As mentioned earlier, most of the literature has focused on currency crises. Developing techniques to assess the vulnerability of fixed exchange rates and
predicting their collapse has been a research program
at least since the 1980s. But currency crises are difficult
to predict. It might be that the wrong variables or models
are being chosen or that the measurement techniques
adopted are not up to the task. In ‘A Perspective on Predicting Currency Crises,’ Flood, Marion, and Yepez survey the empirical literature on predicting this particular
type of crisis. It analyzes the three most important
branches developed in this area: structural models,
discrete-variable techniques using panel data, and signaling methods. The authors find that all methods perform well when fundamentals move abruptly, but all
do badly when fundamentals are merely highly volatile.