Corruption at centralized and decentralized levels

Yesterday I attended at IESE the presentation of the paper on centralization and government accountability by Giacomo Ponzetto from Universitat Pompeu Fabra. The article, co-authored by Piolatto and Boffa, presents a model where less informed regions import institutional quality from more informed regions in a centralized system. They test their main prediction with data about the Clean Air Act in the US, where regions with less newspaper circulation saw larger reductions in pollution than other regions. The literature surveyed by the paper is very interesting, and I touched on similar issues in the past working both on federalism and on capture or corruption. Although the main mechanisms identified by the article are plausible, I have doubts that the model can be used to make a generic statement that corruption is higher at decentralized levels, as the anecdotal evidence presented seems to indicate. There is old work by economists Bardhan and Mookerjee, also both theoretical and empirical, showing that it is not possible to say whether capture or corruption are higher or lower at different governing levels. It depends on the type of corruption, and on the details of the media and political markets. I sympathize with the idea that countries like Spain import institutional quality when they are integrated in a larger Europe, and it is true that New Deal politics and other centralizing reforms eroded practices like machine politics and patronage in US local politics. But centralization replaces some corrupt practices by others, like massive legal or  illegal party financing or revolving doors. If Europe, as is desirable, keeps making progress towards a true federal state, it will have to be careful not to create newly adapted forms of capture.