动态经济系统分析的经济计量模型与方法  

Econometric model and method for analysis of dynamic economic system

作　　者：姚峰[1] 

机构地区：[1]香川大学经济学部

出　　处：《管理科学学报》2003年第2期74-80,共7页Journal of Management Sciences in China

基　　金：日本文部科学省1998-2000年度科研费资助项目(国际学术研究10045016)

摘　　要：综述可有效阐明动态经济系统长期关系和因果关系的因果测度理论.首先简要介绍多变量时间序列的协整过程及与此相关的若干概念,并总结了在经济计量学领域评价较高的多变量自回归模型的统计识别方法.基于多变量时间序列协整过程的向量自回归模型,较详细讨论了多变量时间序列间各种因果测度的定义及其沃尔德检验.所述单方向因果测度及其统计检验理论作为C W.J.Granger非因果性理论的扩张,不仅可以检验两组时间序列间的因果影响存在与否,还可以定量描述影响的程度.单方向因果测度理论为分析复杂经济系统提供了一种有效手段.In this paper we mainly discuss the one_way effect causal measures which can be applied to the analysis of long_run and short_run as well as causal characteristics of dynamic economic system. We first summarized the related concepts of cointegration, and then showed the procedure of cointegrated model identification. Based on the identified error correction model, we discussed causal measures in time domain and frequency domain. The presented Wald test of the causal measure is incorporation Johansen's algorithm for the maximum likelihood estimates and the likelihood ratio tests. Using the Wald statistics, the paper also showed the computational algorithm of confidence_set construction for the overall causal measures. For the purposes of testing long_run or short_run characterization of causal relation, a variety of causal measures are introduced by means of the integral of the frequency_wise measure of one_way effect on specific frequency bands. In contract to the conventional tests of Granger's non_causality which amount to testing the hypothesis of zero restriction of a certain set of autoregressive coefficients, the approach of this paper enables us to test not only Granger's non_causality but also the strength of the one_way effect. The one_way effect causal analysis can be considered as an effective approach for the investigation of complex economic system.

关 键 词：动态经济系统 经济计量模型 协整分析 长期经济关系 因果关系 

分 类 号：F224.12[经济管理—国民经济]