Analyzing and Forecasting the Exchange Rate of USD/CNY

Abstract

In this study, based on the characteristics of the ARIMA models and ETS models, respectively, that the former focuses more on autocorrelation between data, while the latter focuses more on trends and seasonality of data sets. These two forms of models are used to forecast the USD/CNY exchange rate. This study used the monthly average USD/CNY exchange rate from January 2010 to June 2023, which taken from the website of the China Foreign Exchange Trade System (CFETS) , which data is provided by People's Bank of China, and used computer software to forecast and test the results using each model. Ultimately, it was found that the forecasts using the Autoregressive Integrated Moving Average (ARIMA) and Exponential Smoothing (ETS) models, were close to each other, with both showing a flattening trend over the long term. In the short term, ETS(M, Ad, N) forecasts an upward trend in the USD/CNY exchange rate while ARIMA(0,1,2) forecasts an almost flat trend. ARIMA (0,1,2) forecasts that the USD/CNY exchange rate shown the shape that finally stabilized at around 6.85, while ETS (M, Ad, N) forecasts around 7.6, and comparatively ARIMA model gives more reliable forecasts on the test set, and fit the training set better. This study compares the accuracy of the ETS model and the ARIMA model in fitting and subsequently forecasting the USD/CNY exchange rate for the period 2010-2023. Future research could build on this by discussing the impact of particular specific events and policies on the forecasts at this stage and experimenting with optimization.