【作者】 朱海霞；

【导师】 顾海英；

【作者基本信息】 上海交通大学 ， 企业管理， 2008， 博士

【摘要】 边境效应,(border effect或home bias),是指边境对跨边境经济行为的影响,这种影响与边境特有的政治、经济、文化、社会等属性密切相关的。基于引力模型的边境效应方程简单清晰地反映了一国内各地区之间贸易量和不同国家两个地区之间贸易量的差异,同时考虑了距离等贸易壁垒,目前已被成功地应用在双边贸易量的研究中,用来衡量国家或地区的开放度和贸易壁垒等问题。一方面,通过农产品边境效应研究,可以深入分析中国的农产品贸易流量、流向及其影响因素和变化趋势,在提升中国农产品国际竞争力、促进国内经济一体化,扩大中国农产品出口等方面具有重要的现实意义:伴随中国经济国际化进程的加速,中国农产品的生产贸易和消费状况对世界农产品贸易格局和市场状况的影响越来越重要。虽然中国是世界农产品第一生产大国,但其出口总额在过去的二十年里仅占世界农产品出口总额的2%-4%,有必要对中国农产品贸易的国际、国内流量和流向进行深入研究,寻求扩大中国农产品国际出口、缓解农产品贸易赤字的对策建议。美国虽然与中国距离遥远,但近十年美国占中国农产品出口总额中的份额迅速增加,立刻成为中国农产品贸易主要伙伴,同时中国对美国的农产品贸易赤字压力很大且争端不断;另一方面,通过中美农产品贸易的边境效应研究可以巩固现有引力模型的理论基础,在模型的鲁棒性检验和在农产品行业贸易的解释性、中国农产品贸易模式分析、贸易政策的提出等方面具有一定的理论意义。然而,边境效应引力模型还需要进一步研究,主要表现在以下几个方面:1)模型的理论基础仍相对薄弱:不同学者根据自己的偏好和研究对象采用不同的模型,得出的结果也有所差异。因此需要对现有引力模型及边境效应模型进行总结和比较研究,寻求不同模型的特点和适用范围,为实证研究提供坚实的理论基础。2)现有的引力模型已经对世界很多国家或地区之间的边境效应问题做了较为细致地研究,但是绝大多数研究是针对两国或两个地区之间的贸易总量,也有部分文献利用SITC分类进行行业分析,与人们生活直接相关的农产品贸易的边境效应却较少有人问津,引力模型是否适用于农产品贸易,哪些因素会影响农产品双边贸易量,影响程度有多大都有待研究,因此需要加强农产品贸易引力模型理论方面的研究。3)从研究对象看,国外的实证研究多为发达国家或地区,对于发展中国家和发达国家之间的边境效应研究尚处于起步阶段,发达国家和发展中国家之间的边境效应是否与现有的研究结果相同也有待研究;目前中国的农产品贸易研究局限在省际或市际贸易,因此在新时代有必要对两个农业大国-中国和美国之间的农产品贸易模式及发展趋势进行研究。4)目前的边境效应研究集中在边境效应的存在性及其成因方面,有必要进一步研究边境效应的显著性和变化趋势。同时,贸易流向和地区差异究竟会对边境效应产生多大影响也有待实证检验。5)由于引力模型的发展经历了先实证研究后理论研究独特的发展轨迹,因此需要在目前文献研究的基础上对引力模型的检验进行归纳总结,并对检验结果进行比较分析。为此,本文在总结国内外现有引力模型和边境效应研究成果的基础上,以中国和美国农产品贸易为研究对象,以统计学、贸易理论为研究工具,以定量实证研究为主,结合定性分析讨论,集中归纳总结了基于引力模型的边境效应理论及模型推导,提出了基于引力模型的、适用于农产品贸易的边境效应模型:1)基于贸易客观实际和计量方法根据模型截距个数分为单一截距模型、固定截距模型和变截距模型;2)基于贸易流向和地区差异将每个模型进一步分为:标准模型(对称同质)和非齐次模型(非对称性、异质性和变截距)。分别给出了每个模型各变量之间关系的分析比较,阐述了各模型的性质、特点和应用范围,建立了中美农产品贸易边境效应度量方法,评价了多个模型在反映中美农产品边境效应的绩效,采用多种方法检验模型的鲁棒性,并在此基础上具体阐述了中美农产品贸易边境效应的存在性、显著性和变化趋势,探悉了中美农产品贸易边境效应形成和变化的原因,最后提出相应的政策建议,完善当前现有边境效应引力模型的理论和实证研究。本文的主要研究工作及成果可归纳如下:第一章绪论本章在系统地阐述边境效应引力模型研究现状的基础上,分析了研究中存在的不足,在此基础上提出了本文所要解决的问题。第二章基于引力模型的边境效应基础理论本章重点阐述了边境效应的研究发展,在详细分析贸易引力模型理论的基础上,介绍了贸易引力方程在不同贸易理论前提下的推导过程,深入讨论了边境效应引力模型的三种基本形式的推导过程,为后文的实证研究奠定理论基础。并根据目前边境效应引力模型的实证研究,总结了这一模型的应用范畴。第三章中美农产品贸易边境效应标准模型研究及实证检验本章运用引力模型研究了中国和美国在农产品贸易领域的边境效应问题。在简要介绍了引力模型推导的基础上,通过引入与农产品贸易相关的变量,构建了适合于本文实证研究的农产品贸易边境效应引力模型的标准模型。然后详细介绍了实证研究中各种数据的选取方法及数据特征。并在此基础上,采用1987至2005年的截面数据,回归了中美农产品贸易的边境效应,研究中美农产品贸易边境效应的显著性和变化趋势,并对模型进行了鲁棒性检验和非线性检验。结果表明:1)同其它总量贸易的研究比较,边境效应引力模型也同样适用于单一行业的农产品贸易的边境效应研究。2)同其它贸易产品一样,引力模型中的主要解释变量-经济规模和距离也会对农产品贸易产生影响:分别同双边贸易量正相关和负相关;农业总产值占GDP的份额越大、农业人口比重越小,农产品出口就会越多。3)中美两国在农产品贸易的边境效应显著存在,且呈下降趋势,由1987年的87.4下降到2005年的7.0。边境效应非常显著,但通过同其它农产品贸易边境效应的研究相比,可以发现经济差异越大的国家之间的农产品边境效应数值相对越小。中美两国之间的农产品贸易边境效应可能是由于政府的农业保护政策、贸易成本的差异和中美两国在农产品生产、技术差异等方面的原因造成的。随着中国加入WTO、经济的飞速发展和对三农问题的逐渐重视,农产品贸易的边境效应也有所下降。4)五种不同的模型鲁棒性检验可以证明本章实证模型的可靠性。模型中距离变量是否平方并不影响模型的最终结果,对数线性的引力模型可以较好地反映农产品边境效应。第四章中美农产品贸易边境效应非齐次模型研究及实证检验本章在第三章边境效应引力标准模型理论和实证检验的基础上,根据客观贸易状况,首先构建了非齐次模型,即根据国际贸易流向和地区差异构建边境效应的非对称同质模型、对称异质模型和非对称异质模型,用来研究中美农产品贸易流向和地区差异对模型回归结果的影响和边境效应的显著性和变化趋势。然后,根据每一对贸易伙伴的个体差异构建了变截距模型,采用固定效应法回归了四种中美边境形式下的模型,并同单一截距模型和固定截距模型的结果进行了比较分析。最后,根据1987-2005年的面板数据,分别采用普通最小二乘法和混合最小二乘法对中美农产品贸易流量、流向、中国各地区和每对贸易伙伴的差异,以及中美农产品贸易边境效应显著性进行了实证研究,最后比较分析了单一截距模型、固定截距模型和变截距模型的异同。研究结果表明:1)模型的回归结果同目前的文献研究结果和第四章结论基本相同:经济发展水平-农业总产值和人均农业总产值正面影响双边农产品贸易量,距离变量虽然对双边贸易量有负面影响,但其影响力逐年下降,表明中美两国之间农产品贸易成本的不断降低和贸易障碍的不断缩小。同时,所有模型都表明出口地区的农业总产值占其GDP的份额越大,贸易伙伴之间的距离越近,进口地区的土地面积越小,出口地区的农业人口比重越小,双边农产品贸易量越多。而进口地区的农业总产值占GDP的份额、农业人口比重和和出口地区的土地面积对双边贸易量的影响则不确定。2)非齐次模型,尤其是变截距模型中的异质模型能够更加准确地反映中美农产品贸易模式,因此边境效应的研究更具理论和实证依据。3)各模型的的比较:同质模型和异质模型的差异在于同质模型的常数项大于异质模型,异质模型的距离变量的负面影响大于同质模型;固定截距模型、变截距模型和单一截距模型的差异在于固定截距模型的常数项符号不确定,但后两个模型的常数项则大于0;固定截距模型的距离变量对双边贸易量的影响程度最小,系数都在0--1之间,而变截距模型的距离变量对贸易量的影响程度最大,系数都小于-1。4)边境效应值在各个模型中都很显著,但是个体差异越大,边境效应的显著性水平相对越高。中国和中国各地区对美国的农产品出口边境效应都显著大于其对美国的农产品进口边境效应,反映了中国各地区的农产品出口更多流向国内其它地区,而中国和中国各地区的农产品国内进口和从美国进口的差距却相对较小;西北、中部地区和东部沿海的边境效应相对最大,北部沿海和南部沿海的边境效应最小,说明经济发展水平同边境效应具有一定的相关性,经济发展水平越高,交通越便利的地区对美国的进出口边境效应越小。第五章中美农产品贸易边境效应的分析本章在前两章中美农产品贸易理论和实证研究的基础上,深入分析了中美农产品贸易边境效应的显著性和变化趋势的成因,主要因素包括农产品易腐蚀、变质和不易保存的自身特点导致农产品国际贸易成本增大,两国各级政府不同时期不同程度的农业保护政策增加国际贸易壁垒,同时中美农产品贸易在新时期迅速增加以及汇率的波动、人们生活水平的提高和偏好的变化都导致中美两国的国际贸易相对于中国的农产品国内贸易增幅较大且不平衡,中国经济区域的不均衡发展导致中国各地区对美国的农产品进出口边境效应存在较大差异,且在过去的二十年里发生巨大变化,因此边境效应虽然显著但呈现总体下降趋势且存在边境效应进出口上的差异。第六章中美农产品贸易进一步发展的对策建议本章在前三章定量研究和定性分析的基础上,首先就当前促进中国农产品贸易的措施进行了回顾评述,然后提出了基于本论文提出的农产品边境效应引力模型,提出中美农产品贸易进一步发展的对策建议,为促进中国对美国的农产品出口和政府农业政策的制定提供依据。根据模型中的自变量和因变量之间的关系,主要提出四点建议:扩大中国的经济规模和农业经济规模、加快农村劳动力的转移以降低农业人口比重、减少农用土地的流失保证农业生产和农民收入、通过科技发展创新和提高人口素质缩小中美差距、均衡发展中国各地区经济实现经济一体化。第七章结论与研究展望本章对全文进行了总结,并对今后的研究做出了展望。本文主要创新点有:1.综合分析比较贸易引力模型和派生的边境效应模型的理论基础,建立基于引力模型的边境效应模型的分析框架,总结评述边境效应引力模型的应用范围及其绩效。现有国内关于引力模型的研究相对较少,主要集中在对模型中的“距离”变量的探讨及对中国贸易流量的实证研究,而引力模型的理论基础研究没有得到足够的重视,对派生出来的边境效应研究更是相对匮乏。这样边境效应引力模型作为政策分析工具的功能就会受到质疑。本论文对引力模型和边境效应模型进行了综合地分析比较,建立基于引力模型的边境效应模型的理论分析框架。2.根据农产品贸易的特点,构建两国边境效应模型,并采用这一模型实证检验中国和美国之间的农产品贸易边境效应,考察两个农业大国之间的边境效应存在性、显著性和变化趋势。在现有的边境效应文献中,本论文首次将引力模型同中美农产品贸易相结合,通过引入与农产品生产和贸易的相关因素,构建了边境效应的两国模型。由于中国相对经济发展不平衡,各地区在农业生产和农产品贸易上存在着巨大的差异,为了更好地揭示现实的经济现象,模型又进一步分为标准模型和非齐次模型:由于中美两国在政治、经济、文化、人口等方面的差异使得两国的农产品贸易存在着较大的差异,因此边境对两国的农产品贸易会产生不同的影响,同时,中国国内各地区的农业发展和需求存在着很大差异,模型还充分考虑了边境效应的非齐次性,在不同贸易方向上和在不同样本地区上的差异,即非对称性和异质性,并采用固定效应法检验中美两国每一对贸易伙伴之间的固定因素(模型截距或常数项),以考虑任何其它不可度量因素对每一对贸易伙伴农产品双边贸易产生的影响。3.对实证模型进行鲁棒性检验和非线性回归检验,考察本文构建的边境效应引力模型的可靠性,为解释和预测中美两国的农产品贸易客观经济现象提供良好的依据。目前的边境效应研究实证研究虽然很多,但是对模型进行检验的相对较少,本文首先总结评述现有的模型鲁棒性检验方法,并在此基础上采用多种不同的方法分别对本文的实证模型进行检验,同时采用解释变量不同次数的回归检验,进一步考察线性模型的可靠性,为解释和预测中美两国的农产品贸易模式奠定基础。本文通过边境效应引力模型探析中美两国间农产品贸易的变化因素和特征,以期判断中美两国农产品贸易领域的合作竞争关系,为加强双方在农产品贸易领域的合作交流提供政策建议。更多还原

【Abstract】 “Border effect”or“home bias”refers to the extent to which volume of domestic trade exceeds the volume of international trade. In other words, two different countries trade much less with each other than do two regions within one country, taking into account income, size and distance. One way to estimate the border effect is through Gravity Model of bilateral trade flows. The deduced border effect gravity model has been used successfully to measure the degree of fragmentation or integration of the economy, given that it will be observed if the access to the market in a specific region is different whether we consider local producers or producers of other regions.On the one hand, with China’s fast economic development, its agricultural production, trade and sales have an increasing effect upon world agricultural trade pattern and market condition. Though China is world first major agricultural producer, its agricultural export accounted for only 2-4% of the world total agricultural export in the past two decades; U.S. is comparatively far away from China, but it turns out to be one of the major agricultural trade partners during the past decade, causing Chinese large number of trade deficit in agriculture. Thus it’s necessary to study Chinese trade flow and direction of agriculture within China and with U.S., their influential factors and evolution over time in order to promote Chinese agricultural export to U.S. and increase international competitiveness of Chinese agricultural goods. On the other hand, this study helps to consolidate the theorectical foundation of Gravity Model and provide effective measures and policies in agricultural trade.But the border effect model is not without its problem. Firstly, current literatures about border effect gravity model prefer empirical study to theorectical study, so this model has been criticized by many scholars for its lack of therectical foundation. Different scholars apply different forms of model specification and different methods of variables’calculation to study the same issue, resulting in various magnitudes of border effects. Secondly, although there is vast literature about border effect in different countries or regions, they mainly focus on total merchandise trade. The study of agricultural trade using border effect gravity model needs more attention due to the non-substitutability of agricultural trade. Whether the same independent variables are involved in the Gravity Model and whether the border effects of agriclutral trade show the same pattern in magnigude and evolution as that of total merchandise trade remain unknown so far. Thirdly, many studies have applied the model to developed countries or regions, but the border effect study of developing countries or regions which enjoys a great percentage of the world both in polulation and in agriculture seems to have been ignored so far. Whether there are any differences in border effects between developing countries and developed countries and those among developed countries needs to be studied. China is the largest developing country and U.S. is the largest developed country and both are major countries in agriculture, but the study of these two countries seems to be untouched upon in past literature. Finally, most of the empirical studies of border effect provide the explanation of causes in the great magnitude of border effects, but how border effects change over time and across regions, and how different model specification and data choices influence the model results, namely robustness checks, need to be further studied to promote the reliability of the established model.By concluding the current literature, this dissertation establishes the border effect framework based on gravity model. Using regression method-OLS and PLS as research methods, this paper first presents two classifications of the model: 1) four border effect gravity models based on the direction of international trade and regional differences; 2) three models according to the number of the intercept to discriminate the discrepancies in each trading-pair over time, then it investigates the parameter estimation method, analyzes the relationships between dependent variables and independent variables, evaluates different models in interpretating border effects, presents five kinds of robustness check of the model, generalizes border effects in magnitudes and evolution over time and across regions of each model, and finally puts forward the causes of China-U.S. agricultural trade border effects and corresponding policy suggestions, which have been not touched upon in the current literature.The main contents and conclusions are summarized as follows:1 This dissertation analyzes the shortage in the current literature after setting forth the status of gravity model and border effect, and introduces the possible solutions provided by this paper.2 The second chapter shows the historical development of gravity model, introduces the concept, arrives at the gravity model based on different trade theories, studies and evaluates four kinds of border effect gravity model, discusses the calculation methods of border effect in detail. By comparing the existing method, it puts forward the differences of the models and their application in reality. Border effect gravity model is a new trade model which can precisely measure the difference between internantional trade and intra-national trade though it has its limitation. 3 The third chapter explores McCallum’s standard border effect equation derived from the Gravity Model both in theorectical and empirical study. First, border effect gravity model is derived to fit the two-country agricultural trade by adding some variables capturing agricultural characteristics, which serves as the theoretical base of the empirical study. Then it uses the panel data covering from 1987 to 2005 to study the China-U.S. border effect of agricultural trade in magnitude and evolution over time using the Gravity Model, which has not been touched upon so far in the past literature. The results first indicate that border effect gravity model fits well in agricultural trade as well as total merchandise trade. Second, the main determinants of trade in a gravity framework--namely economic size and distance--have the same impact on agricultural trade as they have on total merchandise trade. As for new variables specific to agricultural trade, the estimates show that the exporter’s higher share of agriculture in GDP, the importer’s fewer land area and a lower rural population density are associated with higher bilateral trade flows of agricultural products. Border effect does exist between U.S. and Chinese agricultural trade. It is large in magnitude and tends to drop from 87.4 in year 1987 to 7.0 in year 2005. The border effect might result from the great geopolitical characteristics of two nations and historical characteristics of agriculture. China faces great challenges in order to comply with WTO rules and deal with agricultural issues. It has to strike a balance between economic reforms and inherent adjustment costs and to limit local government’s economic interventionism.4 The fourth chapter further revises the standard border effect gravity model and establishes three other fixed-intercept model namely asymmetric homogeneous model, symmetric hetereogeneous model and asymmetric hetereogeneous model according to the direction of international trade and regional differences. From the panel data of year 1987 to 2005, the empirical study of China-U.S. agricultural trade reaches several conclusions. Firstly, the main determinants of trade in the gravity model--namely economic size and distance--have the same impact on agricultural trade as they have on total merchandise trade. For new variables capturing features of agricultural trade, the estimates show that a higher share of agriculture in GDP and a lower rural population density are associated with higher bilateral trade flows of agricultural products. Secondly, it’s found that the China-U.S. agricultural trade is asymmetric and differs greatly from regions. What’s more, regional discrepancies play a more important role in deciding the intercept and coefficients of independent variables of the model than the direction of international trade. Thirdly, the import and export border effects between U.S. and Chinese agricultural trade both exist over the period and are significant. Eight Chinese regions share the same pattern of border effect in magnitude and evolution over time as the border effect found in the fourth chapter, but border effects vary across regions. Generally speaking, the level of economic development, the convenience of transportation is negatively related to the magnitude of border effect. On the basis of fixed-interpect model, the changed-intercept model is put forward to catch unique characteristics of each trading-pair and the single-intercept model is compared with the former. Both models are also classified into symmetric homogeneous model, asymmetric homogeneous model, symmetric hetereogeneous model and asymmetric hetereogeneous model acooding to the types of the China-U.S. border. By using fixed-effects and pooled least square method, empirical study indicates that border effects are still large in all revised models and variables like economic sizes, distance, the share of gross agricultural product in GDP, land area and rural population density continue to affect the bilateral trade. Conclusions can also be made that border effects tend to be larger in the changed-intercept model than in the other two models. The more regional differences are concerned by the model, the bigger influence distance will have upon the bilateral trade. Finally, robustness check provides the reliability of the models put forward in this chapter.5 Based on both theorectical and empirical study of Chapter Three and Four, the fifth chapter generalizes what factors lead to the great magnitude and evolutionn over time across regions of border effects in all models by exploring the internal and external causes. It is found that the inherent characteristics of agricultural trade, government’s protection policies, new change of China-U.S. agricultural trade, the non-equilibrium in development of Chinese domestic regions and the fluctuating exchange rate between China and U.S. constitute major factors resulting in the great magnitudes and evolution of border effects between China and U.S. agricultural trade.6 The sixth chapter first generalizes current suggetions on how to promote Chinese agricultural trade and then provides the possible solutions dealing with the issue of promoting Chinese agricultural export to U.S. and bilateral agricultural trade based on the border effect gravity model analysis, namely, expanding the scales of economies, especially those of agricultural economy, accelerating and facilitating rural labor shift, reducing soil erosion and bridging the international and domestic gap in farm.7 The seventh chapter concludes the dissertation and provides some prospects for the future study. The primary innovations include:1 Presenting the development of Gravity Model, arriving at the equations of border effect model in different trade theories and market conditions, establishing a framework of border effect gravity model followed by the introduction to its application in different fields.2 Putting forward two-county border effect model capturing the characteristics of agricultural trade, and further revising the model into three models according to the number of intercept and into four models according to the types of China-U.S. border in order to study the border effect of China-U.S. agricultural trade in magnitude and evolution over time.3 Generalizing various methods of robustness check of border effect gravity model in current literature and providing five kinds of robustness check to evaluate the reliability and linearity of border effect gravity model used in the empirical study of this paper, paving the way for interpreting and forcasting the bilateral agricultural trade volume and pattern.更多还原