Group action on a lattice and an application to time series analysis (Q1209651)

The \(k\)-th \((\geq 2)\) cumulant spectral density, which gives us information concerning the structure of the underlying process, of a stationary process with mean zero is the \((k-1)\) dimensional Fourier transform of the \(k\)-th cumulants of the stationary process with mean zero. The procedure for obtaining a consistent estimator of the fourth order cumulant spectral density is to first estimate product moments, then obtain estimates of cumulants by using the relation between cumulants and product moments, and finally take the Fourier transform of these estimated cumulants. This paper is concerned with obtaining the asymptotic variance of the proposed fourth order cumulant spectral density estimator.