Bivariate first-order random coefficient integer-valued autoregressive processes

In this paper, we propose a new bivariate first-order random coefficient integer-valued autoregressive (BRCINAR(1)) process with dependent innovations. Some basic probabilistic and statistical properties of this model are obtained. Estimators of unknown parameters are derived by using Yule-Walker, conditional least squares and conditional maximum likelihood methods. The asymptotic properties of the estimators are established. The performance of these estimators is compared through a simulation experiment. Moreover, the coherent forecasting for BRCINAR(1) model is addressed. Finally, an application to a real data example is investigated to assess the performance of the model. (C) 2019 Elsevier B.V. All rights reserved.