With the (I guess still impending) fall of Qadaffi and the future of Libya moving front and center on to the radar screens of many, there will inevitably by many discussions in the coming months of the “resource curse”:http://en.wikipedia.org/wiki/Resource_curse. With that in mind, we are pleased to welcome the following guest post from “Samuel Greene”:http://www.nes.ru/en/people/catalog/g/sgreene of the “New Economic School”:http://www.nes.ru/en/about/ in Moscow, Russia.
For reasons that continue to escape me, but that apparently have more to do with nutrition than taste, shredded seaweed salad is incredibly popular here in Russia. Aware of this, a former colleague of my wife would force himself every few months over the course of several years to try the stuff, only to reconfirm that, yes, it’s still disgusting.
That’s a roundabout way of saying that, six months months later, I still haven’t learned to like the “APSR piece”:http://journals.cambridge.org/abstract_S0003055410000584 by Stephen Haber and Victor Menaldo on the resource curse. Really, I’ve tried.
In a nutshell, the article argues that the conventional wisdom (if there is any) on the so-called ‘resource curse’ – which holds that significant endowments of natural resources skew both economic and, more importantly (to the readers of this blog, at least), political development towards the ineffective and authoritarian, predominantly through a rentier effect – is wrong. And they prove that it’s wrong with numbers.
Other people in other forums have already pointed out the various ways in which their analysis doesn’t make much sense, so I won’t dwell on that here (if you’re interested, click “here”:http://www.freakonomics.com/2011/04/04/is-the-natural-resource-curse-not-quite-true/, “here”:http://blogs.cfr.org/levi/2011/07/20/is-there-a-resource-curse/ and “here”:http://www.plawlotic.com/?p=8). The key bone of contention is that a sample of resource-rich countries that includes Belarus, Ukraine and all three of the Baltic States is not exactly the one that serious resource-curse theorists had in mind and is, to say the least, biased towards not finding any significant effects.
Equally problematically, they set this difficult analysis against those whom they accuse of “making sweeping, law-like statements” about the effects of the resource curse. Who is making sweeping, law-like statements? Not Michael Ross, or any of the other theorists I’ve read. No, the citation for that accusation is Tom Friedman’s “The First Law of Petropolitics”. A worthy opponent, to be sure, but not one usually taken on in the pages of APSR.
It’s also worth noting that this has been done before, and better. Two years ago, Michael Alexeev and Robert Conrad published “The Elusive Curse of Oil “:http://www.mitpressjournals.org/doi/abs/10.1162/rest.91.3.586, demonstrating (using more sensible models and data sets) that the effect of the resource curse was in fact not systematic and making the logical argument that if you took oil away from, say, Saudi Arabia, you’d basically end up with Yemen (or substitute Russia and Ukraine, respectively, if that’s more to your liking).
But both of these articles suffer from a more fundamental misconception, which is really what this post was meant to be about: just because you can’t quantify an effect doesn’t mean it doesn’t exist. Put slightly more academically, quantitative tests of significance are not always the best way to evaluate every causal argument.
Regression analysis in its various permutations – and, indeed, almost any analysis based on statistics – requires a broadly linear conception of causality, in which starting points are well defined and intervening variables can be clearly identified and reasonably controlled. This, combined with the need to collect a sufficiently large number of sufficiently similar observations to achieve statistical significance, leads to an oversimplification of reality. Even Alexeev and Conrad’s more circumspect article misses this point, when they try to use Belarus, Russia and Ukraine as a “natural experiment”, claiming that the three emerged from the Soviet period virtually identical, differing significantly primarily in terms of their natural resource endowments (never mind that Haber and Menaldo consider all three to be resource rich). Alexeev and Conrad write that “the comparison of Belarus, Russia, and Ukraine is instructive because of its stark divergence from the prediction based on the natural resource curse”, i.e., from the prediction that Russia’s natural resources should have hampered Russian institutional development versus that of Belarus and Ukraine but, according to World Bank and other data, apparently did not. The problem is, despite all their cultural, linguistic, structural and historical commonalities, the three countries went through remarkably different processes of post-Soviet nation- and state-building, inherited very different ethnic and confessional cleavages, systems of center-periphery relations, administrative and coercive apparatuses and so on and so forth. Not many political scientists would argue that any of these factors should be subordinated in an encompassing formal model to resource endowments (or ignored altogether), yet that is precisely the test that statistical analysis requires.
Quantitative analysis also requires a capacity for generalization that is not necessarily inherent in every valid political science question. Tests of significance require large numbers of observations, and the entire exercise of creating statistical models of reality is predicated on the idea that we can extend the relevance of our models to at least an important chunk of the real world. But the theory behind the resource curse – similarly, I would argue, to the bodies of theory that describe social movements, collective action, and political party system development, to name a few – is not of that nature. Such theories seek to understand phenomena that occur sufficiently frequently (and/or with sufficiently striking consequences) to be of interest but that may not occur in every case, and where the actual trigger may never clearly be defined. Indeed, it is quantitative modeling, not ‘thick’ qualitative analysis, that most frequently engages in making “sweeping, law-like statements”. The two approaches can interact fruitfully, suggesting new avenues of inquiry and probing the validity of earlier wisdom, but the attempts described here to blow the resource curse argument out of the water, as it were, are less helpful.
Returning to the actual resource curse argument – which describes a mechanism by which resource rents can (but need not always) distort political developments – the messiness of reality is not a problem. Ross and others have looked carefully at cases in which resources do seem to matter; “a recent book”:http://www.amazon.com/Oil-Not-Curse-Institutions-Comparative/dp/0521148081/ref=sr_1_1?ie=UTF8&qid=1313567434&sr=8-1 by Pauline Jones Luong and Erika Weinthal looks similarly carefully at cases that suggest that political structure, rather than oil, is key, and that oil can actually be a powerful tool for modernization. It is a contentious and fascinating debate and will remain so for quite some time. But no progress is made when we apply the wrong methodology.