| 1,011 | 12 | 250 |
| 下载次数 | 被引频次 | 阅读次数 |
传统识别SVAR模型的方法包括两类,一类是约束模型中的结构参数,另一类是约束脉冲响应函数,但多为严格的等式约束,符号约束则基于先验理论限定脉冲响应的方向,用较为宽松的不等式约束实现模型识别,能有效降低主观因素影响;同时随着经济结构的变化,SVAR模型的参数估计值有随时间变化的趋势,固定的参数估计值已不能有效刻画不同时期的经济发展状态。本文基于Gibbs抽样思想与贝叶斯统计推断理论,系统介绍符号约束下时变参数SVAR模型的贝叶斯估计方法,使用中国和美国数据,分别估计VAR模型、Sign-SVAR模型和Sign-TVP-SVAR模型。实证结果发现,符号约束能够有效避免脉冲响应的方向性偏误,时变参数能够更好刻画不同时期内经济变量的结构时变特征,在货币政策分析中具有明显优势。
Abstract：Traditionally,SVAR model is often identified through two methods: one is imposing constraints on structure parameters,the other is on the impulse response functions,most of which are strict equalities. The sign restrictions just defines the impulse response functions' direction based on the priori theory,and identifies the model with a more relaxed inequality constraints,which can effectively reduce the influence of subjective factors. With the changes of the economic structure,the SVAR model's parameters are changing over time,and the fixed parameters can't effectively portray the economic development characteristics in different periods. Basing on the Gibbs sampling and the Bayesian inference theory,this paper introduces the detailed process of parameters' estimation of the SVAR model with the time varying parameters. It respectively estimates VAR model,Sign-SVAR model and Sign-TVP-SVAR model by using Chinese and US data. The results show that sign constraints can effectively avoid the impulse response functions ' directional bias,and the time varying parameters can better characterize the structural variation of economic variables in different periods. It also proves that the Sign-TVP-SVAR model has obvious advantages in monetary policy analysis.
[1]Blanchard O,Quah D.The Dynamic Effects of Aggregate Demand and Supply Disturbances[J].American Economic Review,1989,79(4):655-673.
[2]王曦,邹文理.货币政策对股票市场的冲击[J].统计研究,2011(12):55-65.
[3]魏强,陈华帅.关于我国央行独立性与通货膨胀的研究[J].数量经济技术经济研究,2009(09):32-42.
[4]Sargent T J.Rational expectations,econometric exogeneity and consumption[J].Journal of Political Economy,1978,86(4):673-700.
[5]Litterman R B.Forecasting With Bayesian Vector AutoregressionsFive Years of Experience[J].Journal of Business&Economic Statistics,1986,4(1):25-38.
[6]朱慧明.基于Minnesota共轭先验分布的贝叶斯VAR(p)预测模型[J].统计研究,2004(01):44-48.
[7]沈悦,李善燊,马续涛.VAR宏观计量经济模型的演变与最新发展——基于2011年诺贝尔经济学奖得主Smis研究成果的拓展脉络[J].数量经济技术经济研究,2012(10):150-160.
[8]Granville B,Mallick S.Monetary Policy in Russia:Identifying exchange rate shocks[J].Economic Modelling,2010,27(1):432-444.
[9]Fry R,Pagan A.Sign Restrictions in Structural Vector Autoregressions:A Critical Review[J].Journal of Economic Literature,2011,49(4):938-960.
[10]Uhlig H.What are the Effects of Monetary Policy on Output?Results from an Agnostic Identification Procedure[J].Journal of Monetary Economics,2005,52(2):381-419.
[11]Sims C A.Rational expectations modeling with seasonally adjusted data[J].Journal of Econometrics,1993:9-19.
[12]Canova F.Modelling and forecasting exchange rates with a Bayesian time-varying coefficient model[J].Journal of Economic Dynamics and Control,1993:233-261.
[13]Cogley T,Sargent T J.Evolving Post-World War II U.S.Inflation Dynamics[J].NBER Macroeconomics Annual,2002.
[14]Cogley T.How fast can the new economy grow?A Bayesian analysis of the evolution of trend growth[J].Journal of Macroeconomics,2005,27(2):179-207.
[15]Cogley T,Sargent T J.Drifts and Volatilities:Monetary Policies and Outcomes in the Post WWII U.S.[J].Review of Economic Dynamics,2003,8(2):262-302.
[16]Primiceri G E.Time Varying Structural Vector Autoregressions and Monetary Policy[J].The Review of Economic Studies,2005,72(3):821-852.
[17]Blanchard O J,Watson M W.Are Business Cycles All Alike[M].University of Chicago Press,1986.
[18]Blanchard O J,Quah D.The Dynamic Effects of Aggregate Demand and Supply Disturbances[J].American Economic Review,1988,79(4):655-73.
[19]Rigobon R.Identification Through Heteroskedasticity[J].The Review of Economics and Statistics,2003,85(4):777-792.
[20]Rigobon R,Sack B P.The effects of war risk on US financial markets[J].Journal of Banking and Finance,2005,29(7):1769-1789.
[21]Cochrane J H,Piazzesi M.The Fed and Interest Rates-A HighFrequency Identification[J].American Economic Review,2002,92(2):90-95.
[22]Faust J,et al.Identifying VARs based on high frequency futures data[J].Journal of Monetary Economics,2004,51(6):1107-1131.
[23]Bernanke B S,Kuttner K N.What Explains the Stock Market's Reaction to Federal Reserve Policy?[J].Journal of Finance,2005,60(3):1221-1257.
[24]Romer C D,Romer D.Does monetary policy matters?A new test in the spirit of Friedman and Schwartz[J].NBER Macroeconomic Annual,1989:121-170.
[25]Romer C D,Romer D H.A New Measure of Monetary Shocks:Derivation and Implications[J].The American Economic Review,2003,94(4):1055-1084.
[26]Romer C D,Romer D H.The FOMC versus the Staff:Where Can Monetary Policymakers Add Value?[J].The American Economic Review,2008,98(2):230-35.
[27]Ramey V A,Shapiro M D.Costly Capital Reallocation and the Effects of Government Spending[J].Journal of Monetary Economics,1999,48(1):145-194.
[28]Fujita S,Ramey G.The cyclicality of separation and job finding rates[J].International Economic Review,2009,50(2):415-430.
[29]Faust J.The robustness of identified VAR conclusions about money[J].Carnegie-Rochester Series on Public Policy,1998:207-244.
[30]Carter C,Kohn R.On Gibbs sampling for state space models[J].Biometrika,1994,81(3):541-553.
[31]Kim C J,Nelson R N.State Space Models with Regime Switching:Classical and Gibbs-Sampling Approaches with Applications[M].Cambridge:the MIT Press,1999.
[32]Koop G,Potter S M.Forecasting in large macroeconomic panels using Bayesian Model Averaging[EB/OL].Federal Reserve Bank of New York Staff Reports no.163,March 2003.
[33]Canova F.Methods for Applied Macroeconomic Research[M].Princeton:Princeton University Press,2007.
(1)克里斯托夫·A·西蒙斯(Christopher A.Sims)教授是2011年度诺贝尔经济学奖得主之一,现就职于美国普林斯顿大学,其研究领域主要涉及向量自回归分析等方法研究、理性疏忽理论及政策研究。
(2)共轭先验属于有信息先验,结合共轭先验分布与似然函数可得到与先验分布形式相同的后验分布,便于充分利用经济理论和现有研究成果。
(1)给参数设定初始值,但同时给其一个自由变动的范围,而不是锁定该值。
(1)比如利率上升,对产出增长率有负向影响,对通货膨胀率有负向影响,而对利率本身则是正向影响,分别对应于响应矩阵中相应行的各元素。
(2)冲击有两个方向,只要结构影响矩阵中冲击对应行满足一个方向的符号约束即可。
(1)Minnesota先验因起源于明尼苏达联邦储备银行而得名,该方法假定VAR模型中的内生变量服从随机游走过程,因首先由Litterman等(1986)使用而闻名。
(2)Normal-inverse Wishart先验假定VAR模型中的系数服从正态分布,方差协方差矩阵中的元素服从IW分布,这种设置意味着模型中不同方程的误差项之间存在相互影响。
基本信息:
DOI：10.19343/j.cnki.11-1302/c.2016.10.012
中图分类号:F224
引用信息:
[1]苏治,位雪丽,赵宣凯.符号约束与时变参数SVAR模型的贝叶斯估计实现[J].统计研究,2016,33(10):100-112.DOI:10.19343/j.cnki.11-1302/c.2016.10.012.
基金信息:
国家社会科学基金重大项目“‘互联网+’推动经济转型机理与对策研究”(15ZDC024);; 2013年度教育部“新世纪优秀人才支持计划”“从货币总量到信用总量:一个新的全球经济分析与宏观政策调控的框架”(NCET-13-1055);; 中央财经大学博士研究生重点选题支持计划“基于贝叶斯统计推断方法的非常规货币政策研究:有效性、调控机制及退出路径”;; 中央财经大学研究生科研创新基金“基于贝叶斯参数估计的Sign-TVP-SVAR模型及其在货币政策分析中的应用”;; 国家自然科学基金面上项目“货币总量转向信用总量:全球虚拟经济与实体经济背离机理与宏观政策应对”(71473279)资助
2016-10-15
2016-10-15