Some Alternatives for Robust Estimation of the Spectrum in Stationary Processes

Abstract

This paper is dedicated to estimation of the spectral density of stationary linear processes in the presence of additive outliers. We suggest the use of robust periodogram proposed by Fajardo et. al. (2009) (LPR) with different smoothing windows. Empirical results showed the robustness of the estimator under additive outliers. A real data application is presented with IGP-DI series.