...currency crises focused on explaining currency crises ex post, and we are still uncertain whether it is possible to forecast currency crises. The issue is particularly relevant for policy analysis: if there was a method to provide signals on future currency crises, and on their nature, it would be easier for international organizations like the International Monetary Fund (IMF) to prevent them. Recent effort has been devoted to develop new "Early Warning System" methodologies to monitor the behavior of macro-economic fundamentals and evaluate the likelihood of currency crises. (1) But which economic variables should we include? How good are these methodologies?

This paper provides an empirical analysis of traditional leading indicators (LIs) for forecasting currency crises, as well as a New Early Warning System that uses a large number of economic variables to monitor each country. We focus on the recent East-Asian currency crises, and we are interested in answering the following questions. Can we improve our forecasts of currency crises by using a large number of predictors? Which variables are the most important to predict currency crises? An answer to these questions would have practical implications for preventing currency crises, as well as provide guidance on the theoretical modeling of these events and which economic policies would be most effective. To answer these questions, we use the recently developed method of Diffusion Indexes (DIs; Stock and Watson 1998, 2002). The DI extracts a few factors from a large database; the factors are chosen so that they explain most of the variability in the data and are orthogonal to each other. We analyze DIs as an alternative method to the few selected LIs usually used in the literature because each currency crisis has its own characteristics, which were difficult to predict ex ante by using previously developed economic models.

Empirical evidence shows that while the conditional mean of exchange rate returns is un-predictable, their variance is conditionally heteroskedastic. We therefore, model the volatility of exchange returns following Andersen et al. (2001, 2003), who showed that natural logarithms of squared returns are approximately normal. The main advantage of their finding is that it immediately translates into easy and empirically valid distributional assumptions for predicting the distribution of the forecast of the exchange rates, which in turn can be used to evaluate the probability of a crises. Thus, their method is particularly convenient to achieve our objective of forecasting currency crises.

The main body of empirical evidence on forecasting currency crises in the current literature builds on Kaminsky and Reinhart (1999; KR hereafter) and Frankel and Rose (1996). KR identify macro-economic variables (LIs) that signal future crises by using a noise to signal ratio methodology. Frankel and Rose instead estimate a panel probit model, with economic variables dictated by well-established economic models. There are two main differences between our approach and that in the current literature. First, the above methodologies are cross-sectional in spirit, and the LIs are chosen based on the historical data, whereas we examine the performance of these LIs in real time and from a truly out of sample forecasting perspective. Thus, while the existing analyses are very useful for understanding the dynamics of currency crises ex post, nobody has investigated their usefulness to monitor the likelihood of currency crises in real time, which is one of the objectives of this paper. Furthermore, a drawback of the threshold values traditionally chosen for the LIs in the existing literature is that they are independent of the forecaster's loss function (see Elliott and Lieli 2004). This paper attempts instead to link the measure of the likelihood of a crises to a measure of risk (as in Kilian and Manganelli 2003, 2004, and in the Value at Risk literature). Second, we consider a new Early Warning System based on DIs that, unlike the current literature that focuses on a small number of LIs, uses a large number of predictors. In the DI approach, we do not have to know ex ante that economic variables to use, and we can...

NOTE: All illustrations and photos have been removed from this article.



Looking for additional articles?
Search our database of over 3 million articles.

Looking for more in-depth information on this industry?
Search our complete database of Industry & Market reports by text, subject, publication name or publication date.

About Goliath
Whether you're looking for sales prospects, competitive information, company analysis or best practices in managing your organization, Goliath can help you meet your business needs.

Our extensive business information databases empower business professionals with both the breadth and depth of credible, authoritative information they need to support their business goals. Whether it be strategic planning, sales prospecting, company research or defining management best practices - Goliath is your leading source for accurate information.