Paper Title
BINAR(1) Model with Geometric marginals: Application to Intra-Day Stock Transactions in Mauritius

Abstract
This paper proposes a novel bivariate integer- valued autoregressive process of order 1 (BINAR(1)) where the counting series are Geometric while their corresponding innovations follow a mixed Geometric distribution with non-stationary moments due to the effect of time-dependent covariates. In contrast to the existing stationary BINAR(1) with marginal Geometric model, this paper assumes the cross-correlation between the series is induced by the correlated innovations only. As for the inferential part, the regression parameters are estimated by the generalized quasi-likelihood (GQL) method while the dependence parameters by the method of moments based on a robust autocorrelation specification. A simulation study is designed to assess the estimation procedures. The new model is also applied to analyse the trading intensity or volume of intra-day transactions of the two major banking institutions in Mauritius, namely Mauritius Commercial Bank Group Limited (MCB) and State Bank of Mauritius Holdings Ltd (SBMH), that are listed in the Stock Exchange of Mauritius (SEM). The root mean square errors (RMSEs) are computed and compared with the BINAR(1) with Poisson innovations. Index Terms - BINAR(1), Geometric, GQL, Non-Stationary ,Over- Dispersion