I 摘要 随着经济社会的飞速发展，国家竞争和区域竞争愈演愈烈，大量存在的企业 集群已经成为当今社会普遍的经济现象，我国各级政府将其作为区域经济发展的 重要途径和政策工具，学术界也将其纳入了研究领域的热点课题。 企业集群在广义上是指大量具有相互间联系的企业在地理上的集中，着重强 调经济活动主体在空间上的配置形式、运行关联以及企业之间竞合联系的内在机 理。创新技术是决定企业间关系的重要因素，失去创新与核心技术的企业无法立 足于竞争激烈的市场，迟早会被淘汰。众多集群研究者们也意识到了创新对于集 群发展的重要影响，在集群内大力倡导企业走技术创新之路。各级政府这双“看得 见的手”也在努力为推动集群企业的创新发挥重要作用。 但是，随着研究的深入，许多从理论上秉持着传统理性假设的研究成果在运 用到现实时不尽如人意，甚至产生了与预期完全相反的结果。不可否认，“理性经 济人假设”一度被誉为经济学最重要的假设论断，但在现实中面对复杂的市场环境， 决策者受制于心理偏差、认知偏差、计算能力和不完全信息等因素，无法进行完 全理性的决策，相反他们通常是在“有限理性”的前提下进行决策。“有限理性”至今 仍是一个没有完全界定的概念，学者们通常借鉴研究生物种群性状进化和稳定机 制的方法来分析人类的经济行为，并模拟博弈均衡解的动态实现过程。实际上， 动态演化分析的关键是确定博弈参与者的学习机制和策略变化的过程，如何根据 具体情况构造动态机制来模拟现实经济社会中博弈参与者的经济行为变化是国内 外学者一直致力解决的问题。20世纪80年代以来，越来越多的学者们将动态演化 思想引入到经济学领域，成功将其应用于社会制度变迁、产业演化以及股票市场 等；进入21世纪以后，国内的学者也开始关注动态演化理论，不仅在基本理论概 念和相关内容上做了更多延伸，还将基于演化理论的分析方法更为广泛地应用到 电力市场竞争问题、零售市场竞争问题、保险市场竞争问题以及政企关系研究等。 但是，这些研究成果仍然存在一定的问题。首先，演化博弈并不是演化观点 和博弈思想的简单相加，动态演化也并不一定就是演化博弈模型，许多文献对此 产生了混淆。其次，有些文献只是运用非常简单的演化博弈模型，研究不够深入。 由此，本文以集群中的主导企业为研究对象，基于有限理性的假设，构建集群寡 头主导企业在三个博弈阶段——创新决策阶段、创新研发竞争阶段以及市场竞争 阶段的动态博弈模型，结合演化博弈理论和非线性动力系统理论，针对有限理性 的寡头企业在各阶段博弈中的演化表现进行研究分析，得出具有现实经济意义的 结论与启示。主要研究工作及成果总结如下： 重庆大学博士学位论文 II （1）在创新决策阶段，引入“技术能力势差”概念，考虑在由一类龙头企业和 一类中小企业组成的创新集群核心价值网络里，技术知识从高位势龙头企业流向 低位势中小企业时，集群溢出系数、企业间位势差以及企业创新效率对于创新决 策的影响，并从动态视角分析了集群企业的创新决策演化过程。研究表明：集群 溢出系数和位势差对集群两类企业的创新动力呈现相反的激励作用；当龙头企业 和中小企业的创新效率处于某一特定区域时，因为博弈双方的相互作用，系统会 出现多种可能的创新状态。为达到集群最优创新状态，政府可以采用干预措施， 适度采用创新补贴、技术支持、知识产权保护等措施，帮助激发企业创新动力， 改善集群创新状态。 （2）在创新研发竞争阶段，构建带有政府创新补贴的有限理性双寡头创新企 业创新竞争动态模型，并对其进行稳定性分析，运用matlab2016a进行数值仿真， 研究分析在不同补贴力度下集群寡头企业在创新竞争中的动态表现。仿真结果表 明：为了维持产量博弈的纳什均衡，政府创新补贴率的取值受到需求函数、产量 调整速度以及产品成本的约束；当补贴率超过某一临界值时，产量的竞争就会进 入倍岔或混沌状态；当补贴率处于稳定区域时，在引入创新补贴政策的初期，双 寡头企业创新投入和利润将产生明显波动，并且增大其达到纳什均衡的难度（耗 时更长）。在补贴实施后，有限理性的双寡头企业均衡创新投入明显增加，远大于 其利润增幅。从平均创新投入和平均利润的角度来看，若补贴力度不当，双寡头 市场中创新投入调速较慢的企业将会经历一个创新投入递增而利润却递减的阶段， 并在整个竞争过程中处于劣势地位。由此，集群寡头企业需在关注补贴政策的同 时，在政策实行初期处于短暂不均衡状态时坚定自身创新投入的调整策略；而当 企业长期处于创新竞争劣势地位时，则可适当增加创新投入调速以求改变现状。 （3）在市场竞争阶段，分别构建因理性程度不同而具有各异行为决策规则的 异质多寡头价格竞争与产量竞争动态模型。之后以产量竞争模型为例，对其进行 稳定性分析，运用matlab2016a软件进行数值仿真，从中度理性的企业作为观察点， 深入分析不同行为规则给集群寡头企业在市场竞争中的行为动态变化带来的重大 影响。研究表明：需求函数、产品边际成本以及GD型企业特有的产量调整系数 共同决定整个产量动态演化轨迹及其最终状态。较小的产量调速有助于系统较早 达到Nash均衡点且在均衡点处保持稳定，而不合理的边际成本或产量调速则会导 致市场无序而进入混沌状态。延迟决策能够将系统从混沌无序的状态引回稳定均 衡点，并且能够通过调整控制因子来达到控制混沌产生时机的目的。 关键词：企业集群，技术创新，有限理性，动态演化 英文摘要 III ABSTRACT With the rapid development of economy, competition of nations and regions has become increasingly intense, and enterprise clusters have become a porpular economic phenomenon, receiving considerable attention. Now, national government has regarded this normal as the important access and political tool to contribute to economic development, and academic circles has taken it to be a hot research topic as well. Enterprise clusters refers to a large number of interconnected enterprises geographically concentrated in a broad sense, emphasizes the configuration form, operational association and internal mechanism inside the co-opetition of different enterprises. In fact, all the activities conducted among enterprises related to technological innovation, without which no one could survive in the highly competitive market. Many researchers have realized the significant impact of innovation on the development of enterprise clusters, and vigorously encouraging firms carry out activities of innovation to enhance the technology levels. Meanwhile, as the visible hand, government plays a vital role in the technological innovation of cluster. However, with the in-depth study, many research findings based on the traditional rational assumptions have proved to be failed on practical application, and some of them have even come out to be contrary to expectation. These signs that faced with the complicated market environment in reality, decision makers are subject to many influence factors, such as psychological bias, cognitive bias, computing ability, incomplete information and so on, so that unable to make completely rational decisions, although no one can deny that the hypothesis of economic man is the basis of microeconomics and the foundation of game theory. Hence, it is reasonable to say, most enterprise managers make decisions with bonded rationality. Therefore, based on bonded rationality hypothesis, this paper establishes three types of dynamic game model in three different competition stages, namely innovation decision making, R&D investment and quantity competition, combined with the evolutionary game theory and nonlinear dynamic system theory, aiming at analyzing the evolutionary performances of oligopolies with bonded rationality in game stages. At the end, conclusions and implications of practical economic significance are drawn from the research above. The main research work and findings are summarized as follows: 重庆大学博士学位论文 IV In the stage of technology innovation decision-making, this paper introduced the concept of “technology potential difference, and analyzed the significant influence of potential difference 、Tech-innovation spillover and innovation efficiency on the strategies selection of cluster enterprises with potential difference from views of static game and dynamic game separately. In addition, the factors caused market failures of innovation were explored and the effective solutions to encourage innovation among cluster enterprises were pursued. The study found that the motivation of leader firms for innovative activities has positive relationship with potential difference and negative relationship with knowledge spillover, while SMEs are just the opposite. And enterprise innovation efficiency coefficient shows complexity of the innovation game more directly. In the end, this paper offered reasonable suggestions on policy to optimize the game result, aiming to provide theoretical reference to decision-makers and regulators. In the stage of R&D investment, this paper created a dynamic model of R&D competition with government R&D subsi