I was talking to my girlfriend last night, in my usual service as assistant for the “International Debate” class she’s currently taking and the subject of kleptocracy came up, I believe in the context of a larger discussion about the IMF and World Bank. I don’t rightly recall, actually, but it was something like that.
In any case, I managed to sort of stumble upon this working paper by Oguzhan C. Dincer, Christopher J. Ellis and Glen R. Waddell. Full disclosure, I checked the UO Economics website because Professor Ellis once mentioned to me he was doing some work on kleptocracy and I started wondering what ever became of it. The working paper linked above is on the negative relationship between corruption and decentralization of the power to tax and spend. That is, as power is decentralized there appears to be less corruption. From the introduction:
Recent works by several contributors point to a significant negative relationship between the degree to which the powers to tax and spend are decentralized in an economy, and the overall level of governmental corruption (e.g., Treisman, 2000; de Mello, 2000; de Mello and M. Barenstein, 2001; Fisman and Gatti, 2002a; Arikan, 2004).1 On the surface this is quite surprising, as it might be anticipated that local politicians or bureaucrats would possess detailed knowledge of any opportunities for corruption that might arise in their jurisdictions. They might thus be expected to extract any corruption rents available more efficiently than less well informed national counterparts.
In what follows, we both propose a theoretical explanation for, and find further evidence in support of, a negative correlation between empirical measures of corruption and decentralization. We argue that because there are more independent taxation and expenditure decisions made in a decentralized economy, there are more opportunities for local populations to make cross-jurisdictional comparisons of politician or bureaucrat performance. The poor performance of a government in one jurisdiction might be attributed to a number of factors including corruption. An increase in the number of comparative observations made by the populations of jurisdictions has two effects relevant to the relationship between corruption and decentralization. First, as the number of comparative observations made by a jurisdiction’s population increases the inferences they make about the causes of a particular observed outcome become increasingly precise. Second, as the number of observations increase the likelihood that particular inferences will be arrived at also change. As we shall subsequently see, this latter effect proves to be crucial in generating a negative theoretical relationship between corruption and decentralization.
The rest of the paper goes on to outline the theoretical and empirical case for lower levels of corruption as power becomes more decentralized. This is relevant to libertarians in that it lends verifiable data to our claim that a minimal state with many decentralized branches of power will likely result in the best outcomes. Personally, I also don’t find the conclusion particularly shocking.
I’m sure this has been address elsewhere, but I don’t exactly have the time this evening to educate myself on the last 10 years of kleptocracy research, but I would suspect that centralized authorities would be a bit easier to bribe because you can get more for your marginal corruption dollar. If you have a fixed amount to spend on bribe money, and a whole host of petty bureaucrats to bribe you’ll, obviously, be able to spend less per bureaucrat and thusly have a harder time achieving your ends than you would if you could spend the entire graft budget on the one fellow in charge.
