| Over the last decades, as barriers to the mobility of capital have fallen, international capital flows have become increasingly important. While foreign capital undoubtedly offers development opportunities for many countries, free mobility of capital has also been blamed for financial crises, which have affected developed countries, e.g. Britain in 1992, and, much more severely, developing countries, e.g. Argentina in 2001--02.; Despite their importance, a literature review quickly reveals that crises are poorly understood. Available models provide valuable insights, but crises are still extremely hard to predict, and there is no consensus as to how to alleviate their impact once they arrive. In this thesis, I develop new models to explore two important, unresolved questions, to what extent crises are self-fulfilling and what determines exchange rates after crises.; My first model is a general equilibrium model where the government chooses how fundamentals evolve and when/whether to abandon a peg. I assume that the government cannot commit to future actions and study when self-fulfilling crises may arise. I find that, for some parameters and initial conditions, the peg survives forever under commitment but collapses if attacked. Further, for pegs that will be abandoned at some point, the collapse can happen at different times, depending on a sunspot. Thus, at least in theory, the model supports self-fulfilling crises. A calibrated version of the model suggests that, in the case of the Italy in 1992, the crisis was bound to happen, but the timing was arbitrary.; I also study bubbles as a possible explanation for exchange rate overshooting after crises. Some instances of overshooting, such as Korea 1997/1998, appear difficult to explain based solely on expectations of future inflation. The model of bubbles of Abreu and Brunnermeier (2003) provides an interesting alternative framework. However, results depend crucially on the unsatisfactory assumption that the price (or exchange rate) does not reflect changes in selling pressure unless it exceeds a certain threshold, at which point the bubble bursts. I relax this assumption, and show that bubbles still arise when prices reflect selling pressure monotonically, if there is an unobservable random component in selling pressure. |
