A Comparative Study of Euro-RMB Exchange Rate Value-at-Risk Calculation Methods --Based on GARCH, EWMA and quantile regression methods

Details

Supervisors

Leonardo Iania

Faculty

Louvain School of Management

Degree label

Master [120] en sciences de gestion, à finalité spécialisée

Abstract

With the implementation of the Belt and Road policy and the continuous development of EU-China relationship, China has become one of the most important trading partners of EU countries. Therefore, the measurement of volatility risk of Euro-RMB exchange rate is of great significance to the economic development of EU and China. VaR, an important tool for measuring risk of financial assets, is being used by a wide range of financial institutions. In this thesis, based on volatility characteristics of the Euro-RMB exchange rate yield, we will pre-process the data and prove that the yield has the modeling conditions through a series of tests. Then, based on the assumptions of normal distribution, student-t distribution, and GED distribution, the VaR of the EUR-RMB exchange rate return at different confidence levels is measured by the traditional GARCH model, EWMA model and quantile regression model, respectively. Finally, after obtaining the results of VaR measurement, a comparative analysis of the above VaR results by back testing using the failure rate test is performed to compare the accuracy of the three models for the VaR of the Euro-RMB exchange rate return and rank them. According to the results of the empirical study, the euro exchange rate return series has the typical characteristics of a financial time series which has sharp peaks and thick tails, showing a left-skewed trend, basically does not obey the normal distribution. In addition, the test results can be intuitively judged to be closer to the form of GED distribution. The results of the back test, from the accuracy of the three types of VaR measurement models for the VaR of the euro exchange rate return series, the quantile regression model performs the best, and has obvious advantages compared with the other two types of models at the 1%, 5% and 10% significance levels. In contrast, the EWMA model has a slight lead over the traditional GARCH model at high confidence levels, and the two models perform similarly at lower confidence levels. This also indicates that the quantile regression model has good results for the study of risk measures of the euro exchange rate and can be applied more frequently in the study of related fluctuations of exchange rate.