产业技术与制度的共同演化分析——基于多主体的学习过程

黄凯南,乔元波. 产业技术与制度的共同演化分析——基于多主体的学习过程 [J]. 经济研究，2018，12：161－176.

摘要：技术与制度存在较为复杂的关联, 本文尝试从多主体学习过程来考察产业技术与制度的共同演化, 推动该领域的研究。在阐释技术和制度共同演化机制基础上, 构建理论模型, 运用一个包含技术创新的复制者动态模型来描述企业内部的技术演化过程, 考察作为企业内部共同学习规则的制度对企业研发投资决策的影响, 进而影响企业内部的技术演化和企业自身的适应度, 在产业层面, 技术结构和制度结构处于共同演化中。通过仿真模拟分析, 考察不同关键参数下产业技术和制度的共同演化, 剖析共同演化过程的创新效应、模仿效应和规模效应及其对产业总体绩效的影响, 探讨相关产业政策, 最后进一步展望未来研究方向及其对我国产业转型升级的启示。

关键词：共同演化;多主体;学习过程

Co-evolution between Industrial Technologies and Institutions: Based on Multi-agent Learning Process

Abstract：

At present, China is actively transforming the mechanism driving its growth from factor investment to innovation, by promoting industrial transformation and upgrading and promoting high-quality economic development. Industrial transformation and upgrading is not only accompanied by upgrading product structure, factor structure, and demand structure, but also by a coordinated upgrading of technological and institutional structure that involves complex interaction and learning among multiple levels and agents. Analyzing the co-evolution of industrial technologies and institutions will not only reveal the internal dynamic of industrial evolution, but also has theoretical and practical significance for accelerating China's industrial transformation and upgrading. This paper explores the co-evolution of technologies and institutions from the perspective of multi-agent learning. Technological evolution includes not only the expansion of technological boundaries, but also changes in the shares of different technologies in a given area. The former is innovation-driven and the latter is driven by structural change. Compared with the emphasis of new institutional economics on reducing transaction costs, evolutionary economics emphasizes cognitions of institutions. Institutions not only influence interaction among participants by influencing transaction costs, but also influence their learning behavior by shaping the rules of learning. Institutional evolution includes not only the emergence of new institutions that change the original institutional space, but also changes in the proportions of different types in a given institutional space. As technology and institutions jointly determine the fitness or economic performance of participants, the evolution of the two will lead to co-evolution. In the proposed theoretical model, technology and institutions as a common learning rule establish a co-evolution relationship by influencing an enterprise's fitness. The evolution of the technology share affects the learning rule share by affecting the enterprise's fitness, and the evolution of the learning rule share affects the technology share by the same route. Conservative enterprises that develop at a constant capital growth rate without learning and innovative enterprises that apply trial-and-error methods lag behind conservative enterprises in the early stage of development and surpass them in later stages. The greater the R&D investment, the higher the capital growth rate in later stages. Enterprises that imitate the most innovative enterprises develop fastest among the three types of enterprises. The broader the scope of imitation, the higher the capital growth rate. The industrial technology share is ultimately determined by the technology share of innovative and imitative enterprises, and imitative enterprises play a major rule. The greater the technology restructuring rate, the higher the new technology share. The institutional share of conservative enterprises increases first and then decreases; the institutional share of innovative enterprises decreases first and then increases; and the institutional share of imitative enterprises increases all the time. The R&D effects and imitation effects promote the growth rate of the industry. When the number of enterprises is small, scale effects promote industrial growth, but when the number of enterprises is large, the industry growth rate fluctuates as the number of enterprises increases. Simulation results verify the theoretical analysis. Co-evolution is the process of diversity generation and reduction. Under the influence of the selection mechanism, highly adaptive technologies and institutions stand out, poorly adaptive technologies and institutions are weeded out, and thus diversity is gradually reduced. In the early stage of industrial development, the government must give appropriate subsidies to innovative enterprises or increase the imitation cost of imitative enterprises. Once the enterprises and industry have escaped the new technology trap, industrial policy should switch from incentives for innovative enterprises to incentives for imitative enterprises. Industrial development policies must trade off current efficiency and future efficiency to improve dynamic adaptive efficiency.

Keyword：

Co-evolution; Multi-agent; Learning Process