Bank of Korea launches warning model to improve foreign exchange crisis prediction ability

Superior to existing models by applying AI and machine learning

As instability in the won-dollar exchange rate continues this month, a model is being developed that can provide advance warning of signs of a financial or foreign exchange crisis.

The Bank of Korea announced on the 24th in its report on ‘Data-based financial and foreign exchange early warning model’, “We have recently developed a reliable early warning model by applying artificial intelligence (AI) and machine learning algorithms.” It is a model that can identify periods of crisis and market instability based on the Compound Financial Pressure Index (CFPI), which reflects banks, bonds, stocks, foreign exchange, etc., and then capture the crisis occurrence mechanism.

As a result of the Bank of Korea comparing several early warning models and ML algorithm application models from 1998 to 2023, the ML model had the best predictive power. In this model, the alertness index rose markedly several months ahead of the crisis. In particular, compared to the 2008 global financial crisis or the 2022 Legoland incident, the index rose sharply in a short period of time during the 2020 COVID-19 crisis.

Park Jeong-hee, head of the Bank of Korea’s Digital Innovation Office, explained, “AI and ML technologies, which have recently been used in various areas, are believed to be useful in developing early warning models with better predictive power compared to existing methods.”

< Copyright ⓒ Seoul Economic Daily, unauthorized reproduction and redistribution prohibited >