Good questions ... The methodology looks thorough. I'm not an econemetrician, so can't say exactly that it's correct. But it's like climate science - you go with the repeated, peer-reviewed stuff - and we now have two major reports, OECD and IMF, arriving at the same point using different methods/data sets, so it looks rigorous.

It's a good question, Ben, and the quickest answer in this case is that the OECD have done the timing so that it has to be that inequality leads to reduced growth, not the other way around. Their data takes rising inequality 1985-2005, then looks at what happened with growth 1990-2010 (i.e. allowing a five-year lag for the effects of rising inequality to be felt).