Methodological sophistication in development economics

Following up on Chris Blattman's post, here's a question for my fellow development economists out there: of the important things we know about development in the real world, are there any of them that required more than relatively simple econometrics (along with good data and possibly a clever identification strategy) to discover? In other words, is there anything important that we wouldn't know without sophisticated econometric methods (say, beyond the level of instrumental variables)?