Financial Globalization and Financial Deregulation Trend

In the background of financial globalization and financial deregulation trend, the frequency of capital flow and scale of the capital markets have risen sharply. Market risk, such as interest rate risk, exchange rate risk, equity risk, seem to be considered as an important factor for the financial institution, and financial risk management become the core of modern financial institution management. The VaR model can easily and clearly represent the various risks faced by asset portfolio, and it is widely used and developed in the international financial industry.

   The definition of VaR is, in normal market conditions and the given confidence level (confidence interval, usually 95% or 95%), held at a given period, a portfolio expected possible maximum loss. For example, the portfolio at the 99% confidence level in the next six months of VaR value is $100000, the probability of the loss of more than $100000 per day is l %, in other words, every 100 days, only one day that the loss will be more than $100000.

According to Basle Accord, banks could establish their own internal risk management models to meet capital adequacy requirements. Banks can use VaR to measure the market risk they face, and supervisors require banks to calculate capital requirements using an internal control model. However, capital adequacy requirements may directly affect the profitability of financial institutions. From a purely profit-driven perspective, financial institutions tend to choose the parameters of the lower VaR level of the internal control model, thus reducing capital reserves required. This would make supervisors suspicious of the reliability of the VaR. Based on such potential problem, Basle Accord requires banks to routinely test internal models. "Backtesting" is a commonly used method for evaluating a company's risk measurement model. It is to compare the results of actual transactions with the risk values generated by the model to confirm and verify the reliability of VaR risk measurement methods. Since VaR is only a statistical prediction model established by historical data or assumed statistical parameters and distributions, it is necessary to check whether the prediction of future risks is accurate and effective. The main method of inspection is to return the test. Backtesting in statistics refers to the process of entering actual data into a tested model, and then testing whether the predicted value of the model is the same as the actual result. For example, a VaR model to measure the results for the risk of a portfolio, the portfolio at the 99% confidence level in the next six months of VaR value is $100000, the probability of the loss of more than $100000 a day for l %. Backtesting the predictive value of VaR for the actual transaction data of past 100 days, if the number of days lost more than $100000 is not more than one day, then it illustrates the effectiveness of the model, if the loss of $100000 more than the number of days is two days or more, it is doubtful that the effectiveness of the model.