High-frequency sampling and kernel estimation for continuous-time moving average processes

Interest in continuous-time processes has increased rapidly in recent years, largely because of high-frequency data available in many applications. We develop a method for estimating the kernel function $g$ of a second-order stationary L\'evy-driven continuous-time moving average (CMA) process $Y$ based on observations of the discrete-time process $Y^\Delta$ obtained by sampling $Y$ at $\Delta, 2\Delta,...,n\Delta$ for small $\Delta$. We approximate $g$ by $g^\Delta$ based on the Wold representation and prove its pointwise convergence to $g$ as $\Delta\rightarrow 0$ for $\CARMA(p,q)$ processes. Two non-parametric estimators of $g^\Delta$, based on the innovations algorithm and the Durbin-Levinson algorithm, are proposed to estimate $g$. For a Gaussian CARMA process we give conditions on the sample size $n$ and the grid-spacing $\Delta(n)$ under which the innovations estimator is consistent and asymptotically normal as $n\rightarrow\infty$. The estimators can be calculated from sampled observations of {\it any} CMA process and simulations suggest that they perform well even outside the class of CARMA processes. We illustrate their performance for simulated data and apply them to the Brookhaven turbulent wind speed data. Finally we extend results of \citet{bfk:2011:1} for sampled CARMA processes to a much wider class of CMA processes.