Using Logistic Regression Model To Analyse And Compare Factors Affecting Poverty For the National Household Budget Poverty Survey Data 2014-2015

Abstract

The phenomenon of poverty has become a great global economic and social problem that affects most of the countries of the world, especially the developing countries. The objective of this study is to identify the most important demographic, social and economic factors that influence poverty household in Sudan. The problem of the study is represented by how to identify the main variables that affect poverty household and individuals in the country. The data for the study was obtained from the National Household Budget and Poverty Survey that was undertaken by the Central Bureau of Statistics in 2014/2015, which gathered 11953 households in three-stage and, 50% were randomly selected for researchers about 5965 households. A logistic regression model was used to estimate and determine which variables might be significant in explaining poverty. The dependent variable is the probability of a household being poor or not and a set of demography and socio-economic variables as the explanatory variables. Households are classified as either poor or non-poor on the basis of per capita annual spending of 6082 SDG as the poverty threshold and a daily energy intake of 2110 calories per person identified by Central Bureau of Statistics. The study reveals that nearly 36.1% of the sample households live below the poverty line with an average of 10.3 poverty gap, while about 67.2%for the poorest state (Central Darfur). Data were analyzed by statistical packages for social sciences SPSS and STATA. The results showed that the variables which are positively correlated and significantly explain with the probability of being poor are: place of residence, household size, dependency ratio, crop farming and “khalowa” education. While the gender, age of household head, university level and marital status (married/widowed) was negatively associated and significantly explain with the probability of being poor. Moreover, the model correctly predicted 78.9% of the observations. The study recommends using the optimal model of the logistic regression to predict household poverty in future through the variables that affect it and, for the government should be focusing on improving the livelihood situation, education, and intensification of family planning programmers in Sudan.