Economic Development of EU Countries – Multivariate Outlier Detection

Abstract

The measure of a country‟s economic development is one of the most critical and highly debated issues in economic research. Different approaches have been applied and numerous indicators have been used in the process to do so. The aim of this research is to present one synthesized indicator that can quantitatively point out any country‟s development level. In this paper, the statistical I-distance method is thoroughly explained and applied to 28 EU countries. Furthermore, an improved method of I-distance (stepwise I-distance) procedure has been used in order to examine statistical significance of each development indicator. This procedure allows iterative exclusion of statistically insignificant indicators, whereupon the final results are obtained using the most important development indicators. Through defined procedures for the elimination and/or selection of indicators, the new method seeks to form an optimal set of indicators, while reducing the dimension of the complex problem at hand. The stepwise I-distance method takes into account the discriminatory power of each of the indicators used. Accordingly, a unique I-distance value is formed for each observation from the observed set. The research results show that this method can be used to detect multivariate outliers.