Analysis of 2008 Central and East European Currency Crisis Using Early Warning Model

Triggered by the failure of the Lehman Brothers in the fall of 2008, a global financial crisis affected many countries. In this critical situation, many emerging countries in Central and Eastern Europe faced crisis situations, while many countries which had been through @past currency crises, like Asian countries or Latin American countries, were able to avoid such crises. This paper investigates this phenomenon using an empirical approach. The idea is to answer the question, "Why were many Central and Eastern European countries affected by a global financial crisis at this time?" In order to answer this question, I use the DCSD model by Berg, Borensztein, Millesi-Feretti and Patillo (1999). The DCSD model is a popular Early Warning Model which has improved upon the previous best model which was able to predict Asian currency crises most effectively. In this paper, I extend the samples of countries to include some Central and Eastern European countries, and use the latest data available. Furthermore, I use a global crisis dummy variable on the assumption that the recent crisis in Central and Eastern European countries was different from many past crises because it was originally affected by a global financial crisis. The possible significance of my research is based on these two points since existing research has not yet analyzed them. In order to know more about this special area, I analyze the empirical results by checking the determination of crises, the extent to which the model can predict the crises, and changes in significant explanatory variables. The model consists of three steps. For the first step, I use the EMP (Exchange Market Pressure) system in order to test whether a crisis occurred or not. In addition to this normal model, I use a "potential crisis model" that eases the crisis criteria by half. This model enables us to check potential risks for the entire Euro zone. In the second step, I use a multivariate panel probit regression technique, using the crisis index which was derived from the first step as a dependent variable. At the final step, I calculate the crisis probability of each period by using the loss function indicator derived from the coefficients obtained by the second step's regression analysis. In my analysis using the method described above, I obtained the following five conclusions. (1) The recent situation in Central and Eastern Europe was judged to be a crisis. (2) The potential crisis model criteria were attained not only in emerging Central and Eastern European countries that asked for support from the IMF but also in other European countries (Euro zone crisis). (3) From the dummy variable analysis, it was not clear whether the recent crisis in Central and Eastern Europe was of a special nature compared with previous crises. (4) The two explanatory variables - current account deficit and short-term foreign debt - had statistically significant results, as well as previous crises. (5) However, compared with the Asian currency crisis, the Central and East European crisis had a lower model accuracy. Furthermore, explanatory variables that reflect economic fundamentals (foreign reserves and exports) were not statistically significant. The fifth conclusion is a particularly interesting result. This result shows that the rate of crises that can be explained by the five explanatory variables of the DCSD model has decreased. From the results of deterioration in the significance of explanatory variables reflecting economic fundamentals, we can know the impact of worsening economic fundamentals on crises has diminished. Therefore, we need to develop a new analytical model that can predict crises more accurately.