Volatility forecasting using adaptive exponential smoothing versus GARCH modeling
(2005)Department of Economics
- Abstract
- Different models for predicting volatility in financial returns have been proposed during the years. This paper focuses on two different ways of volatility modeling, generalized autoregressive conditional heteroskedasticity (GARCH) models and exponential smoothing models. In a previous study 2004, James W. Taylor introduced an adaptive version of the traditionally exponential smoothing method with transition variables in order to capture the well known asymmetry of stock return volatility, the leverage effect. Like non-linear GARCH models, the sign and size of past shocks are used as transition variables to allow smoothing parameters to adapt to changes of the time series characteristics over a period of time. This paper compares the yet... (More)
- Different models for predicting volatility in financial returns have been proposed during the years. This paper focuses on two different ways of volatility modeling, generalized autoregressive conditional heteroskedasticity (GARCH) models and exponential smoothing models. In a previous study 2004, James W. Taylor introduced an adaptive version of the traditionally exponential smoothing method with transition variables in order to capture the well known asymmetry of stock return volatility, the leverage effect. Like non-linear GARCH models, the sign and size of past shocks are used as transition variables to allow smoothing parameters to adapt to changes of the time series characteristics over a period of time. This paper compares the yet common GARCH(1,1) and two asymmetric GARCH models with four smoothing methods using this new adaptive approach. The performance of the models are evaluated by making volatility forecasts of seven stock indices from different countries. However, the new adaptive smoothing methods are not found to be significantly better than traditional exponential smoothing, neither did they outperform GARCH(1,1). (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/1644140
- author
- Almbladh, Per
- supervisor
- organization
- year
- 2005
- type
- M2 - Bachelor Degree
- subject
- keywords
- economic theory, econometrics, Economics, nonlinear GARCH, smooth transition, adaptive exponential smoothing, Volatility forecasting, economic systems, economic policy, Nationalekonomi, ekonometri, ekonomisk teori, ekonomiska system, ekonomisk politik
- language
- English
- id
- 1644140
- date added to LUP
- 2005-04-21 00:00:00
- date last changed
- 2012-01-10 09:52:10
@misc{1644140,
abstract = {{Different models for predicting volatility in financial returns have been proposed during the years. This paper focuses on two different ways of volatility modeling, generalized autoregressive conditional heteroskedasticity (GARCH) models and exponential smoothing models. In a previous study 2004, James W. Taylor introduced an adaptive version of the traditionally exponential smoothing method with transition variables in order to capture the well known asymmetry of stock return volatility, the leverage effect. Like non-linear GARCH models, the sign and size of past shocks are used as transition variables to allow smoothing parameters to adapt to changes of the time series characteristics over a period of time. This paper compares the yet common GARCH(1,1) and two asymmetric GARCH models with four smoothing methods using this new adaptive approach. The performance of the models are evaluated by making volatility forecasts of seven stock indices from different countries. However, the new adaptive smoothing methods are not found to be significantly better than traditional exponential smoothing, neither did they outperform GARCH(1,1).}},
author = {{Almbladh, Per}},
language = {{eng}},
note = {{Student Paper}},
title = {{Volatility forecasting using adaptive exponential smoothing versus GARCH modeling}},
year = {{2005}},
}