The Pilot Environmental Performance Index 2006 as a Tool for Optimising Environmental Sustainability at the Country Level

Abstract

The Environmental Performance Index (EPI) constitutes one of the most influential as well as prevalent measures of environmental sustainability (ES) currently in use. In this piece of research, the earliest version of the EPI 2006 (the Pilot EPI) is employed for (a) uncovering the characteristics of the maximally sustainable country, and (b) assisting any interested country (either participating in the creation of the Pilot EPI, or not) on how to maximally increase its ES. A genetic algorithm (GA) is employed to this end, where (i) each of the three constructs of the Pilot EPI hierarchy (raw data, Proximity-to-Target data, and Policy Categories), (ii) the Pilot EPI scores, and (iii) the underlying relationship between (i) and (ii), are encoded in the (I) chromosomes, (II) fitness values and (III) fitness function, respectively, of the GA. Following GA convergence, the fittest chromosome(s) express the characteristics—in terms of EPI constructs—of the country(ies) of practically maximally attainable ES, with their fitness (Pilot EPI score) approximating the maximum value of 100 i.e. attaining almost ideal ES. Further to confirming the agreement between maximum Pilot EPI score and maximal fitness (which has not always been the case with other ES-related indices), the evolution per se of the GA population can be exploited for steering any interested country towards attaining maximal ES in a gradual, tailor-made manner. This performed as follows: at each step of that process, a chromosome (coming from any GA generation) that is sufficiently similar to the current construct values of the interested country, yet which demonstrates a higher fitness (Pilot EPI score) than the interested country, is selected; the necessary modifications in the interested country’s ES policies are implemented so that its construct values are made to approximate the values of the genes of the selected chromosome, thereby causing an increase in the country’s ES. The potential of selecting—at each step—the most convenient (viable for the interested country) as well as effective (causing maximal increase in ES via, either the easiest to implement, or the smallest) changes in the chromosome, adds efficiency and flexibility to the methodology; at the same time, alternative paths to the gradual improvement of the ES of the interested country are made available. By repeating this process as many times as desired, the interested country’s ES moves in a gradual manner towards the (theoretically established) maximal level of ES. The fitness function of the GA is represented via the most accurate approximation(s) from among (A) the methodology reported in the literature (whenever available), (B) polynomial approximation of various degrees, and (C) artificial neural networks (ANNs) of the general regression architecture (GRNN). Although neither approximation is capable of perfectly reproducing the relationship between each of the Pilot EPI constructs and the EPI scores (which puts into question the means of construction of the Pilot EPI), the GA is found successful in guiding the step-by-step changes that are deemed appropriate for gradually as well as maximally improving the level of ES of any country of interest.