摘要
研究目标:提出线性与非线性MS-DSGE模型的条件最优粒子滤波(COPF)以及参数的贝叶斯估计方法。研究方法:利用MS-DSGE模型状态空间表示的条件线性特征来构造COPF,基于上述算法进一步提出粒子马尔科夫链蒙特卡洛(PMCMC)方法,实现参数的贝叶斯估计。最后构建可变货币政策规则的新凯恩斯理论模型,运用数值模拟方法检验了COPF与参数贝叶斯估计的效果,同时对中国实际数据进行了估计与分析。研究发现:在粒子滤波的低观测误差设定与小样本条件下,模型深度参数的贝叶斯估计结果表现精确,所有参数的真值都在其后验分布的[5%,95%]分位数区间内;实际应用结果显示,中国货币政策存在积极与被动这两种区制,且政策的区制转换会导致价格通胀波动幅度上升1%左右;2011年之后,中国货币政策区制表现的非常稳健,对通胀的响应程度较高。研究创新:提出了非线性MS-DSGE的COPF以及PMCMC方法,从而实现深度参数的贝叶斯估计。研究价值:所构建的粒子滤波与贝叶斯估计方法为一般的MS-DSGE模型提供了实证分析工具,对未来内生区制转换、系统性风险传染等问题的研究奠定了基础。
Reserch Ojectives:Proposing the conditional optimal particle filter and bayesian estimation method for linear and nonlinear MS-DSGE models.Reserch Methods:Using the conditional linear feature of state space representation from MS-DSGE model to construct the conditional optimal particle filter.Based on the above algorithm,Particle Markov Chain Monte Carlo method is further proposed to do the Bayesian estimation of parameters.We build a New Keynesian model with Markov regime-switching monetary policy to examine the performance of conditional optimal particle filter and Bayesian estimation of parameters.At the same time,the actual data of China are used for estimation and analysis.Reserch Findings:Under the conditions of low observation error and small sample,the Bayesian estimation results of model depth parameters are accurate,the true value of all the parameters are in the estimated[5,95]percentile of posterior distribution.The results of practical application shows that there are two different regimes of monetary policy in China:active and passive.After 2011,monetary policy rule becomes very stable and has been in the active regime,which has a high response to inflation.Reserch Innovations:The conditional optimal particle filter and particle Markov Chain Monte Carlo method of MS-DSGE model are proposed to do the Bayesian estimation of depth parameters.Reserch Values:The conditional optimal particle filter and Bayesian estimation methods constructed in this paper provide an effective empirical tool for the general MS-DSGE model,and lay a foundation for the future research on endogenous regime and systemic risk contagion.
作者
周上尧
Zhou Shangyao(School of Economics and Management,Wuhan University)
出处
《数量经济技术经济研究》
CSSCI
CSCD
北大核心
2021年第3期160-180,共21页
Journal of Quantitative & Technological Economics
