Value at Risk revolutionized risk management ever since J.P Morgan introduced the concept in their RiskMetrics in the 1980s. VaR has changed and improved through the years and today it is an acknowledged tool in risk management and used daily in many financial institutions.
Value at Risk measures an estimate of the experienced risk in monetary terms under a given confidence interval and time period, which makes it easy to understand and communicate to people without special financial knowledge.
The parametric Value at Risk, analyzed in the thesis, is one of the different methods to estimate Value at Risk. It is the standard model of the Basel III recommendations to financial institutions. The parametric VaR model uses volatility and correlations of assets or portfolios, under the normality assumption, to estimate the VaR value.
The thesis found that the normality assumption was rejected for all 11 assets used in the VaR calculations. That broke one the models essential assumptions and the estimates was therefore analyzed with that in mind.
The analysis revealed that the parametric VaR estimates had a failure rate above the expected 1% given the chosen 99% confidence interval. It could therefore be concluded that the rejection of the normality assumption had worsened the models performance. The crisis periods furthermore worsened this failure rate.
In the work with portfolio compositions a clear trend in preferred assets under different market conditions was shown. Stocks was preferable in calm periods, while bonds were prefered in crisis periods. The commodities were used to control the portfolio volatility due to their low correlation with paper assets.
All in all the parametric VaR model works well as a risk management tool. It is easy to implement and the results are easy to communicate. The model has a weakness in its normality assumption, which rarely are true in real scenarios, but the estimates are still useful if you use them with caution.
|Educations||MSc in Finance and Accounting, (Graduate Programme) Final Thesis|
|Number of pages||91|