Using AI and Machine Learning to Predict Future Financial Crises

At CBC, we use machine-learning techniques to uncover three intuitive liquidity-based drivers of future financial crises – (1) cross-border capital flows; (2) investors’ risk appetite and (3) Central Bank policy. Adding these factors to a probit model warns of upcoming problems in 2018, largely because of reversing cross-border flows out of the US dollar. Reverse QE by Central Banks is unhelpful in this context, but it is unlikely by itself to trigger a crisis. Liquidity is important, but cross-border flows are the paramount factor to watch.