Growth Fragility and Systemic Risk Under Model Uncertainty

Abstract

The link between systemic risk and economic growth is hard to study because the relationship is believed to be nonlinear and systemic risk is unobservable. The myriad of measures proposed in the literature add model uncertainty as an additional difficulty. I use a Bayesian quantile regression to study the relevance of 33 systemic risk indicators to explain lower quantiles of output growth. Model uncertainty is tackled with sparse and dense modelling techniques that perform both model selection and shrinkage. I find that systemic risk indicators add value to quantile forecasts of GDP growth, in-sample and out-of-sample. However, less than halve of the indicators considered are selected to explain the different quantiles of economic growth.

Keywords: Quantile Regression, Systemic Risk, Growth Fragility, Forecasting

JEL Classification: G10, E37, C11, C32, C58