Macroeconomic tail events with non-linear Bayesian VARs

Motivated by the desire to probe macroeconomic tail events and to capture non-linear economic dynamics, we estimate two types of regime switching models: threshold VAR and Markov switching VAR. For each of the models, we estimate regimes which carry the interpretation of recessionary/normal and financially stressful/stable periods. Using the recursiveness assumption and conditional on shocks of one standard deviation, we show that (i) financial shocks hitting during times of recessions create disproportionately more severe contractions in output; (ii) output growth shocks hitting in financially stressful times result in disproportionately further financial stress; (iii) monetary policy shocks hitting in recessionary times create more severe contractions in output. We also demonstrate the power of a feedback loop between real and financial sectors when extremely large shocks hit the economy in normal/financially stable periods. Afterwards, we perform out-of-sample forecasting exercises, and find that the threshold VAR model has the potential to predict tail events in conditional forecasting. Overall, our findings provide strong evidence of nonlinearities and shock amplification mechanisms in the UK data, as well as empirical support to the theoretical findings of Brunnermeier and Sannikov (2014).