Poverty Indicators Measuring Method And Model Researching Based On Small Area Estimation

Poverty is a serious problem that influences the development of human kind. Poverty still harassed most developing countries and even some developed countries. Research on poverty will make people understand the causes, mechanisms and dynamics of poverty. Poverty research is how to understand and measure poverty. The difficulty is that countries or regions lacked of census data reflecting poverty,And access to relevant data through sampling surveys extremely is difficult. So poverty indicators measure become a bottleneck study the poverty problem.In this context, Using small area estimation method can solve poverty indicators measuring when data is insufficient.This article tracked small area estimation theoretical frontier and searched poverty’indicators on the condition of insufficient sample. We take the problem of poverty in china as the object of study. The study is based on the Empirical Theory of Poverty by Amartya Sen, and the analysis frame based on FGT model and using small area estimation method for estimating poverty indicators Nonlinear small field overall. To estimate the results were further evaluated for small area estimation, we use Monte Carlo simulation and bootstrap method to calculate the MSE of the estimators, and stochastic simulation of China applied to analysis of poverty indicators. With the simulation analysis, the Empirical Bayesian Predictor of the poverty indicator has a good performance in its bias and MSE.In the article we draw the conclusions based on combining empirical analysis and Simulated Experiments method:(1) The use of small area estimators and the auxiliary estimator by the "simulation" census was possible to reduce the poverty gap MSE;(2) It is possible to reduce the poverty gap MSE; auxiliary estimatoators of small area and use the bootstrap method to obtain the MSE of estimators, the conclusions show empirical Bayes method is superior to the direct estimation method and ELL;(3) Empirical Bayes method is a method based- designed and based model, it can be used to estimate the various values of small area (such aspoverty gap), but empirical Bayes model relies heavily on the effectiveness, the model test and the choice is particularly important.In the article we made several innovative searching:1. The estimation of the poverty in the SAE solves the problems when the data is not enough, incomplete or even no available data at all;2. When there is no census data, we can use the SAE method together with the Monte Carlo simulated experiment and parametric bootstrap as an auxiliary method for a robust estimator. Improvement in the poverty indicator leads to a more accurate estimation and a smaller range of the MSE;3. Introduce the SAE method for the poverty indicator for the SAE method which will save man power and material resources to make it possible for a dynamic monitoring of poverty.