技术新颖性从何而来?——基于纳米技术专利的分析

Identifying the sources of technological novelty:Empirical analysis based on nanotechnology patents

本研究以知识组合理论为基础,将技术新颖性看作是知识单元组合的产物,结合社会网络理论和演化理论的观点,提出知识单元的多样性和依赖度影响技术新颖性的理论框架。知识多样性和依赖度的影响分别通过组合的潜力和组合的实现两种机制予以传导。知识多样性一方面通过提高知识丰富性以及可能组合的方式来增加新颖组合产生的可能性,另一方面会提高吸收、消化这些异质性知识的成本和难度。知识依赖度的影响也存在正反两个方面,正向的影响表现在可以提供以往知识组合的经验从而降低组合的难度,反向的影响表现在路径依赖和僵化制约了新颖的知识组合产生的可能性。以专利数据为样本的负二项回归结果表明,知识多样性和知识依赖度分别对技术新颖性有正向和倒U型的影响,且两者的交互项对技术新颖性的影响显著为正。这一结论在稳健性检验中得到进一步的支持。本研究的贡献在于:第一,打破以个体、团队或组织为对象的研究将创新活动同质化处理的假定,探索了微观的知识单元特征对技术创新活动的影响,对现有的研究是一个重要的补充。第二,从内容(知识多样性)和结构(知识依赖度)两个维度定量分析了不同技术创新活动新颖性差异的原因,揭示了技术新颖性的来源及作用机制。

Innovation has been considered as the result of a combination or reorganization of existing knowledge since Schumpeter. Existing research extensively discusses the impact of individual and organizational characteristics on individual or organizational innovation performance, which helps to explain the differences in performance between individuals or organizations. But these studies usually assume that the innovation or invention activities of individuals or organizations within a certain observation period are homogeneous. In fact, even different inventions carried out by the same person may differ in their novelty or usefulness. Existing studies rarely explain this phenomenon. The purpose of this study is to reveal the micro-foundation of the heterogeneity of different inventions and to explain the path of novel inventions through theoretical and empirical analysis. Since any patent can be decomposed into knowledge components with different technical functions, patent data provides the possibility to track the connection relationship between knowledge components. Based on the existing research, this study regards patents as a group of combined knowledge components, and each knowledge component is coded as a classification number. New technologies arise from the combination of previously unconnected knowledge components or the reorganization of already connected knowledge components in new ways. Evolutionary theory has long believed that the diversity of components and their relationships are important factors that determine the direction and performance of evolution. Social network theory emphasizes the functional diversity of network nodes and the impact of the connections between nodes on innovation results. Combining evolutionary theory and Social network theory this paper proposes that the diversity and dependence of knowledge components may affect the degree of technological novelty by affecting the potential of the combination and realization. The higher the diversity of knowledge components is, the more abundant the effective knowledge has, and the greater the potential of the possible combinations is. At the same time, diversified knowledge may cause a relative lack of absorptive capacity, thereby increasing the difficulty of knowledge combination realization and increasing the cost of knowledge combination. The higher the degree of knowledge dependence, the more mature the path of using knowledge, and the lower the uncertainty of the knowledge combination, but it will increase the rigidity of creating new combinations and strengthen the path dependence. The net effect of the final impact depends on the trade-off between positive and negative effects. Therefore, this study suggests that knowledge diversity and dependency have a nonlinear effect on technological novelty. And the interaction between the two may also have a significant positive effect on novelty. In this study, 9328 invention patents in the field of nanotechnology applied for and finally authorized by the United States Patent and Trademark Office(USPTO) from 1972 to 2010 were used as samples to test the theoretical framework. The novelty of a patent is measured by the number of back citations of the patent, and the diversity of knowledge is measured by 1 minus the Herfindahl index calculated according to the technical field where the patent classification number is located. The measurement of knowledge dependency is relatively complicated and is implemented in two steps. First, calculate the degree of ease of each knowledge component combination in the patent, and divide the number of all sub-classification numbers in the previous patent that have a co-occurrence relationship with the sub-classification number i by the number of patents in the previous patent that include the sub-classification number i. Secondly, the arithmetic average of all the knowledge components calculated in the first step and the reciprocal is the value of the patent’s knowledge dependency. The negative binomial regression results show that: the negative binomial regression results using patent data as a sample demonstrate that knowledge diversity and knowledge dependency have a positive and inverted U-shape effect on technology novelty respectively, and the interactive terms of the two have an effect on technology. The impact on novelty is significantly positive. This conclusion is further supported in the robustness test. Except that the knowledge diversity hypothesis is only partially supported, the other two hypotheses are both fully supported. The contribution of this research lies in: First, it breaks the normal research assumption on individuals, teams, or organizations homogenizes innovation activities, and explores the impact of micro-knowledge component characteristics on technological innovation activities as an important additional research. Second, it quantitatively analyze the causes of the differences in the novelty of different technological innovation activities from the two dimensions of content(knowledge diversity) and structure(knowledge dependency), and reveal the source and mechanism of technological novelty.