Determinants of stock market classifications

摘要

We use discriminant analysis to describe and predict market classifications. Potential discriminators are derived from relevant characteristics of market indices, in particular from the returns' volatility. Using a training data set, an initial screening on the predictors is carried out and a model-based simple rule is constructed with 96.6% of correct classifications. 10 new markets are allocated to one of the previously defined groups: Developed, Emerging, or Frontier, with only one misclassification. The quantitative approach was able to anticipate classification reviews.