Computational Modelling and Simulation Using a Full Correlation BINAR(1) Process

A full correlation first-order bivariate integervalued autoregressive (BINAR(1)) process with Poisson innovations is modelled and simulated computationally in this paper. The present series observation and the previous-lagged observations of the other series interact to locally produce crosscorrelation in this proposed model, although the paired innovations are taken to be independent. The paper also uses the Conditional Maximum Likelihood (CML) approach to estimate the model parameters and the performance of the CML is assessed through a simulation study.