrer communicated the precise details of its nuclear capabilities and strategy, this information might well be of limited value to a potential aggressor (and therefore be relatively uncostly for the deterrer to convey) since, in practical terms, there might be little the potential aggressor could do to block the deterrer’s blow or mitigate its consequences. The principal obstacle faced by a potential aggressor in “designing around” a nuclear threat is unlikely to be information regarding the deterrer’s capability or strategy, but fundamental military-technical realities, such as the technological difficulties of tracking ballistic-missile submarines or destroying ballistic missiles in flight or the economic costs of hardening industry and dispersing population.

In conventional deterrence, by contrast, both the amount of information that must be communicated and the cost of communicating this information are far higher. The credibility of a conventional threat is likely to depend on constantly changing technology, strategy, tactics, training, morale, and political will—all of which must therefore be communicated for the threat to be effective—and this information about the deterrer’s capabilities and strategy is likely to suggest solutions to a potential aggressor.

Indeed, this observation is likely to particular relevant in the American situation. U.S. reliance on high-technology weaponry, on highly trained warfighters, on superior logistics and mobility, and on the leverage provided by information superiority, tends to exacerbate the informational difficulties of conventional deterrence and increase both the amount of information required to make conventional deterrent threats credible and the cost of communicating this information. Unlike conventional threats involving sheer mass and already-demonstrated technological capabilities, American conventional threats are both hard-to-believe and potentially vulnerable to unanticipated countermeasures. Without demonstrations like Desert Storm of the effectiveness of American military power, American threats may be incredible; demonstrations like Desert Storm, however, also suggest to potential aggressors strategies for negating American superiorities.

It is particularly difficult to convey deterrent threats with forces that rely on intangibles—such as “the quality of the individual soldier, technological proficiency, superior organization, the ability to mobilize a reserve army quickly and then operate it, and the tactical ability to substitute the movement of force for its accumulation”—for their effectiveness, and these sources of superiority are particularly likely to be vulnerable to countermeasures by an informed adversary. “Although these components of superiority may make for battlefield victory, they may also render deterrence difficult; with conventional forces, victory and deterrence can be mutually exclusive.” [Shimshoni (1988), 227]

It is useful to note that trends in American strategic thinking, most clearly those outlined in the “2020 Vision” statement issued by the Chairman of the Joint Chiefs of Staff, General John Shalikashvili, promise to exacerbate this problem. The idea of using information superiority, precision engagement, and dominant maneuver to transform American warfare in the next century, thereby reducing the cost and casualties of future conflicts, is appealing. The capacity to wage this sort of warfare, however, and the kind of impact it would have on an opposing society or government are not self-evident.

Obviously, this need to communicate both the specifics of the threat (what is threatened and in response to what provocations) and the capacity to execute it creates a painful trade-off for the deterrer. Particularly with regard to military capabilities, but to a lesser degree with regard to political will and diplomatic arrangements, there are real costs to sharing information with a potential adversary, and a deterrer may choose to err on the side of withholding information. “The defender is in a real dilemma: To deter, he must appear to be ex post superior, capable of executing his deterrent threats. But to really be ex post superior, he must keep most of his capabilities secret.” [Shimshoni (1988), 18] In a nutshell, “there are strong incentives for both the challenger and defender to create ambiguities and deliberately destroy common knowledge,” [Shimshoni (1988), 213] and this creates major problems for a relationship like one of deterrence, which rests precisely on the existence of common knowledge about military options and capabilities. [Harknett (1994), 95-97]

Thus, even a deterrer with decisive capabilities may, for fear of compromising that decisive superiority, fail to communicate the capabilities on which deterrence rests. If capabilities were not continuously in flux, this might not be a critical problem. But such is not the case. Both sides’ capabilities are constantly in motion, making accurate assessment difficult and increasing the danger that the deterrer will miss developments until it is too late to respond or that the potential aggressor will overlook changes in the deterrer’s capacity that would give it the ability to block newly developed options. “Given the dynamic nature of conventional forces, both challenger and defender are apt constantly to change their force structures, doctrines, and tactics in response to inaccurately perceived conditions on the other side. The inaccuracy is inevitable because of deception compounded by the need for prescience. Sooner or later the challenger is likely to perceive a window of opportunity, real or not.” [Shimshoni (1988), 18] This suggests a second structural reality that works to the detriment of a deterrer.

Conventional deterrence is not a static condition: it is, rather, an ongoing, dynamic, and strategic process. Potential aggressors are constantly searching for new options to achieve their political objectives, and deterrers are constantly searching for credible threats to make these new options unattractive. Because of the capacity of an intelligent adversary to evaluate the deterrent threat and continuously to develop new political or military options, a deterrer must constantly update both its threat and its capabilities to keep pace with the evolving challenge. This is a never-ending struggle: so long as the potential aggressor remains strongly motivated to change the status quo, it can work to discover the loopholes or to exploit the weaknesses in the deterrent strategy. [Shimshoni (1988); Harknett (1994); Wirtz (1993)]

Obviously, there is a critical interaction between information and the strategic dynamic. Potential aggressors are constantly exploiting the information communicated by the deterrer to transform the situation and redesign the potential challenge so as to undermine or defeat the deterrer’s threat. The longer and more intense the interaction between deterrer and potential aggressor, and the more effective the deterrer is in communicating its capabilities and in deterring at time t, the better able is the potential aggressor to defeat the deterrer’s threat at time t+1. There is an irony here, of course: “Not only does the structure of a relationship undergo constant natural change, but the very engagements that produce reputation actually accelerate such dynamics. To establish a reputation for capability requires that relative superiority be established, in technological and doctrinal details. The very act of establishing such superiority—violent engagement—is also the first step in its own undermining. Every engagement is a lesson, instructing the challenger in his own and in the defender’s relative strengths and weaknesses, suggesting ways to get around them in the next round.” [Shimshoni (1988), 23]

In 1939, French efforts to deter Germany failed because the Germans were able to use information communicated by France regarding France’s defensive denial strategy and to design a military solution that circumvented it. Similarly, in 1973 Israeli efforts to deter Egypt failed because the Egyptians, again relying on knowledge communicated regarding likely Israeli responses, designed a conventional strategy that negated and, indeed, even exploited Israeli strengths. [Harknett (1994), 97-98]

The pace of conventional conflict—the fact that, unlike a massive nuclear war, in which unbearable pain might be inflicted within the first hours, conventional war might drag on for weeks, months, or even years before the deterrer is able to counter the aggression or impose significant sanctions—also has important implications. A potential aggressor may reason that it will have the ability to adapt to the deterrer’s strategy even during the course of a conventional conflict. “From the challenger’s perspective, time and relative capabilities combine to promote the perception, if not the reality, that the costs threatened by a deterrer are open to manipulation. They do not have to be taken at face value. In fact, they may be contested either [sic] through exploiting time, new tactics, and improved counter-weaponry. A conventional deterrence environment is conditioned by the possibility of adaptation. It is possible to adapt to and, therefore, contest conventional costs even while under attack. Germany’s ability during the Second World War to adapt to the Allied strategic bombing campaign by decentralizing its industrial base, finding appropriate substitutes for lost resources, and converting its entire economy to a total war footing is a classic example. The initial adaptation to the Allied conventional onslaught was so successful that as the tonnage of bombs dropped on Germany increased so did the volume of German war production.” [Harknett (1994), 92]

But the challenge for the deterrer is not simply that the potential aggressor’s strategy changes over time and that the potential aggressor may count on being able to adapt to the deterrer’s actions. The deterrer must also face the problem that the potential aggressor is likely to be developing multiple, alternative options for achieving its goals at any given time.

“Nations interested in changing the status quo normally have more than one option for doing so. The relevance of this observation to the design of defenders’ deterrence policies is self-evident. The defender’s strategy must be made relevant to the range of alternative options possibly available to the initiator. A deterrence policy which discourages an opponent from employing some options but not others is incomplete and may not prevent a failure of deterrence.” [George and Smoke (1974), 520-21]

In this regard, it is useful to note that measures taken to deter one option may not only be ineffective but even counterproductive in deterring other options. In some hypothetical future Persian Gulf scenario, for example, stationing U.S. troops ashore might be useful in deterring a radical regional power from classical cross-border invasion, while at the same time enhancing that radical state’s ability to achieve its political goals through violent internal subversion.

It should be apparent that potential aggressors can pursue either or both of two general approaches to multiplying their options for changing the status quo. The first approach is to develop new types of threats—subversion rather than direct aggression, attacks with weapons of mass destruction rather than traditional armaments, coalitional rather than unilateral aggression, limited rather than all-out war, surprise attack rather than a fully mobilized assault, and so on.

In designing new types of threats, potential aggressors can take advantage not only of the particular capabilities of the deterrer’s military forces but of the particular attributes of the deterrer’s political culture and constitution. Given certain similarities to the United States, the Israeli experience is relevant in this regard. “Perhaps the aspect of Israel’s political culture that has most weakened deterrence has been her Western democratic nature. This label subsumes a number of important attributes: decision making that is highly centralized yet subject to public and private debate and pressure; a highly disciplined army; a free and rather adversarial press; high social and political sensitivities to costs in resources, lives, and lifestyle; and political sensitivity to the breakdown of consensus on security questions.... The Israeli threat of reprisal for low-level harassment has been substantially undermined by the nature of decision making and execution.... The credibility of the Israeli threat to wage large-scale offensive war has been similarly undermined. The need for consensus, public accountability, and the sensitivity to costs have made Israel’s few decisions to wage general war very difficult even under the most clear-cut strategic conditions.... The same traits have made it difficult to project a credible crazy-state image by threatening to respond irrationally to small-scale provocation.... Finally, as open and Western democracy, Israel has been in a poor position to threaten frustration through protracted low-level war.” [Shimshoni (1988), 224-225] Potential aggressors are able to create options that exploit these political “weaknesses.”

The second approach available to potential aggressors for multiplying options is to transform existing threats temporally: the same threat can either be generated repeatedly or can be maintained for a prolonged or indefinite period of time. It should be clear that, from a deterrer’s point of view, repeated or protracted crises may pose a fundamentally different problem than single, short crises, even when the military danger involved is exactly the same. Indeed, this is a particular vulnerability for deterrers that rely on mobilization or that, like the United States, face relatively high economic, political, or diplomatic costs in deploying to, or remaining in, theater. For this reason, “the relative skill in speedy mobilization and the relative economic and political ability to sustain the costs of repeated mobilization greatly influence the long-term outcome of deterrence. As it takes only one successful deception to foil deterrence, its long-term prospects under such conditions seem unpromising. Moreover, the idea that costs can be cumulative suggests a challenger’s policy akin to ‘controlled pressure,’ as described by George and Smoke. It may be possible through a series of mobilizations to impose such costs on the defender that his valuation of the object decreases compared with his valuation of peace.” [Shimshoni (1988), 19]

The potential irony here is worth noting: in this situation deterrence may fail either because the deterrer gives up or because the deterrer elects to launch a preventive war. That is, the potential aggressor may succeed in forcing the deterrer to become the “aggressor,” with all the political, diplomatic, and ethical problems this suggests. To the degree that U.S. deterrent capability relies on the temporary surge of forces to theater, “U.S. decision makers will face the policy dilemma that forces either must be used to effectively eliminate the military threat that triggered their deployment or face pressures to leave with the threat still intact.” [Allan (1994), 207]

One final, perhaps nonobvious, observation about the multiple options open to potential aggressors is worth making. These multiple options may reflect multiple objectives or motivations: they may not simply be different ways of achieving the same goal, but different ways of achieving different goals, all of which would be enhanced through aggression. In other words, there may be more than one impetus behind aggression—a potential aggressor may, for example, simultaneously face domestic pressures and perceive economic or strategic gains from aggression. “Ideally, the objective of the deterrer is to develop a strategy that assigns a prohibitive cost to every military option available to the challenger. A deterrence strategy must be resilient enough to adapt to any number of motivations behind a challenge, whether they spring from a sense of opportunity or a feeling of vulnerability.” [Harknett (1994), 91]

There is a fourth structural reality that complicates the task of preventing a potential aggressor from “designing around” a deterrent threat.

In dealing with these evolving, multiple options, deterrers are also handicapped by their need to deal with multiple international audiences. Deterrers like the United States are unlikely to have the luxury of deterring only one potential aggressor at a time. Indeed, deterrers may be faced with the need to deter both sides in a single conflict.

“[T]he United States is likely to face situations that require what might be called dual deterrence—the deterrence of two countries of two countries or factions that are in conflict with each other.... It is easier to convince another state that you are its adversary and will block it at every turn or, conversely, that your interests and those of the other are congruent and that you will protect it. Complex, conditional threats and promises are much more difficult to make credible, and it is not surprising that the United States was unable to do so in this case [North and South Korea prior to the Korean War]. The failure of U.S. policy toward Iraq in the years before the Gulf War has a similar explanation. If all the Unitd States had wanted to do was to deter that country, it might have succeeded. But it wanted to see that neither Iran nor Iraq dominated the area, and this required trying to show each of them that the United States would support it if it were moderate and restrained but oppose it if it sought regional hegemony.” [Jervis (1994), 122-23]

The final structural feature of this strategic interaction, in which the deterrer must reveal critical information about capabilities and strategies while the potential aggressor does not, in which capabilities are constantly evolving, in which potential aggressors are free to develop multiple options, and in which deterrer’s must address multiple audiences, is perhaps the most important, however.

The rules of this strategic competition are profoundly skewed in favor of the potential aggressor. It does not matter how many aggressive options the deterrer blocks or for how long it blocks them: for deterrence to fail, all the potential aggressor needs is to perceive a single workable option in a single window of opportunity.

The fact that potential aggressors may be able to design around conventional deterrent threats has at least two noteworthy implications.

First, perhaps unsurprisingly, the greater the political pressure on the potential aggressor to act, the more likely it is that deterrence will fail.
- As political motivation for action increases, pressure to find an acceptable military option grows and the more inventive a potential aggressor will become. Political leaders may, for example, remove military leadership that is not sufficiently creative in developing new options.
- As political motivation for action increases, willingness to tolerate less-than-certain military options and military options that promise only limited political gains also grows.

In sum, “when a nation is dissatisfied with the status quo, the long-term prospects for deterrence are not promising. The pressure to find an acceptable military solution will be too intense.” [Mearsheimer (1983), 211] Historically this has meant that when sovereignty or political prestige (or, during the Cold War, ideological or moral issues) were at stake for the potential aggressor, deterrent threats were highly likely to be ineffective. [Karsten, Howell, and Allen (1984), 81]

Second, wars not only happen, they may be both necessaryfor conventional deterrence to work and, ironically, an indication that conventional deterrence is working. War and conventional deterrence are not opposites: they are complements, part of a single ongoing, strategic process.
- Periodic wars may be necessary to communicate capabilities and may thus be a prerequisite for establishing or re-establishing deterrence. In this light, wars are not so much a failure of deterrence, but a critical part of the ongoing strategic process that maintains it. [Shimshoni (1988), 26, 215; see also Haffa (1992)]
- Limited war may be an indication that, in some larger sense, deterrence is working. If the aggressor’s challenge is a limited one, this may suggest that the deterrer has successfully prevented broader or graver challenges to the status quo.

This last point is perhaps a counterintuitive one. “If limited war has meaning and not all violence is automatically escalatory and strategic, then challenge at a tactical level may reflect successful deterrence at a higher level, more important to the defender.... Success at one level may actually cause failure at another.” [Shimshoni (1988), 26, 215] Limited military challenges to the status quo are an indication of continued frustration with political realities, but they also indicate successful deterrence of other, more dangerous aggressive acts. State-supported terrorism, for example, may represent a failure of American deterrence policies—but at the same time suggests the success of deterrence policies in foreclosing more direct, potentially more dangerous, challenges to the international order.

Challengers that Irrationally Ignore Deterrers’ Commitments

But potential aggressors may be undeterrable not only because they are willing to tolerate the costs of war or because they believe they have found a way to design around threats but because they are unable to recognize or respond rationally to deterrent threats that they face. Contrary to assumptions of rational behavior typically incorporated into classical deductive deterrence theory, states appear routinely to ignore or misinterpret other states’ commitments, even when deterrers intelligently and actively attempt to communicate these commitments. This failure results from at least three causes.

First and most obviously, potential aggressors may fail to recognize credible deterrent commitments because of analytic biases reflecting cultural differences. [Shimshoni (1988), 224]

For deterrence to operate, the deterrent threat must be communicated, either implicitly or explicitly. Communication, however, is notoriously difficult. Messages, signals, and histories are interpreted through particular cultural lenses, and, even in the best of cases, the communication received is unlikely to be the communication sent. This can go well beyond misunderstood nuances. What “goes without saying” to one side may be implausible to the other—indeed, so implausible that even saying it will not make it credible. And what is diplomatic or military chaff to one set of leaders may be perceived as the vital kernal to another. Certainly the kind of subtle signals that states often wish to send are frequently lost. “Carefully calibrated signals most often fail to make the desired impression because they are based on distinctions that seem obvious to the sender but to which the receiver is oblivious. A striking example of this was the U.S. decision in 1965 to send ashore a ‘light’ marine division instead of a ‘heavy’ army division in order to signal to Hanoi Washington’s limited aims.... Bombing pauses were used later in the war as another means of signaling to Hanoi. Their short duration, fine graduations in targets, and the complex messages they were meant to convey presupposed the existence of a common bargaining framework. This simply was not the case.” [Lebow (1985), 205-206]

Cultural differences pose two distinct obstacles. The first is more obvious: particular words and actions have different power, resonance, or nuance in different languages and cultures. Thus, as Lebow notes with regard to the breakdown of Spanish-American negotiations in 1898, “McKinley’s difficulty in communicating with Spain was due in the first instance to cultural differences; Americans and Spaniards read the same message in contrasting ways because of the different associations they brought to the words in question.” [Lebow (1985), 208] Similarly, the movement of American troops to the Yalu was ascribed very different meaning in Beijing and Washington.

Second, though, different cultures may impose completely different frameworks of understanding: not simply words and actions but the entire situation may be understood differently in different cultures. The Falklands/Malvinas crisis offers a good illustration of the problem. For the Argentinians the whole situation was understood in the context of decolonization and appropriate measures to accomplish it; for the British, it was understood in terms of self-determination and the imperative to deter and punish resorts to violence. Words and actions took on fundamentally different meanings in these different cultural frameworks. [See, for example, Lebow (1985), 210] Similarly, the breakdown of the Anglo-French peace in 1803 reflected a culturally-rooted misunderstanding of the relationship between press and state which led the two sides to misinterpret the other’s level of hostility and willingness to abide by the status quo. [Burrows (1996)]

Cultural barriers may result in miscommunication in two separate areas. First and most obviously, cultural differences may obscure that a commitment exists—that is, may result in misinterpretation of the diplomatic and military signalling of interests and demands.

But second, cultural barriers may also result in miscommunication of capabilities: cultural differences may obscure the deterrer’s ability to carry out its threats. Indeed, this may be particularly relevant in cases in which the deterrent threat involves naval forces or sealift or airlift. Unfamiliarity with these capabilities may lead potential aggressors to discount these significantly. [Arquilla and Davis (1992), 29; on the general point, see also Lebow (1985), 209]

Indeed, this may help to explain a somewhat unexpected statistical finding, which is “the importance of the target’s [that is, potential aggressor’s] perception of the threatener’s capabilities and... the relative unimportance of the actual capabilities of the contending parties.” [Karsten, Howell, and Allen (1984), 70]

Cultural obstacles to communication are, however, only the tip of the iceberg.

Second and more interestingly, a wide range of studies suggest that, particularly during crises, potential aggressors tend to focus on their own internal needs—the domestic political pressures and challenges they face—and to ignore external realities, including the likely behavior of the deterrer.

The evidence here is that, at least in some cases, deterrence fails not because a potential aggressor identifies an international window of opportunity and deliberately jumps through it but because the “aggressor” is pushed into taking the aggressive leap—without looking at the external consequences—by its own internal troubles. “Aggression is less a function of opportunity than it is of need. We discovered a good opportunity for aggression (i.e., a vulnerable commitment) in only about one-third of our cases but found strong needs to pursue an aggressive foreign policy in every instance.” [Lebow (1981), 276]

Ultimately, the argument that potential aggressors may be undeterrable because of their internal preoccupation involves on two separable empirical observations. First, crises are frequently caused by “serious domestic and strategic problems that leaders believed could best be resolved through aggressive foreign policies.” Second, in such crises, potential aggressors become insensitive to deterrent threats: “perceiving the need to pursue a policy of brinkmanship, these leaders rationalized the conditions for its success and sought to insulate themselves from information that challenged their expectation of success.” [Lebow (1981), 274]

The point—and it is an important one, suggesting as it does the potential irrelevance of deterrence policies—is that “Policy makers who risk or actually start wars pay more attention to their own strategic and domestic political interests than they do to the interests and military capabilities of their adversaries. Their strategic and political needs appear to constitute the principal motivation for a resort to force. When these needs are pronounced, policy makers are prone to disregard the ways in which the same kinds of strategic and political needs might compel adversaries to stand firm in defense of their commitments. They may discount an adversary’s resolve even when the state in question has gone to considerable lengths to demonstrate that resolve and to develop the military capabilities needed to defend its commitments.” [Lebow (1985), 216] Empirically, this tendency toward excessive inward-focus is manifested in several important ways, all of which have consequences for deterrers who are attempting to make deterrence work.

Potential aggressors ignore the deterrer’s interests.

Potential aggressors do not appear to be influenced by military (e.g., arms sales), political (alliance), or economic (trade) ties between the deterrer and the target of aggression. [Huth (1988), 82-83] Indeed, even when the potential aggressor correctly perceives the deterrer as having substantial interests at stake, there is no evidence that this affects the potential aggressor’s behavior: “an entirely correct perception by a target [i.e., potential aggressor] of a high level of threatener [i.e., deterrer] interest did not lead to the target’s capitulation if the target’s own resolve, its own ‘independent valuation of the stake,’ was high.” [Karsten, Howell, and Allen (1984), 68]

Potential aggressors ignore the deterrer’s public reputation.

Despite the usual—and seemingly plausible—presumption that potential aggressors take into account the deterrer’s historical proclivity to carry out its threats, the evidence suggests that this is not, generally speaking, the case. There is no statistical support for the contention that potential aggressors are influenced by the overall reputation of the deterrer as established by that deterrer’s behavior in conflicts with third parties. [Huth and Russett (1984), 516; Huth (1988), 75] This suggests that paying a high price to create a general reputation for toughness may be a bad investment.

The evidence whether potential aggressors are influenced by their own past interactions with the deterrer is mixed. Apparently, however, potential aggressors are influenced only to the extent that a history of either bullying or conciliation by the deterrer reduces the likelihood that deterrence will be successful in the present case. [Huth (1988), 75]

Potential aggressors ignore the deterrer’s capabilities.

Even when decision-makers accurately perceive their own military inferiority and the deterrer’s military superiority, they are capable of ignoring this in their decision-making. Instead, decision-makers appear to focus on their own motivation for war rather than on the potential consequences of it. [Zinnes, North, and Koch (1961); Russett (1967), 98-99]

There is, though, a third reason why potential aggressors fail to respond rationally to apparently clearly communicated commitments and threats.

Third and most interestingly, in addition to experiencing cultural barriers to understanding and being inward-focused, potential aggressors are likely to suffer from “motivated biases” in their evaluation of the deterrer’s will or ability. That is, potential aggressors are likely to engage in pervasive wishful thinking, distorting information about the deterrer in order to avoid the stress associated with acknowledging the dangers and trade-offs actually inherent in the situation and chosen policies. Psychological pressures to ignore warning signs will be particularly strong during periods of internal or external crisis. Because they “need” for the deterrer to be weak or timid for their preferred policy to work, and because the consequences of their preferred policy failing are too painful to bear contemplation, potential aggressors will ignore or twist the available evidence to make it consistent with the image they desire.

One major study, for example, concluded that “When policy makers believed in the necessity of challenging commitments of their adversaries, they became predisposed to see their objectives as attainable. They convinced themselves that they would succeed without provoking war. Because they knew the extent to which they were powerless to back down, they expected that their adversaries would accomodate them by doing so. Some of the policy makers involved also took comfort in the illusion that their country would emerge victorious at little cost to itself if the crisis got out of hand and led to war.” [Lebow (1985), 212]

This inability of potential aggressors to recognize deterrent commitments—because of cultural barriers, the aggressor’s inward focus, or its susceptibility to stress and motivated biases in interpreting commitments—has at least four important implications for deterrence policies.

First, if a crisis reaches the point where it is publically apparent that a potential aggressor is seriously considering military action, it may be too late to convey to the potential aggressor of the dangers of this policy.

“To be most effective, threats must be issued before the other side has committed itself—psychologically and politically—to move. Decisions, even tentative ones, are not easily reversed. Once a state’s leader has decided that it is important to take certain acts, he or she will not be easily dissuaded. This is especially true if the leader’s political fortunes have been staked on the course of action. Under these conditions, furthermore, he or she is not likely to accurately perceive any threats that others may issue. Thus, I doubt whether anything Ambassador April Glaspie could have said to Saddam Hussein on the eve of the invasion of Kuwait would have done much good. Similarly, President Kennedy warned Nikita Khrushchev not to deploy missiles in Cuba only after the weapons were on their way.” [Jervis (1994), 129]

It is important to understand the problem is not simply that the stakes increase for potential aggressors after they have made a tentative decision, though this is surely the case. Receptivity to deterrent warnings apparently drops off sharply as international or domestic conditions deteriorate: potential aggressors’ ability to pay attention to deterrent threats is not simply low, but declines when it is most needed. “A challenging state is unlikely to be receptive to threatening messages from the opposing side. In an environment of increasing tension, the challenger’s ability or desire objectively to evaluate the deterrer’s intentions and possible reactions may be reduced. The motivations that tend to drive the cause of war—domestic politics, patriotism, ethnic nationalism, militarism—can cause a level of ‘noise’ that inhibits proper communication between belligerents.” [Harknett (1994), 94] In other words, after an international situation or domestic conditions reach the point that a potential aggressor is seriously considering a resort to violence, cultural obbstacles to clear communication, preoccupation with internal political problems, and psychological unwillingness to face up to the possible dangers of chosen policies will all be elevated.

If potential aggressors are to recognize and understand them, deterrent threats and the capacity to execute them need to be communicated in advance. This suggests the potential value not only of clear diplomacy and pre-existing treaty commitments, but of forward military presence. At the same time, however, it raises more grounds for skepticism about the ability to generate robust conventional deterrence policies. By shifting the ground from immediate deterrence—that is, deterrence of an adversary who has formulated an intention to attack—to general deterrence—deterring an adversary from even thinking about attacking—this requirement creates enormous political difficulties for would-be deterrers. As Jervis observes, “this requires that the United States be able to identify its interests and the potential menaces to them before the dangers have started to emerge. This is probably asking too much of intelligence, in the general sense of the term.... Furthermore, detecting threats in a timely manner is especially difficult if acting on this perception requires a sharp change in policy. This was one of the major problems in the summer of 1990, and it is not likely to be unique. If a state that the United States has supported starts to menace U.S. interests, political and perceptual inhibitions again are likely to prevent a quick recognition of the situation. It is hard for decisionmakers to see that their policy is failing and indeed may have been misguided for quite some time.” [Jervis (1994), 129-30]

Second, efforts to “signal” commitment through diplomatic and military steps, such as deploying forces to the region, are unlikely to have an impact.

Based on analysis of case studies, one major study concluded that “the inner-directedness of policy makers points to one of the ironies of international life: much of the effort made by defenders to impart the appearance of credibility to their commitments is probably wasted because challengers may not pay much attention to them. Even when they do, they may distort or simply not understand the signals being directed their way. Cultural, contextual, and organizational barriers to interpretation abound, not to speak of motivated biases on the part of policy makers already committed to proceeding with a challenge.” [Lebow (1985), 232] A statistical examination of the historical record suggests that “Repositioning one’s forces in conjunction with a threat may help one win the war that ensues if the target refuses to yield, but it has rarely been instrumental in bringing about the success of the threat itself.” [Karsten, Howell, and Allen (1984), 110] Indeed, an examination solely of U.S. cases reveals the counterintuitive finding that “the more force the United States put on line in conveying the threat, the less likely the threat was of being successful.... Once again, the relationship is almost certainly not causal, but it does permit us to conclude that displays of force did not appear to have been instrumental to the success of American threats in the nuclear age.” [Karsten, Howell, and Allen (1984), 84]

A corollary specifically relevant to the U.S. Navy is worth noting explicitly.

The widely accepted notion that repositioning of naval forces is an effective means of communicating national commitment appears without general merit.

It is easy to make ad hoc arguments about the value of naval movements in specific cases and to reason that the potential aggressor “must” have been influenced, particularly if that adversary did in fact eschew actual aggression. In general, though, both statistical analysis and case studies suggest that potential aggressors seem insensitive to this sort of communication. “For those who wonder whether the brandishing of one carrier task group, or two, or a task group and two battalions of Marines, or two groups and four battalions will carry the day, we submit that such fine tuning of the threat forces does not appear to matter.” [Karsten, Howell, and Allen (1984), 73]

Third, it may be extremely difficult to “teach” a potential aggressor the dangers of its policies and the existence and effectiveness of deterrent threats.

Potential aggressors are surprisingly unlikely to learn from experience. Even limited military engagement may be insufficient to teach potential aggressors appropriate lessons about the deterrer’s capacity to defeat or frustrate the potential aggressor: “a large military organization is not likely to decide that it is inferior to an opponent simply because that opponent successfully slaps around a number of its units.... Low-level demonstrations simply appear to be irrelevant to assessments of the general balance of power, positive or negative.” [Shimshoni (1988), 226] Four factors are at work. First, leaders may psychologically resist hearing the enormously painful news that they would lose a war—and that they must therefore abandon cherished goals or policies essential to their continued control of power. Second, bureaucratic organizations may have blinders imposed by their institutionalized cognitive framework—that is, their deeply-rooted conception of how the world works and, more specifically, what another war would look like, may result in unmotivated biases in interpreting new information and insensitivity to indicators that their politico-military strategy will not work. Third, potential aggressors may indeed learn, but learn the wrong lesson. The lesson Egyptian nationalists drew from the 1948 experience, for example, was not that Israel possessed the conventional military wherewithal to defeat Arab efforts to reverse the status quo but that the Egyptian royal regime was corrupt and that a new, national, progressive government would permit the Egyptian state to tap the overwhelming power latent in the Egyptian and Arab nations. [Shimshoni (1988), 232] Fourth, potential aggressors may learn, but conclude that the lesson applies only to highly specific circumstances—circumstances which are in their power to change—and that, therefore, the lesson can be rendered irrelevant. [Shimshoni (1988), 232]

The pessimistic view, then, is that deterrers and potential aggressors do not understand each other and that, because of cognitive and bureaucratic obstacles to learning, even repeated or extended contact is unlikely to change this. “Our analysis of adversarial relationships in this study indicates that the expectations that the two sides have of each other may bear little relationship to reality. Challengers, as has been demonstrated, tend to focus on their own needs and do not seriously consider, or often distort if they do, the needs, interests, and capabilities of their adversaries. Defenders, in turn, may interpret the motives or objectives of a challenger in a manner consistent with their expectations whether or not these expectations are in any way warranted. Both sides, moreover, may also prove insensitive to each other’s signals for a variety of political, cultural, or other reasons. In such circumstances, recurrent deterrence episodes may not facilitate greater mutual understanding. Experience may actually hinder real learning to the extent that it encourages tautological confirmation of misleading or inappropriate lessons.” [Lebow (1985), 231)]

The difference between conventional and nuclear deterrence in this regard is worth underscoring. During the Cold War, nuclear learning promoted a stable Soviet-American relationship, as both sides concluded that provocation and brinkmanship generated unacceptable risk. There is, however, no reason for optimism that this will be the case in conventional relationships: time and experience may not work in deterrence’s favor.

Fourth, efforts to strengthen conventional deterrence by signalling commitment may be not only unproductive but counterproductive.

Two factors are involved.

“Deterrent” steps may convince potential aggressors that conflict is inevitable and therefore provoke the action they were meant to deter.

To whatever degree potential aggressors actually notice signals, there is a real danger that they will misinterpret them, seeing measures such as movement of military forces and strengthening of alliance commitments as proof of hostility or aggressive intent rather than as purely defensive efforts to dissuade aggression. “Most international conflicts.... contain an important element of defensive motivation (i.e., concern for security) on both sides as well as offensive goals. Deterrence becomes less relevant to the degree that a conflict can be described as a mutual security dilemma. Deterrence can even be dysfunctional in such situations because it does nothing to defuse hostility. It may actually intensify it by appearing to confirm mutual suspicions of aggressive intent.” [Lebow (1981), 278]

As this makes clear, there is a danger that steps that may appear useful in the short run—that are designed to underscore the deterrer’s ability and intention to respond to a current crisis—may have a deleterious effect in the longer term, convincing a potential aggressor that the deterrer has ruled out political accomodation and that in the long run a military solution will be necessary. Again, this suggests the long-term instability of conventional deterrence—that short-term successes are likely to be bought at the cost of long-term failure and that, as a politico-military policy, conventional deterrence needs to be understood as a stopgap measure, not a lasting solution.

But this suggests a second insight as well. As deterrers attempt to steer a course between the near-term Scylla of failing to communicate their threats and the longer-term Charybdis of convincing an opponent of the ultimate need for war, it is important to search for policies that deter without challenging the survival or fundamental interests of the adversary. “To avoid war, a status quo power must pursue two distinct and not easily reconcilable objectives. It must maintain its own power and credibility while at the same time working to forestall the development of the kind of unstable conditions that lead to a security dilemma. As Snyder puts it, the status quo power ‘must worry about everyone’s security, not just its own.’ The best way to accomplish this, he indicates is to deploy defensive forces that create no obvious first-strike advantage and thus avoid bringing about a situation of strategic instability.” [Lebow (1985), 226]

“Deterrent” steps may increase the time pressure and stress on potential aggressors’ decision-makers, thereby increasing the danger of an irrational resort to war.

This observation has important policy implications for naval forward presence. “Since a decisionmaker’s susceptibility to stress, rationalization, or other factors is partially a function of constraints on time, a deterrence strategy should be structured so as to encourage a challenger to evaluate the consequences of using force in a measured, comprehensive manner.... One approach is for the deterrer to structure his threat so that the utility of preemption is reduced. If a deterrent threat’s effectiveness is not diminished (in the aggregate) even after absorbing an attack by the challenger, the incentive for the challenger to proceed with such an attack will be low. If the deterrent threat’s effectiveness is based on the necessity of preemption or can be significantly degraded by an opponent’s preemptive attack, the effect of time pressures on decisionmakers will increase. The more that time constraints become a focus of the challenger, the greater the likelihood that when making decisions, he will use stress reducing cognitive short-cuts and act on incomplete or improperly analyzed information. In doing so, the potential for irrational action will increase. A decisionmaker working from a limited, incomplete list of alternatives, under the pressure of psychological stress, may rationalize the slight potential for gaining an advantage by striking first and ignoring the deterrent threat—even though in the long run such a decision may prove counterproductive.” [Harknett (1994), 100-101]

Two points are worth highlighting. First, this suggests the dangers of trying to communicate commitment through “surge” strategies. A potential aggressor, observing or anticipating the surge of forces into theater and feeling under pressure to make a hasty decision, may irrationally leap into war. The development of Air Expeditionary Forces, maintained in continental United States and deployed to prepared fields in allied nations in crises, offers a clear example of the sort of strategy that is likely to stress an adversary unnecessarily and increase the risk of irrational escalation. By contrast, a strategy of communicating commitment through routine presence may be less likely to create this particular problem.

Second, this once again highlights the difficulty of maintaining conventional deterrence over time. The Desert Shield/Desert Storm expereience is illustrative in this regard. One of the lessons that potential regional aggressors have apparently learned is the importance of acting quickly, before American forces have been moved into place. This belief in the value of preemption increases the stress on potential aggressors and the danger that deterrent commitments will be irrationally ignored. Again, we observe a situation in which the demonstration of conventional capabilities actually serves to undermine the long-run effectiveness of deterrent threats.

To Summarize

The picture of conventional deterrence that emerges from the empirical literature is of a fragile, transient condition, a phase in an ongoing strategic interaction during which the deterrer has temporarily successfully convinced a dissatisfied power that it lacks any attractive option for resolving a political dispute through aggressive military action. This phase will pass, though, if the dissatisfied power becomes convinced that the costs of continued stalemate are prohibitive or the costs of defeat are bearable, if it comes to believe that it has designed an option that the deterrer lacks either the ability or will to counter, or if it fails to understand or pay attention to the deterrer’s threats.

II: The Importance of Denying Quick Victory

Despite this cautionary view of conventional deterrence, it is clear from the empirical record that not all deterrence policies are equally ineffective.

To the degree that potential aggressors are deterrable at all—and the preceding discussion suggests reasons for caution on this score—the key to effective conventional deterrence appears to lie in the ability to deny potential aggressors the belief that they have an option that will yield a quick victory and that will prevent events from slipping out of their control.

At first blush, the importance of denying a potential aggressor any option for scoring a quick success seems a logically obvious and indeed almost trivial finding. It is, after all, intuitively apparent that, from a potential aggressor’s point of view, “Short conventional wars are more desirable than lengthy ones because the former generally entail lower costs. Appropriately, war is most likely to start when the potential attacker envisions a quick victory. When the conflict promises to be more prolonged, but ultimate success is still considered probable, deterrence is nevertheless likely to obtain, largely because the protracted nature of the conflict will result in higher costs.” [Mearsheimer (1983), 23-24.]

On closer examination, however, this turns out to be a critically important empirical conclusion, with interesting insights into how policymakers make decisions and with sweeping policy implications for the conduct of conventional deterrence. What it suggests is that, in general, potential aggressors do not consider the full range of options, calculate the costs and benefits, weighted by probabilities, of each option, and then choose the one that offers the highest expected utility. Instead it suggests that potential aggressors are, by and large, uncertainty averse and are loathe to undertake policies that promise to let events slip out of their control.

There is, in fact, a considerable and growing body of evidence and literature, rooted in experimental psychology, on “Prospect Theory” and its implications for interstate aggression. This research strongly indicates that so long as decision-makers view their choices in terms of potential gains, they will tend to be cautious and avoid gambles even when those gambles offer, on average, net gains. By contrast, when they view their situation in terms of potential losses, decision-makers will tend to gamble, even against the odds. [Levy (1992a); Levy (1992b)] These Prospect Theory findings square perfectly with the evidence on conventional deterrence. As noted above, when potential aggressors find themselves under major domestic pressure—that is, when they see themselves as facing the prospect of major internal losses—they tend to be insensitive to evidence that their foreign policies pose enormous risks and, in net, seem likely to be a losing gamble. Ordinarily, however, when decision-makers consider aggression they will be risk averse and will overweight the risks of disaster, preferring a “sure-thing” to a gamble with a higher expected payoff.

Thus unless potential aggressors see themselves suffering a real loss if they do not act—e.g., unless they believe their domestic circumstances will be untenable if they do not undertake external aggression—they will be highly cautious about undertaking aggression. Given that a long war is a gamble, full of uncertainties and unknown and unanticipatable dangers, decision-makers will ordinarily avoid starting a war unless they (correctly or incorrectly!) believe a quick victory is assured.

Obviously, the possibility for self-deception and wishful thinking can not be underestimated. Decision-makers may be able to convince themselves that they can score a quick victory, keep events under control, and retain the initiative and freedom of action when this is not the case. But this nonetheless suggests where a deterrer’s efforts need to be focussed: on convincing a potential aggressor that it can not achieve a quick fait accompli and that it can not be sure a war can be wrapped up quickly.

As Alexander George and Richard Smoke conclude in their pathbreaking study of American deterrence efforts, “In almost every historical case examined, we found evidence that the initiator tried to satisfy himself before acting that the risks of the particular option he chose could be calculated and, perhaps even more importantly, controlled by him so as to give his choice of action the character of a rationally calculated, acceptable risk.... The initiator’s belief that the risks of his action are calculable and that the unacceptable risks of it can be controlled and avoided is, with very few exceptions, a necessary (though not sufficient) condition for a decision to challenge deterrence, i.e., a deterrence failure.... The initiator’s belief that the risks of options available to him are not calculable or controllable is usually a sufficient condition for deterrence success, with respect to those options.... The initiator’s desire to be able to calculate and control the risks of any option he employs will constrain his choice among options.” [George and Smoke (1974), 527, 529, 530]

A long war, by its very nature, offers much less calculability and control: even if the balance of resources indicates that, in the end, the aggression will be successful, the longer the war is expected to be, the more that could go wrong and the greater the danger that, in the end, the total costs will be disastrous. Long wars, in other words, pose the danger of a quagmire of uncertain dimensions and stickiness. This is an issue in conventional conflicts in a way tit is not in nuclear ones. “Costs in a conventional war accumulate in a gradual manner that is often difficult to anticipate. In World War I, the classic example, there was little appreciation beforehand of the losses that lay ahead.... Conventional war... is necessarily more protracted [than nuclear war], thus allowing a nation to adjust to increasing losses. A nation might very well absorb losses in the end that it would have considered prohibitive at the outset of the conflict.” [Mearsheimer (1983), 23]

This desire to control risks and avoid policies which threaten open-ended commitments and costs has important implications for the kinds of military options that potential aggressors will find attractive and unattractive.

Potential aggressors are likely to be deterred if the only military option at their disposal calls for a long war of attrition.
Potential aggressors are most likely to challenge the status quo if they believe they have a military option that promises a quick, decisive victory—a “blitzkrieg.”
Lacking the capacity for a decisive “blitzkrieg” victory, potential aggressors may undertake a limited-aims war if they believe that war aims can be kept under control and that the victim (and its allies) can be induced to make peace after an initial defeat rather than continue the war.

Based on an examination of twelve cases of deterrence success and failure between 1938 and 1979, one major study concludes “If a potential attacker believes that he can secure a decisive victory only by means of an attrition strategy, deterrence is very likely to obtain. Because of the very high costs and the difficulty of projecting the final outcome in a long war, the attrition strategy is the least desirable of the three [i.e., attrition, blitzkrieg, and limited aims wars]—and the one most likely to be associated with deterrence success. Liddell Hart, whose views were profoundly influenced by his experience in the Great War, once remarked, ‘Of what use is decisive victory in battle if we bleed to death as a result of it?’ His rhetorical question lays bare the foundation of conventional deterrence. If a potential attacker cannot effect a blitzkrieg, he will undoubtedly consider employing a limited aims strategy. Despite the lack of real enthusiasm for this strategy, it is an attractive alternative to the attrition strategy. Once the limited aims strategy is deemed worthy of consideration, the key question is: Will a limited victory lead to a war of attrition? If the answer is yes, deterrence is likely to obtain; if the answer is no, deterrence is likely to fail.” [Mearsheimer (1983), 64; see also Huth (1988), 39-41]

The fact that potential aggressors are risk minimizers who are likely to be preoccupied with finding options for quick military victory, and who, unless they see themselves as facing losses unless they act, are likely to eschew options with open-ended dangers, has important implications for the military realities that are important to deterrence success.

Conventional deterrence is affected by “immediate” and “short-term” military balances—that is, the balance of forces able to deny an aggressor its objectives “at the outset and early stages of an armed conflict.” [Huth (1988), 74]
There is no evidence that conventional deterrence is affected by the overall military balance and military-industrial capacity of the deterrer, nor does the possession of nuclear weapons appear relevant to the deterrence of aggression by nonnuclear powers. [Huth (1988), 81-82]
The capacity to inflict retribution is less effective as a conventional deterrent than the capacity to deny a fait accompli.

The most convincing support for these findings comes from Huth’s analysis. [Huth (1988)] Huth distinguishes between the overall balance of military forces and the immediate and short-term balances. Focusing on terrestrial forces able to control disputed territory, he operationalizes the immediate balance in terms of “those land forces of the potential attacker in a position to initiate an attack and those land forces of the defender and protege in a position to repulse such an attack.... To be in a position to engage in combat, the opposing forces must either be concentrated and prepared for action at the point of likely attack or be capable of mobilization for immediate combat from forward positions to the point of attack.” [Huth (1988), 58-59] He operationalizes the short-term balance as “the capacity of the attacker and the defender and protege to augment the immediate balance of forces by mobilizing ground and airforce manpower as well as the first class of trained reserves.... The short-term balance of forces measures the potentially effective strength of the opposing sides in the first few months of war, in which the armed strength of a country depends largely on the mobilization potential of standing peacetime forces.” [Huth (1988), 60]

Based on his probit regression of an identified universe of 58 cases of extended-immediate deterrence (that is, cases of deterrence or attempted deterrence of an actually threatened attack against an ally) between 1885 and 1983, Huth concludes “the immediate and short-term balances of military forces had a significant impact on the probability of deterrence success, whereas the long-term balance of forces had no significant impact.... These results support the argument that potential attackers seek quick and decisive results with the use of force at relatively low cost. It follows that the most effective military deterrent is the capacity of the defender to repulse an attack and deny the adversary its military objectives at the outset and early stages of an armed confrontation. The capabilities of the defender and its protege to prevail in a prolonged struggle are not an effective deterrent because the potential attacker does not initiate the use of force with the intention to engage in a war of attrition. Deterrence based on the threat of denial is much more effective than the threat of punishment in a protracted war.” [Huth (1988), 74]

More specifically, Huth’s probit regression leads him to conclude that “In most cases of deterrence failure either the defender’s forces were not in a position to repulse the initial attack, giving the attacker a decisive advantage in the immediate balance of forces and/or the defender’s capacity to mobilize and reinforce local forces did not alter the short-term balance of forces in its favor. Deterrence failed in only 17% of the cases (4/24) where the defender possessed military capabilities comparable or superior to those of the potential attacker in both the immediate and short-term balance of forces.... In most cases of successful deterrence, in contrast, the defender and protege maintained a favorable balance of forces with reinforcements or buttressed a rough balance of local forces with superior reinforcement. In 88% (30/34) of deterrence successes the defender and protege possessed comparable or superior military capabilities in either the immediate or short-term balance of forces.” [Huth (1988), 75-76]

Huth reaches similar conclusions based on his more detailed analysis of six case studies: “deterrence failure is likely if the potential attacker believes it can achieve a quick and decisive defeat of the protege without becoming embroiled in a larger conflict with the defender.” [Huth (1988), 86] Other studies have underscored this finding that the local balance is more important than the overall balance in shaping the potential aggressor’s decisionmaking—that potential aggressors are influenced by perceived opportunities in the local environment and fail to consider the full weight that can be brought against them. [See, for example, Arquilla and Davis (1992), 29]

Two obvious policy conclusions follow from these findings. First, peacetime engagement and overseas presence are likely to contribute significantly to deterrence during crises. Second, deterrers will need to be politically and logistically prepared to respond militarily to emerging and evolving crises: as a potential aggressor mobilizes, a deterrer must be able to respond. As Huth demonstrates, deterrence success was statistically correlated with a “tit-for-tat” military response to crisis. “The movement of naval and land forces by the defender can erase a local imbalance of forces and establish a formidable trip-wire deterrent capability in which the use of force by the potential attacker against the protege results in the direct engagement of the defender’s forces. The adoption of a tit for tat policy by the defender increased by 33% the probability of successful deterrence as compared with failing to match or exceeding the military escalation of the potential attacker.” [Huth (1988), 79]

There is a tension here worth discussing. Earlier we noted that efforts to signal resolve, and the movement of forces for this purpose, were unlikely to have much impact: potential aggressors, particularly by the time a crisis was underway, were unlikely to be able to interpret such communication effectively. And if these force movements did have an impact, it might well be negative, convincing a potential aggressor of the inevitability of war or pressuring it for a hasty response. Yet here we are suggesting that movement of forces may be important and have a positive impact on deterrence. The important distinction that must be drawn is between the communication of resolve implicit in some sort of show of force—which is likely to be ineffective—and the shift in military realities—which may either foreclose perceived military options or spur the potential aggressor to act quickly. Karsten et al. catch this tension and distinction nicely in their final conclusions: One is... ill-advised to place much faith in a show of force. There may be no harm and some value in a modest, nonbelligeretn military demonstration of resolve (a ‘probe’), but such demonstrations rarely suffice, and we fear that some policymakers may have lost sight of that amid the rustle of ‘options’ and the roar of jets. Other, more substantial troop movements to distant regions may in some instances be considered a sine qua non, demonstrating resolve and credibility and preparing the military for potential future action. But while they are necessary, they are not sufficient conditions for the success of threats, and they may even be counterproductive if they provoke a preemptive reaction on the part of the target.” [Karsten, Howell, and Allen (1984), 114-15]

The point here is that if a potential aggressor believes it has identified an option for achieving a quick victory and timely movement of sufficient military forces by the deterrer convinces the potential aggressor that this window of opportunity has been closed, deterrence may, for the time, hold. Or the deployment may trigger a rational or irrational preemptive response. But the key is actually putting the necessary forces in position to deny the potential aggressor its objective.

In sum, the evidence suggests that if potential aggressors are indeed deterrable—that is, if they are not so strongly motivated that they are willing to pay the price of conventional defeat, if they do not perceive options for designing around threats, and if they are capable of recognizing and responding rationally to threats—then the key to deterrence is likely to lie in denying the capacity for rapid victory. By doing so, the deterrer creates a situation in which the potential aggressor can not remain confident in its ability to control events and in which there is a risk that aggression will lead to unacceptable costs. To create this situation, the deterrer must convince the potential aggressor that it has forces present to prevent a fait accompli, or that it is able to respond in a timely fashion to prevent one.

III: But What if Deterrence Fails?

Obviously, this discussion raises the question of whether, or how, conventional force can be used to terminate a conflict if deterrence fails, short of completely defeating the aggressor. On this subject there has been relatively less research and there exists relatively less consensus. Nonetheless, the most recent and thorough study deserves some discussion here, particularly since its findings are in many ways consistent with the findings regarding deterrence considered above.

Robert Pape explores the problem of intraconflict compellence—attempting to coerce a wartime adversary into concessions rather than going for all out victory—and considers 33 cases between 1917 and 1991 in which strategic airpower was used in an attempt to force concessions. [Pape (1996)] His basic finding is that coercion by punishment does not work, and that strategic bombing (as opposed to theater bombing or close air support) is not typically the most effective way to employ airpower to achieve coercion by denial. “ Punishment strategies will rarely succeed. Inflicting enough pain to subdue the resistance of a determined adversary is normally beyond the capacity of conventional forces. Punishment strategies will work only when core values are not stake.... Denial strategies work best. They succeed if and when the coercer undermines the target state’s military strategy to control the specific territory in dispute.” [Pape (1996), 20; for a different view, emphasizing the vulnerability of states to social pressure, see Quester (1986)] In other words, coercion works by convincing an aggressor that it was wrong to believe that it could achieve a quick or decisive victory and that events will spiral even more completely out of control if it does not reach a political accomodation with the deterrer.

If Pape’s analysis is correct, this has obvious implications for U.S. strategy. Notions of precision engagement and the possibility of “taking out” civilian or leadership “targets that matter” and thereby undermining the political will of the aggressor are unlikely to be effective. Convincing an adversary to throw in the towel is likely to require joint operations that destroy the adversary’s actual military capability in the field. Pape is quite blunt: “Can air power alone do the job? The answer is no. First, coercion is very hard. It hardly ever succeeds by raising costs and risks to civilians. When coercion does work, it is by denying the opponent the ability to achieve its goals on the battlefield. However, even denial does not always work.... Second, strategic bombing does not work. Strategic bombing for punishment and decapitation do not coerce, and strategic bombing is rarely the best way to achieve denial.” [Pape (1996), 314]

If coercion is to be undertaken, it needs to employ air power as an adjunct to joint operations, rather than as independent tool: “using air power for denial entails smashing enemy military forces, weakening them to the point where friendly ground forces can seize disputed territories without suffering unacceptable losses.” [Pape (1996), 69] In highly mobile, blitzkrieg kinds of wars, Pape argues that air power is best used in operational interdiction; in more static wars, it should be used for ground support. [Pape (1996), 76-78]

As noted, Pape concludes that punitive attacks on populations are futile as a means of exerting coercive pressure during war. “Air attack on civilian populations, whether it seeks to kill large numbers or destroy the civilian economy, is not likely to coerce states in serious international disputes. Over more than seventy-five years, the record of air power is replete with efforts to alter the behavior of states by attacking or threatening to attack large numbers of civilians. The incontrovertible conclusion from these campaigns is that air attack does not cause citizens to turn against their government. Air power slaughtered British, German, and Japanese civilians in World War II; threatened Egyptian civilians in the 1970 ‘war of attrition’ with Israel; and depopulated large parts of Afghanistan in the 1980s. In each case, the citizenry remained loyal to its leaders. In fact, in the more than thirty major strategic air campaigns that have thus far been waged, air power has never driven the masses into the streets to demand anything.” [Pape (1996), 68]

Pape is equally dubious about claims that more precisely targeted anti-society campaigns, designed to inflict carefully defined suffering, are more effective than massive strategic bombing. “Electric power grids, internal tranportation networks, and dams were destroyed in Korea; electric power grids, oil refining, and internal transportation were also wrecked in Vietnam; and electric power, oil refining, and internal transportation were demolished in Iraq. In none of these cases, however, did civilian pressure induce governments to surrender.” [Pape (1996), 69]

Ultimately, Pape’s argument about the ineffectiveness of attacks on civilians and leadership hinges not only on his empirical evidence but on a broader theoretical point about mass political behavior. Pape concludes that “the citizenry of the target state is not likely to turn against its government because of civilian punishment. The supposed causal chain—civilian hardship produces public anger which forms political opposition against the government—does not stand up. One reason it does not is that a key assumption behind this argument—that economic deprivation causes popular unrest—is false. As social scientists have shown, economic deprivation does often produce personal frustration, but collective violence against governments requires populations to doubt the moral worth of the political system as a whole, as opposed to specific policies, leaders, or results. Political alienation is more important than economic deprivation as a cause of revolutions.” [Pape (1996), 24] This leads Pape to argue that punishment is not only politically unproductive but counterproductive. “In fact, punishment generates more public anger against the attacker than against the target government. Punishment does produce emotional stress, but this reduces rather than increases collective action against the government, because heavy punishment induces a ‘survival’ response and light punishment, a ‘Pearl Harbor’ effect.” [Pape (1996), 25-26] Only by bombing military targets, and removing tools of repression or inspiring the military to stage a coup, can bombing effect a change in government. [Pape (1996), 26-27]

Revolutions in precision, Pape reasons, do not change these underlying realities. To the extent that PGMs will make a positive contribution, it is by increasing ability to take out theater military targets and increasing attrition rates on the battlefield—thereby, ironically, making strategic bombing less relevant, since the ability to keep a war going until production would have a significant impact is dubious. [Pape (1996), 323-26]

As noted, Pape’s analysis and conclusions are controversial. If, however, they are right, they point again to the difficulty of using conventional force to change political behavior “on the cheap” and to the importance of convincing an adversary that rapid military victory is not possible. Notions of quick, clean, relatively bloodless conflicts, employing American technology and control of air and space to impose American will are cast into doubt. The implication is that political commitments may require major, bloody wars to carry out.

Conclusions

A review of the existing empirical literature thus suggests that conventional deterrence and conventional coercion do not offer much of a panacea. True, under some circumstances, threats of conventional response will be effective in buying time during which peaceful resolution of differences can be achieved. To accomplish this, the key is to demonstrate a commitment and capability to deny the potential aggressor a quick victory and to convince the adversary that events would escape its control. But conventional deterrence does not offer the prospect of the sort of stable stalemate provided by nuclear deterrence, and conventional deterrence and conventional war are likely to be intimately linked.

List of Works Cited:

Allan, Charles T. (1994) “Extended Conventional Deterrence: In From the Cold and Out of the Nuclear Fire?” The Washington Quarterly.

Arquilla, John and Paul K. Davis. (1992) Extended Deterrence, Compellence and the “Old World Order” (Santa Monica: Rand).

Bueno de Mesquita, Bruce. (1981) The War Trap (New Haven: Yale University Press).

Burrows, Simon. (1996) “Culture and Misperception: The Law and the Press in the Outbreak of War in 1803,” International History Review.

George, Alexander L. and David K. Simons. (1994) Limits of Coercive Diplomacy, Second Edition (Boulder: Westview).

George, Alexander L. and Richard Smoke. (1974) Deterrence in American Foreign Policy: Theory and Practice (New York: Columbia University Press).

Haffa, Robert P., Jr. (1992) “The Future of Conventional Deterrence: Strategies and Forces to Underwrite a New World Order,” in Gary L. Guertner, Robert P. Haffa, Jr., and George H. Quester, eds., Conventional Forces and the Future of Deterrence (Carlisle Barracks, PA: U.S. Army War College).

Harknett, Richard. (1994) “The Logic of Conventional Deterrence and the End of the Cold War,” Security Studies.

Huth, Paul K. (1988) Extended Deterrence and the Prevention of War (New Haven: Yale University Press).

Huth, Paul K. and Bruce Russett. (1984) “What Makes Deterrence Work? Cases from 1900 to 1980,” World Politics.

Jervis, Robert. (1994) “What Do We Want to Deter and How Do We Deter It?” in Benjamin Ederington and Michael J. Mazarr, eds., Turning Point: The Gulf War and U.S. Military Strategy (Boulder: Westview).

Jervis, Robert, and Richard Ned Lebow and Janice Gross Stein. (1985) Psychology and Deterrence (Baltimore: Johns Hopkins University Press).

Karsten, Peter, and Peter D. Howell and Artis Frances Allen. (1984) Military Threats: A Sytematic Historical Analysis of the Determinants of Success (Westport, CT: Greenwood).

Lebow, Richard Ned. (1981) Between Peace and War: The Nature of International Crisis (Baltimore: Johns Hopkins University Press).

Lebow, Richard Ned. (1985) “Conclusions,” in Robert Jervis, Richard Ned Lebow, and Janice Gross Stein, Psychology and Deterrence (Baltimore: Johns Hopkins University Press).

Lebow, Richard Ned and Janice Gross Stein. (1990) When Does Deterrence Succeed and How Do We Know? (Ottawa: Canadian Institute for International Peace and Security).

Levy, Jack S. (1989) “Quantitative Studies of Deterrence Success and Failure,” in Paul C. Stern, Robert Axelrod, Robert Jervis, and Roy Radner, eds., Perspectives on Deterrence (New York: Oxford University Press).

Levy, Jack S. (1992a) “An Introduction to Prospect Theory,” Political Psychology.

Levy, Jack S. (1992b) “Prospect Theory and Inernational Relations: Theoretical Applications and Analytical Problems,” Political Psychology.

Maoz, Zeev. (1983) “Resolve, Capabilities, and the Outcomes of Interstate Disputes, 1816-1976,” Journal of Conflict Resolution.

Mearsheimer, John J. (1983) Conventional Deterrence. (Ithaca: Cornell University Press).

Pape, Robert A. (1996) Bombing to Win: Air Power and Coercion in War (Ithaca: Cornell University Press).

Quester, George H. (1986) Deterrence Before Hiroshima, Second Edition (New Brunswick: Transaction).

Russett, Bruce. (1967) “Pearl Harbor: Deterrence Theory and Decision Theory,” Journal of Peace Research.

Shimshoni, Jonathan. (1988) Israel and Conventional Deterrence (Ithaca: Cornell University Press).

Wirtz, James J. (1993) “Strategic Conventional Deterrence: Lessons from the Maritime Strategy,” Security Studies.

Wolf, Barry. (1991) When the Weak Attack the Strong: Failures of Deterrence (Santa Monica: Rand).

Zinnes, D.A., and R.C. North and H.E. Koch, Jr. (1961) “Capability, Threat, and the Outbreak of War,” in James N. Rosenau, ed., International Politics and Foreign Policy (New York: Free Press).