Recurrent Devaluation and Speculative Attacks on the Mexican Peso

We generate an empirical method aimed at predicting the timing and magnitude of devaluations forced by speculative attacks on fixed exchange rate systems. Using the Mexican experience as an example, we produce time-series estimates of the one-period-ahead probability of devaluation, the expected value of the new fixed exchange rate, and the confidence interval of the forecasted exchange rate. The results of the empirical exercise are encouraging. Devaluations, both in and out of sample, did occur when "predicted" by the model. Furthermore, the probabilities of devaluation reached relatively high values prior to actual devaluations.