Predicting Income and Feasible Loan Amount for a Household Unit (Expenditure Analysis of Badulla District, Sri Lanka)

Abstract

Expenditures of a household are growing due to many reasons which came up with complexity of basic needs and downturn of economy. Emerging body of knowledge suggest the necessity of expenditure studies in Sri Lanka. It is difficult to carry out an island wide study. Therefore Badulla was selected since it covers variety of living conditions in a limited extent. Six divisional secretariats were selected randomly and those were Badulla, Passara, Uva-Paranagama, Haldummulla, Soranathota and Meegahakivula. Respondents (household heads) were selected by systematic random sampling. They were interviewed by a structured questionnaire. 95% confidence intervals were calculated for mean costs of basic needs. These means were compared by living areas using one way ANOVA. Result implies that living area can influence on amount of cost. The main objective of this study is predicting total income and feasible loan amount per a household unit. Multiple regression analysis showed 70.9% of the variation of total income was accounted by the explanatory variables. Cost for foods, cost for transport and cost for clothing had a significant (p-value<0.05) effect on total income. Moreover, fitted model for predicting feasible loan amount explained 87.0% of total variation. Cost for electricity, cost for water, cost for education, cost for health and cost for social activities were significant (p-value<0.05) variables of the best fitted model. It can be concluded that multiple linear regression model which was fitted for predicting feasible amount of loan performs better than fitted model for predicting total income. This would be useful for administrative divisions and financial sector. Because before a person go for a higher amount of loan it can be explained by taking basic details of him or her what is the feasible amount of loan with his or her economic status. It would be good for loan receiver and loan providers.

Key Words: Expenditure, household income, stratified random sampling, multiple linear regression
DOI: 10.7176/DCS/9-6-11

Publication date:June 30^th 2019