Because I have posted a lot on the general topic, I noticed that Megan McArdle’s guest blogger Tristan Reed has a post up today that references an academic study purporting to show, analytically, that governmental decentralization tends not to work very well. Of course this study comes across as a not-very-subtle parody of econometrics. The author has conducted a regression analysis to evaluate the impact of decentralization on the quality of government. His unit of analysis is the nation, so he has a total of 166 data points to model all of the various drivers across planet earth that might explain good vs. bad government. Needless to say, this is not entirely persuasive.