空间非平稳数据检验及空间误差修正模型的贝叶斯估计

发布时间：2017-12-27 20:22

本文关键词：空间非平稳数据检验及空间误差修正模型的贝叶斯估计　出处：《新疆大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着计量经济学的发展,特别是时间序列的单位根过程和协整理论的发展,给主流计量经济学的发展带来一场革命,时间序列分析已经建立了相对完善的理论分析框架,并在众多经济领域取得很好的应用.空间计量经济学作为计量经济学的一个分支,在处理横截面数据和面板数据的空间相互作用和空间结构变化方面发挥独特优势,并发展出了空间误差模型、空间滞后模型和地理加权回归模型等经典模型来处理空间自相关性和空间异质性.空间计量与地理信息系统(GIS)的结合广泛应用于房地产经济、环境和资源经济等领域.但是空间非平稳计量经济学在空间计量经济学中的发展却不够完善。本文首先介绍时间序列的单位根和协整理论,然后讨论关于非平稳空间数据的研究,包括空间单位根和协整理论的定义及其检验方法;空间单位根过程的数据生成方法及检验,以及空间误差修正模型的不同形式和参数估计;类似于时间序列的DF检验,空间非平稳检验的临界值也被提出;Moran’s I是研究一个变量的集聚效应,本文也介绍关于两个空间过程的相关性的研究.空间面板数据的单位根的研究已经有成熟的理论,但空间截面数据的平稳性研究仍需完善,所以本文的研究都是基于空间截面数据。本文创新在于用时间序列的方法检验了空间非平稳检验临界值的合理性,然后建议用两步拉格朗日检验法检验空间数据的平稳性;考虑到数据生成过程会对检验结果造成影响,本文对两类造成非平稳的空间过程即含趋势项的平稳过程和含漂移项的随机游走进行检验区分,通过模拟实验可以区分上述两种不同非平稳过程;鉴于工具变量法估计空间误差修正模型时在参数边界上误差较大,本文使用贝叶斯方法对空间误差修正模型进行估计,结果显示贝叶斯方法整体上优于工具变量法,即使在参数边界上估计结果也比较好。
[Abstract]:With the development of econometrics, especially time series unit root process and the development of the theory, for the development of the mainstream of econometrics has brought a revolution to the time series analysis has established a relatively complete theoretical framework for analysis, and achieved very good application in many economic fields. Spatial econometrics is a branch of econometrics the advantages of interaction in the treatment of cross section and panel data of space and spatial structure change, and to develop the spatial lag model and spatial error model, geographically weighted regression model and classical model to deal with the spatial autocorrelation and spatial heterogeneity. The spatial econometric and geographic information system (GIS) combined with widely used in the real estate economy, environment and resources, economic and other fields. But the spatial non-stationary econometrics in spatial econometrics in development But it is not perfect. This paper first introduces the time series unit root and cointegration theory, and then discuss the research on non-stationary spatial data, including spatial definition of unit root and cointegration theory and test method; spatial data generation method of unit root process and inspection, and different forms of space and parameter error correction model estimation; DF test is similar to the time series the non critical value of stable test are also presented; "Moran s I is the agglomeration of a variable, this paper also introduces the research on the correlation between the two spatial process. Research on spatial panel data unit root has mature theory, but the stability of the space research section data still needs to be improved therefore, this study is based on cross-sectional data space. The innovation of this paper lies in using time series method to test the rationality of non smooth space check critical value, and then proposed the stationarity test of spatial data with the two step Lagrange test method; taking into account the data generation process will affect the test results, the paper on the two kinds of stationary process caused by non spatial smooth process including the trend and with random walk test to distinguish, through the simulation experiment can distinguish these two different kinds of non stationary process; given the estimated error correction model in the parameter space boundary error of instrumental variable method, this paper uses Bayesian methods to estimate the spatial error correction model, results show that Bayesian method is better than the method of instrumental variables, even if the bounds on the parameters estimation results are relatively good.
【学位授予单位】：新疆大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O212.8

【相似文献】