Activity Number:
|
65
|
Type:
|
Topic Contributed
|
Date/Time:
|
Monday, August 12, 2002 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Business & Economics Statistics Section*
|
Abstract - #300778 |
Title:
|
Noninvertible Moving Average Modeling Via All-Pass Filters
|
Author(s):
|
Jay Breidt*+ and Richard Davis and Beth Andrews
|
Affiliation(s):
|
Colorado State University and Colorado State University and Colorado State University
|
Address:
|
201 Statistics, Fort Collins, Colorado, 80523-1877, USA
|
Keywords:
|
non-Gaussian time series ; quasi-likelihood ; least absolute deviations
|
Abstract:
|
All-pass models are autoregressive-moving average models in which all of the roots of the autoregressive polynomial are reciprocals of roots of the moving average polynomial and vice versa. All-pass models generate uncorrelated (white noise) time series, but these series are not independent in the non-Gaussian case. If an all-pass is driven with heavy-tailed noise, then its marginal distribution will also have heavy tails, and the process will exhibit "volatility clustering" much like a nonlinear model such as ARCH. All-pass models are useful in identifying and estimating noninvertible moving averages. Theoretical properties of alternative estimation procedures for all-pass, based on a likelihood approximation and related criterion functions, are described. Behavior of the estimators in finite samples is studied via simulation.
|