Early Warning Models For Anticipating Crisis: Insights From Sri Lanka

The study investigates how well early warning systems can predict currency crises in Sri Lanka, with special attention given to deviations in credit and business cycles as potential indicators. The research incorporates domestic and international factors, as well as credit and business cycle indicators, to anticipate currency crises. It uses multiple approaches such as exchange rate pressure index for crisis identification, signalling approach and logit regression for examining the predictive capacity of the indicators. Using quarterly data from 1997 to 2022, the study identifies seven crisis events in the country. Our empirical findings show that before the currency crises, credit cycle often deviates from business cycle. Moreover, the deviations of credit and GDP from their trends show considerable predictive strength, surpassing their respective level forms. Furthermore, global factors like Volatility Index and oil prices proved to be strong indicators, effectively signaling impending currency crises.