Dating currency crises and designing early warning systems: Meta‐possibilistic fuzzy index functions

In order to analyse the currency crises in Turkey over the period of January 1990 and October 2019, we first dated currency crises with meta‐possibilistic fuzzy index functions. Then, we determined the significant predictors or leading indicators of currency crisis with logistic regression. Finally, we tried to measure and compare the in‐sample and out‐of‐sample performances of our method with an index generated by principal component analysis (PCA). We found that the models using the currency crisis index generated by our method have higher in‐sample and out‐of‐sample performances than the models using the currency crisis index generated by PCA. We concluded that the change in real exchange rate, bank loans over bank deposits, bank reserves over bank assets, growth in foreign reserves, growth in central bank foreign assets, M2 over foreign assets, change in exports, change in imports, industrial production index, M1 growth, M2 growth, foreign reserves over M1, central bank foreign assets over M1 and public sector credit growth are the leading indicators of currency crisis among 20 explanatory variables. As policy implications, we recommend that government and the monetary authority should strictly monitor the ratio of bank loans over bank deposits, public sector credit growth, the volatility of foreign trade, ratio of foreign reserves and central bank foreign assets over money supply among significant explanatory variables to avoid currency crisis as policy implications.