In my post last week, I went after a bunch of recent blog commentary that asserted there wasn’t much evidence that corruption mattered for development (and that also asserted the allegedly disproportionate focus on corruption was a “fetish” of self-satisfied Western countries). My emphasis in that post was on presenting the evidence that corruption is indeed a serious problem, one that citizens in developing countries care about. But I didn’t say much directly in response to the evidence that the targets of my screed (Christopher Blattman, Michael Dowdle, Jason Hickel) had offered for the claim that corruption’s not all it’s cracked up to be as a development problem. Of the posts I went after, Blattman’s made the most effort to ground the corruption’s-not-so-important claim in academic research. So I think his most detailed post on the subject deserves closer scrutiny than I gave it in my original polemic. (By the way, Blattman also included an interesting direct response to my post here; my reply is in his comments section.)
After reviewing Blattman’s post and the research he cites (and links to) in support of his argument that there’s little evidence that corruption matters very much for economic growth, my conclusions are largely unchanged. Indeed, I think that the academic papers on which Blattman relies tend to undermine his point more than they support it.
First, Blattman notes that a 2010 survey paper by Nauro Campos, Ralitza Dimova, and Ahmad Saleh “reviewed 41 different cross-country studies” and found that two-thirds of them “don’t even find a negative correlation” between corruption and growth. This point was echoed in a more recent post by Charles Kenny, who cites the Campos et al. study as evidence that there is only a “weak link” between corruption and growth over time.
Blattman and Kenny’s description of the Campos et al. study is partly accurate, but highly misleading. As Campos et al. note in their introductory section:
“[A]bout 32 percent of [the] estimates [reviewed] support a significant and negative impact of corruption on growth, 62 percent suggest a statistically insignificant relationship, while approximately only 6 percent support a positive and significant relation.”
But Campos et al. go on to explain that the whole purpose of their paper is to “use meta-analysis and meta-regression techniques to establish the depth, extent and the reasons behind this inconclusiveness.” And the main substantive result of their meta-analysis — clearly stated in the abstract of the paper as well as the introduction — is that “there is plenty of evidence supporting a genuine negative effect of corruption on growth.” Neither Blattman nor Kenny bothers to mention that seemingly important detail.
Blattman then notes that “cross-country studies” like those reviewed in Campos et al. “have mostly bad data and bad empirics,” so we should not rely on them; instead we should look at “broader evidence.” I agree wholeheartedly. In terms of what that broader evidence has to say, Blattman cites a terrific (though somewhat outdated) 2005 review paper by Jakob Svensson. According to Blattman, Svensson’s review of the “broader evidence” leads him to “draw the … conclusion that there’s not much to show that corruption reduces growth on net.”
Not true — or at best half true. Svensson does note that the cross-country evidence is “inconclusive” on the corruption-growth link. But what about the “broader evidence” for which Blattman cites the paper in the first place? Here Svensson states quite clearly that “the case study and micro evidence suggest that corruption severely retards development,” and he speculates that the failure of macro-level studies to confirm that result is likely due to “econometric problems” with those studies and/or to the coarseness of the cross-country corruption data (which fails to distinguish different types of corruption). So Blattman’s citation of Svensson for “broader evidence” that shows corruption doesn’t matter much for growth is seriously misleading: In fact, Svensson does almost exactly the opposite: he, like Blattman, reads the macro-studies as inconclusive, but finds that the “broader evidence” in fact refutes rather than confirms Blattman’s position.
Finally, to support the claim that corruption may not be that big of a problem because “corruption can also grease the wheels of prosperity”, Blattman cites a 1997 review paper by Pranap Bardhan. Bardhan does indeed discuss the economic arguments as to why corruption might (sometimes) be efficient (or more efficient than the alternatives). But he also discusses the economic arguments for why corruption might be inefficient. Indeed, Bardhan severely critiques the “efficient corruption” arguments and the models on which they are based; Bardhan describes these arguments as “fraught with general problems, even though in individual instances some redeeming features of corruption may be present.” That’s hardly a ringing endorsement of the “efficient grease” hypothesis. And in any event, Bardhan’s discussion is purely theoretical, not empirical, and the Campos et al. review noted above found very little evidence to support the claim that corruption is generally efficiency-enhancing.
So, two lessons to draw from this, one specific and the other more general:
Lesson 1: I can stand by my original claim that Blattman has not offered any serious evidence against the claim that corruption has significant adverse effects on development.
Lesson 2: It’s always a good idea to go back and check the original sources cited in support of propositions thrown around in the blogosphere (or elsewhere). And I certainly wouldn’t exempt myself from this scrutiny. In my earlier post critiquing Blattman and others, for instance, I linked to a whole bunch of academic studies as support for my claims. I think my citations were all accurate, but perhaps not — or perhaps my presentation was in some way misleading or incomplete. I hope interested readers will check the sources and call me out if I’ve misunderstood or misrepresented any of this material. I’d welcome the corrections. After all, the goal here is not to win a debate. It’s to figure out what to do about a very serious set of problems.