Logistic and Augmented Modeling of Poverty Profiles and Forecasting of the Food Crops Balance Sheet (Case of Lake Victoria Basin, Kenya)

Abstract

The problem of poverty is one of the core issues concerning developing countries like Kenya. The formulation of an adequate programme to combat poverty is of importance for any meaningful development plan. The key features relevant are the construction of an appropriate poverty index and proper estimation of the measure. The different dimensions of poverty add to the problem of choosing the appropriate poverty measure and indicators. What is the appropriate measure to estimate the incidence of poverty? In other words, what criteria should be used to define and measure poverty? What is missing from previous studies is an analysis of different poverty measures. The study sought to propose a model that takes care of the multi-faceted nature of poverty and also look into the trends of food security in the Lake Victoria basin in three ways: Firstly, we come up with the poverty line of the region using the consumption data, secondly we look at the two models and estimate the best model for investigating poverty including a wide range of independent variables to reflect the contribution of each to a household being poor and lastly forecasting food insecurity using the food crops balance sheet. The assessment involved analysis using the augmented regression model and the stepwise model analysis for variable reduction, construction of logit models for different poverty proxies and application of the models in classification of households by poverty status. Further, assessment of poverty was made using assets, a multi-dimensional approach. Further analysis was done on the food balance sheet in order to obtain projections on food production and consumption patterns in the region. In the results we precisely state the asymptotic properties of maximum likelihood estimators for logistic regression models and additionally we show that the maximum likelihood estimators converge, under conditions of fixed number of predictor variables, to the real value of the parameters as the number of observations tends to infinity. The results also indicated that the parameters estimates are normal in distribution by plotting the quantile plots and undertaking the Kolmogorov-Smirnov and the Shapiro-Wilks test for normality, and conclude that parameters came from a normal distribution.The thesis comes up with some theoretical as well as empirical contributions taking into consideration various aspects of poverty measurement in the context of Lake Victoria basin, Kenya. A significant development for research has been the improvement in constructing a coherent framework for measuring poverty in multidimensional environment. This framework provides a new insight into particular elements of poverty that is useful and relevant to poverty interventions. The projections in this work are not statements of what will happen, but of what might happen, given the assumptions and methods used. These projections provide a policy-neutral starting point that can be used to analyze national and counties food requirement and policy initiatives.