RMB Exchange Rate Risk Measurement Based On Expected Quantile Regression Model

In recent years,the global financial market has been development rapidly due to economic globalization,financial integration,financial innovation and the innovation of science and technlogy,and so on.Countries gradually began to realize the importance of financial risk management after Asian financial crisis,the US subprime mortgage crisis,etc.Financial regulators and investors have to be confronted with a problem that how to effectively manage and control financial risk.The exchange rate between different countries is an important form to transfer the financial crisis,the risk of exchange rate is divided into transaction risk and currency risk of foreign currency translation.It is important to avoid the risk through strengthening risk awareness,measuring risk accurately,identifying sufficiently.Currently,it is imperfect for foreign exchange market risk management and control system in our country,the financial crisis broke out frequently and alarmed to the institutions and enterprises on account of the collapse of many famous international financial institutions and businesses,thus,the risk of exchange rate markets is increasingly becoming the dominant factor and measuring risk accurately is the most critical factors for identification and risk prevention.This paper has established three different models of exchange rate risk measurement,and selected preferable measurement model by comparative analysis methods.Firstly,this paper establishing the GARCH model aim at the properties such as peak and thick tail of exchange rate yield sequence,then established nine kinds of models of GARCH(1,1),TGARCH(1,1)and EGARCH(1,1),etc.By compare analysis random variables which were normally distributed,t distribution and GED distribution respectively,finally,this paper choose the model of EGARCH(1,1)-GED,and calculated VaR by estimated value of the conditional standard deviation.Although EGARCH-GED model can accuratly discribe peak and thick tail characteristics of exchange rate return series,it can't measure accuratlly the risk of abnormal loss point data.Secondly,Taylor(1999)proposed quantile regression theory because of the defect of EGARCH-GED model in measuring the defects of exchange rate risk yield,this paper based on the quantile regression model calculation VaR method to measure comprehensivly by selecting the different sub-sites on the risk of exchange rate yields and make up for the lack of the model EGARCH-GED.While quantile regression model having better robustness and the advantage that needn't to make strictassumptions for variable distribution,however,the Va R value that based on the quantile regression model can't accurately reflect and insensitive to yield numerical size of the reaction sequence of the tail when the probability of extreme losses are equal while the value is not equal.Thirdly,this article adopt a method of calculating VaR value based on expectile quantile regression model to compensate the defects of EGARCH-GED model and quantile regression defect in the exchange rate risk measure.Expectile quantile regression quantile regression model are more sensitive than quantile regression models,VaR based on expectile quantile regression model can reflect obviously when the probability of extreme losses on tail are same while the value is not equal.In addition,expectile quantile regression model has an advantages of the consistency of risk measurement compared with EGARCH-GED model and quantile regression model.Finally,this paper analyze the superiority-inferiority of the three models in measure the pros and cons of exchange rate risk by the empirical analysis and Kupiec failure rate test methods.The empirical results show that the three models that calculate VaR are all through the test failure rate and it is significant under the condition of the confidence level of 95%,both failure rate and LR statistic aspect,expectile measure quantile regression model was better than EGARCH-GED model and quantile regression model,the abnormal loss point data have better resistance to corrosion and better interval coverage for the real yield.At the same time,expectile quantile regression model show that EVaR have a broad space in financial risk measurement application areas,which provide a new approach to measure and manage risks for managers and investors in our currency market.