Building a composite index using the multi-objective approach: An application to the case of human development

The composite indicator (CI) is a powerful tool to observe the performance of various alternatives when several dimensions are in question. Normalisation, weighting and aggregation are known as the three obligatory steps to build a CI. The current paper proposes a new and unconventional method, that does not require any of these steps to construct a CI. In this method, we suggest the use of Pareto frontiers, a concept in multi-objective optimisation, to set the alternatives (e.g., countries, firms) in an order. Pareto frontiers help to collect similar alternatives together regarding their performance on several dimensions, sequentially. Each Pareto frontier represents a different achievement level. However, it is not possible to evaluate the alternatives located in the same Pareto frontier. To identify the performance of each alternative within the Pareto frontiers, we implement the ordered logit regression model by considering Pareto frontiers as the thresholds. The estimations of this model are then used to predict the latent variable that represents a specific problem. We use this methodology to re-measure human development. A comparison was carried out between the Human Development Index (HDI) and our proposed method. According to the results, the country ranking using the proposed method shows a significant difference in comparison to HDI, although there is a high Spearman’s correlation between them. The proposed method is strong and reliable because it does not require any vague selection of the so-called obligatory steps to build a CI. Moreover, it provides full non-compensatory results and effective ranking of the alternatives.