High-speed rail and regional innovation: How well is it measured?

• Systematic review critically evaluates methodologies in HSR-innovation research. • Quasi-experimental methods (DID) dominate, but causal validity remains a challenge. • The ’black box’ of mechanisms like tacit knowledge is rarely validated empirically. • Advocates methodological pluralism to generate more robust policy insights. Establishing a causal link between high-speed rail (HSR) and regional innovation, and capturing the relationship’s complexity, presents significant methodological challenges. While HSR is theorised to boost innovation via enhanced connectivity, proving this link robustly requires navigating issues like non-random network placement, spatial spillovers, network effects, and appropriate measurement of both HSR exposure and innovation outcomes. This study systematically reviews recent literature to critically evaluate how this relationship is measured. Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, this systematic literature review synthesises methodological insights from a final analysis set of 35 key studies (32 empirical, 3 conceptual/review papers). Results show a clear evolution towards quasi-experimental methods, particularly difference-in-differences and its spatial variants often combined with instrumental variables to address endogeneity. However, significant challenges remain: establishing causal validity (parallel trends, instrument validity), adequately measuring HSR exposure beyond simple connectivity, capturing heterogeneous effects, modelling complex spatial dynamics (concentration, decay), and empirically validating intermediate mechanisms like tacit knowledge transfer, which often remain a theoretical 'black box'. In addition, most methodological explorations were conducted in the context of Chinese HSR, raising concerns about external validity. We conclude that while methodological sophistication is increasing, current approaches struggle to fully capture the systemic complexity and provide uncontroversial causal evidence. Future progress requires methodological pluralism (use of multiple methods), integrating advanced econometrics with tools like agent-based modelling, network science, machine learning, and qualitative methods, alongside richer data and comparative research beyond China, to provide more robust and nuanced insights for policy.