Modeling household expenditure in Rwanda using Neural Network

Abstract The household expenditure patterns and the factors related to them caught the attention of policymakers for social programs. Household consumption is the key component for living standards and a good economy. The objective of this research is to estimate the factors influencing household expenditure patterns concerning household residence area for each household on basis of the social-economic as well as demographic factors. The data was obtained from the Fifth Integrated Household Living Conditions Survey (EICV 5) 2016/2017. The neural network method has been used in the analysis of the data. The results on household characteristics and component expenditure, independent variables play a crucial role in discriminating the household expenditure categories i.e low and high expenditure in Rwanda. The component expenditure analysis on rural and urban residence areas revealed that there is a disparity between rural and urban residential areas. In rural residence areas, the household expenditure is significantly estimated by Household assets (durable goods), food, and household size while in the urban residence area, the household expenditure is significantly estimated by Household assets (durable goods), non-food, and education costs. When the residence area is taken as an explanatory variable, the order of importance has been changed as follows Household assets (durable goods), non-food, education cost, food, household size, and Age. All independent variables considered in this study are important in estimating household expenditure.