Harald Müller. Handbook of International Relations. Editor: Walter Carlsnaes, Thomas Risse, Beth A Simmons. Sage Publication. 2002.
This chapter starts with definitions of the meaning and scope of the term security. It then gives an overview of the types of security cooperation in international relations. The third section defines puzzles that emerge from these cooperative endeavors among actually or potentially mutually hostile actors. The main body of the chapter tries to disentangle the puzzle from the perspectives of the major theoretical approaches in international relations. It describes how the theories explain the phenomena of states cooperating in the security field, and what opportunities and limits the theory would set for such cooperation. Each section also discusses the shortcomings–in the view of the author–of the respective explanation.
The understanding of security as the absence of existential threats to the state emerging from another state is under heavy fire from two directions. First, it has been stated that the state is not the correct, or at least not the single, subject of security. Social, ethnical, religious or otherwise cultural minorities (‘societal security’) (see Waever et al., 1993), individuals with their basic needs (‘human security’) (Suhrke, 1999) or else the world community or humankind (‘global or world security’) (Klare and Thomas, 1994), have been proposed as carriers of security needs of equal rights with states. Second, it has been emphasized that the exclusive attention devoted to the physical–or political–dimension of security of territorial entities is misplaced, at least in an age of complex interdependence. Other aspects relating to human life are as important in security terms: the availability of economic resources, stability, institutions and relations to provide for an adequate level of welfare (‘economic security’) (Borrus et al., 1992; B. Crawford, 1993, 1994; Kapstein, 1992; Orme, 1997/8; Sen, 1990), the integrity of the systems that supply and process information on which modern society is dependent (‘information security’) (Feaver, 1998; Keohane and Nye, 1998; Soo Hoo et al., 1997) and the natural environment as the basis for all life, but also the supplier of resources on which societies and states thrive (‘environmental security’). (Among countless contributions on this subject, consult Homer-Dixon, 1994; Levy, 1995.) It has been argued that this debate on the meaning of security is a reflection of new conditions of global politics engendered by increasing globalization (Cha, 2000).
The generalization and universalization of the term security has been criticized. What is the differencia specifica, and thus the analytical utility, of this term if it covers everything on earth? (See, for example, Deudney, 1990.) For the sake of argumentation economy, the chapter applies a rather conventional understanding of security: security between states, and related mainly to the organized instruments for applying force–the military in the first instance (Betts, 1997; Buzan, 1987). Even so, it must be noted that the concept is not simple. States are not unitary objects. Barry Buzan differentiates between the idea of the state, its physical base and its institutional representation. Security risks may apply to each of these three incarnations (Buzan, 1991: 65).
Cooperation among Friends and Cooperation among Rivals
Security cooperation is usually understood as collaboration between conflictuous parties; this is the reason why alliances are usually dealt with as entities sui generis. This distinction, however, is not completely convincing. If international relations are seen as anarchy wherein conflict is principally possible between any pair of actors, then alliances tend to be problematic. If the temporary security cooperation, undertaken to fend off an imminent threat, emboldens the transient partner too much, cooperation may end in less security rather than more, and in loss of sovereignty or extinction in the worst case. In other words, if we look at alliance partners of today as potential enemies of tomorrow, the latent conflict dimension becomes visible. The difference to other forms of security cooperation, for example the non-proliferation regime is then more one of latency versus visibility, and thus more gradual than principal.
Security cooperation implies relying for an essential objective, national survival, on the resources, intentions and activities of other states, which is hard to reconcile with the notion of security being guaranteed exclusively by self-help. In addition, security cooperation entails some loss of freedom of action, some constraint on one’s ability to accumulate as much military power as resources permit, some sacrifice in options; for instance, enhanced transparency may mean reduced chances of achieving surprise, a premium value in most strategic writings. Arms control, for instance, aims at reducing offensive, destabilizing options. In wartime alliances, the mutual dependency can be extremely high. Since in war the existence of a state is easily at stake, relying on someone else is an existential issue; strategic choices have to be most closely coordinated, freedom of action might be entirely lost. Peacetime alliances can be much more relaxed and may be hardly more than token promises that might be broken in the first second of the hour of truth. But they may also entail intimate structural cooperation or integration. The degree of mutual knowledge, transparency and dependence within NATO was and is breathtaking when seen from the perspective of potential hostilities in the future. In that sense, NATO member states put much more on the line than participants in average arms control or non-proliferation agreements. If there is a puzzle to be solved in explaining security cooperation, it concerns alliances as well.
Post-war Security Arrangements, Concerts and Collective Security
There are three more variants of security cooperation that merit a brief discussion. The first is the relation former enemies must establish after a postwar settlement. The enemy is suddenly turned into a partner, but memories of past hostilities are still vivid, the risk of reversal looms large, and personal animosities and grievances overlap with genuine concerns over cheating, hidden weapons and the like.
A different type of security cooperation is the work of the United Nations Security Council, notably its permanent members. The Charter has invested considerable responsibilities and rights in this body. The idea was to create an institutional framework wherein the big powers could sort out their problems with each other and join their considerable clout to prevent others from stirring up trouble. Historically, the performance of the UNSC fell short of expectations most of the time. Yet cooperation has occurred; given the differences in geopolitical interests, this remains a remarkable fact.
Finally, collective security is a normative proposition to replace the dangers and uncertainty of balancing. The states belonging to a collective security system promise to come to the aid of the attacked victim if one of them would dare to start an aggression: Note that the attacker is presumed to be a member of the community, not an outsider (in which case it would be an alliance, not a collective security system) and that the aggressor is not known in advance. Both the League of Nations and the UN Charter have tried to incorporate elements of collective security, with limited success so far (see Betts, 1992; Claude, 1971; Downs, 1994; Kupchan and Kupchan, 1991; for surprising positive results of a computer simulation, see Cusack and Stoll, 1994).
Security cooperation has been in existence and continues to exist. During the East–West conflict, the growth of cooperative endeavors was impressive even while the rivalry was continuing (George et al., 1988). The end of that conflict resulted in an unprecedented intensification of that collaboration, and was influenced and accelerated by it. From the mid-1990s on, the dynamic slowed down, without, however, grinding to a halt. Meanwhile, global cooperation such as the Nuclear Nonproliferation Treaty, the Biological and Chemical Weapons Convention, the Conventional Arms Register, or the Anti-Personnel Mines Convention had long since elevated security cooperation from the narrow confines of the East–West conflict. Peacekeeping has usually brought together countries from all parts of the world. Also, regional efforts have grown. Nuclear weapon-free zones have been agreed in Latin America, South Pacific, Africa, Southeast Asia. Regional activities to deal with small arms have become a major topic for cooperation in such different places as the OSCE, West Africa and Latin America. Different types of confidence-building measures have been or are being tried in the Middle East, Central Asia and through the ASEAN Regional Forum. The list is not exhaustive, but illustrative.
The puzzle that theories try to solve is the possibility of future hostility confronting today’s partners in security cooperation. This poses a challenge of virtual schizophrenia for security collaborators. Security cooperation does not come cost-free, as shown above. States opting for security cooperation sacrifice a security asset to gain higher security by obtaining another asset that, they believe, helps them better to provide for their security: the collaboration of their potential enemies and the pursuant agreements and organizations. How states develop interests, interpretations and perceptions that permit them to jump into security cooperation is the challenge that all explanations are facing.
International Relations Theory and Security Cooperation
Realism sends us into confusion. Its reformulation by Kenneth Waltz (Waltz, 1979) has claimed pride of place for this theory because it follows the standards of natural science (never mind that it is Newton’s physics rather than Einstein’s or Planck’s). We thus expect the coherence of axioms and strict deduction, only to discover that realists deduce contradictory hypotheses from the same body of abstract theory.
Classical realists analyze state behavior as built upon the innate human striving for cumulating power. Competitive cumulation in an anarchical environment creates inevitable problems for state survival, for which, in turn, accumulating further power is the obvious solution. Neorealists waive the anthropological assumption in favor of a systemic imperative: the need to survive in anarchy where no sovereign grants security means that states are on their own for providing the necessary means to ensure their continued existence. While the starting assumptions vary between classical and neorealism, the theory-relevant consequences, the character of the system as anarchic and dangerous, and the formation of states’ preferences are very similar.
The Problem of relative gains
In its parsimonious formulation by Kenneth Waltz, and the ensuing interpretations by Joseph Grieco (Grieco, 1990) and Mearsheimer (Mearsheimer, 1990, 1994/5), realism is pessimistic about the prospects for cooperation beyond that based on hegemonic guarantee, persuasion and/or imposition because of the preferences states are assumed to hold as a consequence of the imperative of survival in anarchy (for example, Frankel, 1993; for a critical analysis compare Grunenberg, 1990). The problem lies in the shifting course of history. New polarities emerge from changes in capabilities, a new system structure evolves, and this may make our friend, ally and cooperation partner of today our competitor, rival or enemy of tomorrow. Neither do we know what another government really wants; we can know even less about its intentions tomorrow. We have therefore to take care that the spoils from cooperation are not distributed in a way that may disadvantage us–however marginally–since the margin might be used by today’s cooperation partner to impose his will upon us tomorrow. It is not that we can extract gains from cooperation which is the controlling factor influencing our strategic choice, but the distinct risk that the other could gain more: this is, in short, the problem of ‘relative gains’ that minimizes neorealists’ expectations for security cooperation. Of course, equality of relative gains does not require symmetric distribution among asymmetric partners mathematically; but the point is moot in security cooperation among rivals, where denying the other a superiority of resources that may pose dangers tomorrow is imperative. Since symmetric distributions are extremely hard to calculate reliably, and even harder to execute, the opportunities for successful cooperation are naturally very limited. Institutions do not help much; they are epiphenomena that wax and wane with the interest of the primary actors, the states. They have no remarkable influence on their own on these interests and the ensuing behavior (Waltz, 1986: 336).
The one type of security cooperation that plays a major role in realist accounts is the alliance. The alliance is a necessary tool of balancing when states’ own resources are insufficient to create an appropriate counterweight to the hegemonial endeavors of one state or a group of states. Alliances are problematic under the ‘relative gain’ assumption as well, but they are inevitable tools of convenience if the alternative–succumbing to the risk of an overwhelming power–is clear and immediate. Stephen Walt has modified the ‘balancing-against-capabilities’ hypothesis by proving that states usually balance against threat (see Walt, 1987) since threat is both a behavioral and a perceptual variable rather than a material one. Realist theory is stretched to the limits here. From another realist perspective, Randall Schweller has shown that bandwagoning–joining the stronger side rather than balancing it–is compatible with realist premises if we assume that the objective of this type of alliance policy is not security, but gain (see Schweller, 1994).
The most elaborate and elegant body of security cooperation theory within the realist paradigm is Glenn Snyder’s theory of alliances (Snyder, 1984, 1997). It stays well within the basic realist assumptions and creates a couple of dynamic ‘laws of alliance’ that catch well the contradictory impulses that mutual security dependence produces. According to Snyder, there is an intra-alliance equivalent of the security dilemma. Since the interests of allies are never completely congruent, a state allied to another will quickly face the choice between supporting its ally in a conflict where the state itself has no real interest, or keeping out of this conflict. Opting for support may drag that state into a deadly contest it may wish to avoid. Staying out may risk the defection of the ally. And vice versa, in the conflict in which the state’s national interest is most intimately involved, confronting the enemy very hard may scare the ally away, while accommodating the enemy may induce the ally to pursue its own version of appeasement. As Snyder has shown, even within the basic assumptions of realism, allying is far from being the straightforward and uncomplicated balancing act that simplified and popular versions of the theory would have it. Rather, it is a complex set of relations, fraught with contradictions and dilemmas. Interestingly enough, Snyder, though writing in realist terms most of the time, finds it necessary to introduce, ad hoc, the intrinsic binding quality of norms as a factor persuading states to enter alliances and live up to their commitments, a notion hardly compatible with basic realist assumptions (G. Snyder, 1997: 8, 35, 350, 355).
Optimist realists and the disappearance of the relative gains problem
Security cooperation among allies is certainly the easier part of the realist account of security cooperation. Dealing with cooperation among enemies is more challenging. Charles Glaser has refined the realist argument further, creating a new brand of defensive realism that claims to account better for the cooperation we observe in the real world (Glaser, 1994/5, 1997). Defensive realism does not refute the possibility of long-term security cooperation even under the assumption that anarchy is the structure of the international system and that the state actors, relying on self-help, strive to pursue their security as the overriding national interest. Glaser starts by defining cooperation as an instrument of self-help rather than its opposite; this definition, though, remains problematic as cooperation is always reliant on somebody else’s assets, intentions and behavior, which does not fit well with a common-sense notion of self-help. Self-help, it should be recalled, is put up as insurance against the uncertainties of an anarchic environment. Security cooperation does not eliminate the uncertainty about the present and future intentions of the partner(s). As far as a state’s security depends critically on the cooperation of others, the certainties of autarchic self-help are replaced by the uncertainties inherent in anarchy, as long as we accept the basic premises of realist theory.
Glaser argues further that the risks of cooperation must be weighed against the risks of non-cooperation, such as risky arms races that could be lost. He does not note, though, that this is a view that can be ascribed readily to weaker parties in highly asymmetrical races. A strong party can be confident to outrace its competitor(s), while in close calls an incremental investment might suffice to stay even and thus not to lose. Glaser notes–correctly–that what counts for security in the first place is not power as such, but how power translates into military capabilities. However, this reformulation of the balancing needs in an anarchic world helps cooperation only marginally. For one, ‘reserve power resources’ can always be transformed into additional military capabilities–if we believe in the universal fungibility of power resources, as realism does. And secondly, as balances of military capabilities are precarious, rational actor governments will wish to have a margin of insurance against a possible miscalculation of the balance, a margin that might look threatening from the other side. If all other realist assumptions hold, renouncing this insurance margin for the sake of cooperation might look very risky indeed.
Glaser opens another front against the conventional realist argument that uncertainty about intentions works as a show-stopper for cooperation in maintaining that signaling of defensive intentions is possible by foregoing offensive options and sacrificing offensive capabilities. He agrees that this supposes a clear distinction between offensive and defensive, and that, while the signal will become all the clearer the more advantages the military technology of the day gives to the offensive–because the more options the signaler would actually forego–giving the signals would become ever riskier, as the clearest signals would probably mean serious risks to the signaler if the other side would not join the cooperation train (also Van Evera, 2000; Christensen, 1997). The major problem, however, is the intrinsic difficulty in sending signals that will be read correctly. For one, the distinction between offensive and defensive is never really clear, and interpretations may deviate considerably from reality–see the outset of the First and Second World Wars. What is meant as an unequivocally defensive move–for example, NATO’s renunciation of conventional parity in Europe and the introduction of short-range tactical nuclear weapons as a stopgap measure–may be interpreted by the other side as outrageously offensive. We witness exactly this process presently in the context of the planning for theater missile defense in East Asia. Secondly, particularly in hostile relationships, the common ground for interpretation may be lacking. The West did not at all understand Khruschev’s bold attempt to reduce the Red Army, and did not respond in kind, a factor that accelerated the Secretary-General’s fall from power (see Grinevsky, 1996). In contrast, it learned to understand Gorbachev’s signaling, but how much did it take to reach this effect? Gorbachev not only reduced unilaterally Soviet offensive capabilities, he opened up Soviet society, freed dissidents and stopped competition in troublespots around the world. Even so, Chancellor Kohl reacted to his moves by calling him ‘another Goebbels’, and NATO came close to another round of nuclear deployments as late as 1989. The crucial point theoretically is that the receiver can never be certain that there is not a piece of private information retained by the sender which might discard the benign interpretation, such as unknown priorities, different standards for evaluation, or hidden assets.
In summary, while ‘defensive realism’ is certainly much better suited to come up with more convincing explanations why major powers succeed in cooperating in the field of security, it appears to–almost invisibly–introduce change into basic realist assumptions about actors and structure that are somehow alien to the realist creed: to view cooperation as a self-help strategy takes it for granted that the mutual compliance by agreed rules is a viable model of behavior in international relations. In the end, betting on compliance would rely on trust in the exclusion of risks by deception, strategic cheating, lulling and surprise. Likewise, the assumption that signaling will succeed must rest on a notion of common language and common interpretative culture which is non-trivial and certainly not part and parcel of the realist model of international relations: to the contrary, it could be read as drawing constructivist elements into realism through the back door.
Are realists still realists?
A particular puzzle for the realist creed is the considerable robustness of the nuclear non-proliferation regime (Davis, 1993). The quest for nuclear weapons as the ‘ultimate insurance’ against unforeseeable or uncertain threats should still be the primary choice for states acting in such an environment. If one sees the risk emerging from unwanted escalation, rather than from conscious power policies of actual or potential rivals, one breaks out significantly from the sort of calculus both classical and structural realism view as the rule for prudent statesmen. Relying on nuclear guarantees from others begs the question whether these others may be reliable, or shy away from making good on alliance promises, or may turn their superiority into blackmailing policies in the future. All these arguments are not just hypothetical. They have influenced states’ choices in practice: ‘Deterring the uncertain’ is very much French and British nuclear doctrine today; the reliability of allies has haunted NATO throughout the Cold War, and has influenced French and Israeli nuclear choices. Being subject to an ally’s nuclear blackmail was very much in de Gaulle’s mind. It is hard to see how non-proliferation can be accounted for within the realist theoretical body without borrowing elements such as trust, the intrinsic binding force of norms, and the like, from other theories.
Joseph Grieco himself, initially an impressive ‘pessimist’, has later modified realist theory so as better to accommodate the growing reality of cooperation. He has specified the conditions under which the ‘relative gains’ orientation may be relaxed, and thus cooperation may have a better chance of being realized. The most original, and also most problematic, deviation from standard realist theory is the admittance that the past–whether the other state has been for an extended period friend or foe–influences threat perception, and, consequently, the inclination to measure the relationship against the standard of relative gains or against some more relaxed referential system (Grieco, 1988). The crucial point here is that perception, interpretation of experience and the transfer thereof from one generation of policy-makers to the rest are arguments outside the realm of realism proper. They bear no causal relation to anarchy, the distribution of capabilities and polarity. They look breathtakingly like elements from a constructivist handbook. There is little doubt that they help to account better for variations in political reality, but clearly at the cost of consistency at the paradigmatic level.
The same verdict would apply to Randall Schweller’s ‘motivational realism’. He finds that the security dilemma does not rest in structure, but in the existence of predatory states, thereby moving the causes of insecurity from the system to the unit level and thereby, consequently, opening the road to problem-free security cooperation among non-predatory powers (Schweller, 1996). Correspondingly, Andrew Kydd argues that ‘security seekers’ create peaceful environments and thereby conditions for security cooperation (Kydd, 1997).
Even more compromising with other–neoinstitutionalist and liberal–approaches, Robert Jervis, whose ‘Cooperation under the Security Dilemma’ is the classic statement of the realist cooperation problematic (Jervis, 1978), and who has enumerated the formidable obstacles that anarchy puts in the way of security cooperation (Jervis, 1983, 1985, 1988), has recognized the capability of institutions to change preferences over outcomes, a strong break with the realist creed that states preferences, fixed by the survival imperative in anarchy, are virtually unchangeable (Jervis, 1999: 58–62).
The contradiction with regard to our subject divides ‘pessimistic’ and ‘optimistic’ realists (for the debates within the realist approach, see Brooks, 1997; Schweller and Priess, 1997). To assign sympathies to these different schools remains a difficult task. Pessimists deserve distinction as they preserve an optimum of theoretical coherence. Yet optimists, heroically sacrificing a measure of coherence, at least are able to explain some of the empirical security cooperation that we see every day (Legro and Moravcsik, 1999).
Realist and, even more so, neorealist thinking can be modeled in rational choice terms, even though the origins of that theory did not emerge from the rational choice paradigm. Neoinstitutionalist analysts are firmly rooted in that paradigm; its most important foundational texts start from rational choice (Keohane, 1984; Oye, 1986), and in the academic debate, rational choice has clearly preserved dominance in the neoinstitutionalist discourse (Lake and Powell, 1999). Like realists, they start from the assumption of self-interested actors working in an international structure of anarchy. However, their assumption about the consequences of anarchy are more forgiving and less fixed on the imperative of survival in an environment of danger. Consequently, their assumptions about states’ preferences are also not fixed on the relative gain assumption; they admit easily the desire of states to achieve absolute gains in welfare and security. Some confusion emerges from the neoinstitutionalist claim that their assumptions were identical with those of realism; they are not. In what follows it is shown that, if the harsh propositions of realism are adopted, neoinstitutionalist reasoning has a hard time explaining the emergence of cooperation, while if those assumptions are relaxed, it can contribute well to such an explanation.
Situation structure and problem structure as constraints on cooperation
Institutionalists have identified factors that influence the propensity of given constellations of interactions for the creation or not of international cooperation, security cooperation included. In particular, the ‘Tübingen school’ in Germany has devised two relevant typologies. First, the type of ‘situation structure’, that is, the game that is played, influences the likelihood that cooperation will emerge. The game ‘leader’, in which a player will always earn its best payoff from cooperation, the moves of that player’s partner notwithstanding, is obviously most favorable to cooperation. Prisoners’ dilemma is at the center, while ‘Rambo’, where one player always reaps his best payoff by defecting, is least probable to engender cooperation. The second typology is the value at stake in the conflict at hand. Where parties struggle for absolutely assessed goods, cooperation is relatively easy. Next is conflict about means to achieve an objective appreciated by either player. Third is conflict about relatively assessed goods, and least favorable to cooperation is conflict about values. Since security falls normally in the third category, it is obvious that cooperation is not easily achieved (Rittberger, 1990; Zürn, 1992).
Communication in neoinstitutionalism
The first neoinstitutionalist argument why cooperation in the security field is not only possible, but even likely under conditions of anarchy is that rational actors are not prevented from communicating as long as the exchange of information does not involve prohibitive cost. In indicating to their interlocutor their own interpretation about the situation both parties find themselves in, and in enlightening each other about their preferences, they may approach an outcome that is close to a Pareto-optimum and which neither has difficulties of accepting, since it serves the security interests of both parties well (Kydd and Snidal, 1993; Morrow, 1994).
The problem here is that security talk is not cheap in a realist world. Since Sun Tzu, the great classical Chinese strategist, it has been part and parcel of strategic thinking that surprise is the key to victory–and, conversely, falling victim to surprise may be the beginning of annihilation–while deception is a most useful instrument to achieve surprise in the first place. If this is so, then listening to information from somebody who is supposed to be a potential enemy is a treacherous endeavor. The intention might be sincere. But the intention might also be to filter in false information about the situation–for example to pretend to be weaker (or stronger) than one really is; this may lull the ‘cooperation partner’ into complacency before an onslaught; or it may intimidate the ‘partner’ into virtual surrender and the acceptance of an unequal distribution of whatever spoils the security cooperation might offer. Now countries considering security cooperation know of this danger; and they know the other side knows; and they know the other side knows they know, and so on ad infinitum. It is very hard to see, as long as we assume acute survival risks within anarchy, how cheap talk that is not really cheap helps us to arrive at cooperation (Johnson, 1993).
In another use of the communication argument, neoinstitutionalists explain that the gap from cooperation motivation to actual agreement can be overcome exactly if the partners can prove credibly their commitment to cooperation by giving signals that are costly to themselves; Sadat’s speech before the Knesset is a vivid example (Fearon, 1997; Kydd, 2000; Morrow, 1999: 87–9). However, two questions arise: first, how would a rational actor give such a signal first, not knowing if the other side reciprocates, but having to pay the heavy costs anyway? The willingness to do that, for sure, enhances credibility. But the calculus for making this move remains unclear: as George Downs and David Rocke have proved, uncertainty about a prospective partner’s responsiveness and a short time horizon make an initial cooperative step inadvisable under rational choice assumptions (Downs and Rocke, 1990: 201–5). Secondly, successful signaling in this sense presumes a common reference system for evaluating the content of this signal (compare the reasoning about ‘defensive realism’). This condition is by no means sure, as research about security cultures (see below) has revealed, and the emergence of such an ideational superstructure appears to be exogeneous to neoinstitutionalist theory based on rational choice.
The shadow of the future
A second argument is the shadow of the future (Axelrod, 1984; Axelrod and Keohane, 1986; Goldstein, 1995). The prospect of repeated games relaxes the risks of the first round; the higher the net present value of the future, the less imposing the value of ‘defecting’ will appear at present. Three problems arise here: first, even if the game is sure to be continued into the distant future, getting a good deal becomes even more important and may induce negotiators to bargain harder, leading to standoff rather than to cooperation (Fearon, 1998). Second, because survival is at stake, the future is heavily discounted; this diminishes the expected payoff of cooperation and thus reduces the incentive to cooperate (Stein, 1990: ch. 4). Third, it is rather uncertain how many further rounds of the game will take place; if the security stakes are high–as they usually are–then the risk of the ‘partner’s’ defection with asymmetrical gains in its pocket loom very large indeed (Lipson, 1984). A leading proponent of neoinstitutionalism has countered that what is needed is not absolute certainty that the game will be played in future rounds, but only a ‘certain’ probability that this will be the case, and that in more benign situations–certain alliance types, for example–an expectation iteration is not needed at all (Keck, 1997: 50–2). Unfortunately, the threshold for this probability was not quantified; but even if it were, the solution begs more questions than it answers. If there is a ‘certain’ probability (p) that a further round will be played, then there is the complementary probability (1-p) that it will not. My expected utility is then diminished by exactly that probability times the expected damage which a defection will cause. A rational actor, desiring to neutralize this risk, will look for insurance against it. But incidentally, the assumption of anarchy eliminates also the practical institution of a neutral insurance company. The two insurance possibilities remaining are either to provide for an ally in case of defection of the ‘partner’ or to set aside some resources to make up for the loss. These two options, however, push us back into the well-known vicious cycle of the security dilemma: either enhances my own capabilities and must look somehow ominous to my partner-in-spe. Will not the enhancement of my capabilities–at the outset of security cooperation!–look like a preparation for defection? And if so, would not the partner feel compelled to reciprocate?
In fact, the partner would feel compelled to take some measures even before he discovered my own insurance moves. Being in exactly the same situation, and being a rational actor, the partner would as well have to consider insurance action before entering security cooperation for good. In other words, the evil 1-p probability of defection, however small or big p might be, propels us back into the security dilemma which we desired to leave through the door of security cooperation in the first place.
The relativity of the relative gains problem
A third argument put forward by rational choice institutionalists is that the problem of relative gains does not kill the possibility of cooperation as long as the asymmetries of gains are so small that they cannot be translated into decisive military advantage (Powell, 1991, 1994; Snidal, 1991a, 1991b). This sounds reasonable. It echoes the famous dictum of Henry Kissinger in the heated debates, about the SALT I Interim Agreement: ‘What in heaven is nuclear superiority?’ Nevertheless, we have to note that in this debate the other side believed that the small inequalities inbuilt in the marginally higher upper limits for the Soviet Union were relevant indeed; the partisans of this view, led by the late Senator Henry ‘Scoop’ Jackson, prevailed in inserting the prescription into the ratification resolution that no future nuclear arms control treaty must establish lower limits for US forces. This episode serves as a reminder that the asymmetry problem is anything but trivial.
The problem is twofold. First, it is intrinsically hard to quantify cooperation gains in the first place. Given the fluid situation in international politics, the permanent possibility of allies changing sides, the evolution of military technology and the virtually insuperable difficulties of assessing their meaning (see above), distributing the spoils of security cooperation in an equal way is close to impossible; this was discussed in some detail in connection with realism. The argument applies here as well.
Second, the future consequence of a present distribution of gains remains bogged down in uncertainty. Every investment can carry enhanced profits. Small asymmetries may be amplified through smart use by the better winner, and used to secure even more unequal distribution outcomes in later rounds. A slightly stronger party may be able to press its advantage, either against immediate rivals, or to force a weaker party to bandwagon. It is all very well to maintain that relative gains only matter if they accumulate in the future to dangerous levels (Matthews, 1996, 112–46), but how can a state be sure they will not? In other words, the shadow of the future, usually seen as a benevolent element for security cooperation, may turn upside down and impede, rather than foster, collaboration among self-interested parties that behave strictly within the rules of rational choice textbooks. Even if we assume that there is somewhere a threshold between small and big gains in asymmetries, small asymmetries are still good enough to make governments nervous as long as the rational choice paradigm is ruling within a ‘realist’ type of anarchic environment. Only if we relax the assumption about the threat of survival and/or introduce common knowledge about the persistence of non-threatening preferences among the states can institutionalist assumptions help explain the emergence of security cooperation.
The perseverance of cooperative regimes
Rational choice institutionalism is in a much better position to analyze and explain the perseverance of security cooperation once it has been established in the first place (Keohane, 1984). Security institutions are a tool to reduce transaction costs among parties. They open communication channels with established rules: parties are told what they have to report, what they can expect to receive from their partners in terms of relevant information, and they are given standards of evaluation to review and scrutinize that information (Mitchell, 1998). Games played over extended rounds within the framework of an established institution prolong the shadow of the future considerably and positively. Means are available to clarify ambiguities–an evil problem in an unregulated security environment. In multilateral frameworks, parties can rely on the solidarity of regime communities if ambiguities turn out to be real and effective breaches of the rules. Lowered transaction costs, provision of information and institutionalized enforcement are important incentives to cooperate (on enforcement, see Downs et al., 1996). In comparison with the ‘standing alone’ posture in an uninstitutionalized world, these are tangible advantages. They explain why countries most of the time abide by institutions’ rules–even if in their absence they would have pursued different policies–and support them as long as the overwhelming majority of their partners do the same (for example Duffield, 1992, 1994).
Another rationalist–institutionalist argument helps us understand why institutions show an astonishing robustness despite challenging changes in the international structures that existed when they were created. For instance, the Nuclear Non-proliferation Treaty has survived the breakdown of bipolarity, the structure instrumental in bringing it to life, the rule-breaking by both Iraq and the Democratic People’s Republic of Korea, and the defiance by India and Pakistan demonstrating their nuclear weapons status, and thus the failure of the NPT to achieve the objective of full universality. A second impressive example is the Treaty on Conventional Forces in Europe (CFE), concluded to establish a balance of forces between the two alliances of the East–West conflict. The conflict has gone, one alliance has disappeared, the other one has swallowed part of the former’s membership. One could surmise that the CFE Treaty had outlived its mission, and yet member states have worked with enthusiasm to adapt this apparently obsolete instrument to the new circumstances! Finally, the perseverance of NATO even after the enemy against which this alliance was established had disappeared is comprehensible in an institutionalist framework much better than in a realist one (Hellmann and Wolf, 1993; Lepgold, 1998; McCalla, 1996; Skalnes, 1998).
Rationalist institutionalists tell us that governments calculate the utility of existing institutions against the investment costs for new ones. This calculation, in most cases, results in unequivocal support for the–however imperfect–institutional structures that are there. To summarize, neoinstitutionalism, built upon the rational choice paradigm, gives a good explanation why countries should wish to cooperate in the security sector (Keohane and Martin, 1995). Their rigorous models force us to think through thoroughly our assumptions about actors’ preferences and the constellations they create.2 It is also apt to overcome the domestic/international divide by constructing two-level games (Evans et al., 1993; Zangl, 1994) or to model national leaders as replaceable ‘principal-agent’ of domestic politics (Bueno de Mesquita and Siverson, 1995; Morrow, 1991). However, it has great difficulties in explaining convincingly how actors can overcome the considerable barriers that the structural constraints rationalism buys from realism–unfettered anarchy, self-interested actors–pose to the jump from the mere motivation to cooperate to real cooperation. The call for including cognitive factors into rationalist cooperation theory has thus been heard repeatedly, including in one of the most meticulous rationalist case studies on security regime formation (Bernauer, 1993: ch. 7; see also Hasenclever et al., 1997, ch. 6).
Liberal theory seeks the roots and causes of external behavior in domestic structures and processes. Accordingly, different domestic structures will cause different preferences in security policy, notably variations in the inclination to enter security cooperation. At the most general level, then, liberal theory would analyze security cooperation as the result of a convergence of benign, cooperation-prone national preferences, engendered by domestic coalitions for which such cooperation obtains priority (Risse-Kappen, 1991).
Liberal cooperation theory: democratic peace
There is a particular branch of liberalism that has developed substantive hypotheses about preference-formation in democracies bearing on security cooperation: this is the theory of ‘democratic peace’, first developed by Immanuel Kant, but now with a very solid body of IR theorizing behind it. In its ‘monadic’ variant, based more on the Lockean and commercial foundations, it is hypothesized that democracies prefer peace to war because of citizens’ basic interests in self-preservation and welfare and wish to avoid costly (and sinful) external violent adventures. Another root cause is the value orientation of the democratic citizen, which prefers non-violent means of conflict management and solution to the decision by the sword, because he and she appreciate the value of human life. In the same vein, democracies and their citizens are disinterested in the costs of maintaining highly armed standing armies, preferring the minimalist defense posture of militias instead (Russett, 1990, 1993).
It is not the objective here to discuss the contribution of democracy to peace, but what follows for security cooperation. Deducing from democratic peace argumentation, we would expect democracies to seek understandings with potential enemies–democracies or not–to provide for procedures to settle conflicts peacefully, externalizing the internal modes of conflict management, and to find ways to agree on minimalist military postures in order to prevent costly arms races. Kant himself, and consequently the modern version of the theory, has proposed international organization as a way to frame the international interactions between the republics (B. Crawford, 1994; for a promising variant compare Deudney, 1995; more critical on this point is Remmer, 1998). But the theory is ambivalent on whether this mode of cooperation is also appropriate for relations between democracies and non-democracies. However, since their intrinsic peacefulness would compel democracies to seek ways to settle their problems with non-democratic states also in a non-violent manner, it would appear that in the monadic version of this theory organized security cooperation and arms control/disarmament in a heteronomous setting would be a logical deduction (Russett et al., 1998; for additional empirical evidence see Leeds and Davis, 1999). In other words, what appears a puzzle for other theories–the alliance version of security cooperation as well as cooperative efforts to come to collaborate with rivals and enemies–liberal theory sees as the inevitable and consequent outgrowth of a particular form of internal rule. The reasoning would allow for all forms of security cooperation we know of, be it alliance (among democracies), arms control, broader security regimes and collective security arrangements.
The dyadic variant maintains that the ‘peacefulness’ argument applies only to relation among democracies. Initially, this insight was forced upon theorists by empirical findings that the warlikeness of democracies was not so different statistically from that of non-democracies, but that ‘democratic wars’ were fought almost exclusively against non-democratic states, while among democracies a ‘zone of peace’ prevailed. However, there are voices–and some empirical support–suggesting that democracies may be generically less war-prone than other systems (Ray, 1995). There is some debate on the causes of this difference. One school argues for institutional affinity: open, welfare-orientated societies are more prone to develop patterns of cooperation with each other than other types of political systems. Cooperation intensifies and extends its scope, ultimately into the security area–a neofunctionalist argument. In addition, transparency, an essential condition for deep and intense security collaboration, is an attribute of democracy, but not of autocratic or totalitarian states. Consequently, democracies lose much less in opening up their military sector to the eyes of their partners than these other states, and feel more comfortable as transparency rises. We should thus expect security cooperation in all forms to be much stronger among democracies than between them and non-democracies.
Another school argues with the mechanism of mirror-image empathy embedded in political culture. Democracies have an image of themselves as peaceful and rational; they project this image onto fellow democracies. Being democratic is interpreted as a valid marker for peaceful intentions. As a consequence, the security dilemma shrinks, and the risk of being cheated or of getting unequal gains out of cooperation is seen as negligible (Owen, 1994; Risse-Kappen, 1995a; Weart, 1994, 299–316). Democracies are thus not only able to develop far-reaching regimes of cooperation among each other across the security sector, or to enter long-term military alliances, but to take serious steps at integration, such as joint commands, joint procurement, joint planning or even joint services (such as the integrated navy of the Benelux states). Democracies have much better prospects to develop long-lasting, intimate alliances (Risse-Kappen, 1995b; Starr, 1997; more critical on the joint democratic organization thesis are Lai and Reiter, 2000) and even security communities, an advanced type of security cooperation that will be described in more detail in the next section.
As a pragmatic consequence of liberal theory, democratization appears as the necessary tool to extend the zone of peace over the globe. If democracies are more cooperation-prone and peaceful than others, or if at least security cooperation can be taken for granted as the prevailing mood within and among democratic states, then the more democracies, all the better for security cooperation! In fact, this conclusion has long trespassed across the boundary of academic discussion and has become the creed of Western security policy, providing a powerful self-image–and ideology–for the Western alliance (Muravchik, 1991).
Empirical evidence gives a first-cut confirmation of aspects of both versions of liberal theory. For monadic liberalism, the growth of security cooperation–bilateral, regional and global–has seen democracies in the driver’s seat. The most essential attempts at providing security jointly, the League of Nations and the United Nations, emerged from concepts developed in democratic countries (notably the United States). Equally, arms control and nonproliferation were ‘invented’ in the United States and quickly picked up by other democratic states who persuaded others to follow suit. The dyadic version finds some affirmation in the fact that security cooperation is much broader in scope and more intense and deep between democracies than between them and non-democracies, not to mention security relations among non-democratic states (Mousseau, February 1997, 73–87). NATO and the EU are the cases in point. Indeed, it is plausible that the transfer of the structural and normative principles that guide conflict management inside democracies to the security cooperation institutions of the Western world are primary causes for both their relatively smooth operation and their longevity (Ikenberry, 1998/9).
Why do democracies behave differently in security cooperation?
Nevertheless, liberal theory leaves open some very important questions. To start with, one should assume–on the basis of the theoretical argument and the causal chains assumed by either version–that the behavior of democracies in security cooperation should be rather similar, and that variations for the same democracy over time should be limited. This, however, is not the case. If we compare the present inclination of democratic states to expand the realm of security cooperation in ways that imply further constraints on national sovereignty, we find countries like Canada, Sweden, the Netherlands or Germany in the forefront, the United Kingdom somewhere in the middle, and France, and even more the United States, towards the end. For the United States, the commitment to security cooperation evolves through waves, with the early 1980s and the late 1990s showing particularly strong unilateralism, and the early 1970s and early 1990s displaying determined efforts to cooperate.
Within liberal theory itself, we find some clues to explain this deviation from standard expectation. Democratic peace theory is a particular mode of liberal theory that predicts conclusions on expected behavior from fundamental system attributes. However, other variations of liberal theory rely on more detailed differentiations between various types of democratic society and their relation to the political system, or between different types of democratic institutions. If we take this as a starting point, we may expect variations in security cooperation depending on the strengths and weaknesses of countervailing forces within a democratic country: the pluralistic structure of policy-making gives various interests different chances of access and influence, and the outcome then depends on the balance of chances of the pro- and contra-security cooperation forces within a country (for the argument and additional empirical studies, compare Risse-Kappen, 1994 and Knopf, 1998).
An example for such a preference-biasing factor might be the ‘military–industrial complex’. This socio-political formation is thought to present a coalition of the defense industry, the military services, security think-tanks, military research and development establishments, defense-dependent local or regional governments and their representatives in national parliaments. While this formation is not necessarily opposed in principle to all forms of security cooperation, it would request constraints on it if it threatened priority armament projects, the financial resources allocated to defense, or research and development options. To preserve a considerable degree of freedom of choice in defense is thus a basic interest of the MIC that would tend to keep security cooperation limited unless more cooperation-minded forces were to overwhelm the collective MIC influence.
Can democracies cooperate with non-democracies?
There is another difficulty for the dyadic democratic peace strand of liberal theory in both versions: to come up with explanations as to how security cooperation is possible with non-democracies. This appears so easy to the monadic version, as democracies are supposed to pursue collaboration ‘tous azimuts’(all directions). However, since this theory bases the roots and prerequisites of cooperative behavior so exclusively in the structure of the political system of democracy, it can hardly explain where motivations to reciprocate should emerge from in non-democratic systems. It must thus be assumed that democracies are offering such outrageously asymmetrical deals in their own disfavor that the asymmetrical gains accruing from such agreements satisfy the nasty interest calculus of the non-democratic governments; but then it is difficult to understand how such cooperation results could resonate well in an enlightened electorate.
The dyadic version is facing a different problem with similar results. The increasing in-group feeling of democratic states–facilitated through their extraordinary inclination to join together in international organizations, including security alliances, an important pillar of the democratic peace–produces a commensurately growing aversion against the non-democratic ‘other’. Personification of this hostility and demonization of the chosen personality (Saddam Hussein, Slobodan Milosevic) creates the image of a dangerous and powerful, alien enemy who must be vanquished before he can do too much damage to the democracies and their security environment. This explains well why democratic states gang up together to solve conflicts with non-democracies–Iraq, Yugoslavia–by force if needed. It falls short of a convincing explanation of how this aversion can be overcome so that collaboration is possible with the loathed ‘other’; however, as we have seen, such collaboration has been seen and begs explanation. Again, the most obvious answer is a rational choice calculus; another would be customs or cultural habits. None of these answers are really rooted in liberal theory. The first intrudes into rationalism, the second into constructivism. It thus appears that it would be difficult for liberalism to stand alone in the battle for good explanations as to why security cooperation is possible.
Antinomies within the theory
Three further puzzles and contradictions deserve mentioning. The first concerns an internal contradiction between norm dynamics in democratic systems and the requirements of security cooperation. Privacy and ownership rights count among important normative prescriptions and proscription in democracies. Transparency requirements, however, cut deeply into these rights of private actors. The difficulties to arrive at a consensus in the endgame of the Chemical Weapons Convention negotiations, the reservations introduced by the US Senate into the ratification decisions which, inter alia, prohibit taking chemical samples from US industry outside US borders (in strict contradiction to the stipulations of the CWC verification system) and the barriers to a successful negotiation of a transparency and verification protocol for the Biological Weapons Convention pay witness to this fundamental problem.
The second concerns a vicious dynamic that applies to both monadic and dyadic democratic peace theory. If democracies are shy to risk their citizens’ lives in war, and bear sympathetic feelings even to the subjects of an enemy–having internalized the value of human rights and human lives–but still regard the possibility of being attacked as real, then an important conclusion emerges for their armament policy: they must try to develop weapons technologies that would be capable of protecting their own country while limiting victims on both sides in war, and helping to finish war very quickly. Unfortunately, such arms technology providing superiority runs counter to the requirements of security cooperation through arms control regimes and is powerfully feeding the security dilemma. In other words, the very motivations that are an important drive for democratic peace may tend to cause an armament research, development and procurement policy that prevents security cooperation. The prominence of National Missile Defense and the Revolution in Military Affairs fits this hypothesis well.
Finally, there might be a last contradictory dynamic that is both part of democratic peace-driven security cooperation while at the same time obviating it. The extension of the zone of democratic peace is, by the logic of the theory, the only secure way to work towards eternal global peace. However, an increasing number of democratic states working together intimately in the security field may look very threatening for those non-democratic states who are not–or much less intensely–participating in security cooperation with the community of democratic countries. Feeling overwhelmed, these countries may then seek their security in increasingly unilateralist armament moves. Again, some statements coming from Moscow in response to NATO enlargement, and echoed in Beijing, may point to such a vicious mechanism.
Constructivists do not believe in fixed, quasi-natural structures or individual preferences that give way to quasi-mechanic laws of system development and actor behavior. They see agents’ preferences as being in eternal development, and structures as historically evolved and thus malleable. However, malleability at both the agent and structure sides is not limitless. Structures change usually only slowly, and what agents think and do is influenced by constraints on and opportunities for action that structures afford. Nevertheless, there are leeways. Situations where action is required retain a degree of indeterminacy. Several options are open, and if agents opt continuously for options at the limit of what structure admits, structure will be subject to change. As Alexander Wendt has shown, this theoretical model for agent–structure change can be applied to the security dilemma. Since the environment is not depicted, as in realism, as necessarily survival-threatening, states are in a position to choose non-aggressive, defensive options for national security policy and cooperative options for inter-state security interactions. By continuing this choice over time, state-agents are capable of changing their security environment–with little risk to come–from highly competitive (and deadly dangerous) to highly cooperative (and much more forgiving).
In the constructivist paradigm, the emergence of security cooperation, thus, rests on two requirements:
- A structure that permits to an actor moderate, non-aggressive moves without the immediate risk of perishing; this requirement, one can boldly state, was given throughout the ‘Westphalian’ age, and more and more so as the twentieth century approached its close.
- Agents that would discover such options and choose them as better alternatives to a perpetual rivalry for superiority, with all the unpleasant risks such a course of behavior entails.
The malleability of security concerns
Structures in the constructivist paradigm are not primarily material. Matter can only contribute to structural features if it acquires social meaning. If it does, it becomes part of the structure. The primary quality of structure, however, consists of the meaning ascribed to it by the agents whose perpetual practice reproduces and changes it. This explains why the normative and the habitual dimensions loom so large in constructivist thinking. Habits tell actors what to do in most unproblematic and problematic situations in which they find themselves. Norms tell them how to consider possible courses of actions in situations where mere habits are of no help. But as said before, norms and habits are not necessarily determining behavior in all and every situation. Indicators of where actors are and stand may be ambiguous or blurred. A residual creativity is thus always implied in any new action, and this means that habits and norms are by themselves changeable. In this, constructivist theory is analogous to international law that maintains that legal rules change if states show deviant behavior in large numbers over a considerable span of time.
On this ontological basis, security cooperation is less of a puzzle than for some other approaches. It is, in principle, a plausible and thus possible mode of interaction (Ruggie, 1998). Constructivist theory can take two approaches to inquire into the subject: it can reconstruct, in the abstract, the conditions in the structure–actor interface that can bring cooperation about and maintain it; and it can analyze, in empirical history, how cooperation actually did develop, persist and decay.
In his seminal article ‘Anarchy is what states make of it’, Alexander Wendt (1992) elaborated the first approach. He constructed a hypothetical interaction between two actors and described how a collaborative structure would emerge from it. He positions two actors into a not-yet-structured environment; their first encounter, incidentally, emerges through a series of mutual, more or less amicable, moves. He then shows how, as a result of a couple of such encounters, mutual expectations develop that the partner will behave in a cooperative way, and that a norm to behave cooperatively ensues. What was spontaneous reciprocity in the first place becomes mutually expected habit. A normative structure that fosters cooperation develops, rather than prescriptions for extremely cautious behavior driven by fear, as in the classical security dilemma among states.
Security cooperation has, of course, not emerged in a no-man’s land, but in a very violent and competitive international environment. Constructivists put much emphasis on change; individual and collective learning, defined as a change of basic ideas about security, and an ensuing adaptation of the constitutive and regulative norms that shape the environment, is the basic mechanism by which change emerges. The constructivist ontology, in which neither structure nor agency are unchangeably fixed, but mutually modifiable and changeable, supports this emphasis (Koslowski, and Kratochwil, 1994; Lebow, 1994; Stein, 1994). The dynamic relation between material and ideational factors has been shown by Steve Weber’s analysis of détente. Nuclear weapons, in his view, changed the international structure by eliminating concerns about relative gains. The first détente, however failed because the two antagonists developed different understandings of what it meant. Only when ideas converged on the notion of ‘joint custodianship’ could broad security cooperation evolve (Weber, 1990; similar results are reported by Bonham et al., 1997).
Initiation of a change in the normative structure
The initiation of learning and change needs ideational/normative entrepreneurs who offer an alternative discourse to pure power politics. Much of the empirical work by constructivists has been devoted to identifying such initiators and to analyzing the type of activities by which they succeeded in persuading decision-makers to adopt their suggestions.
Frequently, ideas for cooperation did not originate with governments–supposed to calculate cooly national interests–but from non-governmental individuals and organizations. The model of the ‘epistemic community’–a transnational group of experts who come to share specific ideas about a policy field and derive action programs for cooperation from these ideas–has been applied to the evolution of arms control regimes in the East–West contexts (Adler, 1992): security experts from the United States and the Soviet Union, starting from quite different positions, developed a common framework for regulating the risky arms competition. This epistemic arms control community succeeded in establishing a new paradigm on security thinking (mutual assured destruction stabilized by agreements) as the leading–if not uncontested–norm for the two superpowers (Checkel, 1998; Evangelista, 1995, 146–89; Risse-Kappen, 1994). From there, important agreements such as the SALT Treaties or the ABM Treaty emerged (Kubbig, 1996); such experts, on the Soviet side, later became close advisors to Mikhail Gorbachev and helped shape his ‘New Thinking’, the basis for the growth in security cooperation in the late 1980s and early 1990s.
A second source for initiating security cooperation is disarmament-specialized non-governmental organizations. Such organizations follow a logic of action different from that of government. Almost universally, they start from a humanistic value orientation that leads them to demand and support disarmament steps (Keck and Sikkink, 1998; Smith et al., 1997). An outstanding success by such a nongovernmental movement was the landmine campaign (Price, 1998, 613–44). Joining more than 2,000 organizations across the globe, the campaign managed to persuade a group of governments to take the lead for an international convention on antipersonnel landmines, and retained their influence throughout the negotiations with a view to resisting the watering down of a full prohibition of those weapons. Non-governmental organizations, thus, not only initiated a new agenda, but helped shape the emerging cooperative norm even in detail (for several case studies compare Risse-Kappen, 1995c). A third initiator is the benevolent individual. Martha Finnemore has traced the origins of modern humanitarian law of war–one of the most striking civilizational projects in security cooperation–back to the efforts of a single person, Henri Dunant, the founder of the Committee of the Red Cross (Finnemore, 1996). With immense energy and commitment, Dunant managed to project his ideas on limiting carnage in wartime and excessive cruelty into high politics; concerns by the military and deep skepticism in various governments were overcome. A self-sustaining process was started which led to a continuing broadening and deepening of the norms up to the present system of the Geneva Conventions and the Convention on Certain Conventional Weapons. Remarkably for such a value-driven project, transcultural boundaries were easily broken, as proven by the existence and growth of the Red Crescent, the Red Cross’s Islamic partner organization. Humanitarian law of war regimes have changed profoundly our understanding of right and wrong in a conflict situation, and have equally changed our definition of who is counted as a legitimate state in international affairs and who falls into the category of a ‘rogue’.
Identity and the process of preference- shaping
This points to an important difference between rationalist and constructivist approaches to security regimes. For constructivists, regimes are not just instrumental systems of rules designed to regulate issue-specific state behavior with a view to enhance all participants’ interests, with preferences and identities given. Rather, they tend to reshape preferences themselves. The norms they contain tell states not only what they should do but what they are supposed to wish; at an even deeper level, it impacts upon what states believe they should be. A decent actor in international affairs is one that does not even think of employing atrocities in war for the sake of victory, one that supports the strengthening and sharpening of the respective international rules, and one that does not have too amiable a relationship with states whose compliance with these rules is not assured. Norms that have become generally accepted and have enjoyed validity over an extended period of time will thus shape states’ identities (Wendt, 1994). As an example, the embracing of the idea of democracy and human rights by the CSCE in its Paris Charter in 1990 established the democratic state as the ideal identity in this region, depicted the non-democratic state as a security concern and opened a new practice–hitherto seen as illegitimate or at least contested–that interference in non-democratic internal affairs was not only desirable, but necessary to overcome possible risks to security and peace (Flynn and Farrell, 1999).
Germany’s relationship to the Nuclear Non-proliferation Treaty is a case in point. Germany felt initially target and victim of this regime. It did its best to restrict the scope and impact of the rules and behaved initially accordingly. Over time, Germany came to tolerate, then to determinedly apply the rules. When reunification occurred, Germany had unlearned its opposition and instead understood non-nuclear status as part and parcel of its own identity. This explains why the parliament voted unanimously to support indefinite extension of the Treaty (and thus infinite perpetuation of Germany’s ‘lower status’), helping to eliminate a weakness of the Treaty that German diplomats had worked hard to insert some twenty-five years before. German non-proliferation policy, in the late 1990s, was devised and executed not from the role of a grudging anti-regime rebel, but from an identity of committed non-nuclear weapon state; acting within the regime had profoundly changed the way in which German political élites looked upon the identity of their own state (Berger, 1998; Duffield, 1999; Müller, 1999; Müller and Kötter, 1990).
Security cultures and subcultures
Under normal circumstances, policy aims at making the international and the domestic discourse on security compatible, while being shaped by both. The stabilizing structure that emerges over time from the interaction of the two discourses is ‘security culture’, a set of values, norms, rules and practices with regard to security that gives thinking and acting in the security field of a specific state a particular, sometimes singular, pattern; states in a given region may share a set of cultural values or a security culture as a whole. Security cultures define identity and thereby shape preferences. Thus they frame action and reaction in specific situations and help explain the considerable variation in security policies among states that may share the same security environment and/or are endowed with fairly similar domestic institutions (Katzenstein, 1996; Krause, 1998).
Even specific sub-cultures may influence the capability of states to enter and maintain cooperative relationships. Traditions, value-orientations and deep-seated priorities by the military can take a particular impact upon security policy and facilitate or impede security cooperation (Farrell, 1998; Johnston, 1995; Klein, 1991; J. Snyder, 1990). Either the military as the decisive expert group can exert a virtual veto over decisions for or against cooperating in the security area (Legro, 1995), or their views may spread over society as to influence the whole security culture of a country, such as in the ‘cult of the offensive’ so popular at the outset of the First World War (J. Snyder, 1984; Van Evera, 1990).
When benign security cultures converge, interests are interpreted in a similar way and cooperative institutions abound, states may form security communities as specific institutions for security cooperation. The concept of the security community, initially originating from the functionalist approach (Deutsch et al., 1957), has been appropriated by constructivists as entities uniquely fit for studying the institutional effects of converging interpretations, values and security cultures, that is, of variables emphasized by constructivism (Adler and Barnett 1998). Members of such communities develop mutual images of each other that make the thought of violent conflict unthinkable; in addition, the ‘other’ is represented within one’s own mind. Perpetual, dense communication helps in addition to include the partners’ interests, wishes and preferences within one’s own decision-making (Risse-Kappen, 1995a). Research has shown that democratic countries have displayed a strong inclination to form such communities–this may be the most important contribution so far by constructivism to the debate on the ‘democratic peace’. Yet it appears that cultural affinity is another variable that helps to bring about at least a weaker form of security communities, such as in Latin America or the Gulf Cooperation Council, thus putting considerable weight on the cultural variable.
Constructivism contributes strong arguments for explaining both stability and change, not only in policy outcomes but, more importantly, in preference formation and–at a deeper level–identity. It thereby makes up for some blind spots in rationalist theories. Its weakness, however, is that it has not yet managed to specify the contexts in which one or the other set of identity/preferences is more likely. This may not be intrinsic to either the epistemology or the ontology of constructivism, but it has not been done. The approach is thus lacking predictive power, and some of its interpretations of past events may appear to a skeptical observer arbitrary.
While constructivists pay due attention to security discourses, they are also interested in how these discourses relate to the observable material reality of power relations ‘out there’. In contrast, postmodernism focuses on discourses exclusively. While the existence of a reality ‘out there’ is not denied, its intelligibility is refuted. Since the ‘out there’ is represented for us in perception and interpretation through language only, we cannot deal with it other than through these representational artifacts. We can describe and analyze discourses only, without the vain hope of discovering behind the words the real thing: there is no corresponding theory of truth. Nevertheless, concepts that are central to mainstream international relations, and particularly the security debates, take pride of place in postmodernist theory, too: power and hegemony. However, these concepts do not address the relationship between states, classes, or individuals ‘out there’, but the subordination and domination relationships between the discourses themselves. Postmodernists ask for the history of present interpretations and understandings (‘archeology’, ‘genealogy’), trying to dismantle the aura of necessity and incontra-dictability that surrounds them. In the course of this, they identify alternative views of the issue in question that have been suppressed–silenced–as the presently hegemonic discourse made its way to ascendancy (Foucault, 1973, 1980).
Given the prioritization and the sense of urgency that common security discourses contain, they are a natural target for postmodern analysis: since a security discourse integrates, explicitly or implicitly, the notion of fear and the threat of force, it is uniquely placed to function in a discursive practice of power and hegemony (Walker, 1986).
The deconstruction of the security discourse has been developed into a sophisticated theory as well as methodology of analysis by the Copenhagen School, notably Ole Waever (Waever et al., 1993). In line with the core postmodernist argument, objective analysis of a constellation with a view to identify security threats to a given actor is refuted. Since a given constellation can motivate a set of very different, equally legitimate interpretations, since the world ‘out there’ is always perceived and analyzed through the lenses of language-bound cultural filters, the only object of analysis can be the discourse of the actors. This is in full agreement with the postmodernist argument: ‘Danger is not an objective condition. It is not a thing which exists independently of those to whom it may become a threat … Danger is an effect of interpretation’ (Campbell, 1992: 1–2).
Security analysis has thus to focus on discourses of ‘securitization’. Constellations are not a security problem by themselves; they become one by the application of specific speech acts by actors. The speech act of securitization consists of identifiable components. These are (a) the declaration of an existential threat to a given community to which the speaker belongs, requiring priority attention for dealing with this situation and (b) the request for extraordinary, non-routine measures to cope with the threat identified in (a). The process of securitization is only complete, however, if the performative effect intended by the speaker obtains: the speech act is directed to an audience and is only successful if the audience accepts the validity of (a) and (b), consequently putting (b) into action (Waever, 1989). The move is all but interest-free. Since it gives the state the legitimate claim to go beyond the limits of routine political behavior, and since securitization claims are most frequently made by the power-holders of the state, the act serves to reaffirm their rule over society. By invoking the extraordinary measures for defending against the alleged threat, securitization opens the specter of enhanced fears abroad; the dangerous spiral of the security dilemma, at its core, is thus not a matter of accumulating hardware, but of applying specific terms of language to interpretable situations.
The way out of the dilemma cannot be sought in ever more ‘national security policies’ with the big apparatuses of military defense, intelligence and so on at its core that tend to perpetuate the securitization process. Rather, desecuritization helps to overcome it. To remove issues out of the realm of security language, and submitting it to the normal (non-extraordinary) instruments of everyday politics, is the only way out of the trap created by the securitization speech act. Only if we have stopped talking (or even thinking) in security terms have we become truly secure. The job of the researcher is thus to identify securitization processes, to analyze and reveal the consequences, and to point to desecuritization as the alternative path (Waever, 1995).
No doubt the securitization approach is, again, a most useful tool to deal with the risks embedded in setting security agendas. It creates a clear operational standard with which security discourses can be analyzed. Yet it begs two questions that, as an exclusive approach to the field, make it less than satisfactory.
For one, we cannot be satisfied with exclusively analyzing speech acts and their consequences. If we want to predict the consequences of securitization, we must have a model for how the ensuing action affects the relation with other actors in the field. For this, we need also an image of these actors. This amounts to nothing less than a fairly comprehensive analysis of structures and processes ‘out there’ with which our discourse is supposed to interact.
The second criticism is that security cooperation cannot be analyzed properly with the securitization approach. The opposite to securitization is desecuritization, removing an issue from the language of security (Buzan et al., 1997). But this is not what partners in security cooperation do. Rather, they recognize that an issue is security-relevant, containing risks to the existence of all of them; nuclear weapons is the most obvious example. It is the mutual recognition that the existential risks existed in the interaction system created by hair-trigger ready nuclear arms on both sides that motivated the astonishing series of negotiations and agreements which established the field of nuclear arms control. These were indeed ‘extraordinary measures’, namely close cooperation with the rival in the most sensitive field. The field was not desecuritized at all, but repeated speech acts of securitization were used to continue to motivate further steps towards cooperatively disentangling the nuclear stand-off, as they were used by the critics to request the end of arms control.
According to postmodernist analysis, the notion of security has to rely on a sharp polar differentiation between ‘self’ and ‘other’. This distinction is very central to the whole deconstructivist endeavor in postmodernism, particularly in the version going back to Derrida. The insistence on the centrality of the subject, the hard core of modernist philosophy and everyday thinking and talking, can only be upheld if that subject, the self, is sharply separated from the alien ‘other’ against which it defines and defends its subjective identity. This distinction is one of inclusion–that which belongs–and exclusion–that which is alien. The same process evolves at the collective level of the state (Walker, 1988).
The state itself, as an artificial construct of identity, cannot exist without this discursive mechanism of inclusion and exclusion. It is the determined exclusion of the ‘other’ that construes the bounds between those included in the territorially enshrined boundaries of the construct. And it is the ascription of danger (threat, disorder, anarchy) to the excluded ‘other’ that produces the necessary discipline to subordinate the inhabitants of that territory to the prevailing identity discourse centered on the state (Ashley, 1998; Campbell, 1992). In fact, human beings could adopt many identities, and the permanent risk of defection from the hegemonic one has to be contained; the security discourse emerging from the self–other distinction serves that very function.
The inclusion–exclusion scheme applies to agent as well as to structure. And in both cases, it is not just a value-free description of different entities, but it contains heavy ascriptions of positive and negative values. The included is seen in a positive light: the internal structure offers order, stability, safety, security. The external disorder–anarchy in realist parlance–vibrates with threat, risk and danger. The security discourse thus is instrumental to rally strong support around the state, and to direct defensive energies against that which is outside.
The deconstructivist approach along the self/other, inclusion/exclusion border is a powerful and useful tool to uncover the origins of enemy images, and to reveal the abuse of propaganda and public influence in order to exaggerate threats and risks. It has, however, serious shortcomings. It is ill-equipped to look for, and find, polarities containing more than two poles, overlapping identities, partial compatibilities though all these are theoretical possibilities and can be found in practice and are indeed the foundation on which security cooperation is built. For example, Simon Dalby analyses the process by which the Committee of the Present Danger established a security discourse that became hegemonic over, and eventually destroyed, the détente discourse that prevailed previously (Dalby, 1990). But it is hard to explain from the same vantage point how the détente discourse could emerge from the Cold War in the first place. If the self/other divide is necessarily dominating the statist security discourse–being the precondition for the existence of the state–how could a ‘third area’, that of common interest, ever be recognized, codified and even expanded? For the area of cooperation rests on a (partial) amalgamation of self and other into a ‘we’, even though, in other respects, the distinction continued as sharply as before in the still unregulated spheres of international competition. It appears from this discussion that postmodernist deconstructivism contains an inbuilt bias that makes it difficult to account for inter-state security cooperation at all; significantly, Campbell’s study of US foreign policy identifies all possible exclusions and constructions of enemies; it contains not a single word on the panoply of cooperative relationships in which the American state has been involved–and has quite frequently initiated–in the course of its foreign policy history (Campbell, 1992).
Does postmodernism recognize cooperation when it reads it?
Reading through postmodern security studies one discovers a virtual complete lacuna of analysis of the many instances of states cooperating in the security sector. The few that we find are problematic in their application of the inclusion/exclusion and hegemony/subordination schemes. Given the basic orientation of postmodernist analysis it is perhaps not incidental that certain types of security regimes that are ostensibly more symmetric than the hegemony/subordination divide would permit are not found, such as the Latin American Nuclear Weapon Free Zone, the Berlin Four Powers Agreement, CSCE, the CFE Treaty or superpower arms control.
Richard Price has tackled the genealogy of the Chemical Weapons Taboo as the growth and finally successful imposition of an image developed from what behavior is appropriate for a (Western) civilized country (Price, 1997). ‘Belonging’ to the included ‘inside’ means to behave according to the rules of this taboo. For all the sophisticated analysis that brings out this result, though, the question remains open why countries that have shown considerable resistance to Western attempts to impose a civilizational model, embrace the taboo in its ultimate form, the Chemical Weapons Convention; countries such as Iran, India, Indonesia or China.
The discriminatory nature of the nuclear non-proliferation regime has motivated postmodernist analysts to devote some attention to its analysis. It is seen by James Keeley as a hegemonic imposition of a proliferation discourse over an alternative–but suppressed–discourse emphasizing nuclear disarmament. This hegemony, obviously, serves the power interests of the nuclear weapon states and is reproduced through the practices of regime members (Keeley, 1990).
Along very much similar lines (but curiously unaware of Keeley’s study), David Mutimer has reconstructed the non-proliferation regime (Mutimer, 2000). While Keeley emphasizes directly its roots in the power hierarchy, Mutimer traces it back to an image of self-spreading technology. Technology, in his view, is thus at the heart of the–academic and political–‘proliferation discourse’, and the inside/outside division is thus provided by the membership or non-membership in the various suppliers’ groups; this, in turn, reveals the discourse as hegemonic along a North–South divide. Mutimer proposes a disarmament discourse as a dissenting alternative. Both accounts are seriously flawed. For one, the disarmament discourse has been dominant in international diplomacy (though not in nuclear weapon states’ activities) throughout the existence of the NPT, and it became clearly dominant in the 1990s, as the discursive contests during the 1995 and 2000 Extension and Review Conferences revealed (Johnson, 2000). Secondly, while parts of US academia and the various administrations overemphasized the technical aspects, there is an abundance of scholarly analysis of nuclear proliferation and non-proliferation that focuses on the political factors and requests addressing them as first priority (for example Goldblat, 1985; Müller, 1987; Reiss, 1988; Snyder and Wells, 1985). Indeed, non-proliferation diplomacy has made continuous efforts to deal with the political issues; however, in some regions they proved intractable to the diplomatic instruments at hand. Third, some of the most difficult controversies about nuclear supplies arose within the members of the suppliers groups themselves–for example, between the United States and the European Union–for the simple reason that this group does not succumb easily to a dichotomic divide: all suppliers are recipients as well. Fourth, there is no fixed dichotomy between the ‘non-proliferation’ and ‘disarmament’ discourses. Countries with an impeccable disarmament pedigree such as Sweden, Ireland, New Zealand, South Africa or Brazil participate actively in the suppliers group. The participation of the two latter countries also shows that the North–South dichotomy has long become blurred. Mutimer’s attempt to identify the ‘proliferation discourse’ as defined by him in other regimes as well is even less convincing. The Chemical Weapons Convention is emphatically a disarmament treaty, requesting in the first place the notification and dismantlement under strict verification of existing chemical arsenals. The same applies to one of his other examples, the Ottawa Convention banning antipersonnel landmines. That either Convention contains stipulations on verification and export surveillance is inevitable given the goals of disarmament: parties to such a convention want to make sure that those who have not signed cannot get easy access to the prohibited weapons and related technologies, and they want to assure that those subscribing to the rules abide by them and, if not, the rest get an early warning. Without verification and export controls, these goals cannot be satisfied.
The conclusion is very clear: the scheme of discursive inquiry, consisting of two dichotomies–inclusion/exclusion and hegemony/subordination–leads analysts to overlook those aspects of the discourses which don’t fit: symmetries, compromises, cross-cutting identities, ambivalences, tri- or multipolar discourse structures, which are all more or less conducive to security cooperation–and therefore found in the respective regimes–than the structures postmodernists look for.6 Rescue from the security quagmire in which the reification of the modern state has put mankind is not expected from state discourse and related actions, but from the ascendancy of the alternative discourse promoted by the postmodernists and for which they seek societal carriers in, inter alia, civil society (Dalby, 1997; Klein, 1988). If highest state representatives carry out such a changed discourse themselves–as Mikhail Gorbachev did–postmodernist analysis can at best describe it; how it could happen cannot be explained on the basis of the postmodernist approach, except by simply appointing Mikhail Gorbachev a ‘critical strategic theorist!’ Very ironically, the structures that poststructuralist analysis has claimed to find as typical in modernist discourses are used to construct a world of discourses that looks awfully similar to the world as constructed by realists: a world of sharp and dangerous hostility between ‘we’ and ‘them’, and a world of hegemony, suppression and struggle against it. What remains on the positive side is the deconstructive method that is very useful to disentangle the ideologies of national security policy and their academic analysis.
None of the theoretical perspectives on security cooperation discussed in the previous paragraphs gives full satisfaction. Realism is useful by warning against well-minded illusions about cooperative possibilities in the light of the barriers posed by power politics, but is short on explaining why, under the circumstances, security cooperation has been relatively successful if not abundant. Its counterpart at the other end of the spectrum, postmodernism, mirrors curiously that skepticism. It supplies us with useful tools to uncover the ideological roots of much of present security policy, but–as realism–puts so much emphasis on hostility (that is, the inclusion/exclusion divide) that real cooperative relationships can be hardly seen as anything else but hegemony imposed on the carriers of suppressed discourses. Rational choice informs us about the constellations of preferences that are more or less favorable to security cooperation, and about the useful role of institutions in reducing uncertainty and transaction costs. But it leaves us wondering how the states, despite these prospects, can overcome the profound impediments to engaging in those institutions under the dire circumstances of the struggle for survival in anarchy in the first place. Liberalism, in the form of democratic peace theory, offers a rich theory of how security cooperation is motivated and maintained. But its one (dyadic) version falls short of explaining how democracies come to cooperate with non-democracies, and both versions have difficulties in accounting for the vast variation in cooperative behavior among democracies. Finally, constructivism, with its emphasis on ideas and the cultural grounding of behavior, its treatment of the interplay between material and ideational factors and between structure and agency, may be best fitted to explain security cooperation. But it does so only in hindsight; the theory is much too indeterminate at present to allow for the development of distinct hypothesis, let alone prediction. Attempts at synthesizing various strands of theory are in too early a stage to be fairly assessed (Hasenclever et al., 1997: ch. 6 and Deudney, 2000).