This project will develop procedures for computer simulation in risk measurement. The procedures will apply to simulating coherent risk measures, which are an advanced class of risk measures that have advantages over older measures that are widely in use, such as standard deviation and value-at-risk. The procedures will be precise, computationally efficient, adaptive to the problem at hand, and robust. The research will enhance existing procedures for simulating a certain class of coherent risk measures, by making them more adaptive and robust. It will also produce new, more precise procedures for simulating one particular coherent risk measure known as conditional tail expectation, which is a risk management standard for some insurance policies that are linked to the stock market. Tests of the computational efficiency and statistical validity of the procedures will be conducted on examples of financial portfolio management.
This work is expected to contribute to improvements in risk management by banks, investment firms, and insurers. The availability of good computer simulation procedures will make it more attractive for risk managers to adopt state-of-the-art risk measures. Coherent risk measures, unlike standard deviation and value-at-risk, can provide a clear quantification of the magnitude of the largest potential losses. The research has the potential to change risk management practice in a way that reduces the likelihood of bank failures, insolvency of insurers, and catastrophic disruption of financial markets.
