Assessing macroeconomic tail risk

Abstract Real gross domestic product and industrial production in the United States display substantial asymmetry and tail risk. Is this asymmetry driven by a specific structural shock? Our empirical approach, based on quantile regressions and local projections, suggests otherwise. We find that the tenth percentile of predictive growth distributions responds between three and six times more than the median to monetary policy shocks, financial shocks, uncertainty shocks, and oil price shocks, indicating a common transmission mechanism. We present two data-generating processes that are capable of matching this finding: a threshold vector autoregression model and a non-linear equilibrium model.