Constructing Multidimensional Indexes of Development: A Factor Analysis Approach

Ganegodage, K. Renuka (2008). Constructing Multidimensional Indexes of Development: A Factor Analysis Approach PhD Thesis, School of Economics, The University of Queensland.

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
n40674546_phd_abstract.pdf n40674546_phd_abstract.pdf application/pdf 151.59KB 25
n40674546_phd_content.pdf n40674546_phd_content.pdf application/pdf 2.89MB 6
n40674546_phd_front.pdf n40674546_phd_front.pdf application/pdf 219.96KB 82
n40674546_phd_totalthesis.pdf n40674546_phd_totalthesis.pdf application/pdf 28.87MB 28
Author Ganegodage, K. Renuka
Thesis Title Constructing Multidimensional Indexes of Development: A Factor Analysis Approach
School, Centre or Institute School of Economics
Institution The University of Queensland
Publication date 2008-09
Thesis type PhD Thesis
Supervisor Rambaldi, Alicia N.
Rao, Prasad
Tang, Kam K.
Subjects 340000 Economics
Formatted abstract
The study aims at constructing an index that can capture the multi-dimensional process of development to fill the gap in the existing literature. What motivated such an index is the view that development is a dynamic process involving major transformations in economic, institutional, and social structures, and that it is characterized by multi-dimensional changes in the composition of production, sectoral redistribution of resources, changes in income inequality, and reduction in poverty. Based on this view, a large array of measurements on different characteristics is required to gauge the level of development. Adelman and Morris (1967) pioneered the use of factor analysis in condensing diverse development indicators into a single index. This thesis extends their approach with several significant modifications to overcome methodological deficiencies, and with a substantially larger sample of countries that covers all income categories and a more recent sample period.
Factor analysis is a data reduction technique to handle a large number of correlated variables. We select image analysis as the estimation method for factor analysis due to its sound statistical properties. However, we also perform the estimation using principal components analysis as a robustness check. Though factor analysis is a very popular technique, it lacks proper statistical procedures to evaluate a series of decisions that have to be made in its implementation as well as in finalizing the results. The thesis proposes an estimation of standard errors through a jackknife procedure to overcome these problems.
The analysis is conducted for 97 countries using data compiled for the period 1995 to 2004. The results of the two factor analysis methods provide similar results with slight differences. The study extracts four measures (factors) from 42 variables to quantify the social, economic, technological and institutional structures that underpin development. Based on our first factor, technology and institution quality is found to be the most important development measure, especially for countries at the upper level of development. Our second factor shows that basic social development is important for the countries at the bottom level of development. These results imply not only that development is a multi-dimensional concept, but also that the notion of development changes with the stage of development. Below a certain threshold level, development is characterized by survival capabilities, and above that, it is characterized by advancement capabilities. This finding is in contrast to the one-size-fit-all philosophy that underpins many development measures, such as the Human Development Index (HDI) and the Physical Quality of Life Index (PQLI). The third factor provides information on capital formation while the fourth factor provides information on HIV incidence. These extracted factors are considered as a set of sub-indexes which can be used as measures of development. Moreover, our results indicate that some of the popular choices of variables used in the literature to measure development might not be capturing the aspects presumed by the researchers. Primary school enrolment rates as proxy for educational attainment, and industrial value added, another popular proxy in the literature, were not significantly loaded on any of the factors.
These factors are used to construct two new indexes: the first one, the Development Index (DI), is a measure of the level of development and is constructed as a composite of the first two factors; the second one, the Growth Potential Index (GPI), is a measure of the potential for future development and is constructed based on the third factor. The DI is used to rank countries according to the level of development and is validated using income per capita and the HDI. The DI is able to tease out the differences between countries that have similar income level or HDI scores. This is because it makes use of a much richer set of information from a large array of variables compared to those made with the HDI and income. A further validation of DI is through a cluster analysis exercise.
Countries are clustered based on the four factors for two separate periods 1995–1999, and 2000– 2004. The clustering results are used to identify possible drivers and constraints of the development process. The findings show the importance of technology and institutions as well as the basic level of development in determining the development levels of countries at different stages. Though capital formation could play a major role in economic growth, it alone is not sufficient for the achievement of higher development targets; appropriate levels of institutional, technological and basic development are also required. Overall, the findings suggest the crucial importance of identifying the stage of development of a country and the characteristics of its structure, in formulating development policies for it. The analysis clearly shows the ability of the proposed new index, DI, and the set of sub-indexes to provide policy directions. The study offers a new way to compare the level of development of countries and a possibly new direction for setting policy priorities.
Keyword economic development, development indexes, factor analysis, Jackknife, cluster

Citation counts: Google Scholar Search Google Scholar
Created: Tue, 23 Sep 2008, 10:44:28 EST by Noela Stallard on behalf of Library - Information Access Service