The Basel Accord altered regulatory environment by establishing international minimum capital guidelines that linked banks’ capital requirements to their credit risk exposures. More recently, the Basel Supervisory Committee(BSC) extended the 1988 Basel Accord to include risk based capital requirement for the market risks in banking trading accounts.
At present, the BSC market risk capital standards consist of two alternative approaches for setting regulatory capital requirements for trading positions. One is the Standard Approach and the other is Internal Models Approach. Under the internal models approach, a bank uses its proprietary risk measurement models to estimate a measure of its trading account’s market risk exposure. The Internal Models Approach should be approved by regulators.
The Internal Models Approach will be implement on Dec.31.2001 in Korea. The VaR estimates are used to determine the banks’ market risk capital requirements. This development clearly indicates the importance of evaluating the accuracy of VaR estimates from a regulatory perspective.
In this paper, three methods for evaluating VaR estimates are discussed. The binomial method, currently the quantitative standard in the market risk amendment, and the interval forecast method are both based on a hypothesis-testing framework and are used to test the null hypothesis that the reported VaR estimates are “acceptably accurate,” where accuracy is defined by the test conducted. The power of this test can be low if small sample size such as a year. This result does not negate their usefulness, but it does indicate that the inference drawn from them should be questioned and examined more carefully for regulatory purpose.
The loss function method is based on assigning numerical scores to the performance of the VaR estimates under a loss function that reflects the concerns of the regulators. Furthermore, it allows the evaluation to be tailored to specific interests that regulators may have, such as the magnitude of the observed exceptions. According to capital can only provide protection against unexpected losses, the magnitude of loss should be considered in calculating the VaR accuracy. Since loss function provide complementary information, especially lower tail, they should be useful in regulatory evaluation of VaR estimates.
The Pre-Commitment Approach can use the bank’s market market risk capital charge. Also the VaR estimates underlying banks’ capital charges will be based on a uniform set of prudential parameters and will accurately reflect the assumptions and specifications of each banks’ internal model.
I suggest that banks make public the results of their backtests to gain information about the performance and accuracy of their models over time and loss function can be used to evaluate on internal model approach for market risk measurement.