Analysis of RMB Exchange Rate Fluctuation Using ARMA-GARCH Model

Abstract

This paper selects central parity rate of RMB against US dollar on each trading day from January 2, 1995 to December 31, 2020 as empirical research data, takes the logarithm and then differentiates the data to obtain return series for a stable RMB exchange rate against US dollar. Descriptive statistical analysis is performed on the series, and it is concluded that distribution of the serious had obvious characteristics of "leptokurtosis and heavy tails”, compared with normal distribution. Through test of stationarity, autocorrelation and heteroscedasticity, it is confirmed that there is a phenomenon of fluctuation clustering in return series, so it is reasonable and feasible to use GARCH model to fit the return model of RMB exchange rate against US dollar. In this paper, ARMA model is established to obtain the mean equation to eliminate autocorrelation of the series. At the same time, TGARCH and EGARCH models are used to fit the asymmetric effects in return series. By comparison, it is concluded that the model with the best fitting effect is ARMA (2,1)-EGARCH (1,2) model. At the same time, it is concluded that RMB depreciation will cause a larger fluctuation in RMB exchange rate compared with RMB appreciation, which also indicates that RMB exchange rate has characteristics of volatility clustering and asymmetric effects. The feasibility of model is tested by in-sample prediction of results of the constructed EGARCH model. Finally, based on volatility of RMB exchange rate, policy recommendations such as controlling rhythm of exchange rate changes, improving RMB exchange rate system and exchange rate derivatives market are put forward.