As the opening of capital markets and the innovation of financial markets accelerate, the volatility of risk factors such as interest rate, exchange rate, equity price and commodity price has greatly increased. Recently banks tend to take aggressively financial risks differently from the hedging through derivatives transactions in the past.
Banks and regulators now realize that it is essential to the stability and the soundness of banking system to precisely understand risk profiles and measure sources of market risk.
This research consists of the following subjects.
○ the outlook of the standard model by Basle Committee proposals and the internal model(or VAR model) by advanced banks
○ survey of three estimation techniques for Value-at-Risk (Delta-Normal, Historical Simulation and Structured Monte Carlo Method)
○ the example of VAR estimation and the analysis of the estimated VAR,
○ evaluation on VAR models,
○ recommendation on bank supervision for banks' risk management
Specifically, Gauss program, typical statistical software package, has been adopted and hypothetical portfolio of a bank in Korea has been constructed for VAR model design. In this research, capital requirements by VAR model appeared to be much compared with those by standard model reflecting the unripeness of our capital and foreign exchange market. In other words, risk diversification effect within and between portfolio seems very low, which means that the incentive of using VAR models doesn't come out so strong. And in its essence, the result of VAR estimation could be very divergent according to the estimation technique, confidence level, portfolio holding period and the estimation method of correlation.
Notwithstanding, VAR models still have various utilities as an information reporting tool, resource allocation tool and performance evaluation tool. Considering the increasing trend of banks' using VAR models, bank regulators need to develop the evaluation techniques on the accuracy of VAR models. Recently, regulators in USA have proposed the method of VAR probability forecasts evaluation which can be tailored to their interests. Besides there are three hypothesis testing method - VAR estimates based on the binomial distribution, VAR interval forecasts, VAR distribution.
Meanwhile banks have to do their best in the set-up and well-functioning of their risk management system in that the mitigation of firewall and the evolution of merger & acquisition in financial sector are expected in the near future.
Our bank regulators will hurry on with the establishment of concrete supervision standards in respect to capital requirements, derivatives transactions, financial conglomerates, internal control system and early warning system on the unexpected emergency in the international context. Especially the development and application of VAR models by bank groups or common basic model can be studied in consideration of our circumstances under the initiative of Office of Bank Supervision.