Advanced

VaR for a portfolio of Swedish Index-bonds - An empiricial evaluation

Johansson, Magnus LU (2011) NEKM07 20111
Department of Economics
Abstract
Purpose: The purpose of this paper is to empirically evaluate the performance of seven different methods that are used when estimating Value-at-Risk for a portfolio of Swedish index-bonds with different maturities. As a supplementary objective, the paper tries to determine the history that one needs to account for when calculating VaR.

Methodology: In order to calculate the VaR, a portfolio consisting of five Swedish index-bonds is constructed. For this portfolio, one-day VaR-estimates are calculated using three different windows and seven methods; Basic historical simulation, Age-weighted historical simulation, Volatility-weighted historical simulation, Normal distribution, Student’s t–distribution, Asymmetric slope and Symmetric... (More)
Purpose: The purpose of this paper is to empirically evaluate the performance of seven different methods that are used when estimating Value-at-Risk for a portfolio of Swedish index-bonds with different maturities. As a supplementary objective, the paper tries to determine the history that one needs to account for when calculating VaR.

Methodology: In order to calculate the VaR, a portfolio consisting of five Swedish index-bonds is constructed. For this portfolio, one-day VaR-estimates are calculated using three different windows and seven methods; Basic historical simulation, Age-weighted historical simulation, Volatility-weighted historical simulation, Normal distribution, Student’s t–distribution, Asymmetric slope and Symmetric absolute value. The methods are evaluated using the Kupiec’s and Christoffersen’s test for unconditional and conditional coverage.

Conclusion: The empirical results show that neither method is able to exactly capture the amount of acceptable exceedances. Though, it is evident that several of the estimated methods are accepted according to the Basel Rules. Also, it can be seen that the volatility-weighted approach is the most suitable method to use when calculating VaR for a portfolio of Swedish Index-bonds during the investigated period. The efficient window of historical observations is deemed to be equal to one year. (Less)
Please use this url to cite or link to this publication:
author
Johansson, Magnus LU
supervisor
organization
course
NEKM07 20111
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Value-at-Risk, Conditional Autoregressive Value-at-Risk, Age-weighted historical simulation, Volatility-weighted historical simulation, GARCH (1, 1), Normal distribution, Student’s t-distribution, Christoffersen’s test
language
English
id
2167277
date added to LUP
2011-09-27 08:04:42
date last changed
2011-09-27 08:04:42
@misc{2167277,
  abstract     = {Purpose: The purpose of this paper is to empirically evaluate the performance of seven different methods that are used when estimating Value-at-Risk for a portfolio of Swedish index-bonds with different maturities. As a supplementary objective, the paper tries to determine the history that one needs to account for when calculating VaR.
 
Methodology: In order to calculate the VaR, a portfolio consisting of five Swedish index-bonds  is constructed. For this portfolio, one-day VaR-estimates are calculated using three different windows and seven methods; Basic historical simulation, Age-weighted historical simulation, Volatility-weighted historical simulation, Normal distribution, Student’s t–distribution, Asymmetric slope and Symmetric absolute value. The methods are evaluated using the Kupiec’s and Christoffersen’s test for unconditional and conditional coverage. 

Conclusion: The empirical results show that neither method is able to exactly capture the amount of acceptable exceedances. Though, it is evident that several of the estimated methods are accepted according to the Basel Rules. Also, it can be seen that the volatility-weighted approach is the most suitable method to use when calculating VaR for a portfolio of Swedish Index-bonds during the investigated period. The efficient window of historical observations is deemed to be equal to one year.},
  author       = {Johansson, Magnus},
  keyword      = {Value-at-Risk,Conditional Autoregressive Value-at-Risk,Age-weighted historical simulation,Volatility-weighted historical simulation,GARCH (1,1),Normal distribution,Student’s t-distribution,Christoffersen’s test},
  language     = {eng},
  note         = {Student Paper},
  title        = {VaR for a portfolio of Swedish Index-bonds - An empiricial evaluation},
  year         = {2011},
}