巴基斯坦产出缺口的分析和估算

发布时间：2018-01-23 22:09

本文关键词： Monetary Policy Bayesian State-Space Method Kalman Filter Output Gap　出处：《吉林大学》2014年博士论文　论文类型：学位论文

【摘要】：货币政策的分析和实现取决于许多宏观经济变量的动态行为和联系。作为巴基斯坦央行(巴基斯坦国家银行-State Bank of Pakistan-SBP),研究群体的成员，我得到了塑造巴基斯坦经济的动态实践知识的特权。在SBP，我的责任范围是评估巴基斯坦的房地产业，,因此对于我来说，货币政策实施与实体经济的动态链接，以及实体经济的长期增长轨迹似乎显得非常有趣。此外,在巴基斯坦国家银行(SBP)我参与了巴基斯坦经济业务周期的分析和评估，通过分析这两大领域，让我重新考虑研究巴基斯坦经济的增长潜力。产出缺口和潜在产出是直接链接商业周期分析评估以及影响货币政策实施的两个潜在的宏观经济变量，产出缺口的稳健估计提供了有关实体经济周期和趋势的独特信息。一个显著的正向产出缺口表明通货膨胀倾向的存在（即高总需求），而负向产出缺口的存在则表明经济中存在产能过剩（即总供给过剩）。我相信本文研究的目的便是这种关系的相关问题研究。在本文中，我通过抽样检查和实施各种方法去估计巴基斯坦的产出缺口，而先前在这个方面的研究仅旨在估计产出缺口年度频率。这是由于在巴基斯坦的国民收入核算（NIA）中缺乏高频数据。我提出并实施获得巴基斯坦实际国内生产总值的Quarterly Temporal Disaggregation实现理论相一致的方法，，在随后的分析，实际GDP季度估计是利用基于“结构”和“产出缺口”估计单变量的方法。第1章包含对产出缺口的概念引入，研究动机以及巴基斯坦产出缺口的理性讨论研究。在第一篇（2章）我详细说明用各种单变量的方法来估计潜在产出和产出缺口，这些方法在文献中有良好的基础，而且也是各中央银行在对产出缺口的估计中积极使用的方法。运动估计基于年度和季度的频率，这对于实时性政策输入和分析是特别有用的。在第二篇（3章）我基于状态空间方法，使用菲利普斯曲线估计和预测巴基斯坦季度频率的产出缺口。产出缺口和巴基斯坦商业周期的实证估计有类似的解释和方向。我也使用布兰查德和柯（1979）的精华—VAR结构估计产出缺口。在这篇文章中还包括采用基宏（2008）的精华—小波滤波分析产出缺口。所有的三种方法的结果有可靠的经济解释。我还使用自回归（AR），ARIMA和结构回归方法预测巴基斯坦的产出缺口。商业周期分析的文献表明，估计潜在的宏观经济变量，如潜在产出和产出缺口在高数据频率方面有货币和财政政策的意义。与以往不同,本文旨在规避季度短缺的问题，对巴基斯坦国民收入账户（NIA）的估计，基于一个强健的代表真正的总产出的大型制造业指数（LSM指数）。本文实现了使用三种常用的方法对产出缺口的估算，包括：状态空间模型，小波滤波器，VAR结构模型。我认为，在巴基斯坦宏观经济总量线中，自从2009第三季度以来，需求压力已经平缓下来，除此之外，自2011第三年度以来，由于实体经济即总供给放缓，我国的负向产出缺口也平缓下来。最后，我预测当前总需求低的趋势预计将持续到2015年第四季度。第4章包含在一个结构性货币和信贷变量的环境中分析和预测产出缺口与通货膨胀，我使用了许多贝叶斯变量，采用适当的分化先验（3种不同的先验），以预测产出缺口和通货膨胀。结果表明在本文所采用的方法中可以避免多重共线性，估计误差高和回归模型的参数化的问题。此外，通过像Korobilis（2009）那样编程一个大都市-黑斯廷斯算法，我说明在后验概率密度得到的情况下，这样的密度是不可用的解析。我也使用吉布斯采样器获取预测密度和推理。详细的说,我试图使用贝叶斯VAR方法预测巴基斯坦的通货膨胀和产出缺口。为此我使用了三种不同的先验方法，本文分析是使用货币和信贷的宏观变量进行的。此次产出缺口分析是在状态空间框架下使用卡尔曼滤波器进行的估计。文献表明，贝叶斯收缩是一个使用大量宏观经济变量进行预测的合适的工具，另外，适当的先验选择是贝叶斯VARs稳健预测的基础。在这种背景下，我们分析中3种类型的先验检验：1、明尼苏达先验检验；2、独立的诺曼-威沙特检验；3、独立的明尼苏达-威沙特检验。估计和预测与库普(Koop)和Korobilis（2009）相一致。诊断方法的贝叶斯VAR模型与稳健预测估计表明，贝叶斯估计VARs提供稳健的预测和合适的构造解释,特别是，结果表明独立的诺曼-威沙特先验和独立的明尼苏达-威沙特先验比明尼苏达先验具有更好的预测性能。第四篇（5章）阐述了使用贝叶斯VAR方法预测通货膨胀。与第4章相比，我实现了7中不同的贝叶斯VAR估计的先验方法。结果显示，在预测三个月以上的时间中，SSVS(Stochastic Search Variable Selection Methods1and2)和方法与独立诺曼-威沙特先验方法一样，能提供更好的预测性能。另外，其他的贝叶斯变量也表现出可敬的结果。
[Abstract]:The dynamic behavior and the relationship of monetary policy analysis and implementation depends on many macroeconomic variables. As Pakistan's central bank (the National Bank of Pakistan -State Bank of Pakistan-SBP), members of the research group, I got the privilege of shaping the dynamic practice knowledge of Pakistan's economy. In SBP, my responsibility is to assess the real estate industry in Pakistan. So, for me, the dynamic link of monetary policy and the real economy, long-term growth path and the real economy appears to be very interesting. In addition, the State Bank of Pakistan (SBP) I participated in the analysis and evaluation of Pakistan business cycle, through the analysis of these two areas, let me think of the Pakistan economy growth potential.
The output gap and the potential output is a direct link to the business cycle analysis of two potential macroeconomic variables and the impact assessment of the implementation of monetary policy, the robust estimation of the output gap provides unique information about the real economy cycle and trend. A positive output gap that inflationary exists (high total demand), and negative to the output gap that exists overcapacity present in the economy (i.e. total excess supply). I believe that the purpose of this study is to study the related problems of this relationship.
In this paper, I through sampling inspection and implementation of various methods to estimate the output gap in Pakistan, and the previous research in this area is only aimed at estimating the annual output gap frequency. This is due to the national income accounts in Pakistan (NIA) in the lack of high frequency data. I proposed and implemented Pakistan real GDP Quarterly the Temporal Disaggregation method to realize the theory consistent, in the subsequent analysis, the actual GDP quarter estimate is based on "structure" and "output gap" estimation of single variable method.
The first chapter includes the introduction of the concept of the output gap, the motivation of research and the rational discussion of the output gap in Pakistan.
In the first part (Chapter 2) I detail to estimate potential output and output gap with various methods of single variable, these methods have a good foundation in the literature, but also the central bank to actively use the method in estimating the output gap in the annual and quarterly. Motion estimation based on the frequency, which is particularly useful for real-time input and policy analysis.
In the second chapter (Chapter 3) based on the state space method, using Phillips curves to estimate and predict the frequency of the output gap in Pakistan quarter. The empirical output gap and Pakistan business cycle are estimated to explain and similar direction. I also use Blanchard and Ke (1979) the essence of VAR structure of output gap estimates are also included in this. Article by motohiro (2008) - the essence of wavelet analysis of output gap. All three methods were reliable interpretation economy. I also use autoregressive (AR) prediction, Pakistan's output gap and the structure of ARIMA regression method.
Business cycle analysis of the literature suggests that the estimation of potential macroeconomic variables, such as potential output and output gap with monetary and fiscal policy in terms of high frequency data. Unlike in the past, this paper aims to avoid the shortage of the Pakistan quarter, the national income accounts (NIA) estimation, large-scale manufacturing industry a strong representative the real output based index (LSM index). This paper realizes the estimation, using three methods of output gap include: state space model, wavelet filter, VAR model. I think in Pakistan macroeconomic line, since the 2009 quarter of third, the pressure of demand has slowed down, in addition to in addition, since the 2011 since the year third, because the real economy is the total supply slowed, China's negative output gap also slow down. Finally, I predict the current total demand trend is expected to be low It continued until the fourth quarter of 2015.
The fourth chapter contains the analysis and forecast of the output gap and inflation in a structural monetary and credit environment variables, I use a lot of Bayesian variables, using appropriate prior differentiation (3 different priors), to predict the output gap and inflation. The results show that the method used in this paper can avoid multicollinearity high, error estimation and regression models parameterized problem. In addition, through Korobilis (2009) as a metropolis Hastings algorithm programming, I explained in the posterior probability density are obtained in the case that the density is not analytic available. I also use the Gibbs sampler to obtain predictive density and reasoning.
Specifically, I try to use the Bayesian VAR method to forecast Pakistan's inflation and output gap. Therefore I use three different methods in this paper is a priori, macro variables using the money and credit for. The output gap analysis is estimated in the state space framework using the Calman filter. The literature suggests that Bayesian shrinkage is a suitable tool for a large number of macroeconomic variables predicted in addition, appropriate prior selection is the basis of Bayesian VARs robust prediction. In this context, we analyze 3 types of prior inspection: 1, Minnesota prior inspection; 2, the independent Norman Wishart test; 3, independent Minnesota - test. Wishart estimation and prediction and Kupchak (Koop) and Korobilis (2009) is consistent. The diagnosis method of Bayesian VAR model and robust estimation show that Bayeux The VARs estimates that it is robust
Prediction and appropriate structural interpretation, in particular, the results show that the prediction performance of the independent Norman Wishart prior and independent Minnesota Wishart prior Bimingni Minnesota prior is better.
The fourth chapter (Chapter 5) describes the use of Bayesian VAR method to forecast inflation. Compared with the fourth chapter, I realized the transcendental method Bayesian VAR different estimates of 7. The results showed that in the prediction of time more than three months, SSVS (Stochastic Search Variable Selection Methods1and2) and Norman - Wishart's transcendental method and independent methods can provide better prediction performance. In addition, other Bayesian variable also showed admirable results.

【学位授予单位】：吉林大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：F135.3

【相似文献】