How does Geographical Proximity Shape U-I Linkages? Evidence from Large-Scale Text Data

• In Event: Geography, Networks, and Regional Innovation

Abstract

The knowledge flow between universities and firms is widely recognized as a key driver of innovation (Feldman, Audretsch, 2004). Empirical studies have highlighted the role of tacit knowledge transfer compared to codified knowledge (Bellucci & Pennacchio, 2016). Given that tacit knowledge is primarily transmitted through face-to-face interactions (Asheim, 2007), informal university-industry (U-I) interactions serve as vital channels for its dissemination. However, the lack of records documenting these informal interactions poses challenges to reveal the impact of geographical proximity on these linkages (Cohen et al., 2002). Consequently, much of the existing empirical work on U-I collaboration focuses on a particular type of formal interaction that reflects codified knowledge flows—paper trails such as patents, academic publications, and citations (Jaffe, 1989; Ponds et al., 2010)—which largely overlooks the role of geographical proximity in shaping U-I linkages. To capture these often unobserved linkages, we employ Large Language Models (LLMs) to extract detailed information on university–industry (U–I) interactions from a large-scale textual database of annual reports of Chinese listed firms spanning 2000-2022. The novel dataset identifies 12 distinct forms of U–I linkages, encompassing both tacit knowledge-based interactions—such as permanent and temporary personnel mobility between academia and industry, the establishment of postdoctoral and academician workstations, and firm-sponsored training programs—and more codified forms of collaboration, including joint research projects, the creation of research platforms, and university-affiliated spin-offs. By linking this comprehensive collaboration dataset with campus-level microdata, we are able to credibly estimate the causal effect of institutional proximity on firms’ propensity to engage in university partnerships. Our empirical analysis yields three main findings. First, in terms of spatial scale, university expansion over a broad radius (up to 500 km) has no discernible effect on firms’ engagement in university–industry (U–I) collaboration, whereas only campus openings within a 25 km radius generate a statistically significant and robust increase in such activities. The effect attenuates rapidly with distance, indicating that—even for large and medium-sized firms—the absorption of university knowledge spillovers remains highly localized. Second, heterogeneity analysis shows that this effect is concentrated in high-tech manufacturing sectors, suggesting that technology-intensive firms are better positioned to internalize academic knowledge. In addition, university expansion primarily stimulates informal collaboration, with comparatively weaker effects on formal partnerships, consistent with the notion that tacit knowledge exchange is particularly sensitive to geographic proximity. Finally, distinguishing between direct and indirect channels, we find that proximity to new campuses not only strengthens direct firm–university linkages but also increases collaboration with other universities, likely by reshaping local firms’ innovation behavior more broadly. These findings suggest that when planning the location of universities, policymakers should prioritize spatial proximity to existing industrial clusters—particularly in high-tech manufacturing—to minimize the costs of informal interaction. Furthermore, policy support should move beyond a narrow focus on contract-based projects toward strengthening informal collaboration channels. This includes institutionalizing personnel mobility, supporting academic exchange platforms, and creating shared spaces to facilitate tacit knowledge flow between university and industry.

Author

• Lu Lin, Peking University
  Presenting Author