基于引力模型和稳定差值法对OECD国家之间的双边贸易研究

发布时间：2018-03-02 00:32

本文关键词： 传统引力模型稳定差值法系数估计　出处：《湘潭大学》2016年硕士论文　论文类型：学位论文

【摘要】：国际双边贸易一直以来都是经济学研究的重点领域。研究方法主要分为理论分析、事前分析(ex-ante)和事后研究(an post)。这三种研究方法各有优点同时又各有不足。其中事后研究主要是通过数据来构建经济因素之间的关系,并用这种关系来检验某些经济理论,所以事后研究成为目前国际贸易研究的主流方法,其中引力模型是最具代表性的事后研究工具。然而引力模型虽然能够很好的拟合那些经济数据,但是很少有学者从数理逻辑上去解释为什么这种线性结构就是合理的结构。为此本文尝试在引力模型的基础上,用推导出来的稳定差值法来从数理逻辑上证明引力模型的正确性,并用实证数据来进行验证。最后本文还用稳定差值法来重新进行引力模型建模,从而得到更加合理的系数估计。本文的论述结构和得出的相应结论可以归纳为:第一章主要是提出本文的研究背景和研究目的,并对相关的经典文献做了简单的梳理。第二章主要是对相关的理论做了系统性的解释和阐述,并对作者的创新算法做了数学推导,为后面的实证部分提供理论支持。第三,四章是本文的实证部分。本文的主要数据来源分别是OECD.org和comtrade.un.org两个网站,所以数据真实可信。获得数据后,作者用第三章的理论架构进行数据建模并证明了只有引力模型这种线性结构才能得到高度拟合的模型,这从数理逻辑上证明了引力模型的正确性。在此基础上,作者更进一步地用第三章的理论架构来重新拟合引力模型并得到了拟合优度更高的模型与其对应的系数估计。最后本文对全文进行了总结,分析了一些不足点,并提出了未来进一步的研究方向。本文的创新之处有两点:传统的计量模型都是单个体模型,也就是一个样本就是一个个体。这个个体的属性决定了它的目标变量的值。比如引力模型把两个国家的GDP、人口、汇率等等变量都看成一个样本的属性来建模。而本文是双个体模型,一个样本由两个个体决定。这样的模型可以把多个属性的数值通过差分给抵消掉。比如法国与美国的贸易减去中国与美国的贸易就可以抵消掉美国的相关属性。这大大的简化了模型。其次,传统的计量模型都是用自变量来预测因变量,而无法用因变量来反向预测自变量。但本文的算法推导出,用因变量可以推算出那些具有交叉作用的自变量。
[Abstract]:International bilateral trade has always been a key area of economic research. Research methods are mainly divided into theoretical analysis. These three research methods have their own advantages and disadvantages, in which ex post research is mainly based on data to construct the relationship between economic factors, and to test some economic theories with this relationship. So ex post study has become the mainstream method of international trade research, and gravity model is the most representative tool of afterwards research. However, although gravity model can fit those economic data very well, But few scholars explain why this kind of linear structure is a reasonable structure from the mathematical logic. The stability difference method is used to prove the correctness of the gravity model mathematically and logically, and the empirical data is used to verify it. Finally, the stable difference method is used to remodel the gravity model. The structure of this paper and the corresponding conclusions can be summarized as follows: the first chapter is mainly to put forward the research background and purpose of this paper. Chapter two is mainly about the systematic explanation and elaboration of the relevant theories, and the mathematical derivation of the author's innovative algorithm, which provides theoretical support for the later empirical part. The four chapters are the empirical part of this paper. The main data sources of this paper are OECD.org and comtrade.un.org, so the data are true and credible. The author uses the theoretical framework of Chapter 3 to model the data and proves that only the linear structure of gravity model can get a highly fitted model, which proves the correctness of the gravity model mathematically and logically. The author further uses the theoretical framework of Chapter 3 to refit the gravity model and obtain the model with higher goodness of fit and its corresponding coefficient estimation. Finally, this paper summarizes the whole paper and analyzes some shortcomings. The innovation of this paper is that the traditional metrological models are single individual models. That is, a sample is an individual. The attributes of the individual determine the value of its target variable. For example, the gravity model takes two countries' GDPs, population, Variables such as exchange rate are modeled as attributes of a sample. A sample is determined by two individuals. Such a model can offset the value of multiple attributes by difference. For example, the trade between France and the United States minus the trade between China and the United States can offset the related attributes of the United States. This greatly simplifies the model. Second, The traditional econometric models use independent variables to predict dependent variables, but it is not possible to use dependent variables to predict independent variables in reverse. However, the algorithm in this paper deduces that dependent variables can be used to calculate those independent variables with cross-action.
【学位授予单位】：湘潭大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F224;F742

【相似文献】