Small area estimation under varying area boundaries using the synthetic estimator

This paper investigates the use of hierarchical models for small area estimation with varying area boundaries, employing the synthetic estimator. The paper shows how area estimates and corresponding MSE estimates can be obtained at a variety of nested and intersecting boundary systems by fitting a model at the lowest possible level. The estimates are obtained by aggregating from the lowest level and are therefore internally consistent. The methodology is illustrated by presenting results of a simulation study that uses hierarchical models built at the lowest area level defined by the UK 1991 census