我国季度GDP实时数据预测与评价

Forecasting and Assessment of China's Quarterly Real-time GDP

传统实证研究中使用的当期特定数据存在滞后信息和噪音信息缺陷,导致模型估计结果存在偏误。应用宏观经济实时数据可以有效的剔除造成模型偏误的滞后信息和噪音信息,得到更为准确的估计结果。MIDAS模型可将低频的关键经济数据与高频数据同时估计,较好的解决了应用一般模型存在的高频数据信息损失问题。本文应用M-MIDAS-DL模型与季度GDP实时数据建立我国季度GDP预测模型,实证表明,应用实时数据与组合预测方法,能及时准确预测出2008年以来中国经济增长率的下滑与反弹走势,能起到较好的提前预警作用,是当前较为有效的经济预测手段之一。

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.