Estimation of poverty indicators for domains with unit-level auxiliary information

Risto Lehtonen, Ari Veijanen

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientific


The paper presents current developments on the estimation of poverty
indicators for domains and small areas in the context of the EU FP7 project
AMELI (Advanced Methodology for European Laeken Indicators). The indicators
considered are the Gini coeffi cient, relative median at-risk-of poverty gap
(poverty gap for short) and quintile share ratio (S20/S80 ratio). The corresponding
estimators to be compared are of direct, synthetic or composite type. It appears
that a direct (default) estimator can be quite ineffi cient, and a more effi cient
indirect synthetic estimator is often seriously biased. Composite estimators
are constructed as a linear combination of the direct and indirect estimators.
Underlying the indirect estimators, we use mixed models with domain-specifi c
random intercepts. Unit-level auxiliary data are incorporated into the estimation
procedure. In our case, these data are based on statistical registers maintained
by Statistics Finland. To cover realistic sampling designs, both equal and unequal
probability sampling designs are covered. By using selected indicators on bias and
accuracy, the relative performance of the estimators is assessed with Monte Carlo
simulation experiments based on real data.
Original languageEnglish
Title of host publicationProceedings of the Q2010 Conference : European Conference on Quality in Official Statistics
Number of pages1
Publication date2010
Publication statusPublished - 2010
MoE publication typeB3 Article in conference proceedings

Fields of Science

  • 112 Statistics and probability

Cite this