Early Warning System for Financial Crisis: Statistical Classification Approach

This chapter discusses a statistical classification approach for developing a financial crisis early warning system (EWS). The main aim of this chapter is to introduce a lag-. l forecasting classifier constructed in accordance with the sequential works by Son et al. (2009), Ahn et al. (2011), and Ahn et al. (2012) and to discuss how to implement this classifier in practice. Our EWS issues a warning by classifying and forecasting the future conditions of a market. The core of the procedure is defining the oracle rule that determines future market conditions, tracing the oracle rule (the equivalent of lag-. l forecasting) with a trained machine learning algorithm, and then linking lag-. l forecasters across various ls. Our methodology is based on self-fulfilling financial crisis or crisis contagion theories for emerging markets.