Estimation of Hyper-Parameters Based on Logit-Normal Model

摘要

In this paper, we deduce a range of the hyper-parameters, and use Gauss-Hermite quadrature introduced in reference [1] to approximate the moment equation, estimating the moment estimators by enumeration method. Simulation experiment carried out to study the accuracy of estimates by Monte Carlo integration. Simulation experiment shows that estimating parameters from the moment equation of logit-normal model by enumeration method can get very accurate estimates, and verified the EB estimator is robust. We also use the method to estimate poverty incidences in Spain.