Currency crises and the rationality of speculative attacks: An application of Markov-switching regime models

The main objective of this Doctoral Thesis is the analysis of the speculative processes that the Spanish Peseta suffered during the time it belonged to the EMS. Some of the questions we try to answer are: Were the Macroeconomic imbalances in Spain so important to trigger these turbulences? Which kind of variables could be behind a wave of turmoil?; Firstly, we describe the features of International, European and domestic framework and the behaviour of the Spanish Peseta/Deutsche Mark during the time the Spanish currency belonged to the European Exchange Rate Mechanism.; The "Currency Crises" literature focuses on two basic kind of models in order to explain turbulences. The "first generation models" stress the relevance of monetary variables on the time and magnitude of the attacks, whereas the "second generation models" considerer a broad range of variables including real variables.; The Markov Switching Regime Model let us to identify in advance three different periods of speculative pressures suffered by the Peseta: after joining the EMS [without realignments], during the monetary storm of the European Monetary System [with three devaluations] and finally, during the spring 1995 with a last devaluation. The extension of the econometric model with Time Varying Transition Probabilities depending on some exogenous variables let us obtain that: (1) The international reserves variation is the more outstanding and significant variable in the origin of the turbulences. (2) Variables related to existence of the Target Zone are also relevant. Exchange Rate Deviation from the upper band is significant whatever dependent variable or econometric procedure we use.; It is possible to conclude that the Peseta turbulences could be explained without the contrivance of real variables and in the line of balance of Payments crises.; Finally, the results and the mixture of econometric approaches are some of the most outstanding contributions of the work. The application of a Binary Dependent Variable Model on the results of a Markov-Switching on daily interest rate differential between Spain and Germany allow us to confirm the relevance of Target Zone Variables.