Spatio-Temporal Fay-Herriot Models in Small Area Estimation to Obtain Factors That Affecting Poverty in Polewali Mandar District

Abstract

Poverty is a classic problem in almost all regions in Indonesia both at the provincial and district levels. Polewali Mandar district is a district with the highest poverty rate in West Sulawesi Province. Handling the problem of poverty requires the availability of data to the smallest level so that the policies taken by the government can be right on target. One method for obtaining poverty estimates reaches the smallest area using the Small Area Estimation (SAE) method. Spatio-Temporal Fay-Herriot models is one of the SAE methods that has considered spatial and time effects. Auxiliary variables are needed in the SAE method to get good estimation results. Availability of auxiliary variables that affect poverty in Polewali Mandar district is very necessary. Therefore the purpose of this study was to obtain auxiliary variables that influence poverty in Polewali Mandar district. Bootstrap parametric procedures are used to obtain factors that influence poverty in Polewali Mandar district, while for spatial weighting matrices use customized contiguity based on the main business fields in the village. The response variable is the percentage of poverty in each village affected by the Susenas sample. The results obtained were the variables that affected poverty in Polewali Mandar district were the percentage of farm worker’s families, the percentage of the population receiving Jamkesmas / Jamkesda, ratio of population to number of micro and small industries, average number of household members, ratio of population to number of health facilities and percentage of families living on river banks.