The paper presents recent developments in the estimation of EU’s poverty indicators (so-called Laeken indicators) for population subgroups or domains and small areas. The indicators include at-risk-of poverty rate, relative median at-risk-of poverty gap (poverty gap for short), quintile share ratio (S20/S80 ratio) and the Gini coefficient. We concentrate here on the estimation of poverty gap for domains. The estimators to be compared are of direct, synthetic or composite type. The direct estimator does not use auxiliary information. Various synthetic estimators that use auxiliary data are introduced. Composite estimators are constructed as a linear combination of direct and indirect estimators. It appears that a design-unbiased direct (default) estimator can be quite inefficient, and more efficient indirect synthetic estimators are often seriously biased. Composite estimators offer a compromise for many situations. We also discuss the relative performance of the estimators under outlier contamination. Linear mixed models with area-specific random terms are fitted for the income variable underlying the indirect estimators. Unit-level auxiliary data are incorporated into the estimation procedure. We discuss here the case of unequal probability sampling. By using quality measures (bias and accuracy), the relative performance of the estimators is assessed with Monte Carlo simulation experiments based on data extracted from statistical registers maintained by Statistics Finland. Research has been conducted in the context of the AMELI project (Ad-vanced Methodology for European Laeken Indicators), which is supported by European Commission funding from the Seventh Framework Programme for Research.
|Otsikko||Book of Abstracts of the ERCIM'10 Conference|
|Tila||Julkaistu - 2010|
|OKM-julkaisutyyppi||B3 Vertaisarvioimaton artikkeli konferenssijulkaisussa|
- 112 Tilastotiede