Composite indicators (or indexes) are very common in economic and business statistics for benchmarking the mutual and relative progress of countries in a variety of policy domains such as industrial competitiveness, sustainable development, globalization and innovation. The proliferation of the production of composite indicators by all the major international organizations is a clear symptom of their political importance and operational relevance in policy-making. As a consequence, improvements in the way these indicators are constructed and used seem to be a very important research issue from both the theoretical and operational points of view. This article aims at contributing to the improvement of the overall quality of composite indicators (or indexes) by looking at one of their technical weaknesses, that is, the aggregation convention used for their construction. For this aim, we build upon concepts coming from multi-criteria decision analysis, measurement theory and social choice. We start from the analysis of the axiomatic system underlying the mathematical modelling commonly used to construct composite indicators. Then a different methodological framework, based on noncompensatory/nonlinear aggregation rules, is developed. Main features of the proposed approach are: (i) the axiomatic system is made completely explicit and (ii) the sources of technical uncertainty and imprecise assessment are reduced to the minimum possible degree.
|Publication status||Published - 17 Jul 2009|