Socioeconomic Correlates of Poverty- An Application of Multivariate Normal Distribution
Abu Jafar Mohammad Sufian

Abstract
Among the various impediments to peace and security as mentioned in the report of the High Level Panel of Experts of the UN Secretary General(Documents, 2005:595), this paper is concerned with the poverty component, defined as the percentage of population living on less than $2 per day. The aim of this paper is to identify socioeconomic correlates of poverty by analyzing aggregate level data collected from the following sources: World Population Data Sheets, 2011 and 2013 (Population Reference Bureau, 2011, 2013); World Fertility Data 2012 (United Nations, Department of Economic and Social Affairs, Population Division, 2013); and Adult Literacy Rate, Female by Country (United Nations Educational, Scientific and Cultural Organization (UNESCO), Institute for Statistics 2003-2013). On the basis of the percentages of people living in poverty, 82 countries have been classified into 3 groups: low poverty, medium poverty, and high poverty. The ‘canonical discriminant analysis’ technique based on the multivariate normal distribution has been used to identify the variables that discriminate among the groups most. The adequacy of the technique is judged by using eigenvalues, canonical correlations, and Wilk’s lambda. To determine the relative weights of the discriminating variables, standardized canonical coefficients, as well as, the structure coefficients have been used. The analysis identified the ‘percentage of dependent population’ as the most influential variable in discriminating among the groups. The second most influential variable is the ‘use of modern methods of contraception. Other important variables are ‘gross national income’, ‘percent urban’, and ‘children ever born’. Policy implications have been discussed.

Full Text: PDF