Measuring Distribution of Wealth in Zambia Using Census Micro Data: An Application of Principal Component Analysis

Abstract

Census data for Zambia was used to estimate the distribution of wealth in Zambia by constructing the Wealth Index as a measure of socioeconomic status using Principal Component Analysis. The reliability of the index is observed from three fronts; coherence, robustness and validity in representing household socioeconomic status. Classifying the households across all quartiles is highly consistent and robust. The index’s performance in predicting the welfare distribution is analogous to established and most widely used methods from Demographic Health Surveys, as evidenced by similarities in the statistical distributions. Unlike other survey estimates, the index has been produced at the subnational level, such as district, enabling the classification of Zambia’s districts according to their socioeconomic status. The index can be used to predict other socioeconomic outcomes, such as education and health, via Small Area Estimation techniques and determine district-level resource allocation by the central government.