实时数据、动态演变与菲利普斯曲线甄别  

Real Time Data,Dynamic Evolution and Phillips Curve Identification

作　　者：林玉婷 孙坚强[2] 陈创练 LIN Yu-ting;SUN Jian-qiang;CHEN Chuang-lian(School of Economics,Guangdong University of Finance and Economics,Guangzhou 510315,China;School of Economics and Finance,South China University of Technology,Guangzhou 510006,China;Institute of Finance,Jinan University,Guangzhou 510632,China;School of Economics,Jinan,University,Guangzhou 510632,China)

机构地区：[1]广东财经大学经济学院,广东广州510315 [2]华南理工大学经济与金融学院,广东广州510006 [3]暨南大学金融研究所,广东广州510632 [4]暨南大学经济学院,广东广州510632

出　　处：《数理统计与管理》2023年第6期1113-1126,共14页Journal of Applied Statistics and Management

基　　金：国家社科基金重大项目(22ZDA050)。

摘　　要：近年来国家统计局实时公布国内生产总值的初步核算数据,为我国通胀的实时预测提供了基础条件。本文根据实时数据和最终数据估计后顾式、前瞻式和混合型三种菲利普斯曲线,并考究实时数据是否有助于改进通胀预测。研究发现,(1)基于最终数据的混合型时变参数菲利普斯曲线更好地拟合我国的通胀运行,不论是常系数模型还是时变参数模型,实时数据的引入并未提供更有价值的信息,反而降低了曲线对通胀的预期力,均方预测误差提高了4.79%。(2)产出缺口的时变参数近年来不断提高且显著为正,存在明显“经济扩张→通胀”的传导关系,且实时数据和最终数据估计的我国产出缺口自2015年以来均为负缺口,因此要警惕负缺口引致通缩的风险。(3)近年来通胀惯性的时变参数趋于平稳,取代通胀预期成为通胀波动的主导因素,管控通胀的货币政策需要考虑到较长的时滞效应。(4)通胀预期的影响高于产出缺口,但时变参数表现出较为明显的波动性特征,可见通胀预期管理政策需要更加实时和灵活。The National Bureau of Statistics of the People's Republic of China has reported the preliminary accounting data of GDP in real time in recent years,which provides the foundations of the real-time infation forecasts.According to the applications of the real-time data and the final data,this paper estimates three different types of Phillips curves which are the backward-looking,the forward-looking and the hybrid Phillips curves respectively,and investigates whether the incorporation of the real-time data helps refine inflation forecasts.The study comes out with four conclusions as follows.First,the constructed hybrid Phillips curve with the application of final data fits the operation regulations of inflation in China more favorably compared to the one with real-time data,whether it is estimated in a constant coefficient model or a time-varying parameter model,which does not contribute to more valuable information but reduces the efficiency of the inflation forecasts of the model with the mean squared forecast error increasing by 4.79%.Second,the time-varying parameters of the output gap has been increasing constantly in recent years which is significantly positive,indicating that there exists obvious transfer mechanisms from economic development to infation.In addition,the estimations of the output gap have been negative since 2015.Hence,the risks of recession caused by the negative output gap should be treated with caution.Third,the time-varying parameters of infation inertia have stabilized in recent years and replace the infation expectations as the dominant factor in inflation.Therefore the long-horizon time-lagged effects should be considered within the monetary policies aiming at adjusting the infation.Last but not least,the impacts of infation expectations with more volatile time-varying parameters in long horizon is much larger than the output gap,which reveals that policies managing the inflation expectations are suspected to be more real-time and flexible.

关 键 词：菲利普斯曲线 状态空间模型 卡尔曼滤波 时变参数 

分 类 号：F31[经济管理—产业经济] E52[军事—军事理论] F82[理学—概率论与数理统计] O212[理学—数学]