Composite indices applications cover a broad gamut. From evaluations of Universities, countries, institutions or firms; their popularity has grown exponentially, and their use expands throughout all domains of research and policy making exercises. Reasonably, there is something desirable in being able to consolidate the breadth of a variety of information into a single metric that serves as a benchmark of how well an entity operates. Regardless, this raises more questions than it may answer. How is this metric constructed and how sound is it eventually to trust it? The reason being trust in these metrics is of paramount importance, and it is often the case that their whole construct is opaque, lacks a sound framework and thus inferences made based on such metrics are bound to be distorted. There are several steps to be meticulously followed in the construction of composite indices, and choices in each and every step could radically alter the metric produced. The most important steps in the methodological framework are those of weighting and aggregating the elementary indicators into the overall composite index. There is a large documented literature on these steps, and a rough division in two groups at both steps. However large the literature and the heated debate are, whether one talks about weighting or aggregating the elementary indicators, there are some things found to be missing or often neglected in both camps’ arguments. These relate to assumptions about representation of preferences (as far as weighting and other parameters are concerned) and conceptual/methodological drawbacks (as far aggregation is concerned). By combining and extending well-known Operations Research methods, this thesis provides two novel methodological proposals aiming at addressing the issues found -or overlooked- in the above-mentioned steps. The metric that is produced offers a more ‘holistic’ output as, contrary to previous approaches in this strand, it is based on information that takes into account a multiplicity of viewpoints and spatial information related to the a unit’s competitive environment. At the same time, it encapsulates traditional uncertainty analysis, permitting the decision-maker to get more insights than simply creating a cardinal construct, something that is often found missing from most composite indices nowadays.
|Date of Award||Sep 2019|
|Supervisor||Alessio Ishizaka (Supervisor), Gianpiero Torrisi (Supervisor) & Salvatore Greco (Supervisor)|
Multiple criteria evaluation of economic units via composite indices
Tasiou, M. (Author). Sep 2019
Student thesis: Doctoral Thesis