Spectral density estimation for nonstationary data with nonzero mean function

We introduce a new approach for nonparametric spectral density estimation based on the subsampling technique, which we apply to important class of nonstationary time series. These are almost periodically correlated sequences. In contrary existing methods, our technique does not require demeaning data. On simulated data examples, compare estimator function with classical one. Additionally, propose modified estimator, allows reduce leakage effect. Moreover, in supplementary materials, provide simulation study and two real economic applications. Supplementary materials this article available online.