China vs USA patents (granted) and relative sectoral strengths
Description
This dataset contains patent-based evidence used to analyze the comparative technological advantages of China and the United States. It consists of 279,231 USPTO granted patents distributed across 13 major technological sectors and 137 sub-sectors, linked to 633 firms in 347 cities across 26 countries. The dataset supports the study Technology-First vs Institution-First Capitalism: Patents, Power, and the US–China Rivalry in Global Innovation, which examines sectoral patterns of innovation through the lens of National Innovation Systems (NIS) and Sectoral Innovation Systems (SIS). Patents are classified into high-technology (e.g., semiconductors, biopharma, software/AI, defense, solar), medium-technology (e.g., automobiles, aviation, chemicals), and low-technology/resource-based sectors (e.g., transport, services, petroleum/coal, mining). Geographic coverage: 26 countries (focus on China and the USA) Sectoral coverage: 13 sectors (high, medium, and low technology) Unit of analysis: Granted patent (USPTO B1/B2) Time coverage: Multiple decades of USPTO patent grants Purpose: To provide large-scale empirical evidence on sectoral specialization and technological rivalry between China and the USA, reframing debates on the “middle-technology trap” and the role of institutional vs technology-first models of development. Keywords: US–China technological rivalry; comparative advantage; patents; National Innovation Systems; Sectoral Innovation Systems; global governance of technology
Files
Steps to reproduce
Steps to Reproduce Patent Data Collection Retrieve granted patents (B1/B2) from the USPTO PatentsView database or bulk data portal. Filter patents granted to applicants in 26 countries, with a focus on China and the United States. Firm and Location Mapping Link patents to 633 firms across 347 cities using applicant and assignee metadata. Standardize firm names and geolocations for consistency. Sectoral Classification Assign patents to 13 major technology sectors and 137 sub-sectors using NACE codes and OECD/Eurostat technology intensity classifications (2011, 2019). Create categorical variables for high-, medium-, and low-technology sectors. Controls and Metadata Record patent type (B1 = granted without prior publication; B2 = granted with prior publication). Add firm age, city cluster size, and national population (log-transformed) as control factors. Analytical Framework Conduct logistic regression for binary dependent variables (sectoral patent presence). Estimate odds ratios comparing China vs. USA patents against the global reference group. Run multivariate regressions across all sectors to test robustness. Generate correlation matrices to identify sectoral clusters. Visualization and Outputs Plot sectoral specialization (Figures 1–3 in the article) using fitted two-way correlations. Produce regression result tables (Tables 4–10) summarizing sectoral odds ratios and robustness checks.