Abstract:
Borrowing ideas from the medical sciences, we propose tentative guidelines for reliable causal inferences that cover aspects related to both the study itself and its fit with background knowledge. We argue that the current paradigm in economics tends to put too much emphasis on internal aspects related solely to the study itself. To substantiate this view, we discuss various excellent studies from different fields of economics, which all express causal and policy-relevant claims. We conclude that policy implications based on single studies are inherently uncertain, even when the respective studies are state-of-the-art.