Risk Measurement Based On Asymmetric-Laplace-AGARCH Model

The rapid development of the global economy has caused the increasinglyfrequency of financial crises, and globalization of the economy then caused widerspread of the financial crisis. The considerably oscillation of global financial market,especially in the global financial crisis caused by the United States, the debt crisis ofUnited States, the Asian financial crisis, credit crisis in Greece, these all have greatlyincreased the market risk management in modern financial risk management status, andhave aroused the scholars and the financial risk supervision departments to attach greatimportance to such risk measure.VaR theory began in1990s, has gradually become the mainstream of financial riskmeasurement methods. The commonly methods used for VaR calculation are historicalsimulation method, parameter method, and Monte Carlo simulation. But because VaRdoes not consider the extreme cases of loss, and it is not the coherent risk measure, so itcan not meet the actual financial risk measurement. The Expected Shortfall (ES)introduced in this paper is a coherent risk measure method, combination of this twomeasures can not only valued the maximum loss under certain confidence level, but alsomeasured the extreme case of loss of size.According to the actual characteristics of financial market data in our country, thisarticle uses asymmetric Laplace distribution to fit the probability distribution of thefinancial data sequence. This method can be improved to be better meeting the fat-tailbiased characteristics. The author has given following calculation model of VaR and ESunder the conditions. Then this paper used AGARCH model to fit the volatility offinancial data, and put forward to asymmetric distribution of Laplace AGARCH modelby combining the probability distribution of the asymmetry and asymmetry of theAGARCH model together. This kinds of model parameter estimation method isintroduced in detail by using the maximum likelihood estimation method.Finally this article selects the stock data to carry out empirical analysis, andcomparatively analyzed two kinds of models of the measurement results to verify thepaper model. Specific results are as follows: asymmetrical Laplace distribution can beused in fitting the financial data series of return rate; It is feasible to forecast the returnseries volatility using AGARCH model; ES model can be better used in solve theproblem of risk assessment in extreme cases, and it was greatly improved the method of VaR tail risk measurement.