eThesis - School of Business electronic theses

» hse_ethesis_12049.pdf   size:2 MB (1526109)

Value-at-Risk has become one of the most popular risk measurement techniques in finance. However, VaR models are useful only if they predict future risks accurately. In order to evaluate the quality of the VaR estimates, the models should always be backtested with appropriate methods. Backtesting is a statistical procedure where actual profits and losses are systematically compared to corresponding VaR estimates.

The main contribution of this thesis consists of empirical studies. The empirical part of the thesis is carried out in close cooperation with a Finnish institutional investor. The primary objective of the study is to examine the accuracy of a VaR model that is being used to calculate VaR figures in the company’s investment management unit. As a secondary objective the empirical research tries to figure out which backtests are the most reliable, and which tests are suitable for forthcoming model validation processes in the company.

The performance of the VaR model is measured by applying several different tests of unconditional coverage and conditional coverage. Three different portfolios (equities, bonds and equity options) with daily VaR estimates for one year time period are used in the backtesting process.

The results of the backtests provide some indication of potential problems within the system. Severe underestimation of risk is discovered, especially for equities and equity options. However, the turbulent market environment causes problems in the evaluation of the backtesting outcomes since VaR models are known to be accurate only under normal market conditions.