No. I would say "take causality seriously or give me death". There are cases where RCTs are impossible, or would be overkill. There are plausible natural experiments. There are good regression discontinuity designs. Heck, there are cases where the independent variable just is plausibly exogenous. Heck, there are cases when you care about prediction not causality (cross-country analyses can fit this sometimes). But here none of these apply. We do want causality. Correlation just isn't good enough. And there's not just some research - there's lots of research - that does this to investigate poverty. I mean, there's literally a book called Poor Economics by Nobel Prize winners who run RCTs.
