Recently, with increasing volatility of foreign exchange rate, risk management becomes more and more important not only for multinational companies and individuals but also for central governments. This paper attempts to build an econometrics model so as to forecast and manage risks in foreign exchange market, especially during the eve of turbulent periods. By following McNeil and Frey’s (2000) two stage approach called conditional EVT to estimate dynamic VaR commonly used in stock and insurance markets, we extend it by applying a more general asymmetric ARMA-GARCH model to analyze daily foreign exchange dollar-denominated trading data from four countries of different development levels across Asia and Europe for a period of more than 10 years from January 03, 2005 to May 29, 2015, which is certainly representative of global markets. Conventionally, different kinds of backtesting methods are implemented ultimately to evaluate how well the model behaves. Inspiringly, test results show that by taking several specific characteristics (including fat-tails, asymmetry and long-range dependence) of the foreign exchange market return data into consideration, the violation ratio of out-of-sample data can be forecasted very well for both fixed and flexible foreign exchange regimes. Moreover, all of the violations are evenly distributed along the whole period which indicates another favorable property of our model. Meanwhile, we find evidence of asymmetry volatility in all of the studied foreign exchange markets even though the magnitudes of the most of them are weak.
El peso mexicano: la gestión de cobertura del riesgo cambiario mediante la teoría de los efectos olvidados