Efficiency of a Kernel Density Estimator under an Autoregressive Dependence Model

The problem of estimating the probability density function of a strictly stationary process is considered. To study the effect of a dependence structure on the efficiency of a kernel density estimator, the mean integrated squared error (MISE) of the Fourier integral estimator (FIE) is derived on the assumption that the observed data are generated by a first-order autoregressive process. Numerical results for the normal and Cauchy densities show that even moderate departures from independence can lead to a considerable loss in efficiency of the FIE. In addition to efficiency considerations, the issue of determining an optimal smoothing parameter for the FIE under the autoregressive model is addressed.