Forecasting China's current account with EEMD-AWNN ensemble learning approach

With the development of China's foreign trade and economic structure reform, it is necessary to explore the evolution of the balance of payments for the macroeconomic regulation, but China is faced with uncertainty of the current account due to the changeable international environment. Based on this background, The EEMD-AWNN model is proposed to predict the current account, including debit and credit. Traditional econometric model can give economic explanation, but it is difficult to adapt to the rapid change of domestic and international economic structure. In addition, the single-variable machine learning algorithm can overcome this problem, but it is not sensitive enough to the external economic variables. Therefore, this paper learns from TEI@I and proposes EEMD-AWNN model to predict the current account. In EEMD-AWNN, the empirical mode decomposition (EEMD) is first used to decompose the variables, and then required exogenous variables are added according to economic significance. Finally, the synthesized adaptive wavelet neural network (AWNN) is used to predict the current account, which obtains higher accuracy and better forecasting performance than econometric models. Based on the proposed model, the results showed that both debit and credit of the current account will grow. However, the debit growth rates of trade in goods and service trade will be higher than their credit growth rate. What's more, the surplus of commodity trade will narrow and travel trade deficit will increase. Therefore, China's current account will continue to run a surplus but the surplus will narrow in next two years.