中国核心通货膨胀的度量研究

发布时间：2018-03-28 21:05

本文选题：核心通货膨胀　切入点：货币政策　出处：《华侨大学》2012年博士论文

【摘要】：目前，世界上越来越多的国家和地区开始关注并构建核心通货膨胀指标，并将其作为制定货币政策的重要参考指标。我国国家统计局和中国人民银行也开始在制定货币政策过程中关注核心通货膨胀，但目前尚无官方度量并定期公布该指标，本文将在该背景下对我国核心通货膨胀的度量问题进行系统地研究。对这一问题的研究不仅可以有助于准确地认识我国通货膨胀的长期潜在趋势，而且对科学地制定货币政策和有效地实施货币政策具有理论和现实意义。全文的研究工作和研究结论主要概括为以下几个方面。首先分析了核心通货膨胀的含义及其度量方法。在充分回顾和理解对核心通货膨胀研究的相关文献基础上，将核心通货膨胀的含义界定为：剔除CPI篮子中各类商品和服务价格指数变动的暂时成分后，余下各类商品和服务价格指数的长期共同成分，反映通货膨胀的潜在的长期趋势，这是货币政策决策部门制定货币政策时重点关注的指标。根据构建核心通货膨胀过程中所使用的数据信息集将度量方法分为基于同期分类横截面数据、基于时间序列数据和基于面板数据的三类方法。同其他消费价格指数相比，CPI是公众熟知的，具有更新最快，容易被公众所理解，而且不容易被中央银行操纵等特点。因此，本文基于CPI数据度量我国核心通货膨胀即核心CPI。本文并将贝叶斯估计与Gibbs Sampler相结合估计状态空间模型，将货币供给这个“硬核”考虑在模型中度量我国核心CPI。结果表明，样本期间（2001年1月至2011年4月）我国核心CPI的均值和标准差均小于CPI，我国核心CPI与CPI呈现出基本一致的波峰、波谷及波动频率。度量的核心CPI与货币供给增长率的相关性较大，符合模型的设定。从算法角度来讲，不仅解决了贝叶斯方法应用过程中多维积分计算问题，而且提高了估计状态空间的速度和准确度。采用动态因子指数（DFI）模型度量我国核心CPI。度量结果表明该核心CPI的波动幅度、均值及标准差均小于CPI，与CPI相比，核心CPI与货币供给增长率的相关性较大，并且进一步得出各类商品价格指数决定核心CPI的比例，衣着消费价格指数和交通与通讯价格指数对核心CPI的贡献较大，食品、娱乐教育与居住价格指数对核心CPI的贡献最小。从不同角度对上述方法度量各种核心CPI进行比较。分别从核心CPI期望性质、波动性、无偏性、追踪通货膨胀趋势值、平稳性、与CPI的相关性、协整关系、因果关系及与产出缺口关系方面比较各种核心CPI的相对优劣。结果显示，在样本期间内，目前没有任何一种核心CPI完全满足所有期望性质；在实际应用中，哪种核心CPI更适合作为制定货币政策的依据取决于对核心CPI各特征的不同偏重以及目的和需要。在同等重视各特征下，较优核心CPI的度量方法是非对称修剪均值法、BGS状态空间模型法及DFI模型法。但从实际应用角度来看，采用上述度量核心CPI的方法虽然具有稳定的理论基础，但不易被公众所理解，所以作为现实通货膨胀的监测目标使用不太适合，这也反映了经济政策研究和经济理论研究对核心通货膨胀的研究存在一定差别。度量核心通货膨胀的重要目的是它有助于预测未来通货膨胀。本文采用最广泛使用的预测模型检验我国核心CPI预测CPI能力时，模型的拟合度均较小，系数为负也并不显著，说明该模型的预测效果并不明显。为了更有效地检验我国核心CPI对CPI的预测能力，，本文建立一个合理的模型，从拟合度和系数的置信区间两方面对各种核心CPI的预测能力进行检验。结果显示，在样本期内，剔除法、加权中位数法、方差权重法及惯性权重法度量的核心CPI对CPI预测几乎不起作用，SVAR模型及共同趋势模型预测区间较小，而非对称修剪法、BGS状态空间法及DFI模型法的预测区间均超过10期。最后，对货币政策关注核心通货膨胀的原因及风险进行理论分析。关注核心通货膨胀不仅有利于维持货币政策的可信度和可靠性，而且有助于货币政策决策部门准确地把握未来通货膨胀的走势，作出科学有效的货币政策。在此基础上，分析了过度关注核心通货膨胀的风险，食品和能源的价格变动可能包含反映未来通货膨胀变动的有用信息，简单地将其从标题通货膨胀中剔除，将不利于有效地控制通货膨胀和预测未来通货膨胀。因此，本文认为，我国在对短期的价格水平变动进行货币政策调整时，应该主要关注标题通货膨胀即CPI，同时关注核心通货膨胀。货币政策对核心通货膨胀同样存在较大的滞后效应，在制定中长期货币政策时应重点关注具有前瞻性的核心通货膨胀。本文的主要创新为：（1）从研究角度上，本文对核心通货膨胀的含义及其度量方法进行了比较全面系统的研究。（2）在度量方法上，本文使用马尔科夫链蒙特卡洛算法（MCMC算法）度量我国的核心CPI。（3）本文在国内学术界首次使用动态因子指数（DFI）模型度量我国的核心CPI。（4）在评价各种核心通货膨胀优劣方面，本文从多角度综合地对各种核心CPI进行比较。（5）提出一个具有合理理论含义的核心CPI预测模型。
[Abstract]:At present, more and more countries and regions of the world and began to focus on core inflation index, which is used as an important reference index for the formulation of monetary policy. China's National Bureau of statistics and the people's Bank of China also began to focus on core inflation in the process of formulating monetary policy, but there is no official measure and regularly publish the index this paper will be in the context of measure of core inflation in China was systematically studied. The potential long-term trend of the research on this problem can not only contribute to the understanding of inflation in China, and to formulate scientific and effective implementation of monetary policy, monetary policy has theoretical and practical significance to the conclusion of this paper. The main research work and research as follows.
First analyzes the meaning and measure of core inflation. In the full review and understand the relevant literature on the basis of research on core inflation, the core inflation definition: temporary components of all kinds of goods and services price index excluding CPI changes in the basket after long-term common components left all kinds of goods and services price index, potential the long-term trend indicator of inflation, the monetary policy decision-making departments focus on monetary policy indicators. According to the measurement methods are divided into the same classification based on cross-sectional data used to build the core inflation data in the information set, based on time series data and based on three kinds of panel data method. Compared with other consumer price index CPI, is known to the public, has updated quickly, easily understood by the public, and not easy to be the central bank. Therefore, this paper is based on CPI data to measure the core inflation in China, that is, the core CPI.
In this paper, and the Bias estimation combined with the Gibbs Sampler estimation of state space model, the money supply the "hard core" considered in the model of moderate amount of our core CPI. results show that the sample period (January 2001 to April 2011) the mean and standard of our country the core difference of CPI was less than CPI, CPI and CPI showed a consistent core in China the peak, trough and fluctuation frequency measurement. The core CPI and the growth rate of the money supply are greatly correlated with model set algorithm. From the point of view, not only solves the problem of calculating method in the application process of multidimensional Bias integral, but also improve the state space estimation accuracy and speed.
The dynamic factor index (DFI) model to measure the core CPI. the result shows that volatility of the core of CPI, the mean and standard deviation were less than CPI, compared with CPI, core CPI and the money supply growth rate greater correlation, and further draw all kinds of commodity price index determines the core ratio of CPI, with the price of clothing consumption the index and the price index of transportation and communication on the core of CPI is big, food, entertainment and residential price index contribution of education on the core of CPI is the minimum.
To compare these methods to measure the various core CPI from different angles. From the core CPI expected properties, volatility unbiasedness, tracking inflation trend value, stability, and the correlation between CPI, cointegration, causality and relationship with the relative merits of the output gap is a variety of core CPI. The results showed that in during the sample period, there is no any kind of core CPI satisfy all desired properties; in practice, what kind of core CPI is more suitable for the formulation of monetary policy depends on the basis of core CPI to the characteristics of the different emphasis and purpose and need. In the equal importance of each feature, the better the core CPI measurement method is a non symmetric trimmed mean method, BGS state space model method and DFI model method. But from the practical point of view, using the method of the measurement of the core CPI has solid theoretical foundations, but it is not easy to be public As we all know, it is not very suitable for monitoring inflation as a realistic inflation target. It also reflects some differences in core inflation research between economic policy research and economic theory research.
An important objective measure of core inflation is that it helps to predict future inflation. The prediction model is the most widely used test for China's core CPI CPI prediction ability, the fitting degree of the model are smaller, a negative coefficient is not significant, the prediction effect of the model is not obvious. In order to test our core CPI prediction ability of CPI, this paper establishes a reasonable model, from the two aspects of the confidence interval fitting degree and coefficient of the prediction ability of various core CPI were tested. The results show that in the sample period, elimination method, weighted median method, variance weight method and CPI core measure of inertia weight on CPI prediction almost no effect, and the common trend of SVAR model prediction interval is smaller, and the asymmetrical pruning method, prediction interval BGS state space method and DFI model method are more than 10.
Finally, to analyze the reason and risk of monetary policy to focus on core inflation. Focus on core inflation is not only conducive to the maintenance of monetary policy credibility and reliability, but also contribute to the monetary policy decision-making departments accurately grasp the future trend of inflation, to make a scientific and effective monetary policy. On this basis, analysis of the risk of excessive attention to core inflation the food and energy price changes may contain useful information about future inflation, simply removing them from the headline inflation, will not be conducive to effectively control inflation and predict future inflation. Therefore, this paper argues that the change of our country to adjust monetary policy in the short-term price level, should focus on the title inflation is CPI, at the same time to focus on core inflation. Monetary policy on inflation as the core There is a large lag effect, and in the formulation of medium and long term monetary policy, we should focus on the forward-looking core inflation.
The main innovation of this paper: (1) from the research point of view, the meaning of core inflation and its measurement methods are studied systematically. (2) in the measurement method, this paper use the Markoff Montecalo algorithm (MCMC algorithm) to measure the core of CPI. (3) for the first time using dynamic factor index the domestic academic circles (DFI) core CPI. model to measure our country (4) in the evaluation of various aspects of the advantages and disadvantages of core inflation, to compare various kinds of core CPI from a comprehensive perspective. (5) proposed a reasonable physical meaning of the core CPI forecast model.

【学位授予单位】：华侨大学
【学位级别】：博士
【学位授予年份】：2012
【分类号】：F224;F822.5

【参考文献】