Dating currency crises in emerging market economies

The first crucial step in studying currency crises empirically is to identify these crises and to date them as precisely as possible. We construct a database of currency crisis dates for 35 emerging economies for the period 1990-2016, using data at a monthly frequency to improve accuracy. We use multiple quantitative definitions, past studies, narrative and expert opinions to date the crises. We then compare the performance of individual crisis definitions versus combinations of definitions and find that the latter perform better. Lastly, we find that several definitions perform worse in floating and free falling exchange rate regimes than in fixed and intermediate regimes.