Causality between banking and currency fragilities: A dynamic panel model

A panel dynamic model both with and without a threshold is specified to reexamine the lead-lag relationship between banking and currency fragilities. We employ banking sector fragility (BSF) and exchange market pressure (EMP) as the proxies for banking and currency fragilities, respectively, where BSF is made up of real deposits, banks' real claims on the domestic private sector and the real foreign liabilities of banks, and EMP uses the weighted average of the exchange rate changes and foreign reserves. Among the banking sector fragilities, we consider three different proxies, namely, BSF1, BSF2 and BSF3, depending on the components used. Our 51 sample countries include 21 industrial and 30 developing countries. When the whole panel dynamic model is used, bilateral causality is found between the two fragilities using all sample countries. When using only industrial country data, the bilateral causation is found only between EMP and BSF2, but no relationship is found between BSF3 and EMP, or between BSF2⁎ and EMP. When developing countries are employed, stronger bilateral causality is found between banking and currency fragilities. When the panel threshold dynamic model is used, the results overwhelmingly suggest that bilateral causality exists.