Andrew J. Oswald (University of Warwick) has posted An Examination of the Reliability of Prestigious Scholarly Journals: Evidence and Implications for Decision-Makers on SSRN.  Here is the abstract:

In universities all over the world, hiring and promotion committees regularly hear the argument: this is important work because it is about to appear in prestigious journal X. Moreover, those who allocate levels of research funding, such as in the multi-billion pound Research Assessment Exercise in UK universities, often come under pressure to assess research quality in a mechanical way by using journal prestige ratings. The results in this paper suggest that such tendencies are dangerous. It uses total citations over a quarter of a century as the criterion. The paper finds that it is far better to publish the best article in an issue of a medium-quality journal like the Oxford Bulletin of Economics and Statistics than to publish the worst article (or often the worst 4 articles) in an issue of a top journal like the American Economic Review. Implications are discussed.