One of the lessons of the treatment effects literature is the lack of consensus about the ability of statistical and econometric methods to replicate experimental estimates. In this paper, we provide new evidence using an unusual unemployment insurance experiment that allows the identification...