Is Ethnic Conflict Inevitable?

Parting Ways on Nationalism and Separatism

by James Habyarimana, Macartan Humphreys, Daniel Posner, Jeremy Weinstein, Richard Rosencrance, Arthur Stein, and Jerry Z. Muller, Foreign Affairs, July/August 2008

Summary:  Critics refute Muller's assumptions about ethnic conflict; Muller responds.

BETTER INSTITUTIONS, NOT PARTITION

James Habyarimana, Macartan Humphreys, Daniel Posner, and Jeremy Weinstein

Jerry Muller ("Us and Them," March/April 2008) tells a disconcerting story about the potential for ethnic diversity to generate violent conflict. He argues that ethnic nationalism -- which stems from a deeply felt need for each people to have its own state -- "will continue to shape the world in the twenty-first century." When state and ethnic-group boundaries do not coincide, "politics is apt to remain ugly."

Muller points to the peace and stability in Europe today as evidence of the triumph of "the ethnonationalist project": it is only because of a half century of violent separation of peoples through expulsions, the redrawing of state boundaries, and the outright destruction of communities too weak to claim territories of their own that Europe today enjoys relative peace. Elsewhere, the correspondence between states and nations is much less neat, and there Muller seems to agree with Winston Churchill that the "mixture of populations [will] ...cause endless trouble." He advocates partition as the best solution to this difficult problem.

If correct, his conclusion has profound implications both for the likelihood of peace in the world and for what might be done to promote it. But is it correct? Do ethnic divisions inevitably generate violence? And why does ethnic diversity sometimes give rise to conflict?

In fact, ethnic differences are not inevitably, or even commonly, linked to violence on a grand scale. The assumption that because conflicts are often ethnic, ethnicity must breed conflict is an example of a classical error sometimes called "the base-rate fallacy." In the area of ethnic conflict and violence, this fallacy is common. To assess the extent to which Muller falls prey to it, one needs some sense of the "base."

How frequently does ethnic conflict occur, and how often does it occur in the context of volatile mismatches between ethnic groups and states? A few years ago, the political scientists James Fearon and David Laitin did the math. They used the best available data on ethnic demography for every country in Africa to calculate the "opportunities" for four types of communal conflict between independence and 1979: ethnic violence (which pits one group against another), irredentism (when one ethnic group attempts to secede to join co-ethnic communities in other states), rebellion (when one group takes action against another to control the political system), and civil war (when violent conflicts are aimed at creating a new ethnically based political system). Fearon and Laitin identified tens of thousands of pairs of ethnic groups that could have been in conflict. But they did not find thousands of conflicts (as might have been expected if ethnic differences consistently led to violence) or hundreds of new states (which partition would have created). Strikingly, for every one thousand such pairs of ethnic groups, they found fewer than three incidents of violent conflict. Moreover, with few exceptions, African state boundaries today look just as they did in 1960. Fearon and Laitin concluded that communal violence, although horrifying, is extremely rare.

The base-rate fallacy is particularly seductive when events are much more visible than nonevents. This is the case with ethnic conflict, and it may have led Muller astray in his account of the triumph of European nationalism. He emphasizes the role of violence in homogenizing European states but overlooks the peaceful consolidation that has resulted from the ability of diverse groups -- the Alsatians, the Bretons, and the Provençals in France; the Finns and the Swedes in Finland; the Genoese, the Tuscans, and the Venetians in Italy -- to live together. By failing to consider the conflicts that did not happen, Muller may have misunderstood the dynamics of those that did.

Of course, ethnic divisions do lead to violent conflict in some instances. Violence may even be so severe that partition is the only workable solution. Yet this extreme response has not been required in most cases in which ethnic divisions have existed. Making sense of when ethnic differences generate conflict -- and knowing how best to attempt to prevent or respond to them when they do -- requires a deeper understanding of how ethnicity works.

Muller offers one explanation for why ethnic identities figure so centrally in political conflict. Corresponding as it does to "enduring propensities of the human spirit," he argues, ethnonationalism "is a crucial source of both solidarity and enmity." This explanation echoes a fairly conventional account of ethnic conflict according to which people tend to prefer members of their own group and, in some cases, have active antipathy toward out-group members, making conflict the inevitable result. This is an appealing narrative. It helps outsiders make sense of the seemingly gratuitous violence of Africa's bloodiest conflicts. It resonates with the demonization of immigrants and the threats of ethnic domination that politicians around the world invoke in election campaigns. It appears consistent with demands for greater autonomy and self-government by ethnic enclaves in eastern Europe and the former Soviet Union. If ethnic diversity generates antipathies so deep that they cannot be realistically resolved, separation becomes the obvious and, perhaps, only feasible antidote, as Muller concludes. But positive feelings toward in-group members and antipathies for out-group members might not be the correct explanation for why political action is often organized along ethnic lines.

Indeed, recent research points to at least two alternative explanations. One argument suggests that members of the same group tend to work together to achieve collective ends not because of their discriminatory preferences but because of efficiency: they speak the same language, have access to the same types of information, and share social networks. In environments with scarce resources, they may even choose to work together against other groups, whether or not they care for or even like their peers. Thus, political coalitions form along ethnic lines not because people care more for their own but simply because it is easier to collaborate with their ethnic peers to achieve collective ends.

A second account emphasizes the norms that may develop within ethnic groups. Even when people see no efficiency gains from working with their co-ethnics and have no discriminatory preferences, they may still favor their own simply because they expect them to discriminate in their favor as well. Such reciprocity is most likely to develop in environments that are devoid of the institutions and practices -- for example, enforceable contracts and impartial state institutions -- that protect people from being taken advantage of by others. In such cases, reciprocity is a protection against being cheated.

Distinguishing these different theories is important because each one suggests a starkly different strategy for dealing with ethnic conflict. If the problem is tribal or national antipathies, there may well be some utility in separating groups. But if it stems from the technological advantages that accrue to members of the same ethnic group, then initiatives that break down barriers to cooperation (for example, Julius Nyerere's introduction of Swahili as a common language in Tanzania in the 1970s) are more likely to bear fruit. If instead discrimination in favor of one's ethnic peers is a coping strategy that individuals employ to compensate for the absence of functional and impartial state institutions, then the best response may be greater investment in formal institutions so that individuals are assured that cheating will be punished and that cooperation across ethnic lines will be reciprocated.

To discern these competing perspectives, we set out to study ethnicity and conflict using experimental games. We put people in strategic interactions with members of their own and other ethnic groups and examined the decisions they made. We carried out our research in Uganda, where differences between ethnic groups have been a basis for political organization and the source of persistent national political crisis and violent conflict since independence.

Remarkably, we found no evidence that people care more for the welfare of individuals from their own ethnic groups than for the welfare of those from other groups. Given the opportunity to make anonymous donations of cash to randomly selected partners, individuals were just as generous to out-group members as they were to their co-ethnics. One could easily tell a story that links Uganda's decades of ethnic conflict to tribal antipathies (and many have), but our research provided no evidence of such antipathies at work among a diverse sample of Ugandans.

We also found only weak evidence that impediments to cooperation across group lines explain the ethnic dynamics of Ugandan politics. In another set of experiments, we randomly matched participants with a partner and confronted the pairs with tasks that put a premium on successful communication and cooperation. We found no relation between the success in completing these tasks and the ethnic identities of the participants; success rates were just as high when individuals were paired with members of their own ethnic groups as when they were paired with people outside their ethnic groups. Hence, efficiency gains alone cannot easily account for the propensity of political coalitions to take on an ethnic character.

Instead, our studies suggested that patterns of favoritism and successful collective action within ethnic groups should be attributed to the practice of reciprocity, which ensures cooperation among group members. Our subjects showed no bias in favor of in-group members when given the opportunity to make cash donations anonymously, but their behavior changed dramatically when they knew that their partners could see who they were. When they knew that other players would know how they behaved, subjects discriminated strongly in favor of their co-ethnics. This shows, at least in our sample of Ugandans, that ethnic differences generate conflict not by triggering antipathy or impeding communication but by making salient a set of reciprocity norms that enable ethnic groups to cooperate for mutual gain.

Our experimental findings -- from a setting quite different from the European context that Muller treats but in which ethnic divisions run equally deep -- reveal that what might look from the outside like an intractable problem of discriminatory preferences may instead reflect norms of reciprocity that develop when individuals have few other institutions they can rely on to police the behavior of others.

Of course, ethnicity may not work in Uganda today the same way that it does in other parts of the world or that it did at other points in history. But our results do point out a need to consider seriously the possibility that the conventional view is at best an incomplete and at worst an incorrect explanation for why ethnic nationalism generates conflict when and where it does.

If ethnic hatreds are not at work, separating groups may not make much sense as a strategy for mitigating the corrosive effects of ethnic divisions. It might be far more important to invest in creating impartial and credible state institutions that facilitate cooperation across ethnic lines. With such institutions in place, citizens would no longer need to rely disproportionately on ethnic networks in the marketplace and in politics. In this respect, modernization may be the antidote to ethnic nationalism rather than its cause.

SEPARATISM'S FINAL COUNTRY

Richard Rosecrance and Arthur Stein

Muller argues that ethnonationalism is the wave of the future and will result in more and more independent states, but this is not likely. One of the most destabilizing ideas throughout human history has been that every separately defined cultural unit should have its own state. Endless disruption and political introversion would follow an attempt to realize such a goal. Woodrow Wilson gave an impetus to further state creation when he argued for "national self-determination" as a means of preventing more nationalist conflict, which he believed was a cause of World War I.

The hope was that if the nations of the Austrian, Ottoman, and Russian empires could become independent states, they would not have to bring the great powers into their conflicts. But Wilson and his counterparts did not concede to each nation its own state. They grouped minorities together in Hungary, Italy, and Yugoslavia, and the Soviet Union ultimately emerged as a veritable empire of nationalities. Economists rightly questioned whether tiny states with small labor forces and limited resources could become viable, particularly given the tariffs that their goods would face in international trade.

More important, the nationalist prospect was and remains hopelessly impractical. In the world today, there are 6,800 different dialects or languages that might gain political recognition as independent linguistic groups. Does anyone seriously suggest that the 200 or so existing states should each, on average, be cut into 34 pieces? The doctrine of national self-determination reaches its reductio ad absurdum at this point.

Furthermore, the one-nation, one-state principle is unlikely to prevail for four good reasons. First, governments today are more responsive to their ethnic minority communities than were the imperial agglomerations of yesteryear, and they also have more resources at their disposal than their predecessors did. Many provinces populated by discontented ethnic groups are located in territories adjacent to national capitals, not overseas. And many governments in this era of globalization have annual budgets equivalent to nearly 50 percent of their GDPs, much of which is spent on social services. They can -- and do -- accommodate the economic needs of their states' differentiated units. They also respond to those units' linguistic requests. Basques, Bretons, Punjabis, Québecois, and Scots live quite well inside the bonds of multinational sovereignty and in some cases better than residents of other provinces with no claims of being a distinct nation.

Second, the achievement of separate sovereignty today depends on external recognition and support. Prospective new states cannot gain independence without military assistance and economic aid from abroad. International recognition, in turn, requires the aspiring nationalist movement to avoid international terrorism as a means of gaining attention. If a separatist group uses terrorism, it tends to be reviled and sidelined. If an ethnic group does not have enough support to win independence by peaceful electoral means inside its country, its resorting to terrorism only calls into question the legitimacy of its quest for independence.

Recognizing this, the Québecois abandoned the terrorist methods of the Quebec Liberation Front. Most Basques castigate Basque Homeland and Freedom (known by its Basque acronym ETA). Enlightened Europeans have withdrawn their support for the Chechen rebels. And the continued terrorist shelling of Israeli cities from a Hamas-dominated Gaza might undermine the previous international consensus in favor of a two-state solution to the Palestinian problem, or at least warrant an exceptional approach to Gaza.

With the possible exception of the Palestinians, the notion that any of these peoples would be better off in smaller and weaker independent states in a hostile neighborhood is unrealistic. Occasionally, dissidents make the case that if they were to leave the state unit, they would be taken into the comforting embrace of the European Union or the North American Free Trade Agreement, thereby gaining access to a large market. But that would depend a great deal on outsider support for their cause. The United Kingdom might not wish to see Scotland in the EU and would be in a position to veto its membership. The United States and Canada might not agree to let an independent Quebec join NAFTA. The belief that when a tiny nation is born it falls automatically into the loving hands of international midwives is questionable. The truth varies from case to case.

Third, although globalization initially stimulated ethnic discontent by creating inequality, it also provides the means for quieting discontents down the road within the fold of the state political system. Distributed economic growth is a palliative for political discontent. Indonesia, Malaysia, Singapore, and Thailand contain different ethnic groups that have largely profited from the intense economic resurgence of their states stimulated by globalization. Northern and southern Vietnam are culturally different, but both have benefited from the country's economic growth. Cambodia has a diverse population, but it has gained greatly from China's move to externalize some of its production.

Fourth, a discontented population may react to ethnic discrimination, but it also responds to economic need, and whatever its concerns, it does not always have to seek independence to alleviate them. It has another safety valve: emigration to another country. The state of Monterrey has not sought independence from Mexico; rather, many of its inhabitants have moved, legally or illegally, to the United States. The huge emigration from the Maghreb to France and Italy reflects a similar attitude and outcome; the dissatisfied populations of North Africa can find greater welfare in Europe. And when Poles move to France or the United Kingdom, they do not secede from the mother country but demonstrate greater satisfaction with French or British rule. Emigration is the overwhelming alternative to secession when the home government does not sufficiently mitigate economic disparities.

Even where the central government has used force to suppress secessionist movements, it has offered carrots at the same time that it has yielded sticks. The province of Aceh has been coaxed, even as it has been subjected to threats, to remain inside the Indonesian republic. Kashmir, facing a balance of restraints and incentives, is unlikely to emerge as an independent state in India. And the Tamil Tigers have lost the sympathy of the world by their slaughter of innocent Sinhalese.

The recent formation of an "independent" Kosovo, which has not yet been recognized by various key countries, does not foretell the similar arrival of other new states. It is unlikely that Abkhazia or South Ossetia, although largely autonomous in fact, will gain full and formal independence from Georgia or that the Albanian areas of Macedonia will secede. Rather, prospective secessionists, dissuaded by both central governments and the international community, are likely to hold back. Indeed, the most plausible future outcome is that both established states and their international supporters will generally act to prevent a proliferation of new states from entering the international system.

Much empirical work, which shows that a province's aspirations for sovereign status can be confined within a state if the province has access to monies from the central government and is represented in the governing elite, supports this conclusion. The Sikh party Akali Dal once sought Punjab's independence from India, but to little effect, partly because Punjabis are heavily represented in the Indian army and because fiscal transfers from New Delhi quieted dissidence in the region. The Québecois benefit from financing from Ottawa, elite connections, flows of private capital into Quebec, and the Canadian government's acceptance of bilingualism in the province. Chechnya remains poor, but if it seeks to remedy its relative neglect through a strategy of terrorism, it will undercut its own legitimacy. Lacking external support, and in the face of Russia's continued firmness, Chechnya has settled into a degree of political stability. In all three cases, the maintenance of the existing national boundaries seems likely, and so, too, does it seem likely in other cases.

The apostles of national self-determination would do well to consider a still more important trend: the return to bigness in the international system. This is happening not only because great powers such as China, India, and the United States are now taking on greater roles in world politics but also because international economics increasingly dwarfs politics. To keep up, states have to get bigger. The international market has always been larger than the domestic ones, but as long as international openness beckoned, even small powers could hope to prosper and attain some degree of economic influence. In the past decade, however, the tariff reductions proposed in the Doha Round of international trade negotiations have failed, industrial duties have not fallen, and agriculture has become even more highly protected than it was in the nineteenth century.

Globalization has clearly distributed economic boons to smaller countries, but these states still require greater political scale to fully realize globalization's benefits. To generate scale, states have negotiated bilateral and multilateral trade preferences with other states regionally and internationally, thereby gaining access to larger markets. The EU has decided to make up in the enlargement of its membership and a bigger free-trade area what it lacks in internal economic growth. The 27 countries of the EU currently have a combined GDP of over $14 trillion, besting the United States' $13 trillion, and the union's expansion is not over yet.

Europe never faced the limits on "manifest destiny" that confronted the United States -- the shores of the Pacific Ocean. Charles de Gaulle was wrong when he heralded a "Europe from the Atlantic to the Urals": the EU has already expanded into the Caucasus. And with at least eight new members, it will proceed into Central Asia. As the borders of Europe approach Russia, even Moscow will seek de facto ties with the increasingly monolithic European giant.

In Asia, current tensions between China and Japan have not prevented proposals for a free-trade zone, a common currency, and an investment bank for the region. Chinese in Indonesia, Malaysia, the Philippines, Singapore, Taiwan, and Vietnam draw their adopted countries toward Beijing. China will not expand territorially (except titularly when Taiwan rejoins the mainland), but it will move to consolidate an economic network that will contain all the elements of production, except, perhaps, raw materials. Japan will adjust to China's primacy, and even South Korea will see the writing on the wall.

This will leave the United States in the uncomfortable position of experiencing unrealized growth and the possible failure of new customs unions in the Western Hemisphere. NAFTA may have been deepened, but a Free Trade Area of the Americas now seems beyond reach because of opposition from Argentina, Bolivia, Brazil, and Venezuela. U.S. politics has also turned, temporarily at least, against such ventures. South American nations have, in recent years, been far more responsive to China and Europe than to the United States. The U.S.-Central American Free Trade Agreement, now in the making, may be the only likely new string to the current U.S. bow.

Some economists contend that great size is not necessary in a fully open international economic system and that even small countries can sell their wares abroad under such conditions. But the international economic system is not open, and the future resides with broad customs unions, which substitute expanded regional markets for restricted international ones. China is seeking bilateral preferential trade arrangements with several other states, and so is the United States. Prospective secessionists will not prosper under such circumstances. They have to depend on international assistance, membership in trade pacts, and the acquiescence of their mother countries. They may have none of these, and they will fail if they use terrorism to advance their causes.

Under the present circumstances, secessionists will generally be better off remaining inside existing states, if only because the international system now advantages larger agglomerations of power. Economies of industrial scale are promoting economies of political size. In U.S. politics, the problem of outsourcing gets much political attention, but how is it possible to prevent that activity when national production and the national market are too small? Only larger political entities can keep production, research and development, and innovation within a single economic zone. Big is back.

MULLER REPLIES

My essay is not agenda-driven or prescriptive. It is meant to suggest that the power of ethnic nationalism in the twentieth century has been greater than is generally recognized and that the probability of its ongoing global impact is greater than is generally appreciated. I argue that Americans often have a distorted sense of substantial areas of the world because they tend to generalize on the basis of their own national experience or, rather, a truncated and idealized version of that experience. Of course, ethnicity (and its conceptual cousin, race) has long played a role in American life and continues to do so, as reflected in everything from residential patterns to voting behavior.

But by and large, ethnic identification in the United States tends to erode across generations, and the notion that different ethnic groups ought to have their own political entities is marginal. (Voting districts drawn along racial lines echo conceptions of ethnic nationalism. And the Chicano vision of the reconstitution of Aztlán -- a lost nation of indigenous Americans said to include Mexico and much oF the American Southwest -- would qualify as ethnonationalist but seems to have limited appeal.) Thus, Americans have a hard time imagining the intensity of the desire that many ethnic groups abroad have for a polity of their own or the determination of others to maintain the ethnic structure of existing polities. If Poles and Ukrainians get along tolerably well in Chicago, why not Sunni Arabs, Kurds, and Turkmen in Kirkuk?

I further argue that this misperception also occurs among educated western Europeans, who project the cooperative and pacific model of the EU onto the rest of the world while losing sight of the history of ethnic disaggregation that seems to have served as a precondition for the comity of contemporary Europe. The propensity to impose on the rest of the world one's own categories and idealized conceptions of one's historical and current experiences leads to a kind of misleading universality, apt to result in misunderstanding and miscalculation.

There are categories of self-definition that are unfamiliar or uncomfortable to some people's sensibilities -- including ethnonational identity, caste (common in India), or tribe (common in much of Africa and the Muslim world). But the fact that some people may find these categories unreal (since they know that beneath the skin humans are ultimately the same: put them in a room together with a game to play, and see how little they differ) does not make them any less real to those who do believe in them.

The problem of taking seriously the diverse ways in which people in different parts of the world define themselves is exacerbated by the universalizing and scientistic pretensions of some streams of academic political science. "Scientism" refers to the endeavor to apply the methods and criteria of the natural sciences to all realms of human experience -- although for some they are inappropriate. This includes the effort to explain all phenomena with simplified theories of human motivation and the attempt to replicate the hard sciences by using laboratory conditions to study political science. History provides a useful source of data with which to study the range and complexity of human behavior. It is a highly imperfect laboratory, where both the data and their interpretation are influenced by the methodological and ideological predispositions of the investigator. But it is often superior to the alternative: apparently scientific forms of explanation.

My claim is not that the violence of the European experience will repeat itself but a more modest one: that ethnic tensions are likely to be exacerbated, rather than eliminated, by the occurrence of similar processes of modernization in other parts of the world. Contrary to what James Habyarimana, Macartan Humphreys, Daniel Posner, and Jeremy Weinstein claim, nowhere do I argue that "ethnic divisions inevitably generate violence." And while I quoted Churchill, I did not endorse his views as a general policy prescription.

What I actually wrote, toward the end of my article, was this: "Sometimes, demands for ethnic autonomy or self-determination can be met within an existing state.... But such arrangements remain precarious and are subject to recurrent renegotiation. In the developing world, accordingly, where states are more recent creations and where the borders often cut across ethnic boundaries, there is likely to be further ethnic disaggregation and communal conflict. And as scholars such as Chaim Kaufmann have noted, once ethnic antagonism has crossed a certain threshold of violence, maintaining the rival groups within a single polity becomes far more difficult.

"... When communal violence escalates to ethnic cleansing, moreover, the return of large numbers of refugees to their place of origin after a cease-fire has been reached is often impractical and even undesirable.

"... Partition may thus be the most humane lasting solution to such intense communal conflicts."

Habyarimana, Humphreys, Posner, and Weinstein continue their misrepresentations by claiming that I attribute ethnic tension merely to "enduring propensities of the human spirit"; in fact, I attribute ethnic tension to "some enduring propensities of the human spirit that are heightened by the process of modern state creation." My explanation, drawn largely from the sociologist Ernest Gellner, is actually echoed by the four co-authors, albeit in a different vocabulary, when they write that members of the same ethnic group tend to come together because "they speak the same language, have access to the same types of information, and share social networks." As so often happens in the social sciences, here is an attempt at product differentiation through rebranding -- recasting known insights in a new vocabulary.

More novel is the authors' belief that their quasi-scientific experiments in Uganda provide useful new avenues for public policy. They say that their game-playing experiments provide insights as to how the diverse ethnic actors would behave when freed of the social and political contexts in which their actions are known to others. Perhaps, but it is precisely the nature of the real world that this would never be the case.

Moreover, their conclusion that the problem lies in a weak institutional environment characterized by an "absence of functional and impartial state institutions" is both true and misleading, for it fails to consider that the very multiplicity of ethnicities is among the major sources of this institutional environment. A reading of Chinua Achebe's 1960 novel, No Longer at Ease -- about the plight of an idealistic young civil servant who tries to embody the ethos of impartiality in a setting in which such norms are at odds with the understanding of his co-ethnics, who regard his bureaucratic position as a form of group property -- would cast more light on the situation than hundreds of experimental games.

This is not the place for a full critique of the much-cited calculations of Fearon and Laitin on the incidence of interethnic violence in Africa from 1960 to 1979. If one lives in a neighborhood where three in a thousand interactions result in violence and one has three interactions per day, one is violently attacked only three times a year. But is that a safe neighborhood or a dangerous one? The assertion that "with few exceptions, African state boundaries today look just as they did in 1960" is also both true and misleading. It attests as much to the ability of the dominant ethnic coalitions to suppress attempts at rebellion as to the absence of ethnic conflict.

The Biafran War (1967-70) counts as only one incident of interethnic violence in Fearon and Laitin's data and resulted in no change of borders. The million or so lives lost do not register in their calculations. Had Fearon and Laitin repeated their computations for the years since 1979, the murder of some 800,000 Rwandans (mostly Tutsis) would also have appeared as a matter of small statistical consequence.

The claim by Richard Rosecrance and Arthur Stein that the ethnonationalist ideal of a separate state for each cultural unit has been a source of instability is true or, at least, a half-truth. That is what a good part of my article is about. But the fact that ethnonationalism is destabilizing has not diminished its appeal or impact. The other half-truth is that the fulfillment of the ethnonationalist ideal has had a stabilizing effect, at least for large groups.

However, as my article notes and as Rosecrance and Stein emphasize, not every ethnonational aspiration can be realized, and ethnonational aspirations for autonomy and self-determination can be realized within larger political units through federalism -- the devolution of power and income to subnational units. As such, federalism represents a form of "semi-partition," as the political scientist Donald Horowitz has noted. It has the very real advantage of permitting participation in larger political and economic units. But, as Horowitz has also noted, "federalism is not cheap. It involves duplication of facilities, functions, personnel and infrastructure" and often entails jurisdictional disputes. Moreover, "states that could benefit from federalism typically come to that realization too late, usually after conflict has intensified."

Rosecrance and Stein may be right that a greater pool of income can alleviate ethnonational aspirations. But it is worth recalling that the governments in a position to distribute sums equivalent to 50 percent of their GDPs are in Europe, whereas the ethnic groups in potential conflict are in Africa, Asia, and Latin America, where there is less wealth and so less GDP available for redistribution. Moreover, massive government redistribution through taxation may itself inhibit economic growth or make capturing the state apparatus too enticing a prize compared with other pursuits.

Various claims by Rosecrance and Stein are questionable, if not clearly mistaken. The authors assert that mass emigration can serve as an "alternative to secession when the home government does not sufficiently mitigate economic disparities." First, this assumes that all discontents are ultimately expressions of individually conceived economic interest, a radical simplification of human motivation that ignores the desire of some people to share a common culture and their perception that protecting that culture requires political autonomy. For example, throughout much of the first half of the twentieth century, French Canadians emigrated from Canada to the United States, where over time they assimilated into the larger population. Québecois nationalism represents a rejection of that path.

Second, the authors' emigration-as-safety-valve strategy ignores the fact that in contrast to the earlier era of globalization (from the late nineteenth century through World War I), the current era of globalization is characterized by governments better able and more inclined to police their borders and, hence, by the comparatively limited mobility of people across national borders. Moreover, discontent in the relatively wealthy states of the West with some recent streams of immigration has already led to pressures for governments to exercise greater control over the movement of people from particular regions. It is far from clear that emigration from the Maghreb to France, for example, will be allowed to continue indefinitely.

Rosecrance and Stein's assertion that a new era of bigness in international economic affairs is here is truer than the implications they draw from it. The economic advantages of the division of labor do expand with the extent of the market, as Adam Smith explained over two centuries ago. But it is simply not true that "to keep up, states have to get bigger." States can negotiate treaties and other forms of association that allow for freer international trade. As the authors note in passing, smaller nations have opted for inclusion in transnational markets and have often prospered as a result.

In short, Rosecrance and Stein assume that a rational economic calculus governs international activity. This simplification of human motivation has the advantage of methodological elegance. But their predictions conflate three very different circumstances: what global actors would do if they rationally calculated their utilities based on a set of preferences much like those of American professors of political science; what global actors would do if they rationally calculated their utilities based on their actual preferences, which may diverge substantially from those of American political scientists; and what may actually happen given the unlikelihood that either American political scientists or global actors would rationally calculate their utilities. That is to say, Rosecrance and Stein have purchased methodological elegance at the expense of explanatory power by radically reducing the range of relevant motivations and interactions.

For a historian, the authors' assertion that "international economics increasingly dwarfs politics" -- like so much of their response to my essay -- is eerily resonant of a British bestseller of a century ago. In 1910, Norman Angell published The Great Illusion, which explained on economic grounds why an extended war between great powers was impossible under the contemporary economic conditions. His argument was logically compelling but wrong. In 1933, Angell published a new edition of his book, in which he suggested that nations could not enrich themselves by conquering their neighbors and that war was therefore futile. He was awarded the Nobel Peace Prize, but his message seems not to have reached all the relevant parties. I fear that the predictions of Rosecrance and Stein about the future of ethnic nationalism will meet the same fate.

Still, Rosecrance and Stein do raise an important issue that I did not explore in my article: the question of external recognition and support for potential new states. What ought to be the response by external countries, such as the United States, to ethnonationalist claims for independence? If one takes seriously the forces leading to the enduring power of ethnonationalism -- rather than dismissing them as archaic, illusory, or subject to elimination by good governance conjured out of the blue -- the implications for policy are by no means self-evident.

I leave aside the purely legal and philosophical issues, since the "right" to self-determination, like so many others, often conflicts with other purported rights. Representatives of existing states are strongly disposed against the redrawing of borders and the formation of new states. They see self-interest in maintaining the international status quo, which may or may not be justified by prudence. To recognize that national self-determination does provide satisfactions of its own and may well result in viable states is not to say that the endless creation of new states is either viable or desirable. Yet there are dangers both in supporting ethnonationalist claims and in denying them prematurely.

One danger of the international recognition of insurgent ethnonationalist claims to sovereignty is that it may lead to unilateral secession (as in the recent case of Kosovo) as opposed to mutually agreed separation. Secession without ethnic partition usually means that the new political entity will include a substantial minority of people whose co-ethnics dominate the state from which the new state has seceded. This provides a ready source for new ethnic tensions within the new state and international tensions between the new state and the old. Mutually agreed partition that separates the rival ethnic groups may be preferable in order to minimize the likelihood of future conflict. Another danger of a greater international willingness to recognize ethnonationalist movements is that it may create an incentive for the governments of existing countries to violently crush incipient ethnic political movements before they can organize.

There are perils, however, in a blanket refusal of the international community to recognize the claims of legitimate ethnonationalist movements. For having deemed secession an impossibility, governments may feel no incentive to respond to the desire of ethnic groups for greater power and self-determination within the confines of the current states. To recognize the enduring power of ethnic nationalism is not to support it or provide a ready recipe for action but to offer a more realistic appreciation of the dilemmas that will continue to arise in the twenty-first century.