Type
Text
Type
Dissertation
Advisor
Zhu, Wei | Wang, Xuefeng | Wu, Song | Xiao, Keli.
Date
2015-05-01
Keywords
Statistics
Department
Department of Applied Mathematics and Statistics.
Language
en_US
Source
This work is sponsored by the Stony Brook University Graduate School in compliance with the requirements for completion of degree.
Identifier
http://hdl.handle.net/11401/77443
Publisher
The Graduate School, Stony Brook University: Stony Brook, NY.
Format
application/pdf
Abstract
Factor-based models have been extensively used in economic and financial time series analyses. The Factor-augmented Error Correction Model (FECM) is a successful generalization of the Factor-augmented Vector Autoregression Model and the Error Correction Model for large panel nonstationary time series data. By combining the factors and error correction terms together, the FECM is able to utilize both the aggregated panel information summarized through the Dynamic Factor Model as well as the long-term equilibrium information introduced by the cointegration relationship. In this thesis we extend the FECM by allowing time-varying model parameters. There are ample evidences from both theoretical and empirical studies supporting the notion that the parameters of economic and financial models often change over time. By relaxing the parameters to be time-varying, the model will be more adaptable to complicated and realistic data structures, such as those with potential structural instability after a recession or crisis. We conclude this thesis by applying the newly developed time-varying FECM to provide more suitable models for PPNR (Pre-Provision Net Revenue) studies, part of the required modeling process in CCAR (Comprehensive Capital Analysis and Review) -- commonly known as the Federal Reserve’s Stress Test on big banks and other financial institutes. | Factor-based models have been extensively used in economic and financial time series analyses. The Factor-augmented Error Correction Model (FECM) is a successful generalization of the Factor-augmented Vector Autoregression Model and the Error Correction Model for large panel nonstationary time series data. By combining the factors and error correction terms together, the FECM is able to utilize both the aggregated panel information summarized through the Dynamic Factor Model as well as the long-term equilibrium information introduced by the cointegration relationship. In this thesis we extend the FECM by allowing time-varying model parameters. There are ample evidences from both theoretical and empirical studies supporting the notion that the parameters of economic and financial models often change over time. By relaxing the parameters to be time-varying, the model will be more adaptable to complicated and realistic data structures, such as those with potential structural instability after a recession or crisis. We conclude this thesis by applying the newly developed time-varying FECM to provide more suitable models for PPNR (Pre-Provision Net Revenue) studies, part of the required modeling process in CCAR (Comprehensive Capital Analysis and Review) -- commonly known as the Federal Reserve’s Stress Test on big banks and other financial institutes. | 89 pages
Recommended Citation
Hao, Xue, "Factor-Augmented Error Correction Model with Time Varying Coefficients" (2015). Stony Brook Theses and Dissertations Collection, 2006-2020 (closed to submissions). 3258.
https://commons.library.stonybrook.edu/stony-brook-theses-and-dissertations-collection/3258