The shortcomings of KMV model in practical application are as follows: first, it focuses on default prediction, ignoring the change of enterprise credit rating, and is only suitable for evaluating the risk of credit assets (basically just loans) directly related to the value of enterprise assets; Secondly, the model is suitable for credit risk assessment of listed companies. Because China's stock market is not an efficient market, the stock price of listed companies often deviates from the actual value of the company, and the asset value of enterprises, especially state-owned enterprises, cannot be fully reflected in the stock market value, thus affecting the accuracy of model prediction. However, this model can be applied to the risk management of credit assets of multinational groups. Most of the credit assets of multinational enterprises are guaranteed by the parent company, and the stock market in the country where the parent company is located is relatively mature and effective. Thirdly, the model is based on the assumption that the asset value obeys the normal distribution, which is inconsistent with the reality, and the model cannot distinguish different types of long-term debts. This model has two advantages: first, it belongs to MTM(market to market) model, and the VaR of credit risk is calculated based on it, which is basically consistent with the business philosophy of state-owned commercial banks; Second, the model introduces the concept of portfolio management into the field of credit risk management for the first time, which is suitable for risk measurement of credit asset portfolios such as commercial credit, bonds, loans, loan commitments, letters of credit, market instruments (swaps, forwards, etc.). ).
The limitations of this model are:
First of all, the evaluation of credit risk by this model depends largely on the change of the borrower's credit rating. Under the existing credit environment in China, the probability of a large number of losses may be high.
Secondly, the model assumes that the probability of credit grade transfer is a stable Markov process, but in practice, credit grade transfer has a high correlation with the past transfer results.
Thirdly, the model assumes that the risk-free interest rate is determined in advance, and China's bond market is underdeveloped, and a reasonable base interest rate has not yet been formed. The base interest rate is an important factor in calculating the present value of loans.
Fourthly, at present, there is no objective and authoritative credit rating company in China, and there is no ready-made data on the conversion probability of enterprise credit rating and the default recovery rate of enterprises with different credit ratings. In the historical loan database of commercial banks, the probability of enterprises with a certain credit rating changing to another credit rating in different periods may be different, and the average default recovery rate of enterprises with a certain credit rating in different periods may also be different. The average value of transition probability and default recovery rate in these different periods constitutes a chaotic time series. If it is assumed that there is no big fluctuation in macroeconomic factors, chaotic time series can be used to predict the short-term future credit rating conversion probability matrix and the average default recovery rate of enterprises. With these data, state-owned commercial banks can use credit measurement model to quantify and manage credit risk.
Fifth, the model needs high-quality staff who can do a good job in credit rating evaluation in practical application. In addition, because the model adopts Monte Carlo simulation, the calculation amount is large. With the existing computer network system of state-owned commercial banks, it takes several hours or even ten hours to calculate the VAR value each time, which may not meet the needs of business development sometimes. The main advantages of this model are: it is easy to calculate the expected loss and volatility of a single bond and bond portfolio by using the mortality table, especially the calculation of bond portfolio is very convenient; The death model is a statistical model from a large number of samples, so fewer parameters are used. The main shortcomings of this model are: (1) the influence of the correlation of different bonds on the calculation results is not considered; Regardless of the impact of macroeconomic environment on mortality, it is necessary to update the mortality table irregularly; Data updating and calculation are very heavy; Cannot handle nonlinear products, such as options and foreign currency swaps.
Significance of credit measurement model
As the framework of the New Basel Accord, the significance of credit measurement model lies in determining the risk level of banks. Reasonable pricing of various financial products such as loans; Rational allocation of bank capital to resist various risks.
Take the risk model based on VaR as an example to illustrate the positive significance of the risk measurement model under the framework of the New Basel Convention.
In 200 1 year, the Basel Committee issued the new Basel Capital Accord (hereinafter referred to as the new Basel Accord) to replace the old Basel Accord. Under this framework, the risks faced by commercial banks are divided into three categories: credit risk, market risk and operational risk.
The application of VaR in risk management of commercial banks begins with the supervision of market risks. Traditional market risk management techniques can be divided into sensitivity analysis and volatility analysis, but these two methods have obvious defects in accuracy, dependence and comprehensiveness. As Jorion pointed out, VaR method uses standardized statistical techniques to comprehensively measure market risk, which makes up for the defects of sensitivity analysis and volatility analysis and promotes market risk management technology to a new height. The Basel Committee also made it clear that the VaR method combined with the internal model method is used to measure the market risk faced by banks.
Credit risk is the most important risk faced by commercial banks. Due to some characteristics of credit risk itself, it is technically difficult to measure it with VaR. However, with the development of quantitative technology, a new generation of financial engineers have established some credit risk measurement models that are more violent than VaR technology by using new modeling techniques and analysis methods. Among them, the famous methods are CreditVaR series proposed by CIBC and CreditMetrics proposed by J.P.Mrgan
Among the risks of commercial banks, operational risk has always lacked a clear definition and sufficient attention. An important modification in the New Basel Accord is to bring operational risk into the calculation and supervision framework of venture capital. There are many methods to calculate the operational risk capital box in the new Basel Accord, among which the more complicated loss allocation method needs to use the VaR method to determine the operational risk capital.