机械工业宏观经济短期预测模型——时间序列分解方法研究

THE MODEL FOR SHORT-TERM MACRO-ECONOMIC OF MECHANICAL INDUSTRY——RESEARCH OF TIME SERIES DECOMPOSITION METHOD

本文运用时间序列分解技术,在CensusⅡ方法的基础上,通过对机械工业宏观经济指标月度序列变化趋势的分析,建立了TSD短期预测模型,其要点是:(1)根据我国节假日特点,设计了新的基准月调整公式;(2)选择适当的移动平均方式作为因子分解工具;(3)根据偏差信号对季节指数进行动态调整。实践表明,TSD模型对短期经济指标滚动预测较为有效。此外本文还从预测精度、预测成本和预测说明三方面对TSD模型进行了评价。

The time series decomposition method is used in this paper. The TSD short-term forecasting model has been built in the paper based on Census Ⅱ method and on the analysis of the dynamical trend of monthly series macroeconomics of mechanical industry. The main points include: (1)Aecording to the characteristics of the holidays in China an new adjusting formula of monthly pattern has been designed. (2)The suitable smoothing methods have been chosen as the tools of the dccompounding factors. (3)According to the error signals the seasonal indexes are adjusted dynamically. In practice the TSD model has been proved to be valid for the short term economic indexes forecasting. In addition, the TSD model has been appraised from the forecasting accuracy, forecasting costs and Forecasting explanation.