The Measure Of RMB Exchange Rate Risk Based On Extreme Value Theory And Dynamic Extreme Value Theory

After the Bretton Woods system of fixed exchange rates collapsed in 1973 and Jamaica Agreement of 1976 formally recognized the legality of the floating exchange rate system, major developed countries abandoned the original fixed exchange rate system and had a floating exchange rate system. Exchange rate btween major currencies fluctuated violently. Asian financial crisis made people realize the great damage caused by exchange rate risk. As a result, regulatory authorities and institutions affected by exchange rate payed more attention to exchange rate risk. After China's accession to WTO. the financial market will gradually be opened, China's foreign exchange market will face even greater risks. Moreover, China has started the well-supervisory floating rates system which was based on supply and demand in the market and adjusted by a basket of currencies, since July 21,2005. Henceforth RMB exchange rate fluctuates frequently, concerning foreign enterprises in China face more and more exchange rate risk. Effective exchange rate risk management has become an urgent task of related structures.Management of exchange rate risk is based on the measure of exchange rate risk. Currently, the most popular risk measure tool in the world is VaR (Value at Risk) model. Research on abroad is very mature. VaR (Value at Risk) has become the standard risk measure method of financial institutions. Some institutions in China also begin to use the VaR model to measure exchange rate risk.In China,only a few academicians research RMB exchange rate risk and the method of calculating VaR is relatively simple.Almost none of academicians use Extreme Value Theory(EVT) to research RMB exchange rate risk. Since the liberalization of foreign exchange rate market, RMB exchange rate has changed a lot,especially the exchange rate of Euro to RMB and the exchange rate of JPY to RMB.However,the traditional methods of calculating VaR are not fit for the events of non-normal fluctuation or extreme fluctuations. Therefore, the paper introduces the Generalized Pareto Distribution(GPD) in EVT to fit the series'tails to estimate the VaR and calculate ES value as supplementary July 25,2005 to March 31,2010 is the sample interval.The VaR values and ES values clearly show RMB exchange rate risk.Considering that many institutions adopt Historical Simulation Approach(HS),the paper uses Historical Simulation Approach to calculate VaR value and utilize Kupiec Back Testing to test the effect of HS and EVT. The results show that, the EVT model is superior to HS in dealing with the exchange rate of Euro to RMB data and the exchange rate of JPY to RMB data.But, EVT model and HS all cannot reflect the risk of the exchange rate of US to RMB and the exchange rate of HKD to RMB especially in the low confidence level because of their mild fluctuations since 2009. Extreme Value Theory calculating VaR value ignores the variability and clustering of financial asset return.To overcome this problem,the paper tries to introduce GARCH model which can reflect the variability and clustering of financial asset return,combining GARCH model with EVT namely GARCH-EVT to measure the exchange rate risk. At present,nobody uses GARCH-EVT to research RMB exchange rate risk.Considering the characteristics of financial time series and the tail distribution of the error term,the paper uses GARCH(1,1)-t-EVT model to measure risk.Then analyze the effect of GARCH(1,1)-t-EVT,EVT and HS using Kupiec Back Testing. The results show that these three methods are all fit for risk measure for the exchange rate of Euro to RMB return series and the exchange rate of JPY to RMB return series. GARCH(1,1)-t-EVT model can reflect the risk of these exchange rates accurately.However,for the exchange rate of US to RMB return series and the exchange rate of HKD to RMB return series,EVT model and HS cannot fit them well.They greatly overestimated the risk with a large risk measurement error.But GARCH(1,1)-t-EVT model reflects changes in the value of actual loss.Overall,in aspect of measuring RMB exchange rate risk,EVT model is better than HS, GARCH(1,1)-t-EVT is better than EVT model. VaR values calculated by GARCH(1,1)-t-EVT model have dynamic characteristics changing in good agreement with the actual loss.