Early warning systems for currency crises: A Markov-switching approach
The turmoil of the 1990s stimulated the development of "early warning systems" (EWS) for currency crises. The two prevalent approaches have been probit/logit modeling, and the "signals" approach of Kaminsky, Lizondo and Reinhart (1997). Several issues can be raised regarding the use of these methodologies: the need for an a priori dating of crises; the sensitivity of these dates to the arbitrary thresholds used for identification; the loss of information that comes with transformation to binary variables; the use of "exclusion windows"; and the lack of theoretical motivation for using these models. This dissertation proposes an alternative framework, a Markov-switching model with time-varying transition probabilities, for use in EWS construction. This framework does not require an a priori dating of crises; instead, the identification (and characterization) of crisis periods are part of the model's output. It can tell us much more about the dynamics of crises--when they tend to occur, what they are like, how long they tend to last and what might bring them to an end. And because one byproduct of estimation is the one-step ahead forecast of crisis occurrence, it lends itself to being used as an EWS. Finally, one can show how the Markov-switching framework can be derived directly from theoretical models of currency crises. The first chapter summarizes the literature on currency crises, discusses the implications for EWS construction, and looks at the relative advantages and disadvantages of probit/logit and "signals" models. The second chapter develops and estimates a baseline EWS using the Markov-switching methodology, using monthly data for Indonesia, Malaysia, the Philippines and Thailand. The third chapter explores extensions of the baseline model, to see whether we can improve the model's performance in identifying and anticipating crisis periods. In the fourth chapter, we show how the Markov-switching specification can be derived directly from a theoretical currency crisis model. We also explore the phenomenon of regional contagion, where comovements among regional currencies increase following a shock. The final chapter summarizes the thesis results and lays out directions for future research.
|Year of publication:||
|Authors:||Abiad, Abdul de Guia|
|Type of publication:||Other|
Dissertations available from ProQuest
Saved in favorites
Similar items by subject
Find similar items by using search terms and synonyms from our Thesaurus for Economics (STW).