Multivariate count autoregression (Q2278669): Difference between revisions

Linear and log-linear models are studied for multivariate count time series data with Poisson marginals. A copula function is introduced on a vector of associated continuous random variables. The conditions for ergodicity and stationarity are derived based on a perturbation approach in Markov chain theory and theory of weak dependence. Quasi-likelihood estimating functions are suggested that yield consistent and asymptotically normal estimators of model parameters. A limited simulation study and a real data example illustrate the results.