Understanding speculative attacks' aftermaths with computational economics

This research seeks to determine which factors are significant in explaining speculative attack's real effects when all the potential factors suggested by the literature are considered together. This dissertation studies speculative attacks' real effects with novel computational techniques designed to analyze non-linear problems with datasets characterized by few observations relative to the number of potential explanatory variables.; Three main results are obtained in this dissertation: First, this research computes a clustering map of speculative attacks based on the similarities of speculative attacks' aftermaths. The clustering map shows that not all speculative attacks' aftermaths are equal. Second, this dissertation concludes that speculative attacks' real effects are amplified through the financial intermediation sector. This research suggests that policies oriented to strengthen ailing banks reduce the likelihood of strong contractions in output after a speculative attack. Third, we conclude that economic fundamentals are not only important in explaining the occurrence of speculative attacks, but also their real effects.