|
Activity Number:
|
22
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, July 29, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #308599 |
|
Title:
|
Bayesian Analysis and Applications of a Generalized Threshold Autoregressive Model
|
|
Author(s):
|
Peng Sun*+
|
|
Companies:
|
Merck & Co., Inc.
|
|
Address:
|
PO Box 1000 UG1D44, North Wales, PA, 19454,
|
|
Keywords:
|
Bayesian ; Gibbs Sampler ; Threshold Autoregressive Model
|
|
Abstract:
|
In this paper, we generalize the self exciting threshold autoregressive (SETAR) model by allowing the threshold variable to be a linear combination of functions of lag values of the series. The generalized model is then analyzed in the Bayesian framework. When applying the model to simulated datasets, we find that the performance of the posterior simulator is highly dependent on the choice of starting values. If the starting value is close to the true parameter value, then the chain can easily achieve convergence. Otherwise, it persists near a local mode of the posterior distribution. The convergence problem is explored in detail and an effective sequential algorithm is developed to address this issue. We present an application of the generalized model to the S&P 500 daily return data. The model fitting results compare favorably to the GARCH(1,1) model fitted to the same dataset.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2007 program |
