Estimating equations in the presence of prior knowledge

Abstract

Godambe's (1960) criterion of optimality for estimating equations is extended to the case where prior knowledge is given. The optimality of the estimating equation ∂logξ(θ| x )/∂θ = 0, where ξ(θ| x ) is the posterior density, is shown under certain assumptions. The nuisance parameter case is also analysed.