A generalized nonlinear model for long memory conditional heteroscedasticity

We study the existence and properties of stationary solution of ARCH-type equation $r_t= \zeta_t \sigma_t$, where $\zeta_t$ are standardized i.i.d. r.v.'s and the conditional variance satisfies an AR(1) equation $\sigma^2_t = Q^2\big(a + \sum_{j=1}^\infty b_j r_{t-j}\big) + \gamma \sigma^2_{t-1}$ with a Lipschitz function $Q(x)$ and real parameters $a, \gamma, b_j$. The paper extends the model and the results in Doukhan et al. (2015) from the case $\gamma = 0$ to the case $0< \gamma < 1$. We also obtain a new condition for the existence of higher moments of $r_t$ which does not include the Rosenthal constant. In particular case when $Q$ is the square root of a quadratic polynomial, we prove that $r_t$ can exhibit a leverage effect and long memory. We also present simulated trajectories and histograms of marginal density of $\sigma_t$ for different values of $\gamma$.