Efficient Estimation of the Headcount Index

Abstract

The headcount index is an important poverty indicator that analyzes the proportion of people classified as poor. The headcount index is often used in comparisons of poverty across countries. In practice, the population headcount index is unknown, and survey data derived from official surveys are used to obtain an estimation of this indicator. The use of more accurate estimation methodologies may provide a better knowledge about the real headcount index. We describe alternative estimation methods for the headcount index. Proposed methods use the known ratio and regression techniques after transforming an auxiliary variable related to the variable of interest into a dummy variable. Monte Carlo simulation studies are carried out to evaluate numerically the performance of the various estimators of the headcount index discussed in this paper. Results indicate that the proposed estimation methods can be more accurate than existing methods. Confidence intervals are also evaluated and desirable coverage rates are obtained. Simulation studies are based upon real survey data extracted from official surveys on income and living conditions.