In this paper total household expenditures estimated on the basis of independent variables for each gender of House Hold (HH) head. Furthermore the role of independent variables on estimating HH expenditures investigated. For this purpose, data taken from Household Integrated Economic Survey (HIES) 2007-08 conducted by Federal Bureau of Statistics FBS, Pakistan. The sample of 15,512 households was selected using stratified random sampling. Multiple Linear Regression Model and MLP Neural Network Model have been used for the analysis purpose and also compared the performance of both. Results revealed that in male headed households; family size, proportion of children less than of age 5, number of earners, proportion of out of school children, assets, income and number of rooms of households played a significant role to estimate total expenditures. In case of female headed households; family size, number of rooms, assets and income of HH were significant. Performance indices showed that MLP neural network is best model to estimate the total expenditures.
Household Expenditures; Multiple Linear Regression; Multilayer Perceptron Neural Network.