A non-simplistic approach to composite indicators and rankings: An illustration by comparing the sustainability of the EU Countries

Composite indicators are very popular, despite being affected by several problems that often result in lack of robustness of the rankings involved. The aim of this paper is to show that composite indicators can be safely used, provided that rankings are built via uncertainty analysis rather than using a single composite. For this purpose, the approach we follow first combines different normalisation, aggregation rules, and weighting systems to calculate many different composites, and then derives the rankings from the frequency distribution of the rankings of each "competitor" according to each composite. Such an approach appears to be a good compromise between the need for a more concise overview when looking at many variables and the loss of relevant information occurring when indicators are aggregated into a single composite indicator. To illustrate the approach, we rank EU Countries in terms of their sustainability.