混频数据与中国宏观经济周期测定

Mixed Frequency Data and Measurement of China Macro Economic Cycles

笔者利用经数据处理后的3个月度指标、3个季度指标、5个半年度指标和4个年度指标,结合改进的混频动态因子模型和经济周期测定的虑子概率方差方法,对1992年1月至2017年8月间中国宏观经济周期进行了测定分析。分析发现:(1)混频动态因子模型支持利用多频度、多指标、非平衡数据进行分析;支持波动协同性弱的指标经调频处理后被选用,并能有效提取此类指标数据中的有效信息和排除无效信息;能够提取有效反映宏观经济波动一致性走势的不可观测动态因子。(2)概率方差法和CH准则法均测定宏观经济包含5个拐点、3个高速增长时期和3个低速增长时期。利用ROC曲线进行比较分析发现概率方差法的测定精准度更高。分析没有发现2011年9月以后宏观经济新的拐点,从而预测中国宏观经济还将处于低速增长时期。(3)利用实时数据检验了混频动态因子模型具有较好的稳健性。

The paper analyzes the macroeconomic cycles of China from January 1992 to August 2017 with the improved mixed frequency dynamic factor model(MF-DFM)and the probabilistic variance method of testing the economic period,adopting 3 monthly,3 quarterly,5 semi-annually,and 4 annually data that are processed.It is found that:(1)The MF-DFM can fit into the multiple frequencies,multiple indicators,and nonequilibrium data,extract the effective information and remove the invalid information from the data that have bad characteristics of the co-fluctuation,and extract the unobserved dynamic factors which effectively reflect the consistent trend of macroeconomic fluctuation.(2)Both the probabilistic variance method and the CH criterion method measure that the macroeconomic cycles of China includes five turning points,three high speed growth periods,and three low speed growth periods.The probabilistic variance method has higher accuracy by using the ROC curves for comparative analysis.There is no new turning point after September 2011.A prediction can be made that the macro economy of China would remain in a period of low speed growth.(3)The real-time data is used to verify the robustness of the MF-DFM well.