Adaptive density estimation for general ARCH models

Abstract

We consider a model $Y\_t=\sigma\_t\eta\_t$ in which $(\sigma\_t)$ is not independent of the noise process $(\eta\_t)$, but $\sigma\_t$ is independent of $\eta\_t$ for each $t$. We assume that $(\sigma\_t)$ is stationary and we propose an adaptive estimator of the density of $\ln(\sigma^2\_t)$ based on the observations $Y\_t$. Under various dependence structures, the rates of this nonparametric estimator coincide with the minimax rates obtained in the i.i.d. case when $(\sigma\_t)$ and $(\eta\_t)$ are independent, in all cases where these minimax rates are known. The results apply to various linear and non linear ARCH processes.