ABSTRACT
In this chapter, a forecasting approach based on the domestic economic situation is proposed. The proposed VECM-based ensemble learning approach includes the following three steps. First, data extraction. According to the existing literature, 16 macroeconomic variables are selected, including import, export and foreign exchange reserves, which are often selected to study the relationship between the exchange rate and the macroeconomic variables. Secondly, data selection. By integrating the Granger causality test, correlation test and grey relational analysis, we rank the correlation of RMB/USD with China’s 16 major macroeconomic variables, and then filter the three variables that have the highest degrees of relevance with the exchange rate to represent the domestic situation. Thirdly, data computing. Based on the domestic situation, VECM is used to forecast the central parity of RMB/USD. And multiple evaluation criteria are employed to comprehensively evaluate the forecasting performance of the proposed approach and benchmarks.
