A multivariate semi-logistic autoregressive process and its characterization (Q553089)

Three new multivariate semi-logistic distributions (denoted by \(\text{MSL}^{(1)}\), \(\text{MSL}^{(2)}\), and GMSL, respectively) have been recently studied by the present author [J. Multivariate Anal. 101, 893--908 (2010; Zbl 1181.62074)], which are more general than and may serve as competitors to the multivariate logistic distributions of \textit{E. J. Gumbel} [J. Am. Stat. Assoc. 56, 335--349 (1961; Zbl 0099.14502)] and \textit{B. C. Arnold} [``Multivariate logistic distributions'', in: Handbook of the logistic distribution. Statistics: Textbooks and Monographs. 123. New York: Marcel Dekker. 237--261 (1992; Zbl 0794.62001)]. Presently, an autoregressive multivariate stochastic model is constructed, which yields a stationary Markov process with \(\text{MSL}^{(1)}\) as a marginal invariant distribution. This model is denoted by \(\text{MSL}^{(1)}\)-AR(1). Several distributional properties of \(\text{MSL}^{(1)}\)-AR(1), as well as a characterization, are derived. Furthermore, an \(\text{MSL}^{(2)}\)-AR(1) process is also constructed and studied.