关键词: 实时数据, MIDAS模型, 经济预测

Abstract: There exist drawbacks of lagging information and noise information in Current-vintage data in empirical research. Application of real-time macroeconomic data can effectively eliminate bias caused by the lagging information and noise information. MIDAS model will estimate low-frequency key economic data and high-frequency data together and solve the problem of information loss of high frequency-data. In this paper, we use M-MIDAS-DL model and the quarterly real-time GDP data to establish a GDP forecasting model. The results show that using real-time data and combination forecast method can timely and accurately forecast the decline and rebound movement in China’s economic growth since 2008, which is a effective means of economic forecasting and would play a better role in early warning.

Key words: Real-time Data, MIDAS Model, Economic Forecasting