Peace in Bosnia and Kosovo, such as it is, has rested these past several years on an uneasy conspiracy to prop up, but never openly discuss, a set of irreconcilable contradictions. Inhabitants and intervenors have conspired to live with political practices that contradict constitutional principles, and to prolong foreign occupation while genuflecting to the aims of democracy and self-determination. The American foreign policy elite on both sides of the political spectrum has been complicit. Clintonites promoted the conspiracy in order to do something like the right thing without overstepping the seeming bounds of domestic support. The new Bush team disapproves of entanglement in peacekeeping, but it wants to maintain American primacy on the world stage--a contradiction of its own that blocks a graceful exit. And now, Albanian guerrillas subverting southern Serbia and northern Macedonia, and Croat rioters in Bosnia, are disturbing the calm that had preserved inertial peacekeeping as the path of least resistance.

In 1995 President Clinton justified sending American troops to Bosnia with the assurance that they would be out within a year. He mistook an exit date for an exit strategy. As a result, for six years in Bosnia and two years in Kosovo, the United States has continued to collaborate with other occupying powers without an exit strategy, far beyond the passing of the exit date. Unlike the occupations of Germany and Japan after 1945, NATO and the UN have settled into operations in the Balkans that are best understood as institutionalized temporizing. There have been noteworthy efforts at economic reconstruction, but attempts at political reconstruction have been limited and confused.

In fact, the confused political status of these areas has been a vital necessity. It lets occupiers and inhabitants pursue separate agendas. The Western presence has been sustainable because it rests on unresolved contradictions between the de jure and de facto settlements of the two wars: Bosnia is in principle a single state but in practice a partitioned one, while Kosovo in principle remains a province of Yugoslavia but in practice is not. These contradictions allow the inhabitants of Bosnia and Kosovo to avoid organizing their societies in the ways that the occupiers want, while allowing the occupiers to pretend that they are supervising a transition to the type of social organization of which the West approves.

Resolving these contradictions has proved too daunting, so temporizing is the result. Rather than face an unpalatable choice between the much stronger efforts that cultivating political stability would require and a withdrawal that might re-ignite war, the United States, NATO and the UN drifted toward open-ended occupation. This has been the path of least resistance, however, only because the costs have been modest--little treasure and no blood. Without U.S. casualties, and with surpluses suppressing urges to wield sharp budgetary pencils, the American public has no reason yet to rue the occupations. The odds that these fortuitous, permissive conditions will continue indefinitely are low.

Around Kosovo, Albanian action to support their kin in Serbia and Macedonia has already undermined postwar stability. In Bosnia, Croats have already challenged the status quo. Economic decline could easily produce greater popular unrest and protests against the outside powers, thus catapulting the region back onto the front pages of American newspapers and highlighting the risk of further trouble. It would raise the dormant question of whether Americans wish to run that risk.

Despite such dangers, some believe that there is no need to rush to a resolution and that, indeed, getting out would be a bad thing. In this view, it is good that unchallenged global dominance enables the United States to contemplate an indefinite mission. Now, in the absence of any serious opposition, is the time to use American power to shape world order. Why not start in the Balkans?

During the Cold War, the United States was often accused of neo-imperialism. At the time, this was a bad rap. U.S. interventions often found the client's tail wagging the patron's dog, as Washington became mired in support of problematic Third World governments, while not having any direct or real governing authority over them. Today, however, we are engaged in real neo-imperialism, although a quite peculiar multilateral and humanitarian form of it. Under the aegis of international organizations, the United States is collaborating with other governments in the direct control of Bosnia and Kosovo, a return of the Western powers to the tutelary administration of backward nations--rather like a League of Nations mandate. There is certainly no economic benefit to the imperial metropoles. Rather, the Western presence represents a new mission civilisatrice for a new imperium. In effect, beneficent recolonization is the regional security strategy that the "international community" offers up at the turn of the twenty-first century. But is this a solution that we should embrace, or a wrong turn from which we should escape?

For its part, and despite rhetorical backing and filling, the Clinton administration embraced the idea. Indeed, it was the implicit rationale for maintaining American primacy that animated the belief shared by Holbrooke, Berger and Albright in the United States as "the indispensable nation." The new Bush administration rejects the enthusiasm for intervention in principle, yet it endorses the importance of American primacy just as forcefully as its predecessor. This makes it awkward to shed current responsibilities. (And although intervention in Bosnia was a Clinton project, the U.S. commitment to protect Kosovo goes back to the administration of Bush the Elder.) Beneficent recolonization can serve primacy if the United States and its rich allies are willing to invest heavily and sacrifice significantly to make it work; otherwise, half-hearted recolonization exposes a hollowness at the core of primacy. The dirty little secret of American foreign policy is that exercising primacy is popular across the domestic political spectrum, but only as long as it is cheap. When national assertion--whether for altruistic or narcissistic purposes--hits a costly snag, people notice that it is not all that much fun, and begin to see more merit in less meddling.

To get out of the Balkans, the United States should aim at achieving six main objectives:

Establish self-government to end the occupation. The United States should not be an imperial power and should not accept an indefinite responsibility to administer foreign countries.

Stabilize external security and peace for local states. The prime motive to intervene in the Balkans was to end the violence there. Withdrawal that allowed war to erupt again would therefore represent failure.

Minimize damage to relations with other great powers. The main reasons for intervention were humanitarian, but good deeds should not incur significant costs in those aspects of international politics that count most.

Withdraw U.S. forces. Aside from the moral interest in ending occupation, we have a material interest in reducing the strains on U.S. military forces--particularly on the personnel rotation system and training in the army--that are imposed by prolonged peacekeeping expeditions.

Honor moral obligations. As long as the cost is low, at least, there is no reason not to keep faith with those for whom we intervened.

Honor legal obligations. Other considerations being equal, it is in the interest of the United States to observe the terms of international agreements that it has signed if it wishes such agreements to be useful instruments in the future. But other considerations in the Balkans are not equal. To realist critics, "legalism and moralism" are often lumped together as impediments to the wise pursuit of material interests. In Kosovo, however, legal obligations to Belgrade conflict starkly with moral obligations to the Albanian population. This is a vexation for realists, a dilemma for idealists.

Some would add preservation of nato's "credibility" and America's "leadership" to this list of objectives. Indeed, some cite these as the most important ones. Mortgaging the mission to these buzzwords, however, is to put the cart before the horse. It reflects a penchant for self-entrapment that is not unique to involvement in the Balkans, but is a problem of U.S. foreign policy in general. We should, by now, recognize that credibility is not served by re-inforcing failure. Just because the costs were measured on an entirely different scale does not mean that Vietnam's lessons on this score are irrelevant. Credibility should serve the pursuit of substantive objectives, but it should not dictate what those objectives are. Leadership means convincing others to want what we want, not changing what we want in order to keep followers faithful. If we could succeed in meeting the six objectives listed above, leadership would be evident and credibility would follow.

The problems with this list of objectives, however, are, first, that each one is hard to achieve in itself, and, second, that it is impossible to achieve some of them without undercutting others. One or more of them will have to be sacrificed. How, then, to proceed?

Sovereignty and Stability

The hinge of a potential solution in the Balkans is forging a connection between sovereign self-government and interstate stability (meshing the first, second and fourth of the objectives listed above). Establishment of self-government has already occurred to a degree, but it is a sharply limited kind reminiscent of colonies in the more enlightened of the old European empires. Self-government in Bosnia and Kosovo so far remains subject to the higher authority of the occupying forces. NATO officials skew elections, disqualify candidates, close down radio and television stations, and so on. The benefit in this is that it prevents self-government from re-energizing local conflict; the cost is that it defers resolution of the essential issue. Genuine self-government requires decolonization--termination of the controlling role of the occupying powers.

Self-government and stability are at odds because it was conflict over the conditions of self-government--the lack of congruence between cultural and political communities--that caused the explosions in the first place. The essential issues are the number, form and boundaries of independent governments--that is, which units constitute the "selves" of self-government--when sovereignty ceases to be limited by occupation. Would the solution be autonomy for ethnically defined territorial areas within a Bosnia and a Yugoslavia that are organized as loose confederations (the current situation in Bosnia, the official aim of the outside powers for Kosovo)? Or self-government for a genuinely unified Bosnia and a Yugoslavia that includes the province of Kosovo, both as multi-ethnic states? Or self-government of smaller, ethnically defined states in formal partitions of the larger units that are currently the juridically legitimate ones?

At present, peace in both Bosnia and Kosovo depends on a blatant contradiction between principle and practice. As long as outside powers continue to run the region, these contradictions can be finessed and may even be useful. If full self-government is ever to develop, however, they have to be resolved. In Bosnia it will be hard to honor legal obligations while meeting the other criteria for success, unless the Dayton agreement is revised. In Kosovo it will be impossible to honor legal obligations to Belgrade without betraying the Albanian population for whom the war was allegedly fought in the first place.

Bosnia remains at peace only because the unified political structure established by the Dayton Accord does not function. Officially the single state is composed of two "entities"--the Muslim-Croat Federation and the Republika Srpska--but each has veto rights over actions of the central government. (There are actually three entities, as the Federation has broken down into Croat and Muslim areas that cooperate only minimally, with the Croat side increasingly asserting its separateness.) What is the real function of the unified state, if any, when the fundamental divisions behind the war remain in place during the peace? As Ivo Daalder observes, "By incorporating rather than resolving the fundamental disagreement among the parties about Bosnia's future, Dayton assured that its implementation would become little more than the continuation of conflict by other means." What makes this situation preferable to formal partition, other than a belief that a hypocritical liberal fiction is better than a legitimized reactionary reality?

Kosovo remains at peace because its internationally recognized status as a component of the Yugoslav state--the price of getting Belgrade to end the 1999 war--is a fiction. The crucial concession that Belgrade got was the elimination of the Rambouillet ultimatum's provision for eventual disposition of sovereignty over Kosovo according to the will of the province's people--a prescription for eventual independence. Annex II of Security Council Resolution 1244, adopted June 10, 1999, affirms the aim of "substantial self-government for Kosovo", but recognizes "the principles of sovereignty and territorial integrity of the Federal Republic of Yugoslavia" and stipulates "the demilitarization of uck." This is an inherent contradiction even greater than that in the Dayton Accord. Only in a wishful scenario of assured self-restraint by Belgrade could Kosovo's juridical status as part of Yugoslavia be made real without betraying the Albanians of Kosovo. Only with a potent army of its own would Kosovo have any reason for confidence that its autonomy would be safe.

Now, in contrast to the scheme embodied in the Rambouillet ultimatum, occupying powers cannot grant Kosovo independence without violating the terms of the agreement that ended the 1999 war. Nor can NATO violate that provision of the agreement without expelling the UN from jurisdiction, since the Security Council has a central legal role in the occupation. And, as Barry Posen has explained, unless Russia and China agree otherwise, the Security Council cannot relinquish responsibility: "Paragraph 19 of the Security Council resolution of June 10, 1999, declares that 'the international civil and security presences are established for an initial period of 12 months, to continue thereafter until the Security Council decides otherwise.' Thus, if either the Chinese or Russians choose not to decide otherwise, insofar as both have veto power, Security Council control over Kosovo will last forever."

The overlapping UN and NATO roles in the Balkans reflect the complex novelty of multilateral humanitarian intervention. Two observers argue that "Trusteeship is a new weapon in the armoury of international intervention, and Bosnia is its first arena." This is only partially correct. After all, one of the original institutions of the United Nations was its Trusteeship Council, toothless for most of its existence, and inactive since 1994, just as the point was reached when such an organ might be useful. Moreover, Bosnia is not trusteeship's "first arena" even in the post-Cold War era; that distinction belongs to the United Nations Transitional Administration for Cambodia (UNTAC) in the early 1990s. In Cambodia, the UN did manage to facilitate a truce among three sets of contenders, preside over a temporary receivership of the country, and organize elections and a new political start. The terms of the un-supervised settlement, however, were imperfectly honored during the UNTAC mission and crumbled rapidly after its termination. Worse, Cambodia did not make the transition to peaceful democratic competition envisioned by the UNTAC plan. Rather, the victory of the strongest of the three main political groups over the others carried the day. Cambodia is not a disastrous precedent for the Balkans, but it is not a good one either.

Models to Emulate--or Avoid

If trusteeship is not a trustworthy solution for problems such as those of the Balkans, what are the remaining options?

Optimism is a bad bet, but not a ridiculous one. Over the past dozen years some bitter, epochal conflicts turned in far more positive directions than most experts would have predicted: the end of the Cold War, the peaceful democratization of South Africa, the ebbing of civil wars in Central America. But where do we look for a basis on which to predict how sectarian and ethnic conflict in the Balkans will be settled? Liberal optimists tend to rely on logic: domestic peace and international aid, both secured by peacekeeping missions, should foster civil cooperation and tolerance because they make better sense than destructive parochialism. Conservative pessimists tend to look more for precedents. What similar cases have yielded the desired result?

If the aim is to make viable multinational states out of riven polities in the region, few encouraging examples spring to mind. Switzerland and the United States may be cited as models, but their achievements are surely too distant in time and circumstances to be convincing analogues. Neither has suffered a war among its constituent groups that still lives in personal memories. Satisfactory political integration in the American South took more than a hundred years after the Civil War, and social integration remains elusive to this day. The settlement in Zimbabwe has crumbled as the Mugabe government expropriates land from white farmers. South Africa so far offers the best example of hope, but even if reconciliation there proves durable, it is arguably less similar to the Balkans than the many examples of failure.

Those who would bank on joining contending ethnic groups into functioning polities bear the burden of providing relevant examples of successful integration. Other ethnically divided states and regions of the twentieth century generally give rise to skepticism about secular integration after bitter civil wars. This is especially true if the states emerging from the resolution are to be democratic and genuinely self-governing. A multisectarian Lebanon is stable now compared to its previous decades of civil strife, but that is largely because Syria keeps the country under its thumb--just as NATO does in the Balkans. Yugoslavia before the 1990s was united and stable in no small part because it was not democratic, and because secular communism suppressed regional particularisms.

Some relevant lessons might be sought among cases of wars ended by partition along ethnic lines, for example, in Palestine, Kashmir and Cyprus. The mention of these unhappy, controversial and unsettled places would undoubtedly prompt many to reject formal partition as a model for Bosnia or Yugoslavia and Kosovo. A closer look, however, leads to a more equivocal conclusion.

The partition of Palestine in 1947 was immediately revised by the 1948 war. It was altered again by the Six-Day War, the Camp David accords, and the Oslo Agreement--and it still remains in question. Does this suggest that Bosnia would do better to avoid partition and insist on an integrated multinational state? Could anything have been much worse than the past half century of tension and periodic war in the Middle East? Well, yes. An internationally enforced creation of an integrated Arab-Jewish state in the 1940s (no harder to imagine at the time than the integration of Serbs and Albanians in Kosovo today) probably would not have been less violent or more viable than what developed.

Kashmir, too, has remained a dangerous cauldron of conflict. In this case either a more careful plan for the partition of India in 1947 that allocated the area to Pakistan (on grounds of ethnic affiliation), or a more decisive war that left it fully within India (as Israel's gains in the 1948 war overcame the nonviable non-contiguity of the partition plan's territorial divisions in Palestine), might have yielded more stability. An independent Kashmir or an accepted division of the area between India and Pakistan are additional hypothetical alternatives, though not real ones. The analogous choices in Kosovo would be union with Albania, re-incorporation into Yugoslavia, or independence. (The first or third of these could include partition of Kosovo itself, with a slice in the north going back to Serbian Yugoslavia.) There is no good analogy in Bosnia, since the Muslims--who have no supporting external state comparable to Croatia or Serbia--create an unbalanced tripolar situation more complicated than Kashmir or Kosovo.

If one takes the UN role in Cyprus seriously, that case presents the model of indefinite peacekeeping; UN forces have been in the country for nearly forty years. During that time the mission has been eased by its irrelevance to the main security issues on Cyprus--it has not had the powerful controlling role of the West in the Balkans, and did not stop either the Greek coup on the island or the Turkish invasion of 1974. More relevant is the unilateral Turkish partition imposed in 1974. Unacceptable as that partition may be on legal grounds (it remains unrecognized by virtually the entire world outside Turkey), it has ensured peace on the island for more than a quarter century. If justice is to take precedence over peace, what is the solution for Cyprus--to return to the unitary state that preceded the Greek coup? If so, what mechanism would protect the Turkish minority more satisfactorily than has Ankara's intervention? If peace is to take precedence over justice, there is a strong case for international recognition of the partition and the legitimacy of the Turkish Republic of Northern Cyprus. If justice and peace are to rank equally, a solution is nowhere in sight--after nearly four decades of impotent UN presence.

Perhaps the best illustration by analogy of choices for Kosovo comes from the untidy periphery of contemporary China. Is Kosovo's future best exemplified by Tibet, Hong Kong or Taiwan? Since 1950 Tibet has suffered the fate which NATO was concerned to save Kosovo from when it went to war in 1999. Hong Kong represents the hope of the temporizers in the Balkans--in that case an escape from the choice between betraying the Kosovar Albanians and violating the agreement that ended the war with Belgrade; in the Hong Kong case the promise of indefinite actual autonomy under nominal Chinese sovereignty. Taiwan represents de facto autonomy without de jure independence; but autonomy guarded by force rather than, as in Hong Kong, by Beijing's sufferance. The analogy would be a Kosovo recognized internationally as a province of Yugoslavia, but armed and able to prevent Belgrade from imposing its writ. Unlike Taiwan, however, Kosovo lacks the geographic conditions (a hundred miles of water) to make self-defense without foreign forces feasible.

With the choice cast in these terms, most outsiders would be drawn to the Hong Kong model. But would Belgrade remain as restrained as Beijing has with respect to Hong Kong? Hong Kong's special status is of great economic importance to China while Kosovo has no such importance for Serbia. Further, China has incentives for good behavior as long as it seeks a peaceful re-incorporation of Taiwan. Could we count on Serbia to abstain from encroachment unless the West extends a security guarantee to Kosovo--in effect negating Yugoslav sovereignty?

Legal issues aside, is formal partition the lesser evil? Clearly, the history of partitions in the twentieth century is mostly a sorry one. Several wars followed the 1947 partitions in Palestine and on the Indian subcontinent, Northern Ireland remains violently unsettled eighty years after its separation, and so forth. The relevant question, however, is the counterfactual: Would history in these sorry cases have been better or worse if the states had not been partitioned? The argument for partition is not that it is good or desirable. It is that it may be less horrendous than keeping the warring communities in the same state, or that it is preferable to the indefinite foreign occupation of an artificial and uneasy confederation.

U.S. National Security

Moral interests were the prime reason for NATO and UN intervention in the Balkans--the humanitarian imperative to suppress atrocities (although why this imperative should be irresistible in Europe, but not in Rwanda, Sudan or other places plagued by even worse atrocities, has never been made clear). Some observers believed that intervention in the Balkans was warranted as well by material interests, traditional security concerns about the international balance of power and the need to keep local chaos from expanding into conflict among major states. This argument, however, has it backwards. Intervention worsened conflict among great powers instead of dampening it. It would be nice if moral and material interests re-inforced each other, but in reality they have been in tension. Moral interests have prevailed in NATO capitals mainly because material interests have not been seriously threatened.

If the objective had been to prevent escalation of the local conflict to confrontation with a major adversary, there was never reason to assume that Western intervention would accomplish this, or would do so more effectively than diplomatic collusion to insulate the conflict by foreswearing intervention by any of the great powers. It is disingenuous to argue that intervention by the West need not aggravate already disagreeable strategic relations between NATO and Russia.

Luckily, worsened relations with Russia are not a crucial problem in today's world, and some may consider them a price worth paying for the moral benefit of keeping the locals from butchering each other. Russia is weak and has few plausible options for responding to its alienation in a way that can threaten NATO. The West does not have to worry about maintaining a balance of power, reassuring Russia about its security, or pandering to Moscow's wounded amour propre. In short, NATO may take advantage of its hegemonic position and leave the Russians to lump it if they don't like it. Indeed, as it was, Moscow had no choice but to accept the Dayton Accord and participate in both occupation missions. Although the Kosovo war infuriated the Russians, there was little they could do about it.

It is useful to recall, however, that the Russian coup de main in seizing the Pristina airport at the end of the war, and the short-circuiting by the British of General Clark's plan to have NATO forces challenge them, raised the specter of potential unintended military confrontation. Beyond that, the most important question is whether NATO should count on indefinite Russian weakness, or attempt to stabilize relations on a more equitable and cooperative basis before an aggrieved and resurgent Russia regains options of its own.

The Kosovo war brought a large and unanticipated cost to America's relations with another potential great power adversary: China. The accidental bombing of the Chinese embassy (which the Chinese will never accept as accidental) had a gratuitously damaging impact not only on diplomatic relations, but on Chinese public opinion. Moreover, the entire rationale for Western intervention in Kosovo represents a threat in principle to Chinese sovereignty. The rationale could just as easily be applied to justify humanitarian intervention on behalf of the ethnically oppressed populations of Tibet and Xinjiang, or Taiwan's claim to autonomy (in the same way as it impugns Russia's sovereign right to pacify Chechnya).

The burden on Sino-American relations notwithstanding, the American public has supported U.S. operations in the Balkans so far because the costs have been low. But they would also have supported them if the apparent benefits were high--especially in shoring up genuine security interests--even if it had been a more costly enterprise. As John Mueller has argued in these pages, the notion that voters will not tolerate any venture that brings Americans home in body bags is a myth. What voters reject are military operations that appear inconclusive, unsuccessful and bloody, for purposes of dubious importance. Thus, if the perceived benefits of policing the Balkans are low, but the costs rise abruptly, public tolerance will wane. Throwaway rhetoric notwithstanding, no administration has seriously tried to convince the public that humanitarian intervention--if it happens to get messy--is an interest that is truly important to Americans. Leaving support for intervention hostage to endless good luck on the cost side of the equation is therefore a fragile basis for long-term involvement. That is why we need to think seriously about finding a way out.

Ways Out

Recognizing a reality that admits of no good strategy, two analysts of the situation are reduced to recommending that we avoid the question: "Kosovo may now have shattered the exit strategy concept. . . . Not only is it impossible to say when NATO troops will leave Kosovo, it is also impossible to specify under what circumstances they will do so. . . . One cannot say; it would be unwise at this point even to try." Such a plaintive note is a sign of false maturity. We do have to try, unless we want to occupy the region for decades.

There is no way out of the Balkans for the United States that does not entail high cost in either effort or honor. There is no evidence of support for a much stronger effort, so the price will probably have to be paid in honor. In that light, there are three general options: worst, bad and not quite so bad.

Inertia: open-ended occupation. This has seemed the path of least resistance, but it puts us at the mercy of events. It is foolish to assume that either the locals or American voters will want us in place forever, or that the costs on the ground will remain low. The Albanian insurgencies in southern Serbia and northern Macedonia suggest the dangers that can arise to complicate the peace that peacekeeping affords.

If there is a rationale for this option other than mindless inertia, it is a long-term tutelage designed to transform the local societies and allow eventual disengagement and durable peace. If a long period of neo-imperial tutelage has high odds of civilizing the locals and making the next generation willing practitioners of secular liberalism, it might be worthwhile to gamble on it, to view institutionalized temporizing as gradual behavior modification.