Abstract
MVUE θ* of the parametric function θ = Σ i–1 h i , exp(b i μ + c i σ2) for a lognormal distribution LN(μ, σ2) is considered. The exact variance var(θ*) is given. Then the variance of the MVUE of the parametric function θ = var (Y) = exp(2μ) {exp(2σ2) – exp(σ2)} is explored: (i) for a random sample from the LN(μ, σ2) distribution, and (ii) for a lognormal regression.