Measuring Technological Complexity Using Japanese Patents

Abstract

As international competition intensifies in technologies, nations need to identify key technologies to foster innovation. However, the identification is challenging due to the independent and inherently complex nature of technologies. Traditionally, analyses of technological portfolios have been limited to simple evaluations, indicating merely whether a technology is specialized. We propose evaluating TCI at the corporate level, which provides finer granularity and more detailed insights than conventional regional evaluations by using Japanese patent data spanning fiscal years 1981 to 2010. Specifically, we analyze a bipartite network composed of 1,939 corporations connected to technological fields categorized into either 35 or 124 classifications. Our findings quantitatively characterize the ubiquity and sophistication of each technological field, reveal detailed technological trends reflecting broader societal contexts, and demonstrate methodological stability even when employing finer technological classifications. Additionally, our corporate-level analysis allows consistent comparisons across different regions and technological fields, clarifying regional advantages in specific technologies. The corporate level analysis also reveals a new possibility for formulating an innovation strategy.

Figures (6)

• Figure 1: (A) The legend of the five classifications defined by Schmoch Schmoch2008, along with an explanation of the technology codes used in Panels (B), (C), and (D). (B) The Ubiquity $K_{T,0}$ versus average diversity $K_{T,1}$. (C) The Ubiquity $K_{T,0}$ versus TCI. (D) The TCI versus average diversity $K_{T,1}$. The black lines in Panels (B), (C), and (D) indicate the mean values of $K_{T,0}$ and $K_{T,1}$, and $\text{TCI} = 0$ is also shown with a black line. Each dot is color‐coded according to (A).
• Figure 2: (A) The TCI scores of 124 IPC the first three-digit classes and 35 Schmoch classes for both prefectures and corporations. The scores have been scaled to values ranging from 0 to 100. Each IPC is plotted against the corresponding Schmoch axis, and in cases of multiple correspondences (i.e., Schmoch class “Chemical Engineering” maps to IPC classes B01–B08, C14, etc.), IPC classes are plotted individually. (B) The distribution of the absolute differences between TCI of IPC and of Schmoch classes.
• Figure 3: Three scatter plots arranged in a single row compare the Corporate TCI between pairs of prefectures. The left panel compares Osaka with Tokyo, the middle panel compares Aichi with Tokyo, and the right panel compares Aichi with Osaka. Each axis ranges from 0 to 100. Black points represent individual classifications or entities, and the red dashed line indicates the diagonal reference line.
• Figure S1: The Pearson correlation between the degree centrality $K_{T,0}$ of technology node $T$ classified by IPC and the average nearest neighbor degree $K_{T,1}$, and the Pearson correlation between the degree centrality $K_{T,0}$ of technology field node $T$ and TCI. In both figures, the black lines indicate the mean values. Each data point is color-coded according to the five classifications defined by SchmochSchmoch2008.
• Figure S2: Changes in the five-year rolling ranking of the TCI based on 124 three-digit IPC classes for (A) the corporate analysis and (B) the regional analysis. The lines color-coded according to the one-digit IPC classes.