Design and Application of Macro Economic Intelligent Prediction Decision Support System Based on Agent System

Abstract This paper closely combines Agent system and macroeconomic forecasting, and proposes an Agent-based decision support system for macroeconomic intelligent forecasting, which provides a strong support for decision making in macroeconomic forecasting. The mathematical model used in the system of this paper is mixed-frequency vector autoregressive model MF-VAR, and Kalman smoothing operator is used in the commonly used low-frequency quarterly variable and high-frequency monthly variable MF-VAR model for macroeconomic indicator forecasting. On the basis of MF-VAR model, dynamic factor model, Bayesian method estimation are introduced respectively, and MF-VAR model combining factor model and BMF-VAR model are further proposed. In the simulation experiments to test the comprehensive validity, the reliability of this paper’s system is always higher than the 98% level, the response time is short, and the decision support error rate is always lower than 1.2%, which is better than the other systems in the comparison. The system in this paper is applied to macroeconomic forecasting and short-term prediction before and after the Xin Guan epidemic in China. Before the epidemic, the MF-VAR model combined with the factor model can improve the accuracy of forecasting to a certain extent when it is not affected by major events. And after the epidemic and when the economy suffers from major shocks, the MF-VAR model combined with the factor model and the BMF-VAR model face a decrease in the forecasting accuracy of the indicators such as the export volume and the import volume, but the disadvantage is not obvious.