Advanced

Volatility Forecasting In the Nordic Stock Market

Hummel, Niklas LU (2015) NEKH01 20142
Department of Economics
Abstract
This paper studies volatility prediction on OMX Stockholm 30, OMX Helsinki 25 and OMX Nordic 40. The models used are a historical variance model, an exponentially weighted moving average model and three models from the GARCH family. These are GARCH(1,1), EGARCH(1,1) and GJR(1,1), with normal and t-distribution respectively. The volatility for 2008-2013 is forecasted with a rolling window technique using historical data from 2002-2013. The models are ranked based on forecasting accuracy. The difference in accuracy is then translated into an average volatility forecasting error reduction. The financial crisis at the end of 2008 is studied separately, again ranking the models and comparing their relative forecasting ability. For the entire... (More)
This paper studies volatility prediction on OMX Stockholm 30, OMX Helsinki 25 and OMX Nordic 40. The models used are a historical variance model, an exponentially weighted moving average model and three models from the GARCH family. These are GARCH(1,1), EGARCH(1,1) and GJR(1,1), with normal and t-distribution respectively. The volatility for 2008-2013 is forecasted with a rolling window technique using historical data from 2002-2013. The models are ranked based on forecasting accuracy. The difference in accuracy is then translated into an average volatility forecasting error reduction. The financial crisis at the end of 2008 is studied separately, again ranking the models and comparing their relative forecasting ability. For the entire period I find that EGARCH and GJR are the superior models, followed by GARCH and EWMA, with the historical variance model yielding the greatest loss. Student’s t-distribution compared to normal distribution yields varying results for the entire period. For the crisis, GJR is the best model in all of the markets, while standard GARCH is the worst performing of all the models. t-distribution yields a substantial forecasting improvement over normal distribution for the crisis. The relative volatility forecasting error between the best and worst of the GARCH models are several times larger during the crisis than during the entire period for Nordic and Stockholm. (Less)
Please use this url to cite or link to this publication:
author
Hummel, Niklas LU
supervisor
organization
course
NEKH01 20142
year
type
M2 - Bachelor Degree
subject
keywords
GARCH, volatility, forecasting
language
English
id
5050030
date added to LUP
2015-02-19 14:06:08
date last changed
2015-02-19 14:06:08
@misc{5050030,
  abstract     = {This paper studies volatility prediction on OMX Stockholm 30, OMX Helsinki 25 and OMX Nordic 40. The models used are a historical variance model, an exponentially weighted moving average model and three models from the GARCH family. These are GARCH(1,1), EGARCH(1,1) and GJR(1,1), with normal and t-distribution respectively. The volatility for 2008-2013 is forecasted with a rolling window technique using historical data from 2002-2013. The models are ranked based on forecasting accuracy. The difference in accuracy is then translated into an average volatility forecasting error reduction. The financial crisis at the end of 2008 is studied separately, again ranking the models and comparing their relative forecasting ability. For the entire period I find that EGARCH and GJR are the superior models, followed by GARCH and EWMA, with the historical variance model yielding the greatest loss. Student’s t-distribution compared to normal distribution yields varying results for the entire period. For the crisis, GJR is the best model in all of the markets, while standard GARCH is the worst performing of all the models. t-distribution yields a substantial forecasting improvement over normal distribution for the crisis. The relative volatility forecasting error between the best and worst of the GARCH models are several times larger during the crisis than during the entire period for Nordic and Stockholm.},
  author       = {Hummel, Niklas},
  keyword      = {GARCH,volatility,forecasting},
  language     = {eng},
  note         = {Student Paper},
  title        = {Volatility Forecasting In the Nordic Stock Market},
  year         = {2015},
}