Forecasting Expected Shortfall: An Extreme Value Approach
(2013) MASY01 20131Mathematical Statistics
- Abstract (Swedish)
- We compare estimates of Value at Risk and Expected Shortfall from AR(1)-GARCH(1,1)-type models (standard GARCH, GJR-GARCH, Component GARCH), to estimates produced using the Peak Over Threshold method on the residuals of these models. We find that the conditional volatility model matters less than the choice of distribution for the innovations in the loss process, for which we compare the normal and the t-distribution. The Peak Over Threshold estimates are found to improve upon the estimates of the original models, particularly in the case of normally distributed innovations.
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/3735573
- author
- Kjellson, Benjamin
- supervisor
- organization
- course
- MASY01 20131
- year
- 2013
- type
- M2 - Bachelor Degree
- subject
- language
- English
- id
- 3735573
- date added to LUP
- 2013-05-06 14:44:14
- date last changed
- 2013-05-07 14:35:33
@misc{3735573, abstract = {{We compare estimates of Value at Risk and Expected Shortfall from AR(1)-GARCH(1,1)-type models (standard GARCH, GJR-GARCH, Component GARCH), to estimates produced using the Peak Over Threshold method on the residuals of these models. We find that the conditional volatility model matters less than the choice of distribution for the innovations in the loss process, for which we compare the normal and the t-distribution. The Peak Over Threshold estimates are found to improve upon the estimates of the original models, particularly in the case of normally distributed innovations.}}, author = {{Kjellson, Benjamin}}, language = {{eng}}, note = {{Student Paper}}, title = {{Forecasting Expected Shortfall: An Extreme Value Approach}}, year = {{2013}}, }