COMPARISON OF SOME NON-LINEAR AUTOREGRESSIVE PROCESSES

From MaRDI portal





The author considers autoregressive processes \(\{X_ n\}\), \(\{Y_ n\}\) satisfying \(X_{n+1}=\lambda (X_ n)+Z_{n+1}\), \(Y_{n+1}=\mu (Y_ n)+Z_{n+1}\), and \(X_ 0=Y_ 0=0\), where \(\lambda\) (\(\cdot)\) and \(\mu\) (\(\cdot)\) are odd increasing functions and \(\{Z_ k\}\) is a sequence of independent random variables with identical unimodal symmetric distributions. The main results of the paper are that \(\{| X_ n| \}\) is a stochastic monotone random sequence and that if \(| \lambda (\cdot)| \leq | \mu (\cdot)|\) and if \(\{Y_ n\}\) admits a stationary distribution \(\rho\), then \(\{X_ n\}\) admits a stationary distribution \(\pi\) and all such \(\pi\) satisfy \[ \int^{\infty}_{0}f(| x|)d\rho (x)\quad \geq \int^{\infty}_{0}f(| x|)d\pi (x) \] for all positive increasing functions f(\(\cdot)\) on [0,\(\infty)\).











This page was built for publication: COMPARISON OF SOME NON-LINEAR AUTOREGRESSIVE PROCESSES

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3761423)