Starting in the 1960s, a number of Western democracies embarked on new experiments in accommodating indigenous peoples, national minorities and immigrant groups. Although these “experiments in multiculturalism” have always been the subject of intense public debate, there was surprisingly little academic attention to them in the 1970s and 1980s. It was only in the 1990s that multiculturalism became a major area of academic inquiry, picked up initially by political theorists.Footnote ¹ Indeed, for much of the 1990s, the academic literature on multiculturalism was heavily normative, dominated by political philosophers who developed idealized theories of a distinctly liberal–democratic and egalitarian form of multicultural citizenship. These theorists were interested in the question of whether multiculturalism was consistent in principle with their ideal theories of justice—for example, whether multiculturalist claims could be defended from within Rawls's theory of liberal justice—and attempted to show how familiar liberal–democratic principles of individual freedom and distributive justice could be invoked to give a principled defense of certain multiculturalist claims.

These normative philosophical accounts of multiculturalism have proven quite influential, shaping debates on multiculturalism not just within the field of philosophy, but more widely in academia and indeed in public life. A good example is the remarkable international influence of Charles Taylor's Reference Taylor and Gutmann1992 essay, “Multiculturalism and the Politics of Recognition,” translated into numerous languages and cited in discussions of multiculturalism from the isolated highlands of Bolivia to the teeming streets of Paris or Tokyo.

An important aspect of these theories is what I will call the “liberal multicultural hypothesis.” According to this hypothesis, states can adopt multiculturalism policies to fairly recognize the legitimate interests of minorities in their identity and culture without eroding core liberal–democratic values. We can represent this hypothesis schematically, as in Figure 1. On the left-hand side, we have a list of multiculturalism policies (hereafter MCPs) of the sort that have in fact been adopted within one or more of the Western democracies since the 1960s in relation to immigrants, national minorities and indigenous peoples.Footnote ² On the right-hand side, we have a list of the three traditional liberal–democratic values—freedom, equality and solidarity—as defined by the French Revolution and reformulated in Rawls's canonical formulation (Rawls, Reference Rawls1971).Footnote ³

Figure 1

The liberal multiculturalist hypothesis, then, is that one can adopt one or more of the policies on the left, without jeopardizing the values on the right. Of course, a normative defense of liberal multiculturalism need not make this sort of empirical prediction. A philosophical defense of liberal multiculturalism could simply argue that members of minority groups have legitimate interests in their languages and cultures, and that justice requires accommodating these legitimate interests through MCPs, without making predictions one way or the other about how such MCPs affect broader liberal–democratic values. Perhaps MCPs are needed to fairly recognize minority's interests in language and culture, but their effect over time is to undermine the stability of democratic procedures, or the viability of a welfare state. If so, then we would be faced with a tragic choice: the need to choose between justice for minorities and broader liberal–democratic values.

However, most normative theorists of liberal multiculturalism do not view MCPs as creating these sorts of tragic choices. On the contrary, normative theorists of liberal multiculturalism typically assume, implicitly or explicitly, that justice for minorities through MCPs can be pursued without threatening core liberal–democratic values and without having to compromise on a society's commitment to individual freedom, equal opportunity and social solidarity. And this assumption has been crucial in explaining the influence of normative theories of liberal multiculturalism. A commitment to liberal multiculturalism is not a commitment to “justice for minorities even if the heavens will fall,” but rather “justice for minorities because the heavens won't fall.” This is the assumption I am formulating as the liberal multiculturalist hypothesis: that one or more MCPs can be adopted without jeopardizing core liberal–democratic values. Indeed, at its strongest, the hypothesis says that a country could adopt all of the policies on the left without jeopardizing any of the values on the right.

Put that way, the hypothesis seems utopian, a highly idealized vision of how one might wish the world to be but not a realistic basis for describing or predicting the effects of multiculturalism in the world as it actually is. And indeed, starting in the mid-1990s, a number of social scientists began to complain that these idealized theories were not adequately informed by empirical findings about the nature of ethnic relations and ethnic politics. From the perspective of these social scientists, it was naive, even irresponsible, to base public policy on these overly normative theories. If we attended to the findings of social science, rather than the fantasies of political philosophers, we would be much more skeptical of the liberal multiculturalist hypothesis.

In particular, social science would have told us (1) that the real-world motivations for making multiculturalist claims were rarely a disinterested or benign quest for liberal justice but rather a self-interested quest by ethnic entrepreneurs to gain greater power and control, often at the expense of group members; and (2) that even if multiculturalist claims were well-intentioned, their adoption would likely create perverse effects, such as exacerbating prejudice or distrust, or reproducing segregation and exclusion. These problems, it was said, would have been clear if policy makers had been informed more by social science than by philosophy, if they had paid less attention to “moral or philosophical justifications” for multiculturalism and more attention to the evidence about its “underlying assumptions about human behaviour” (Reitz, Reference Reitz, Breton, Dion and Dion2009: 2), or about “the strategic and political context in which cultural claims are formulated” (Jung, Reference Jung and Shabani2007: 269).

We can find versions of this complaint from scholars in a wide range of social science disciplines, including sociologists such as Rogers Brubaker and Frederick Cooper (Reference Brubaker and Cooper2000), Christian Joppke (Reference Joppke2001, Reference Joppke2004), or Adrian Favell (Reference Favell1998a, Reference Favell1998b), political scientists such as David Laitin (Reference Laitin1998), James Johnson (Reference Johnson2000), and Courtney Jung (Reference Jung and Shabani2007), and anthropologists such as Terence Turner (Reference Turner1993) or Jane Cowan (Reference Cowan2001), all of whom bemoan the undue influence of normative political philosophy in the multiculturalism debate.Footnote ⁴

The heart of this critique, then, is that the liberal multiculturalist hypothesis needs to be tested against the empirical evidence, with the expectation that this will require some dramatic revision, if not outright refutation, of the hypothesis. And indeed this complaint has helped to inspire almost fifteen years of new empirical research by social scientists. We now have much more evidence from all of these disciplines. High-profile social scientists have become as influential as philosophers in the general debate over multiculturalism, and normative ideas about the principled arguments for liberal multiculturalism have been supplemented, if not displaced, by new ideas and concepts from the social sciences.Footnote ⁵

This dramatic growth in social science is all to the good, since public policy should indeed be based on empirical evidence and not just idealized theories of justice. But it raises the question: what does all of this evidence tell us about the prospects for a liberal–democratic multiculturalism? What does it tell us about whether or when MCPs support or erode liberal–democratic values?

One safe but boring answer is to say that we don't yet have enough evidence. This indeed is the conclusion reached by several recent surveys, which insist that the effects of MCPs on liberal–democratic values remain largely unknown (Marc, Reference Marc, Dani and De Haan2008, Reference Marc2009; Reitz, Reference Reitz, Breton, Dion and Dion2009; Government of British Columbia, 2008).Footnote ⁶ But while recognizing the incompleteness of the evidence, I think it's worth trying to identify at least in a preliminary way the emerging lessons or conclusions of the past 15 years of social science research. After all, policy makers and practitioners need to make decisions now, based on the evidence to hand, so it's worth asking what the available evidence adds up to.

And here I want to raise a bold and surprising possibility: namely, that the evidence to date, far from refuting the liberal multiculturalist hypothesis, actually supports it. Despite the widespread presumption that greater attention to empirical evidence would lead to greater skepticism about the merits of liberal multiculturalism, I want to suggest that the evidence bolsters the case.

For the purposes of this paper, I will focus on the evidence in relation to immigrant multiculturalism, although I think a similar argument could be made in relation to national minorities and indigenous peoples.Footnote ⁷ What then does the evidence tell us about the impact of immigrant MCPs on basic liberal–democratic values of freedom, equality and solidarity? Let me start be reviewing the evidence in the Canadian case, which was the first Western country to adopt an official multiculturalism policy and remains the only country in which multiculturalism is enshrined in the constitution. It therefore provides a good first test case: if multiculturalism has pernicious effects, they should have shown up in Canada.

In reality, however, the evidence suggests that MC is having a positive effect in the domains of freedom, equal opportunity and solidarity. Consider the following findings of recent research.

Political rights. Compared to every other Western democracy, immigrants in Canada are more likely to become citizens (Bloemraad, Reference Bloemraad2006). Nor is this simply a desire to gain the safety or convenience of a Canadian passport. Compared to other countries, these naturalized immigrants are more likely to actually participate in the political process, as voters, party members or even candidates for political office (Howe, Reference Howe, Banting, Courchene and Seidle2007). For example, there are more foreign-born citizens and Canadian-born minorities elected to parliament in Canada than in other Western countries, both in absolute numbers and in terms of parity with their percentage of the population (Adams, Reference Adams2007: 70–74). While the percentage of foreign-born in the Canadian federal Parliament (13 per cent) is lower than the percentage of foreign-born in the overall population (19.3 per cent in the 2001 census), this level of “demographic parity” is (far) higher than in the US (2 per cent foreign-born in the House of Representatives versus 14.7 per cent in population) or Australia (11 per cent versus 23 per cent) or any European countries. (In fact, in France most of the foreign-born members of parliament are the children of French diplomats or colonial settlers, not people of immigrant ethnic origin). Moreover, it's worth noting that the foreign-born MPs in Canada are not only, or even typically, elected in ethnic enclave ridings composed of their own co-ethnics (Adams, Reference Adams2007: 77–80). To be sure, there remain obstacles to the political participation of immigrants and ethnic minorities in Canada (Bird, Reference Bird and Andrew2004, Reference Bird2005, Reference Bird2007). But, compared to other countries, political parties in Canada are more likely to actively recruit minority candidates and to run them in competitive ridings (and not just as token candidates). And once nominated, there is no evidence that voters in Canada discriminate against such candidates (Black and Erickson, Reference Black and Erickson2006; Bird, Reference Bird2009). No doubt there are many factors that explain this, including the fact that Canada tends to select more highly skilled immigrants than other countries. But those scholars who study the political participation of immigrants in Canada in comparison with other countries concur that multiculturalism has indeed enhanced the effective exercise of political liberties.Footnote ⁸ Indeed, Canada's comparative advantage in the political integration of immigrants is often most striking precisely in relation to less skilled immigrants.Footnote ⁹

Equal opportunity. Equal opportunity has two key dimensions: first, to acquire skills; and second, to translate those skills into jobs that are commensurate. In both cases, Canada has a comparative advantage. According to a recent OECD study, the children of immigrants and minorities have better educational outcomes in Canada than in any other Western democracy. Indeed, uniquely among Western countries, second-generation immigrants in Canada actually outperform children of non-immigrant parents (OECD, 2006). Again, this is not solely due to the higher socioeconomic background of immigrants in Canada; on the contrary, immigrant children from lower socio-economic backgrounds also do better in Canada than in other countries. In terms of acquiring employment, immigrants in all Western societies suffer from an “ethnic penalty” in translating their skills into jobs. However, the size of this ethnic penalty differs from country to country, and according to a recent British Academy study, the penalty is lowest in Canada (Heath, Reference Heath, Heath and Cheung2007). Here again, lots of factors explain this comparative record, but there are good reasons to think that MCPs play a role. For example, the massive OECD study that established Canada's comparative advantage in educating immigrant students emphasized that a crucial factor in this success was the presence of specific policies to address issues of cultural and linguistic diversity in the school population, policies that, in the Canadian context, have emerged under the rubric of multiculturalism. These diversity policies help to explain why the children of immigrants do better in Canada even when controlling for the skills, education and income of their parents, in part, because of the way they help children acculturate (Berry et al., Reference Berry, Phinney, Sam and Vedder2006).

Solidarity. Much of the anxiety around multiculturalism has focused not on its effects on political or economic integration, but rather its effects on social integration, and particularly on feelings of mutual identification and acceptance. Yet what is truly striking is the high level of mutual identification among immigrants and native-born Canadians. Canadians view immigrants and demographic diversity as key parts of their own Canadian identity. Compared to every other Western democracy, Canadians are more likely to say that immigration is beneficial, and less likely to believe that immigrants are prone to crime (Focus Canada, 2006). And immigrants and minorities return the compliment; they have very high levels of pride in Canada and are most proud of Canada's freedom and democracy, and its multiculturalism (Adams, Reference Adams2007). This high level of mutual identification is also reflected in the fact that whereas ethnic diversity has been shown to erode social capital or trust in other countries, there appears to be a “Canadian exceptionalism” in this regard (Kazemipur, Reference Kazemipur2009). Here again, there are many factors at work beyond the presence of MCPs, but social scientists working in the field concur that the presence of multicultural norms has played an important role, helping to “normalize” diversity (Harell, Reference Harell2009; Kazemipur, Reference Kazemipur2009). In particular, multiculturalism provides a locus for the high level of mutual identification among native-born citizens and immigrants in Canada. In many countries, native-born citizens with a strong sense of national identity or national pride tend to be more distrusting of immigrants and minorities, who are seen as a threat to their cherished national identity (Sides and Citrin, Reference Sides and Citrin2007). But the fact that Canada has officially defined itself as a multicultural nation means that immigrants are a constituent part of the nation that citizens feel pride in, so multiculturalism helps native-born citizens to link national identity to solidarity with immigrants and minorities. And, conversely, multiculturalism provides a link by which immigrants and minorities come to identify with, and feel pride in, Canada. From their different starting points, there is convergence on high levels of pride and identification with a multicultural conception of Canadian nationhood. Studies show that in the absence of multiculturalism, these links are more difficult to establish, and national identity is more likely to lead to intolerance and xenophobia (Esses et al., Reference Esses, Wagner, Wolf, Preiser and Wilbur2006). Indeed, Canada is the only Western country where strength of national identity correlated positively rather than negatively with pro-immigration attitudes (Laczko, Reference Laczko2007), a striking finding that I believe can only plausibly be explained by the fact that multiculturalism is now part of Canadian national identity.

So there is growing social science evidence that, at least in the Canadian case, MCPs are not only consistent with core liberal–democratic values of freedom, equality and solidarity, but actually enhance them. A skeptic might respond that Canada is somehow an outlier or exception in this regard and that we can't draw any generalized lessons from the Canadian evidence. Fair enough. So let's set aside the Canadian case, and ask which country comes second in these cross-national studies of immigrant political participation, educational attainment, equal opportunity in the marketplace and mutual respect. Interestingly, the answer typically is Australia, which is precisely the country that most quickly and closely followed Canada in adopting a multiculturalism policy.Footnote ¹⁰

A skeptic might retreat further, and argue that Canada and Australia both reflect peculiarities of New World “countries of immigration,” such that evidence from those countries can't be applied to Europe. But if we ask which European country does best on these criteria, it is Sweden, which has been one of the strongest and most consistent proponents of a multicultural approach. Sweden outperforms those countries that never embraced multiculturalism (for example, France, Germany), or that have retreated from earlier commitments (for example, Netherlands, Britain).Footnote ¹¹

So we seem to have a pattern here: countries with strong and consistent policies of multiculturalism outperform the other Western democracies. Of course, this is still just three countries, and perhaps all three are somehow “exceptional,” rather than manifesting a more general rule about the beneficial effects of MCPs. But insofar as we have cross-national data on the effects of MCPs, the preliminary evidence suggests that this indeed is a more generalized effect. For example, John Berry's cross-national study of 13 countries shows that children are better adapted in countries with multiculturalism policies (Berry et al., Reference Berry, Phinney, Sam and Vedder2006); Christel Kesler and Irene Bloemraad's cross-national study of diversity and social capital in 19 countries shows that multiculturalism policies have a positive impact on political participation and social capital (Kesler and Bloemraad, Reference Kesler and Bloemraad2010); Steven Weldon's cross-national study of prejudice shows that multiculturalism policies have a positive effect on reducing prejudice (Weldon, Reference Weldon2006); and earlier cross-national work that I conducted with Keith Banting suggested that multiculturalism policies may have a positive effect on redistribution (Banting and Kymlicka, Reference Banting, Kymlicka, Banting and Kymlicka2006).Footnote ¹²

So it seems that there is at least some plausible evidence in support of the liberal multiculturalist hypothesis. I hasten to add that none of these bits and pieces of evidence adds up to “proof” of that hypothesis. Legitimate questions can be raised about the way these studies define the dependent and independent variables, and about the limitations of the data. Further research will almost certainly require modifying and qualifying their conclusions. Yet, as I said earlier, we often need to make provisional judgments based on the evidence to hand, and so far as I can tell, this evidence strongly (if provisionally) suggests that liberal multiculturalism is a viable approach, worthy of being considered by policy makers.

But if so, this raises a puzzle. For the general tenor of most discussions of multiculturalism, both in academia and policy circles, is of a “failure” or “crisis” of multiculturalism. To take just one example, the Committee of Ministers of the Council of Europe recently issued a report which states, without hesitation or qualification, that multiculturalism has failed in Europe, and indeed that it has proven to be as harmful as the assimilationist approach it replaced (Council of Europe, 2008). This is a remarkable statement: here we have a formal intergovernmental body, speaking on behalf of its 47 member states, stating that there is, in effect, a European consensus that multiculturalism is no longer a viable approach. This is just one of many examples one could give of the now widespread assumption, found among both intergovernmental organizationsFootnote ¹³ and academic commentators,Footnote ¹⁴ that we have moved into a “post-multicultural” age.

What explains this surprisingly widespread judgment that multiculturalism is no longer a viable option? One possibility, of course, is that I have misread the available evidence and that the retreat from multiculturalism reflects a sound example of evidence-based policy making. As I read the debate, however, the retreat from multiculturalism is even less evidence-based than its original advocacy by political theorists. To be sure, empirical evidence is often invoked by critics of multiculturalism, but in my view, this evidence is being invoked in ways that distort rather than illuminate the debate over MCPs.

Let me give two quick examples, which I think have been particularly influential. The first is the remarkable global influence of Robert Putnam's work on diversity and social capital. According to Putnam's widely cited findings, as the ethnic diversity of American cities increases, social capital declines, reflected in lower levels of civic associational life and of interpersonal trust (Putnam, Reference Putnam2007). These findings are widely seen as proof that the multicultural ideal is not working and have been invoked repeatedly by critics of multiculturalism (for instance, Goodhart, Reference Goodhart2004). But even if Putnam's findings turn out to be robust—and the jury is still out on that question—how do they relate to multiculturalism? The assumption often made by critics is that if diversity drives down social capital, then adopting policies that recognize and accommodate diversity will exacerbate the problem. But this is an empirical claim that needs to be tested and, as I noted earlier, the evidence to date suggests that MCPs ameliorate rather than exacerbate the situation. Nor should this is surprising, since of course the whole point of multiculturalism is to normalize diversity. To be sure, this is very preliminary evidence, and we need more research on how MCPs affect social capital (and how social capital in turn affects liberal–democratic values of freedom, equality and solidarity). Unfortunately, too many commentators have jumped from Putnamesque findings about the negative correlations between ethnic diversity and social capital to predictions about the negative effects of MCPs, predictions that are at best empirically unfounded and at worst irresponsible.

A second example, particularly influential in the European debate, concerns the Dutch case. The Netherlands is widely cited as an example of a country where multiculturalism “failed,” with negative effects on liberal–democratic values. Based on this one case, several scholars have jumped to the conclusion that multiculturalism in general has negative effects. A particularly influential version of this argument is the recent book by Paul Sniderman and Luuk Hagendoorn, which lays the blame for prejudice and ethnic polarization in the Netherlands at the feet of multiculturalism, and concludes that other countries should avoid the Netherlands' costly and failed experiment with MCPs (Sniderman and Hagendoorn, Reference Sniderman and Hagendoorn2007). I have elsewhere criticized their analysis, which in fact provides no evidence that MCPs exacerbated rather than ameliorated the problem of prejudice and polarization in the Netherlands and which ignores much evidence to the contrary (Kymlicka, Reference Kymlicka2008).Footnote ¹⁵ But even if they are right about the effect of MCPs in the Dutch case, on what basis can we conclude that the Dutch case is generalizable rather than exceptional? There are, after all, some unique aspects of multiculturalism in the Netherlands, which initially emerged as a policy to encourage migrant workers to return home to their country of origin rather than as a policy to encourage immigrants to feel at home in their new country of residence. It is hardly surprising that a form of multiculturalism that initially arose within such a “returnist” framework has had deleterious consequences. If it is a mistake to generalize the benefits of multiculturalism from the Canadian case, surely it is equally a mistake to generalize the harms of multiculturalism from the Dutch case. What we need is look across the full range of cases and, as we've seen, the bulk of the evidence—not just from Canada, but also Australia and Sweden, and cross-national studies—suggests that multiculturalism remains a viable approach worthy of serious consideration. To dismiss multiculturalism based solely on the Dutch case is at best premature.

These two examples are indicative of a more general failing in the literature. We certainly need good social science evidence to help guide public policy on issues of ethnic diversity. But we need to approach this evidence carefully, with an open mind. Too often, social scientists have started from the premise that because the liberal multiculturalist hypothesis was championed by normative political theorists unschooled in the methods of social science, the hypothesis therefore must be naïve and wrong. But we should at least contemplate the possibility that the liberal multiculturalist hypothesis is valid, at least in some times and places, and then see where the evidence falls.