Monday, September 17, 2012
What's Wrong with Huntington's Clash of Civilizations?
A better question might be "What's not wrong with Huntington's Clash of Civilizations?", but I'll try not to be too pedantic. In light of recent events, interest in Huntington's thesis has been growing. Even the NYTimes is using language that is clearly influenced by Huntington's work, though they don't specifically mention it. So let's review the merits (such as they are) and demerits of this argument.
First, we need to review what Huntington actually said. Many of those who cite his work approvingly seem have precious little idea what it says, save that it speculates at length about a coming conflict between Islam and the West. And, sure, there is some intuitive appeal to such a gross misreading of his work. But I encourage you to read the Foreign Affairs article, or the book that followed, where you'll find some reasonable claims sharing space with concrete predictions about what was then the future—predictions that have proven to be completely wrong. This shouldn't really surprise us, since the entire argument rests upon a hugely problematic assumption, as I'll discuss below.\(^1\)
Huntington's argument was that cultural factors were suppressed by the super powers during the Cold War, but would come to dominate the post-Cold War era. Statistical critiques of his work (see here and here for example) have often been met with the criticism that they only demonstrate what Huntington had already acknowledged—that civilizational conflict was not common in the past (though see here). We're more than 20 years from the end of the Cold War now, and there's still no evidence for the precise claims he made, but it's worth acknowledging from the start that Huntington's argument was always a little more nuanced than it has sometimes been made out to be.
Huntington argued that cultural factors would have a very specific effect—that interstate conflict would be most likely to occur at or near civilizational fault lines. He did not simply say that identity would matter. He identified several essentially monolithic civilizations, and claimed that conflict would be most likely to occur in the future between governments of states that belonged to different civilizations, particularly those that shared a geographic border. These civilizations are: Western, Confucian, Japanese, Islamic, Hindu, Slavic-Orthodox, Latin American, "and possibly African". You'll note that some of these are defined by geography or political-legal boundaries, while others are delineated by religion.
Why are some religions civilizations while others are not? Who knows. Why are some states coterminal with cultural civilizations while others are not? Who knows. Why is there doubt about whether Africa has any civilization? Who knows. Surely there is no racist subtext there!
Note also that Huntington was not primarily talking about the possibility of religious or ethno-linguistic cleavages fueling civil wars, nor transnational terrorism. He was mostly interested in explaining interstate conflict, which he clearly said would be more likely to occur between certain types of countries in certain regions of the globe. He did discuss terrorism, and his claims there have been partially supported (see here), but this was secondary to his main argument. And insofar as he had anything to say about terrorism, his argument anticipates more incidents to cross civilizational boundaries than to occur within any given civilization—and he was dead wrong about that (see the paper in the previous link).
The big problem I, and others, have with the CoC is that conflict has historically been, and continues to be, more common within the boundaries of what he identifies as civilizations than it has been between states of different civilizations. Put differently, the key expectation of his argument is completely at odds with the available evidence. It doesn't just oversimplify—I'd argue that all theories, by necessity, do that to some degree—it fails on its own terms. Conflict, by any measure yet devised by social scientists, is not now, nor has it ever been, more likely to occur along "civilizational fault lines".\(^2\) It is highly unlikely that shared civilization status is itself a cause of conflict, of course, "civilizational" identities appear to be less important, even in the post-Cold War era, than Huntington anticipated.
But there are other parts of Huntington's argument that are pretty reasonable. Let me acknowledge those before discussing the truly fatal flaw in his thesis.
Huntington argued that the clash of civilizations would be fueled by two factors. Neither of these was new, but the Cold War had kept a lid on the pressures created by them, according to Huntington. The first factor is globalization. The decreased cost of communicating and/or traveling across the globe puts people into contact with members of other civilizations more frequently in the post-Cold War era than was true for the decades preceding it. Among other things, this brought American popular culture to people who had previously had little exposure with it. This, in combination with the second factor, creates serious tension. Namely, at the same time that globalization was accelerating, elites around the world made a concerted effort to resist creeping Westernization and modernization, because it threatened their hold on power. Moreover, economic integration was expected to raise awareness of civilizational identities since it often occurred within the context of regional organizations, rather than on a strictly global scale. On these points, I think Huntington sounds pretty reasonable. I'm not an expert on any of the related topics, so I'll leave to others to judge, but I'll at least say that this part of his argument passes the laugh test. When folks like me joke that Huntington's CoC has detracted from the sum total of human knowledge, that it shouldn't even be taught in Intro IR classes because exposing students to it does more harm than good, it's the part I'm about to turn to that we have in mind more so than his claim that economic globalization has shrunk the world, or that economic integration has proceeded more rapidly at the regional level than the global level, or that dictators in the Islamic world might be threatened by the degree to which young people in their countries embraced American pop culture.
No, the big problem here is that Huntington claims that identity is immutable (even though some aspects of his argument clearly imply the opposite). The fundamental problem with any argument linking cultural cleavages with conflict is that identity is malleable, and we have just as much evidence that conflict -> identity cleavages as we do identity cleavages -> conflict.
The notion that identity is malleable is so well-documented that, in certain circles, I wouldn't even need to back that claim up at all. But anyone who has read this deep into a post about Huntington's CoC obviously doesn't think that it goes without saying. So allow me to present a brief case, based on a very limited sample of evidence. Bear in mind, though, that I'm hardly even scraping the surface here.
Stathis Kalyvas has a nice paper on the frequency with which actors choose which side to take in a conflict based on political expedience rather than ethnic solidarity. This doesn't speak to the notion of identity change being caused by conflict, since the behavior he discusses isn't driven by identity at all, but it further casts doubt on the claim that identity is the primary driver of behavior.
In a recent paper, Doug Gibler, Marc Huthison, and Steve Miller show that individuals living in countries that were recently the target of an interstate dispute over territory are more likely to identify along national lines than ethnic lines, while those who live in countries that recently experienced a civil conflict are more likely to identify along ethnic lines than national lines. That is, how people view themselves is influenced by who they consider the "other" to be.
Along similar lines, Maggie Penn has developed a theoretical model of identity choice that focuses specifically on the question of whether people self-identify first and foremost along ethnic lines or by nationality. Those who have little interest in formal theory should at least find her discussion of recent Iraqi politics to be of interest. Hint: the Sunni and the Shi'ia didn't see themselves as so different until after the sectarian conflict that followed Saddam's fall.
Perhaps the clearest evidence that identity choices can be manipulated by political elites comes from a study by Daniel Posner. While this study doesn't look specifically at the impact of armed conflict on identity, it is hard to look at this evidence and cling to the belief that identity is immutable.
Posner takes up the question of why Chewas are Tumbukas are allies in Zambia and adversaries in Malawi. These two ethnic groups straddle a border that was drawn by the British for purely administrative purposes. It just so happens, though, that in Zambia, neither group is large enough, relative to the rest of the population, for a political party that only represented Chewas or only represented Tumbukas to be viable. In Malawi, we have a different story. Both groups are large enough to contest the presidency on their own. Unsurprisingly, then, there are no political parties in Zambia who make appeals to Chewas in which they portray Tumbukas as "other", while there is such a party in Malawi. And, equally unsurprisingly, we see that Chewas and Tumbukas not only support different parties in Malawi while supporting the same party in Zambia, but generally dislike and distrust each other more in Malawi than Zambia. The following graph illustrates this (see the paper for details).
These are but a few examples of the many studies that find that identity is indeed highly responsive to political context. There is an equally rich body of literature demonstrating that identity is responsive to the social and economic context within which people find themselves.
Given the malleable nature of identity, if we observe a correlation between identity and conflict (and remember—there is no correlational evidence of the pattern anticipated by Huntington), we must ask ourselves whether cultural cleavages caused conflict or whether conflict exacerbated cultural cleavages. Insofar as there is conflict between some actors within the Islamic world and some actors in the Western world, we should remember to ask whether the fact that the people on one side of that conflict just so happen to be Muslim and the people on the other side largely are not is the actual cause of that conflict. There are good reasons to believe that the various and sundry political cleavages dividing these actors are no less important than the cultural ones.
1. Assumptions are not, in and of themselves, impermissible. Too often, perfectly reasonable theoretical arguments are criticized for resting upon assumptions that are clearly false. The question we should ask when evaluating the assumptions of a theoretical argument isn't whether they are true or false but whether their truth status is fundamental to the argument. If the same conclusion could be established, albeit with a lot more effort, after relaxing a given assumption, then we shouldn't care whether it is true or not. If the conclusion critically depends upon a patently false assumption, that's a different story. And that's what we're talking about here.
2. By "ever", I obviously mean "during any period for which we have systematic data available", but that doesn't sound as good, so forgive the poetic license. And it should also go without saying that I'm actually only saying that we do not yet have statistical evidence consistent with this expectation. Perhaps there are measurement issues or something that have yet to be resolved. I doubt that's the problem here, but obviously I don't know for sure, since none of us knows what the true data-generating process is.