Designing and implementing policies that reduce the chances of a crisis occurring – as well as its costs – requires the build-up of risk in the economy to be recognized at an early stage. This paper is an attempt to restore some imagination to the economics of crises.
Researchers from the Bank for International Settlements adapt the framework of sequential prediction or online machine learning to forecast systemic financial crises without knowing the “true” model of the economy. The approach can be described as “meta-statistic” since the aim is to make the best possible prediction by aggregating predictions from different models. These models are estimated using the standard macroeconomic variables used in past studies.
Despite its generality and its flexibility, online learning has some limitations in that it will be unable to predict any crisis of a type that has never happened in history. For example, it will not be able to predict a hypothetical financial crisis caused by a cyber attack as we never observed one so far, or a financial crisis caused by a pandemic shock unless it correlates with characteristics of past crises.
Researchers uncover a time-varying subset of models that carry most of the information needed to predict financial crises. Among those models, they also discuss which ones “flash red” at the right time. Using a mix of 26 models for France, Germany, Italy and the United Kingdom – including central bank financial crises models as well as machine learning models – they are able to predict a systemic financial crisis three years ahead out-of-sample, with lower signal-to-noise ratios than in the existing literature.