Efficient and feasible inference for high-dimensional normal copula regression models

The composite likelihood (CL) is amongst the computational methods used for estimation of high-dimensional multivariate normal (MVN) copula models with discrete responses. Its advantage, as a surrogate method, that based on independence univariate marginal regression and non-regression parameters pairwise correlation parameters. Nevertheless, efficiency CL method estimating can be low. For response, weighted versions equations an iterative approach to determine good weight matrices are proposed. general methodology applied MVN ordinal regressions marginals. Efficiency calculations show proposed nearly efficient maximum fully specified models. Illustrations include simulations real data applications regarding longitudinal (low-dimensional) time (high-dimensional) series response covariates. Our studies suggest there substantial gain in via method.