Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Estimation of Value at Risk using conditional copula: The choice of copula

Puusaari, Joakim LU (2010) NEKM02 20102
Department of Economics
Abstract
This study examines the impact of different types of copula on the performance of weekly VaR estimation for a two-asset portfolio composed of stock indices using a conditional copula methodology. The aim is to find out if the choice of copula is important since Fantazzini (2008) found it not to be. The Gaussian copula, the Student’s t copula and the SJC copula was included in this study based on the recommendations in previous research and their type of tail dependence.
In order to investigate the impact of copulas full parametric models were made where the only difference was in the copula used. One-week-ahead out-of-sample VaR forecasts at the 95% and 99% confidence level was estimated for the portfolio by using these models in a moving... (More)
This study examines the impact of different types of copula on the performance of weekly VaR estimation for a two-asset portfolio composed of stock indices using a conditional copula methodology. The aim is to find out if the choice of copula is important since Fantazzini (2008) found it not to be. The Gaussian copula, the Student’s t copula and the SJC copula was included in this study based on the recommendations in previous research and their type of tail dependence.
In order to investigate the impact of copulas full parametric models were made where the only difference was in the copula used. One-week-ahead out-of-sample VaR forecasts at the 95% and 99% confidence level was estimated for the portfolio by using these models in a moving window technique together with Monte Carlo simulation.
The VaR estimation results were similar and suggested that Fantazzini (2008) was correct. However, a study of the tail dependence modelled by the Student’s t copula showed that it was not negligible. Furthermore, a comparison of the tail dependence modelled by the Student’s t copula and the SJC copula indicated that the imposed symmetric tail dependence in the Student’s t copula was not natural.
The conclusion is therefore that the choice of copula is important since choosing one of the elliptical copulas would entail an increased model risk and lead to an overall increased risk of error in VaR estimates. (Less)
Please use this url to cite or link to this publication:
author
Puusaari, Joakim LU
supervisor
organization
course
NEKM02 20102
year
type
H2 - Master's Degree (Two Years)
subject
keywords
portfolio, Value at risk, copula, risk management, stock market
language
English
id
1700580
date added to LUP
2010-11-15 08:39:26
date last changed
2010-11-15 08:39:26
@misc{1700580,
  abstract     = {{This study examines the impact of different types of copula on the performance of weekly VaR estimation for a two-asset portfolio composed of stock indices using a conditional copula methodology. The aim is to find out if the choice of copula is important since Fantazzini (2008) found it not to be. The Gaussian copula, the Student’s t copula and the SJC copula was included in this study based on the recommendations in previous research and their type of tail dependence.
In order to investigate the impact of copulas full parametric models were made where the only difference was in the copula used. One-week-ahead out-of-sample VaR forecasts at the 95% and 99% confidence level was estimated for the portfolio by using these models in a moving window technique together with Monte Carlo simulation.
The VaR estimation results were similar and suggested that Fantazzini (2008) was correct. However, a study of the tail dependence modelled by the Student’s t copula showed that it was not negligible. Furthermore, a comparison of the tail dependence modelled by the Student’s t copula and the SJC copula indicated that the imposed symmetric tail dependence in the Student’s t copula was not natural.
The conclusion is therefore that the choice of copula is important since choosing one of the elliptical copulas would entail an increased model risk and lead to an overall increased risk of error in VaR estimates.}},
  author       = {{Puusaari, Joakim}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Estimation of Value at Risk using conditional copula: The choice of copula}},
  year         = {{2010}},
}