|
Activity Number:
|
113
- Nonlinear and Nonstationary Dependent Processes: Modeling, Inference, and Applications
|
|
Type:
|
Invited
|
|
Date/Time:
|
Monday, August 9, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
Business and Economic Statistics Section
|
|
Abstract #316675
|
|
|
Title:
|
Polyspectral Mean Estimation of General Nonlinear Processes
|
|
Author(s):
|
Dhrubajyoti Ghosh* and Tucker Sprague McElroy and Soumendra Lahiri
|
|
Companies:
|
Washington University and US Census Bureau and Washington University
|
|
Keywords:
|
Polyspectra;
k-th order periodogram;
Polyspectral mean;
Discrete Fourier Transform
|
|
Abstract:
|
Higher-order Spectra or Polyspectra are useful for non-linear and non-Gaussian processes. Polyspectra of order k is defined as the Fourier Transform of k-th order cumulants. Polyspectral means can be defined as weighted averages of the polyspectra over the Fourier frequencies. Estimators of a polyspectral mean can be obtained by using the k-th order periodograms. In this paper, we consider a class of polyspectral mean estimators and derive the asymptotic distribution. Under suitable conditions, we obtain an exact definition of the limit distribution that depends on the weight function and higher-order spectra. We also establish the validity of a frequency domain Bootstrap for polyspectral means. Results from a simulation study are also reported to illustrate the finite sample properties of the asymptotic results.
|
Authors who are presenting talks have a * after their name.
