Thesis (MSc.) - Cyprus International University. Institute of Graduate Studies and Research Economics Department

Includes bibliography (sheets 60-64)

ABSTRACT The governments with a huge population of people that are living in extreme poverty especially in Africa have continuously been losing millions of dollars from time to time. This is as a result of under capacity utilization of potential resources. Some of the resources like money are spent to cure the effects of extreme poverty (social benefits, socio-political and economic crimes) instead of eradicating poverty itself (the Machiavellian conspiracy) as the root cause. The researcher therefore is going to investigate patterns and trends of extreme poverty on 10 countries in the SADC region (Zimbabwe, South Africa, Mozambique, Zambia, Botswana, Tanzania, Malawi, Angola, Lesotho and Namibia). The study uses a balanced panel data with ten countries over a five year time period (2013 - 2017). With seven determinants to extreme poverty namely, income level, unemployment, health care, education level, natural disaster, inequality and population growth, the researcher adopted an econometric model to run random and fixed effects regression model. Model specification test, unit root test, and multicollinearity test were also carried out to investigate whether there were no violations of the normality assumptions. According to the diagnostic tests conducted the data collected was fit without normality assumptions violation. The findings also concluded that the Random effects model was appropriate after rejecting the null hypothesis. Cemented by the Hausman Test with a much high chi probability of 92% thereby accepting the null hypothesis that Random effect model is appropriate. The results concluded by proving wrong the fallacy of composition (what is true for one is true for all). Every determinant in its hierarchy has different influences in every single country. However the implications on these results might arise when temporal variations in the dependent variable are not captured prior to the intercept and the error term, which might lead to biasness. Events like natural disasters normally come as temporary and occur at the same time since countries are within the same regions and might pose fixed effects. Keywords: Extreme Poverty, Fallacy, Fixed Effects Model, Panel Data, Random Effects Model, Regression.