Perhaps one of the most surprising and influential findings in development economics research is the so-called “resource curse”: the idea that a large natural resource endowment (and, consequently, a significant role for natural resource exports in the national economy) actually leads to slower economic growth, and lower per capita incomes (at least in the long term). The resource thesis has the appealing feature that although it’s initially counter-intuitive (and so people like me can seem and feel clever when we point it out), one can immediately think of many salient examples that seem to corroborate the idea, and it’s fairly easy to construct plausible stories as to why it would be true. Although such stories originally focused on exchange rate appreciation (so-called “Dutch Disease”), contemporary research (see, e.g., here and here) tends to focus more on the impact of natural resource abundance on institutional quality, governance, and corruption. The hypothesized causal chain (at least one version) runs roughly as follows: Natural resource wealth creates opportunities for massive economic rents for those who control the government; the competition for these resources fosters corruption, and makes currying favor with the government more important than entrepreneurship or productive investment. Furthermore, and perhaps even more importantly, natural resource wealth enables corrupt or otherwise inefficient governments can use their control over resource rents to secure their power, alleviating pressure that these governments might otherwise feel to reform their institutions and govern more fairly and effectively. And indeed, many studies (see here and here) show a strong negative correlation between natural resource wealth (especially oil wealth) and various measures of institutional quality (including accountability, checks & balances, and control of corruption). The bad institutional environment that natural resource wealth fosters, the argument continues, has adverse effects on long-run economic performance that outweigh the boost to economic performance associated with natural resource wealth. This, the causal chain runs from resource wealth to bad institutions to poor(er) economic performance; absence of resource wealth tends to generate incentives for institutional improvements that ultimately lead to better performance.
The resource curse thesis grows mainly out of quantitative cross country research that finds a negative correlation between resource wealth and GDP growth (controlling for a range of factors). Some more recent research has refined or qualified the thesis in important ways. For example, (see here and here) suggests that the “curse” is only associated with particular sorts of resources, particularly “point source” resources (such as oil or certain minerals). Other research (see here and here) has suggested that countries that already have relatively good institutions prior to the discovery of resource wealth seem immune from the curse. Still, even with these qualifications, the core idea remains: If a relatively poor country, with less robust governance institutions, discovers oil, its economic prospects over the longer term are actually worse—largely because of the relationship between resource wealth and corruption.
But what if that’s all wrong? What if there is no “resource curse”? What if resource wealth—even from point source resources, even in countries with lower levels of transparency and accountability—is, on average, associated with higher rather than lower economic growth? And what if natural resource wealth actually has no consistent discernable impact on institutional quality? For many years I’d been entirely convinced of the resource curse thesis (at least in qualified form). But I recently read an excellent 2009 paper by the economists Michael Alexeev and Robert Conrad which has forced me to reconsider. I’m still not sure exactly what I think, and I hope to spend the next few months delving more into this research (so I may eventually do a follow-up post), but I thought it would be worth discussing the essence of Alexeev & Conrad’s critique and reassessment of the resource curse thesis.
I won’t go into the technical details here. Instead, I’ll focus on what I took to be Alexeev & Conrad’s most important points.
- First, if what we really care about is the overall impact of the discovery of natural resource wealth, it doesn’t make sense to look only at relatively recent economic growth rates. After all, one could imagine that the discovery of natural resources could, in the early years, cause very rapid economic growth that then slows down later on, even if overall the discovery of natural resources caused relatively higher average growth over the whole period (compared to countries without resource wealth). Unfortunately, reliable annual economic growth data is available only from about 1970, even though many of the major oil-exporting countries discovered oil decades before then. Using older GDP data to calculate longer-term growth rates, Alexeev & Conrad claim that in fact oil wealth is associated with larger increases in national income.
- Second, and perhaps even more relevant to the relationship between oil wealth and corruption, Alexeev & Conrad point out that virtually all the statistical studies purporting to show that oil wealth is negatively correlated with various measures of institutional quality (including absence of corruption) control for GDP. At first blush, that seems entirely sensible. Given everything we know (or at least strongly suspect) about the impact of GDP on institutional quality, it would seem crazy not to control for GDP when trying to assess the impact of some other factor. But here’s the problem: If oil wealth increases GDP (as Alexeev & Conrad claim to have found, and which is at least plausible even if one is skeptical of their assessment), then a correlation between oil wealth and corruption, when controlling for GDP, does not necessarily indicate that oil wealth increases corruption; it could mean instead that oil wealth increases GDP without decreasing corruption, even though GDP improvements in countries that don’t have a lot of oil wealth are associated with lower corruption. Think about it this way: Suppose that lower levels of corruption increase GDP, and suppose further that oil wealth also increases GDP, but without affecting corruption. (That is, suppose that GDP increases, at least those associated with oil wealth, don’t reduce corruption, and also that, contrary to the popular version of the resource curse thesis, oil wealth does not increase corruption.) Under these conditions, if we compare countries with similar GDPs, some of which have oil wealth and some of which don’t, the oil-rich countries will have higher corruption on average. But (by assumption), this is not because the oil wealth increased corruption, but rather because the oil wealth increased GDP. Testing which of these alternatives is more accurate is challenging; Alexeev & Conrad come up with a clever technique that uses a set of geographic and demographic factors to predict what GDP would have been in the absence of oil. I won’t go into that here, partly because I’m not (yet) competent to evaluate it. But their larger point seems correct to me. If, in a large sample of countries, GDP is negatively correlated with corruption, and oil wealth is positively correlated with corruption, we can’t automatically assume that oil wealth causes higher corruption, because it could instead be the case that oil wealth increases GDP.
It should go without saying that even if Alexeev & Conrad are right, this does not imply that oil wealth doesn’t create distinctive corruption challenges that policymakers and activists should address. Even if oil wealth increases per capita GDP (without increasing corruption), the corruption that does exist in oil-rich countries may mean that the benefits of this natural resource boom are not equitably shared. Nevertheless, the resource curse thesis, and the causal link (or even the positive correlation) between oil wealth and corruption, is not nearly so airtight as I’d thought. I suspect some readers will shake their heads at my ignorance. After all, the Alexeev & Conrad paper is six years old, and my own cursory additional research has revealed (A) they’re not the only critics, and (B) there’s been quite a bit of subsequent research on the resource curse. So perhaps some readers out there will let me know that their critique has either already been accepted, or thoroughly debunked. But my impression is that many people writing on this topic are still very much where I was a few weeks ago – confident in the empirical accuracy of the resource curse thesis, and largely unaware of the extent and nature of the criticisms. I’m hoping this post will stir up a bit more discussion of the issue, and help all of us clarify our thinking.