Applications of fat-tailed Distributions in financial time series
碩士 === 逢甲大學 === 統計與精算所 === 91 === This paper studies some GARCH models with generalized error distribution (GED) innovations. We accommodate the heavy-tailed characteristic of financial returns by the proposed error distribution. The parameter estimation is performed by a Bayesian approach. We a...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en_US |
| Published: |
2003
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/37063062046163814314 |
| Summary: | 碩士 === 逢甲大學 === 統計與精算所 === 91 === This paper studies some GARCH models with generalized error distribution (GED) innovations.
We accommodate the heavy-tailed characteristic of financial returns by the proposed error distribution.
The parameter estimation is performed by a Bayesian approach. We adopt reversible-jump Markov chain
Monte Carlo methods to compare GARCH models with GED and student-t distributions. Simulations show that
our Bayesian estimator and model selection procedure perform well. Forecasting of volatility and
Value at Risk is also illustrated in the paper. Our methodology is applied to several financial time
series for volatility modeling and forecasting. The violation rates of different markets are also
provided to understand the accuracy of value at risk forecasting.
|
|---|