In global value chains, value concentrates at the two ends of the "smile curve"—upstream technological innovation and downstream brand building—while the manufacturing midpoint captures the least. For China, industrial upgrading is therefore best understood as a simultaneous ascent toward both ends. Yet the empirical lens on innovation has remained asymmetric: patents densely document the technology end, whereas the brand end is comparatively unmeasured. This talk shows how combining global patent and trademark big data—analyzed with artificial intelligence, machine learning, and network methods—opens a unified, geographically explicit window onto both ends of the curve.
The analysis proceeds across three nested scales. At the global scale, geocoded patent and scientific data identify innovation clusters and the transnational networks that structure where innovation is generated. At the national scale, it examines the spatial logic of knowledge production and the spillover effects of local scientific knowledge across China. At the city scale, it reads urban innovation ecosystems as engines of catch-up climbing both ends of the curve, with Shenzhen as a case study. Extending beyond patents, ongoing work with full-sample global trademark data—mined through text and image techniques—captures product, service, and SME innovation, tracing firms' diverse pathways from imitation to brand power and opening a new frontier for measuring innovation and industrial upgrading.

scoco@ceibs.edu