Provincial And Highly Urbanized City Level Estimation Of Average Household Saving Rate Using Spatially Correlated Random Area Effects Model

Type Conference Paper - 13th National Convention on Statistics (NCS)
Title Provincial And Highly Urbanized City Level Estimation Of Average Household Saving Rate Using Spatially Correlated Random Area Effects Model
Author(s)
Publication (Day/Month/Year) 2016
City Mandaluyong City
Country/State Philippines
URL https://psa.gov.ph/sites/default/files/Session 2-6 Provincial and Highly Urbanize City Level​Estimation of Average Household Saving Rate Using Spatially Correlated Random Area Effects Model.pdf
Abstract
This paper uses a spatially correlated random area effects model in estimating the average saving rate of households in all provinces and highly urbanized cities (HUCs) in the Philippines. Specifically, an empirical best linear unbiased (EBLUP) estimates are combined with a spatial autoregressive (SAR) model-based estimates by relating the small area direct estimates to the area specific covariates and taking into account the information provided by neighboring areas. The SAR model suggests that the provincial or HUC level proportion of young dependents (aged 0 to 14 years), proportion of the elderly (aged 65 years and above), proportion of household heads having a college education, and the average family size determine the average household saving rate in a province or HUC. Moreover, neighboring provinces and HUCs within their 120 kilometer radius are captured to have spatial clustering and such is considered in having a statistically significant and optimal spatial autocorrelation. Based on the SEBLUP estimates, most of the provinces and HUCs (39 out of 112) have average household saving rate between 10% and 15% while less than half (46%) of the provinces and HUCs have average household saving rate below 10%. Specific financial programs could be identified to boost the household saving rates of these provinces and HUCs. Among all the provinces and HUCs, 103 (91.96%) of them have estimates which are said to be reliable.

Related studies

»