A0783
Title: A new prewhitening approach in time series variance estimation
Authors: Xu Liu - University of Washington, Seattle (United States) [presenting]
Kin Wai Chan - The Chinese University of Hong Kong (Hong Kong)
Abstract: Prewhitening is a common approach to deal with strong autocorrelation. We propose a new perspective called tail prewhitening to perform it. It uses parametric models to project the neglected tail autocovariances in nonparametric estimators onto the final estimator. This approach bridges the nonparametric variance estimator and the parametric prewhitening model through a scaling factor. It automatically switches between these two arms using a bandwidth parameter, without the need to transform the entire dataset into residuals, as in the standard prewhitening approach. When the tail prewhitening model is well-specified, a parametric rate can be achieved. It is also more robust against prewhitening model misspecification than the standard approach in finite samples. Besides, it avoids severe potential variance inflation or power reduction caused by the recoloring factor in the standard approach. It is shown that multiple parametric models can be used to construct a multiply robust tail prewhitened estimator. It also naturally works for multivariate time series. A real-data example in Markov chain Monte Carlo output analysis is provided.
