Asymptotic efficiencies of three methods of estimation for the inverse Gaussian-Poisson distribution

Abstract

SUMMARY The two parameters of the inverse Gaussian-Poisson distribution may be estimated by (a) sample moments, (b) the proportion of zeros and the mean in the sample, and (c) by maximum likelihood. In the paper the asymptotic variances for all three joint estimation techniques and the corresponding efficiencies for methods (a) and (b) are derived. For 0 > 0 90 and oc 0 90, both methods (a) and (b) are very inefficient and the maximum likelihood estimation technique should be employed.