Dimitris N. Politis
Distinguished Professor
Department of Mathematics & Halicioglu Data Science Institute
University of California, San Diego
Department of Mathematics & Halicioglu Data Science Institute
University of California, San Diego
Session: Technical Session II
Date & Time: Friday, May 22 — 1:30 PM–3:45 PM
Talk Title
Flat-Top Corrected Spectral Density Estimation and a Test for Over-differencing
Abstract
We revisit the setup of testing the null hypothesis of stationarity where the alternative is that the process has a unit root. We construct a test statistic that is based on estimating the spectral density of the differenced series at the origin. The test statistic is found to require a bias correction that is implemented via flat-top kernels. The resulting flat-top corrected spectral density estimator may be interesting in its own right. (Joint work with Somak Maitra.)