Study On Economic Capital Measure Of Insurance Group With Copula-Var Method

This paper is conducted to build up an integrated-risk measurement model of economic capital, according to the growing tendency that quantities of comprehensive financial institutions are arising. As a coalition of bank, insurance and securities, the financial conglomerates should also build up a kind of standard which can be the base of capital management and risk quantification, on the condition that the banking sectors gradually establish a risk management system of which the economic capital is the core. Under the separate supervision policy background, how to make their own to follow this unified standard to quantize risk and manage capital is a focal point of the operation of China’s financial conglomerates. Similarly, in the process of risk management session of insurance group, risk quantification is the basis for risk management and capital allocation, and risk management and capital allocation should also follow some capital constraints, which are considered to meet the return on capital required to achieve business value. In other words, the aggregation of capital, risk and the value of company need to find a balance in risk management, and the economic capital is the best carrier to be able to reach this balance.Economic capital represents the amount of capital which is to take to withstand the risk of unexpected loss of financial groups under certain confidence level in a given period. Economic capital plays a key role in performance evaluation, management decision-making and risk management. In addition, economic capital has become an important tool to assess the risk-adjusted capital adequacy by regulators.For construction of the Group’s economic capital measurement model, considering the Group’s risk diversification effect, this paper adjusts the traditional model in which the total economic capital is simply the summary of the various risks, and builds a new econometric model with Copula-Var method considering the correlation between the various risks. Dependency structure of each type of risk factors is very important in determining economic capital model. One way to measure the dependence structure is Copula function, which broke through the limitations of traditional linear correlation coefficient, it can effectively measure the dependency between risks considering the tail characteristics (in extreme cases). Based on the measurement of VaR of risk portfolio with Copula function, market risk, credit risk and operational risk were measured in order to determine the Ping An Insurance Group’s economic capital. Market risk is mainly reflected in the volatility of external factors, such as interest rates, exchange rates, stock price volatility, the results of empirical analysis in this paper suggests that the fluctuations explanatory variables respond to the market risk yields is realistic, that is to say, interest rates, stock price volatility has a positive impact on the investment income of financial position, exchange rate has a negative impact. Credit risk regression model is established by the relationship between loan yields and credit spreads, the empirical results show that there is a negative correlation between them.In this paper, it uses an elimination process to measure operational risk, excluding the effects of credit risk and market risk factors on Group’s ROE, the residual part is the Group’s operational risk. This measurement method solves the lack of loss data, but cannot accurately reflect the actual operational risk, there are obvious shortcomings compared to the "bottom-up" approach with loss data.Through the three risk regression models, on one hand, by considering the impact of risk factors on risk yields, the results of estimates is more realistic; on the other hand, the sample of risk yields is expanded, which makes a more efficient result for the marginal distribution fitting.After fitting the risk of marginal distribution, by selecting the dependency structure between different risk, portfolio risk VaR values are measured. As the form of empirical distribution of single risk is uncertain, before the simulation, the specific distribution form of risks is determined by using Bowman method. The measurement of VaR adopts a Monte Carlo simulation method, selected three Archimedean Copula functions, which are Gumbel Copula, Clayton Copula and Frank Copula. The results show different VaR values under different Copula functions, VaR of total risk get bigger if the function represented on a higher degree of dependency of the tail. The results also show a large impact of operational risk’s weights on VaR of total risk. In addition, by comparing VaR with simple summary method and VaR with integrated method, it shows that the risk diversification effects of financial groups does exist, it will affect the valuation of the economic capital, under normal circumstances, the integrated VaR is less than the simple added VaR, so estimating economic capital with integrated method is better for capital management in the insurance group.Based on the model building and empirical analysis, the paper also proposes to establish the economic capital measurement system of insurance group from three angles, consisting of database, verification systems and economic capital allocation.Since the data are not readily available, this article is flawed on the single risk measurement. Firstly, the risk regression models cannot reflect the impact of all the risk factor, secondly, the calculation of operational risk is not accuracy enough with the choice of the method of residual analysis in comparison to the method which uses actual data. In addition, the modeling and empirical analysis is based on a static Copula function, dynamic correlation between risks is not considered.However, this paper has innovations in three aspects:integrated measurement methods, estimation methods and empirical analysis. First, this paper introduces the Copula function as a new measurement method of dependency structure, which can better reflect the true relevant circumstances of risks than the simple linear dependence structure method, and builds a3D Copula function model based on the Archimedes Copula function which is easily extended to the case of N per characteristics. In addition, for fitting the marginal distribution, we use kernel density estimation method, for estimating the parameters of Copula function, we use the semi-parametric estimation method. Furthermore, for the calculation of VaR, this paper uses a Monte Carlo simulation method, based on a number of scenarios simulated; Finally, for empirical analysis, we choose to Ping An Insurance Group as the research object, fill the gaps in the field of empirical analysis of economic capital of Insurance Group, and explains the characteristic of the joint distribution of the Group’s overall risk by comparing the empirical results with three kinds of different Copula functions.