Early warning of financial stress events: A credit-regime-switching approach

We propose an early warning model for predicting the likelihood of a financial stress event for a given future time, and examine whether credit plays an important role in the model as a non-linear propagator of shocks. This propagation takes the form of a threshold regression in which a regime change occurs if credit conditions cross a critical threshold. The in-sample and out-of-sample forecasting performances are encouraging. In particular, the out-of-sample forecasting results suggest that the model based on the credit-regime-switching approach outperforms the benchmark models based on a linear regression and signal extraction approach across all forecasting horizons and all criteria considered.