Modelling an early warning system for financial crises: A dynamic probit model approach

Despite noticeable progress in the literature on Ews in past years, the recent financial crisis has revived the interest in the Ews literature among researchers and politicians. An early warning system (Ews) consists of the quantitative definition of crisis and a set of variables that may help predict crises and an econometric method to obtain an early warning signal from those variables. In this thesis financial crises are defined by exchange market pressure Index (EMPI) and the probability of crisis is attempted to be predicted by analyzing the lagged binary crisis variable and some macroeconomic indicators. In this context, a new generation of Ews model is developed in order to determine the leading indicators of the financial crises of 17 developing countries and the dynamic structure of the crises are examined by using the dynamic random effects probit model. This approach allows us to take explicit account of the existence of intertemporal links between crises i.e. source of endogenous crises persistence. The source of endogenous crises persistence includes the lagged value of the binary dependent variable. The lagged value of the binary dependent variable enables analysis of the state dependence. The results of model show that true state dependence has a significant effect on the probability of currency crises in the short term. The another contribution of this paper is to investigate the forecast horizon at which different indicators signal a crisis. Some variables have a short term impact, such as M1 growth, foreign reserves and growth rate of M2 to foreign reserves. Variables such as rate of M2 to foreign reserves, real effective exchange rate overvaluation, growth of domestic credit, current account balance and foreign interest rate have a long term impact (early warning period). Ews models can be useful to identify common macroeconomic problems associated with financial crises and they can be used explain occurred crises ex post. However, due to the endogeneity of crises and evolving economic and financial structures in the global economy, reinforced the view that developing a stable model that can predict financial crises is a challenging task.