作者简介：陶军锋，1973年7月出生。2003年6月6日在中国社会科学院研究生院获经济学博士学位。学科专业为西方经济学，指导教师是刘树成研究员。 摘 要 本文目的在于，在对内生增长理论（或者说是增长理论本身）进行深入研究的基础上，回答如下问题：什么是经济长期增长的源泉？或者说什么是决定经济长期增长绩效的最重要的因素？作为这一问题的延伸，下面这一问题似乎是自然而然的：在政府的边界之内，政府做什么能够促进增长——或者说政府应该做什么去促进经济增长吗？ 一个统一的框架或一条清晰的主线有助于我们很好的把握理论本身。然而，与新古典经济增长理论不同的是，内生增长理论并没有已为多数经济学家所共同接受的基本模型。确切地说，内生增长理论只是一些持有相同或类似观点的经济学家所提出的各种增长模型构成的一个松散的集合体。从对内生增长理论本身的研究现状看，现有研究成果并没有使得原本略显凌乱的内生增长模型变得清晰和易于把握。 国内外研究的一个共同特点是：把内生增长模型的理论特点归结为，强调经济增长不是外部力量（如外生的技术变化），而是经济体系的内部力量（如内生技术变化）作用的结果，并分别对内生增长模型所侧重的某一方面：如知识外溢、边干边学、人力资本投资、R&D、收益递增、劳动分工和专业化、开放经济和垄断化等进行了详细述评。上述观点无疑各自抓住了内生增长理论的一个侧面。然而，在缺乏一个统一的框架或一条清晰的主线的情况下进行理论研究，所得到结论无疑也是片面和不清晰的。 本文认为，构成内生增长理论的诸多模型之间既存在差别，又有一些共同的思想元素，以劳动分工为主线对内生增长理论进行研究不失为一条可行的思路。从某种意义上说，内生增长理论是对古典经济学的回归，其主要思想来源，就是Adam Smith、Marshall以及Allyn Young关于分工与增长的思想。Smith (1976)指出了劳动分工提高生产率的三条途径：（1）提高劳动者的熟练程度（人力资本积累）；（2）节约劳动者在不同工作之间的转换时间；（3）利于“简化和节约劳动的机械”的发明（物质资本积累）。不难看出，如果将亚当·斯密所提出的这三条途径“翻译”成现代经济理论中的术语，那么它们分别是：“干中学”（learning by doing）、“人力资本”（Human Capital）以及“内生技术进步”（endogenous technological progress）。 内生增长理论家一致认为，经济可以实现持续均衡增长，而技术进步是经济增长的决定因素。然而，技术进步又源自何方呢？显然，这与人类的知识积累有关——迄今为止，我们可以把人类知识归为两类：源于实践的经验积累和通过学校等独立教育部门获得的理论知识。对于前者，假设经济以工业为基础的话，我们似乎可以把它归功于工厂内部的分工；而对于后者，这可以视为社会分工的结果：家庭、学校和厂商中的研究部门显然都对其有积极影响。在内生增长模型中，知识（人力资本）积累或是厂商为追求利润对R&D有意识投资的结果（或无意识投资的副产品），或是来源于干中学、学校教育和有目的培训等厂商、家庭或个人对专业化人力资本有意识或无意识的投资。 内生增长理论的逻辑在于：通过分析技术进步的过程以及技术进步产生的原因来使技术产生于经济系统内部。在内生增长模型中，技术与知识或人力资本密切相联，一方面，内生增长模型中把技术、知识与人力资本视为同一概念，探讨其特殊性质所能带来的收益递增；另一方面内生增长理论把知识（专业化人力资本）积累视为取得技术进步的主要原因。 本文的思路是：经济增长与劳动分工之间存在正反馈。一方面，经济的长期增长离不开收益递增，而收益递增源于技术进步，技术进步又是劳动分工深化的结果，从这个意义上说，劳动分工决定了经济长期增长。另一方面，经济增长扩大了市场范围，又有助于加深专业化和劳动分工。基于劳动分工、专业化人力资本积累、收益递增在经济增长中的重要性，我们从一个独特的视角对内生增长理论展开研究：一方面，分工表现为新行业的出现及生产迂回程度的加强，分工经济是一种多样化经济；另一方面分工表现为工人将越来越多的时间用于生产同一产品，即工人专业化程度的加深，分工经济是一种专业化经济。分工的深入会促进人力资本的积累，进而促进发明、创新。社会分工对人力资本的积累的促进包括学校、政府、家庭、厂商对人力资本的直接投入，以及政府政策、家庭和学校对R＆D的直接、间接影响；而工厂内部的分工主要通过干中学以及R＆D部门的贡献促进人力资本的积累。人力资本存量的提高有助于发明和创新的产生。作为技术进步的两种重要形式，发明和创新意味着新产品或新的生产方法的引入，从而导致产品种类的增加和现有产品质量的改进。发明和创新同时意味着原材料或半成品的新的供给来源以及新市场(新地区的出口市场—国际分工或新的类型的市场—例如金融市场)的出现。发明和创新的增加，一方面会引起经济结构和市场结构的改变，新的工业组织——特别是组成托拉斯或某种其他类型的垄断组织会创立，另一方面，市场结构也会影响企业创新行为的选择，进而影响经济长期增长的平衡路径。最后，研究还涉及在厂商(以及国家)之间创新能力、投入效率、产品技术、行为和战略规则之间的不平衡与多样性所导致的技术扩散。 论文首先概述了劳动分工与经济增长的理论脉络与基本模型，然后分两章论述了新产品出现型内生增长模型和专业化加深型内生增长模型；第四章对涉及劳动分工所引起的市场结构的变化进而引起的企业内部创新、新市场的出现（国际分工）与技术扩散、新产业的出现或国际分工中由于工资率的差异引起的人口迁移等方面的内生增长模型作了论述和评价，最后一章是本文的结论和对内生增长理论的一个整体评价。在这一章中，我们对内生增长理论的理论贡献（包括理论突破、实证研究即对现实的解释力以及政策内涵）、分析工具、方法论以及理论框架等内容作了全面评价，并对本文研究的不足之处及值得继续研究的地方作了说明。 本文的特色以及所得到的几个有普遍意义的结论在于： 1.研究视角独特。论文以劳动分工为主线，对自Romer(1986)以来内生增长理论的发展做了梳理；评价了完全竞争与垄断竞争条件下的新熊彼特主义与达尔文主义内生增长模型，并考察了分工引起的市场规模扩大到国际范围的内生经济增长模型；对内生增长理论的最新进展，如：收入分配、金融市场、产业组织与内生增长通过分工与专业化人力资本积累的视角进行了综述；对内生增长理论的实证研究和政策内涵加以了强调。 2. 探讨了长期经济增长的源泉问题，认为劳动分工是报酬递增的根源，是经济增长的出发点，从而也是决定经济长期增长的源泉。经济增长最重要的理论基础在于劳动分工的演进，技术进步之所以可以内生，在于它是这个关系演进的结果。卷入市场的产品种类的增加，市场一体化程度的提高，新企业的出现，生产率的提高，市场的扩大，收入的增加，都是劳动分工加深的若干个侧面。 3. 阐明了经济学的核心问题是生产率和劳动分工的关系，对经济增长理论的基本问题,如：分析工具、方法论以及政策导向等内容作了全面思考。与流行的注重技术进步，建议国家对研究和开发进行补助的观点不同，本文对内生增长理论的研究得出了一些有意思的政策建议：就促进技术进步而言，政府干预并不总是必要的，即使在需要政府有所作为的情况下，对R&D进行补贴是不是最好的方式，也没有统一答案，最优的政策选择应该根据具体情况而定。在某些不常见但可能发生的情况下，合理的政策不是对研究和开发进行补助，而是对R&D的投资进行征税。 关键词： 劳动分工 人力资本 收益递增 内生增长 Abstract Through the research on the endogenous growth theory, this paper addresses the following question: what is the determinant factor of the sustained long-run growth? In addition, should government do something for the long-run growth or how could government policy be used to promote economic growth? In economic research, a basic analysis framework is helpful to understand the theory itself. However, unlike the neoclassical economic growth theory, the endogenous growth theory till now has not established the general analysis framework accepted by most of the economists. In fact, the current research findings on the endogenous growth theory itself are consisted of different models created by those economists having similar opinion on the growth theory and are not clear enough. Current researches on the endogenous growth theory revealed that it is not the exogenous but endogenous variables (such as endogenous technological change) provide the engine for the long-run growth, and conducted in-depth analysis of one particular aspect of endogenous growth models such as the spillover effect of the knowledge, learning by doing, human capital investment, R&D, increasing returns, division of the labor and specialization and monopoly. All these researches undoubtedly, reflect one side of the endogenous growth theory; however, the conclusions are partial and not so clear, because they were done without a general framework. There do exist differences among the endogenous growth models, but still we could find some common points of the theory. The paper maintains it is feasible way for us to develop the research based on the division of labor. To some extent, the endogenous growth theory derives from the classical economics and fertilized by Adam Smith、Marshall and Allyn Young’s thoughts on the division of labor and the growth. Smith (1976) suggested that the division of labor can improve the productivity in three ways: first, improve the workers’ skill (human capital accumulation); second, saving the workers shifting time from one job to another; finally, promoting the machines invention (material capital accumulation). We can interpret the above three ways with the contemporary economic theory term: learning by doing, human Capital and the endogeneization of technological progress. The endogenous growth literature pointed out that the long-run growth rate was determined by the growth rate of technology. But where does the technological progress come from? Obviously, it is related to the knowledge accumulation. As we know, human knowledge falls into the following two categories: knowledge accumulated from practices and experiences, and knowledge gained through independent education such as from school. As to the former, we can owe it to the division of labor inside the firms in the industrial economy. And the latter was affected by the social division of labor: family, school and the R&D department inside the firms obviously have active effect on knowledge accumulation. According to the endogenous growth model, knowledge (human capital) is generated as a product of profit-aimed entrepreneurs’ consciousness investment (or a byproduct of unconsciousness investment), or comes from firms, families and individuals’ consciousness or unconsciousness investment in human capital through learning-by-doing, school education and training. The logic of the endogenous growth theory is that technological progress through the analysis of the process and the generate reason of technological progress. According to the endogenous growth model, technology is closely related to knowledge and human capital. On one hand, the model treated technology, knowledge and human capital as the same concept and focused on the increasing returns derived from the nature of technology ; on the other hand, the endogenous growth theory take knowledge (human capital) accumulation as the main resource generating technological progress. The logic of this paper is as following: The logic of this paper is as follows: there is positive feedback relationship between economic growth and division of labor. Growth of the variables such as increasing returns, technological progress and division of labor are interdependent, as causation run from division of labor to technological progress then increasing returns and output, as well as from growth to market size and division of labor. Based on the importance of the division of labor, the specific human capital accumulation and the increasing returns for the long-run growth, we expand the research from a unique perspective: on one hand, the division of labor take forms as new industries emergence and the reinforce of the product roundabout, so the division economy is a kind of diversity economy; on the other hand, the division of labor also takes forms as specialization economy in which workers spend time on specific producing task, so the development of the division of labor will accelerate human capital accumulation and result in the higher rate of invention and innovation. The innovation will first change the economic structure, then affect the choice of the firm’s innovation behavior, and finally influence the long-run growth path. The first part of the paper gives a brief review on the theory of the division of labor and economic growth, and then discusses the new product emergence-type and the specialization-type endogenous growth model in the following two chapters. The fourth chapter focuses on the market structure changes, the firm’s innovation behavior caused by the division of labor , and international division of labor with technological diffusion and population transference. The final chapter concludes the paper and we give a full appraisal of the endogenous growth theory, include theoretical contribution, empirical research and the explanation ability to the real word. Of course, it also mentioned some of the weakness of the paper and pointed out some valuable research topics. The innovation and major conclusions are as follows: 1. The unique research perspective. Based on the division of labor, the paper established an analysis framework and further takes a lucubration on the endogenous growth theory.2. Proved that the division of labor is the determinant factor of the sustained long-run growth. The evolution of the division of labor stands the most important theoretical fundamental of growth theory. The reason that the technological progress can be endogenize is that the technological progress is one of the products generated by such evolution process. The increase of product variety, the enhancement of market integration, the emergence of new enterprises, the improvement of productivity, the enlargement of market and the per capita growth of income are some aspects of the development of the division of labor. 3. Clarified that the core issue of economics is about the relationship between productivity and division of labor. Also, the paper discussed the basic problems of growth theory , such as its analysis tools, methodology and the policy indication. The research reached some interesting but important conclusions about R&D policies. Contrary to the widespread popular view that countries tend to underinvest in technology and that the government should do something for it, we suggest that despite market failures (because of imperfect competition, externalities, and increasing returns), it does not mean the government should always intervene in R&D and it is not clear that the best way to deal with them is to subsidize R&D. In fact, in case of imperfect competition, the optimal policy to offset the distortion, however, is not R&D subsidy but subsidy to the purchases of the overpriced goods. Another distortion may arise from the externalities within the structure of R&D costs. The problem is that it is not clear whether a new market invention increases or decreases the cost of future inventions. If the cost declines, firms doing R&D tend not to internalize all the benefits of their inventions, so there tends to be underinvestment in R&D. In this case, the correct policy is an R&D subsidy. However, if the costs increase with the number of inventions, current researchers would exert a negative externality on future researchers, so they tend to overinvest. In this case, the required policy becomes an R&D tax rather than an R&D subsidy. Key words: the division of labor, human capital, increasing returns and endogenous growth