Estimation of Complex Small-Area Parameters with Application to Poverty Indicators

• Activity Number: 213
• Type: Invited
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): J.N.K. Rao*+
• Companies: Carleton University
• Keywords: Empirical Bayes ; Hierarchical Bayes ; parametric bootstrap ; Mean squared error ; FGT measures ; Skew normal

Abstract

We study the estimation of complex non-linear small area parameters using empirical best (EB) and hierarchical Bayes (HB) methods. Our methodology is generally applicable, but we focus on measures of poverty indicators, in particualr on the clas of poverty measures called FGT poverty measures used by the World Bank. Small area estimates of the FGT measures for several countries have been released by the World Bank, using methodology of Elbers, Lanjouv and Lanjouv (2003) that combines census and survey data and produces simulated censuses. Estimates for any desired small area are produced from the simulated censuses.We obtain EB and HB estimators under a nested error regression model, assuming normality on a transformed welfare variable. We use a parametric bootstrap method for estimating the mean squared error of the EB estimators. We obtain posterior means and posterior variances, using a grid method that avoids the use of MCMC methods. We also extend the EB and HB methods to skew normal nested error models. We present the results of a model based simualtion study on the relative performance of EB, HB and the World Bank methods and an application to data from a Spanish survey.