I just ran four million regressions

摘要: In this paper I try to move away from the Extreme Bounds method ofidentifying ``robust'' empirical relations in economic growth literature.Instead of analyzing extreme bounds estimates coefficientof a particular variable, analyze entire distribution. My claimin is that, if we do this, picture emerging theempirical literature not pessimistic ``Nothing Robust''that get with bound analysis. Instead, find that asubstantial number variables can be found strongly relatedto growth.