Predicting Currency Crises: A comparative analysis of static and dynamic models across time horizons

Details

Supervisors

Candelon, Bertrand

Faculty

Louvain School of Management

Degree label

Master [120] : ingénieur de gestion, à finalité spécialisée

Abstract

This thesis is a comparative analysis of static and dynamic models to predict currency crises, focusing on their effectiveness across different time horizons in 10 emerging countries. The main findings show that in the short term (1 to 3 months’ time horizon), the models can deliver a True Positive Rate of over 88%. However, as the time horizon extends, the effectiveness of the models reduces. Remarkably, in Argentina, the dynamic model continues to show high effectiveness with an 89% True Positive Rate at a 12-month horizon. This analysis also identifies an anomaly in predictive accuracy during the 2019 to 2021 period, which is attributed to the economic disturbances caused by the COVID-19 pandemic. It distorted the models' parameters and led to false predictions of crises. Overall, the dynamic model, which incorporates lagged crisis data, consistently outperforms the static model across most time horizons and countries.