Some of the research in this area is misleading, I'd even maybe say poorly conducted, and I worry it will be taken too much at face value by policy-makers and administrators.

The conclusion that institutition doesn't matter, for example, seems ridiculous and flies in the face of a lot of IO research. People move for reasons, for example; the question isn't whether productivity changes after a move on average, it's what would have happened to an individual researcher if they had stayed in a given situation relative to what would have happened after a move to another situation. There's a lot of cliques, the moves aren't random, etc.

It's a bit like concluding that you can determine the importance of a paper solely by looking at the impact factor of a journal.