Human Development Over Time: An Empirical Comparison of a Dynamic Index and the Standard HDI

This research uses panel data to explore inferences about human development associated with two different formulations of the Human Development Index (HDI). The first is the standard HDI as computed since the 2010 Human Development Report. The second is based on a linear combination of the core dimensions, with the weights determined by a dynamic factor analysis algorithm called the Two Cycle Conditional Expectation Maximization algorithm. This algorithm is able to exploit all of the cross sectional and temporal information in a panel dataset and is specifically designed to handle panels with short time dimensions. The two HDI specifications are employed as dependent variables in mixed effects models to estimate the rates-of-change of the HDI by groups of countries classified by income and region. Then, a Monte Carlo simulation is run to assess the efficacy of the two methods in detecting turning points in the trajectories of simulated “true measures” of human development. The results show that, with the exception of low income countries, the two HDI measures give rates of change estimates that do not differ statistically. However, the dynamic HDI does better at detecting the presence of turning points in the trajectory of human development.