Value-at-Risk : Historisk simulering som konkurrenskraftig beräkningsmodell

Sammanfattning: Value-at-Risk (VaR) is among financial institutions a commonly used tool for measuring market risk. Several methods to calculate VaR exists and different implementations often results in different VaR forecasts. An interesting implementation is historical simulation, and the purpose of this thesis is to examine whether historical simulation with dynamic volatility updating is useful as a model to calculate VaR and how this differs in regard to type of asset or instrument. To carry out the investigation six different models are implemented, which then are tested for statistical accuracy through Christoffersens test. We find that incorporation of volatility updating into the historical simulation method in many cases improves the model. The model also generates good results compared to other commonly used models, especially if the volatility is predicted through a GARCH(1,1) updating scheme.