Remeasuring the HDI by Data Envelopement Analysis

Abstract

The measurement of human development has a potentially strong impact on how the development gap is viewed and on the formulation of new policies. Therefore correct and fair measurement is of great importance. In this paper, we develop an algorithm to compute comprehensive differentiation rules suitable for measuring human development. We used models from Data Envelopment Analysis (DEA) literature to compare performance in a multiple output setting. The models were evaluated by empirically re-estimating the human development index (HDI). The most notable advantages of DEA models are that they endogenously construct a non-linearly arranged set of best practice countries and that the weights of each indicator entering the HDI is endogenously determined based on an optimization calculus. These weights are allowed to vary thereby accounting for cross-sectional heterogeneity. While country clusters are identified by their similarity, some interesting outliers can also be singled out using DEA. Such outliers are either best practice frontier countries or countries that are locked in an underdevelopment trap.