Small Area Estimation – Some Applications in NSSO Surveys

Abstract

The purpose of this article is to use small area estimation (SAE) method to produce district level estimates for some of the important indicators such as living condition, poverty incidence and working population ratio. For this purpose, data from 68th round (2011-12) of National Sample Survey Office (NSSO) pertaining to Household Consumer Expenditure Survey (HCES) and Employment and Unemployment Survey (EUS) for Uttar Pradesh has been used along with the 2011 Population Census data. The empirical results, evaluated through set of internal and external diagnostics measures, show that the district-level estimates generated through SAE approach are precise than the direct estimates. Spatial maps showing district level inequality in distribution of living condition, poverty incidence and working population ratio in Uttar Pradesh are also produced. These maps and districts level estimates are important for target oriented effective policy planning, monitoring and decision-making. In this article we deliberately consider two types of estimates viz. averages and proportions and use two different survey data of NSSO for producing district level estimates. We then illustrate how the existing survey data can be linked with Census data to produce reliable, timely and cost-effective district-level estimates of averages and proportions. The SAE methodology, illustration and guidelines set out in this paper can be adopted in other existing surveys for generating the disaggregate level estimates.