我国经济增长率动态波动机制研究

发布时间：2018-07-23 20:08

【摘要】：经济增长率的动态波动机制是宏观经济学中的热议话题。许多学者都对这一问题进行了详细探究。早期的研究通过使用自回归序列和协整检验来刻画经济增长率的变化特征。然而，随着经济理论建模的不断发展，简单的自回归序列已无法充分地反映经济增长率的变化机制，因此更多的非线性模型被引入用以刻画经济增长率的波动机制。近年来，越来越多的学者认为：经济系统的结构在不同的水平位置上存在着差异，而随着经济系统结构的改变，系统中各变量的作用机制也可能存在着相应的变化。早期，许多研究使用区制转移的方法对经济增长率的区间进行了划分，但由于样本规模较小，划分区间后却难以对经济系统中变量的作用机制予以刻画。因此，本文将在以往研究的基础上，使用更广义的时变估计方法，从动态的角度刻画经济增长率的波动机制，并直接给出在变化的同时系统中各变量间作用机制的改变。本文秉承了以往的研究思路，通过GDP增长率，通货膨胀增长率，M2增长率以及PPI增长率建立了经典的VAR经济模型。文中首先基于季度数据探讨了VAR均值方程中GDP增长率对系统中各变量的冲击反映，这一阶段的结论大体上与以往大量的研究相吻合。只有在建立模型前的序列检验上与以往研究略有不同，我们在对数据进行平稳性检验时发现，CPI增长率序列并不平稳，我们认为，这是样本区间差异造成的，我们的数据起始于1996年4季度，截止于2012年3季度，共16年整，当数据剔除1996年前三个季度后，CPI增长率数据便体现出非平稳特征，表现为差分平稳序列。并且，它与GDP增长率序列存在着长期协整关系，因此这并不影响我们建立VAR模型。随后我们在基础研究的结果上，进一步加入了本文的研究要素，通过蒙特卡洛模拟法获取了样本的后验分布，并进一步引入了随机波动予以刻画时变特征，建立了时变向量自回归模型（TVP-VAR）。考虑到在每个样本点给出时变估计方程并没有意义，因此本文选取了时变方程的两种经典冲击反应函数来刻画变量间作用机制的时变特征。实证研究结果表明，TVP-VAR模型对时点信息的捕捉能力要显著优于传统的VAR均值方程。此外，我们发现：在样本期间内，通货膨胀与经济增长间的作用机制可能发生过结构性的改变。以往的文献以及我们对2001年3季度和2006年3季度时点信息的研究都表明通货膨胀在短期内会促进经济增长，而后将对经济增长表现出抑制作用。然而在刻画2011年3季度的时点冲击反应函数时我们发现，通货膨胀仅对经济增长表现出抑制作用，即二者间的作用机制发生了改变。最后，我们还认为：当经济增长率处于较高水平时，无论是控制通货膨胀，，改变货币供给量，或是改变投资规模都会对经济增长产生较为强烈的冲击，而当经济增长率处于低位水平时，相应的刺激效果远不如前者显著。
[Abstract]:The dynamic fluctuation mechanism of economic growth rate is a hot topic in macroeconomics. Many scholars have studied the problem in detail. Early studies used autoregressive sequences and cointegration tests to characterize the changing characteristics of economic growth rates. However, with the development of economic theory modeling, simple autoregressive series can not fully reflect the changing mechanism of economic growth rate, so more nonlinear models are introduced to describe the fluctuation mechanism of economic growth rate. In recent years, more and more scholars think that the structure of the economic system is different at different levels, and with the change of the structure of the economic system, the action mechanism of the variables in the system may also have corresponding changes. In the early years, many studies used the method of district system transfer to divide the region of economic growth rate, but because of the small sample size, it is difficult to describe the mechanism of the variables in the economic system after dividing the interval. Therefore, on the basis of previous studies, this paper uses a more generalized time-varying estimation method to describe the fluctuation mechanism of economic growth rate from a dynamic point of view, and directly gives the change of the interaction mechanism between variables in the system while the change is taking place. In this paper, the classical VAR economic model is established by means of GDP growth rate, inflation growth rate M _ 2 growth rate and PPI growth rate. In this paper, based on the quarterly data, the impact of the GDP growth rate on the variables in the VAR mean equation is discussed. The conclusion of this stage is consistent with a lot of previous studies. Only in the model before the series test and previous research slightly different, we in the data when the stationary test found that the CPI growth rate series is not stable, we believe that this is the result of sample interval differences, Our data started in the fourth quarter of 1996 and ended in the third quarter of 2012 for a total of 16 years. Moreover, there is a long-term cointegration relationship with GDP growth rate series, so this does not affect the establishment of VAR model. Then we add the research elements to the results of basic research, obtain the posterior distribution of samples by Monte Carlo simulation method, and further introduce random waves to characterize the time-varying characteristics. A time-varying vector autoregressive model (TVP-VAR) is established. Considering that it is not meaningful to give the time-varying estimation equation at each sample point, two classical impulse response functions of the time-varying equation are selected to characterize the time-varying characteristics of the interaction mechanism between variables. The empirical results show that TVP-VAR model is superior to the traditional VAR mean equation in capturing time point information. In addition, we find that during the sample period, the mechanism of inflation and economic growth may have undergone structural changes. Previous literature and our studies of time points in the third quarter of 2001 and the third quarter of 2006 have shown that inflation promotes economic growth in the short term and then suppresses it. However, in describing the time point shock response function in the third quarter of 2011, we find that inflation only suppresses economic growth, that is, the mechanism of action between the two has changed. Finally, we also believe that when the economic growth rate is at a relatively high level, whether it is controlling inflation, changing the amount of money supply, or changing the scale of investment, it will have a stronger impact on economic growth. And when the economic growth rate is low, the corresponding stimulus effect is far less significant than the former.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F224;F124.1

【参考文献】