Modelling Poverty and its Determinants in Addis Ababa: A Focus on Multinomial Logit Selection Model

Abstract

I have used different econometric models (OLS, probit and multinomial logit selection) to analyse factors leading to poverty. The main emphasis of the study is to model determinants of standard of living in Addis Ababa using two-stage estimation technique. In the first step, a multinomial logit model is applied to distinguish between three socio-economic groups. The second stage regression determinants of standard of living i.e total household expenditure per adult equivalent per month) are identified after incorporating the correction term for sample selectivity using the Lee-Heckman method. Among others, variables such as education, access to credit, employment status. gender, marital status and food shortage experience are significant determinants of welfare.