Abstract
When a new technology is invented, it may absorb new knowledge from basic research. But how to evaluate the affection of basic research on technological invention and innovation accurately is difficult. The paper constructs a model for measuring the relationship between basic research and patented invention by quantitative analysis. The research introduces the concept of technology coupling which can be used for measuring technological complexity in the process of patented invention. The paper develops an empirical research by USPTO patent datum to identify that basic research play a more important role for those patented inventions at high level of technology coupling. And the research indicates a positive correlation between patent value and utilization of basic research.
Introduction
While it is widely accepted that basic research conducted in universities and public labs fuels industrial innovation, measuring that impact precisely is notoriously difficult. As enterprises shift from closed, internal R&D to "open innovation" models, they increasingly rely on external scientific knowledge. This paper addresses a specific gap: the lack of a quantitative framework to understand how the complexity of a technology influences its need for basic research. It seeks to determine whether "harder" problems—those that are technologically complex—require a stronger foundation in basic science to reach a successful, high-value outcome.
Methodology
The researchers build their model on Fleming’s technological invention theory, which treats the process of inventing as a "re-search" across a "technical landscape." In this view, an invention is a new combination of existing technical components.
Key novel components of the methodology include:
- Technology Coupling ($K_i$): The core metric for technological complexity, measuring the difficulty of combining different technical subclasses. A high $K_i$ value indicates that a patent combines subclasses rarely seen together, signifying high complexity.
- Non-Patent Citation (NPC) Method: NPCs (references to academic papers, conference proceedings, and books) serve as a proxy for the utilization of basic research.
- Negative Binomial Regression: Given that patent citation counts are discrete, non-negative integers with high variance, the authors utilize a multivariate negative binomial regression model to control for variables such as time lags and the number of technological fields involved.
The study utilizes a robust empirical dataset from the United States Patent and Trademark Office (USPTO) database, comprising 45,486 granted patents applied for in early 1988, 1993, and 1999. The data spans four major sectors: ICT, Biological Medicine, Machinery, and Chemistry.
Analysis and Results
The authors presented regression results across these four industries to test the relationship between basic research utilization and patent value.
Key Insights
- The "Complexity" Catalyst: Basic research plays a more critical role in inventions with high "technology coupling." When inventors combine disparate or difficult technologies, scientific theory provides the "map" needed to navigate the complex landscape.
- Science Speeds Up Success: The variable $X2$ (time lag between scientific knowledge and the patent) indicates that utilizing new scientific knowledge quickly leads to higher patent value.
- S&T Synergy: There is a positive correlation between the volume of both scientific ($X1$) and technological ($X3$) knowledge cited and the eventual quality of the patent.
- Avoiding "Blind Search": Basic research helps inventors avoid "trial and error" mistakes, making the search for solutions more direct and efficient.
Empirical Findings
In all industries, particularly Bio-med and ICT, there is a statistically significant positive correlation between the use of basic research (NPCs) and the number of forward citations a patent receives. In the Machinery and Chemical industries, the Technology Coupling coefficient ($X7$) was particularly high, suggesting that complexity is a massive driver of impact in these fields. The $R^2$ values (ranging from ~0.27 to 0.31) suggest the chosen variables explain a significant portion of patent value.
Conclusion
This research quantifies the "bridge" between basic science and commercial innovation. While forward citations are a standard proxy for value, they may not capture full economic impact. Future work could explore more recent datasets to see how these dynamics have evolved with the rise of AI and modern biotechnology. For enterprises, this model offers a way to refine R&D strategies by identifying when project complexity necessitates formal partnerships with academic researchers.
Summary updated