|
Activity Number:
|
26
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 6, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Business and Economics Statistics Section
|
| Abstract - #305527 |
|
Title:
|
A New Model for Forecasting Credit Spread Changes: Model Estimation, Prediction, and Inference Procedures
|
|
Author(s):
|
Yang Wang*+
|
|
Companies:
|
The Pennsylvania State University
|
|
Address:
|
325 Thomas Building, University Park, PA, 16802,
|
|
Keywords:
|
varying-coefficient model ; Wilk's phenomenon ; forecasting ; finance ; local linear ; generalized likelihood ratio test
|
|
Abstract:
|
Motivated by both empirical and statistical considerations of potential structural changes of linear regression models, we proposed time-varying-coefficient models, a new modeling technique for financial data, to analyze credit spread data for nine Merrill Lynch corporate bond indexes. Chow test was conducted to motivate the proposed model. Estimation procedures using local linear modeling were explained. We compared the new model fit to that of ordinary least squares regression model. A novel prediction procedure by time-varying-coefficient model was proposed and compared to ordinary least squares regression model. Confidence intervals were constructed and generalized likelihood ratio-type test was applied to test whether the coefficients really vary. Wilks' phenomenon was shown to hold. Monte Carlo simulation studies were used to assess the finite sample performance of the procedure.
|
