Identifying Non-Gaussian Structural Shocks

We develop a robust and tractable method to identify SVARs using non-Gaussian features of structural shocks. The approach combines inequality restrictions on higher-order moments with standard set-identifying constraints such as sign restrictions. To benchmark its performance, we apply it to the identification of monetary policy shocks. Combining standard minimal sign restrictions with a constraint that monetary policy shocks be leptokurtic, we recover several key properties documented in earlier work: a correct output response to policy tightening, salient policy narrative episodes, and a plausible central bank reaction function. We then identify sovereign and geopolitical risk shocks, assuming they are skewed and leptokurtic, and uncover sizable macroeconomic effects that remain hidden under conventional identification schemes.