Abstract:
During review of the trading book in the year 2013, the Basel Committee on Banking
and Supervision proposed movement from 99% Value at Risk to 97.5% Expected
Shortfall with a horizon of one day in quantifying market risk for banks. However, the
accord allowed financial institutions to use internal models in forecasting their risk
levels. In this regard, most risk managers prefer Value at Risk due to its simplicity and
intuitive interpretation. This is despite the many shortcomings of Value at Risk
evidenced in literature. This study looks at estimation of Conditional Value at Risk using
conditional extreme quantile autoregression and how the resulting estimator is applied
in evaluating financial risk in banks and other financial institutions. Performance of the
estimator is compared against two other values at risk estimators using Root Mean
Squared Error. Results from the study indicate that the proposed Conditional Value at
Risk estimator is consistent and leads to more accurate risk estimates.
