Statistical Methods of Predicting Country Debt Crisis
- 59 Downloads
This chapter discusses discriminant analysis, a statistical method for handling classification problem, and applies the analysis to predict sovereign debt crisis by differentiating two groups, “Default” and “Non-default”, based on certain quantitative and qualitative country characteristics. The model is tested on a new country to determine which of the two groups it belongs, and the model correctly predicts default. With the same characteristics for the discriminant function, the logit function, which measures the odds of default in relation to such characteristics, is also estimated. For classification purposes, discriminant analysis uses normal distribution, whereas the logit model assumes a distribution with fatter tails compared to normal distribution, thus making logit analysis more relevant in the presence of abnormal and extreme values in the population.