Modeling barangay [village] food poverty in the Philippines

Food poverty is defined as the inability to access good quality, affordable and nutritious food. One of its measures is food poverty incidence or subsistence incidence which is defined as the proportion of households whose annual income is below the computed food threshold to the total number of households. In this study, direct estimates of food poverty incidence at the barangay [village] level were obtained using 2000 Family Income and Expenditure Survey (FIES). The barangay-level food poverty incidence ranges from 0 to 100 percent. On the average the barangay-level food poverty incidence in the Philippines is equal to 12.52 percent. The procedures resulted to 1,538 direct estimates equal to zero while there were 5 estimates equal to 100. The estimates were found to be unreliable since only 9 barangays obtained a coefficient of variation of at most 10 percent. Indirect estimation techniques in particular, regression synthetic was used to obtained indirect estimates for all barangays in the Philippines with an aim to improve the directs estimates. However, the validity of the regression synthetic estimation technique, depend on the satisfaction of the classical assumptions underlying linear regression method. In this study, no predicting model of the food poverty at the barangay level was identified that satisfy all of the assumptions of linear regression method using the 2000 Census on Population and Housing and other administrative data as auxiliary variables. Assumptions on normality and homoscedasticity on the residuals were violated for all the models that were explored. Variables were transformed, however, the same problems was still encountered. Results also show that such was observed because of the highly skewed distribution of the direct estimates, that is, a number of the estimates take the value zero.