Abstract
This paper describes a methodology for the construction of a novel patent dataset for computing consolidated patent portfolios across patent offices. In particular, relying on two previous contributions – the EPIP database for EPO/PCT patents and the IIP patent database for JPO patent documents – we discuss how to integrate the applicant names of these two databases with the mean of priority information from PATSTAT. This methodology yields significant improvements in the harmonization of applicants name for European firms in IIP patent database and Japanese firms in EPIP database. The paper concludes with a first look to the differences in patenting strategy by European and Japanese firms.
Introduction: The Challenge of Siloed Patent Data
Innovation research frequently encounters significant hurdles due to the fragmented nature of global patent data. Each major patent office—such as the European Patent Office (EPO), the Japan Patent Office (JPO), and the United States Patent and Trademark Office (USPTO)—maintains its own distinct recording standards for applicant names. Consequently, a single entity like "Panasonic" may appear under various aliases (e.g., "Panasonic Corp" or "Matsushita Electric") across different jurisdictions.
While high-quality "cleaned" databases exist for specific regions (e.g., EPIP for Europe, IIP for Japan), they lack interoperability. This lack of synchronization makes it difficult for researchers to track a firm’s global R&D strategy or compare sectoral patenting trends between Eastern and Western markets without intensive manual reconciliation.
Methodology: The "Priority Link" Bridge
The authors propose a robust five-task workflow designed to "propagate" cleaned names across databases by leveraging priority information found in the PATSTAT (Worldwide Patent Statistical Database).
- Name Cleaning & Harmonization: A 12-step cleaning sequence standardizes strings by converting text to uppercase, removing common suffixes (e.g., "INC," "AB"), and standardizing conjunctions.
- Institutional Categorization: Following OECD guidelines, applicants are categorized into Business Organizations, Individuals, or Non-Business Organizations (e.g., universities, hospitals).
- The Priority Link Bridge: This core innovation identifies "patent families." If a patent filed in the EPO claims priority from an earlier filing in the JPO, they are considered part of the same family, implying a common ultimate owner.
- Data Propagation: Verified names from the Japanese IIP database are applied to corresponding records in the European EPIP database, and vice versa, using these family links.
- Business Directory Matching: Cleaned portfolios are linked to external financial identifiers, such as ISIN codes for Japanese firms and the Amadeus database for European firms, facilitating deeper economic analysis.
Data Sources
The study integrates several key datasets:
- EPIP Database: EPO and WIPO/PCT patent documents up to 2009.
- IIP Patent Database: The standard for JPO patent documents.
- PATSTAT (April 2009): The master linkage database for identifying priority connections.
- Amadeus (Bureau Van Dijk): Financial and ownership data for European firms.
- Japanese ISIN List: Used to match 2,376 Japanese listed firms.
Key Insights and Results
The methodology proved highly effective, reducing the name dictionary size by approximately 28.8% by merging thousands of duplicate entries.
Sectoral Specialization and Strategy
- Strategic Filing: High patent volumes in sectors like electronics often indicate "strategic filing" (creating patent thickets) rather than a simple increase in the rate of invention.
- PCT Preference Gap: European firms utilize the international Patent Cooperation Treaty (PCT) route significantly more than Japanese firms, who favor direct filings in the EPO.
- Sectoral Concentration: Japanese firms are heavily concentrated in IT and electrical machinery. In contrast, European firms filing in Japan show prominence in chemicals, pharmaceuticals, and specialized machinery.
- Home Field Effect: Sectoral concentration is typically higher in a firm's home country. For Japanese firms, the top five sectors account for 64.7% of their JPO filings but only 50.6% of their EPO filings.
Linkage Success
The study successfully linked 1,069 distinct Japanese patentee names to EPO records and approximately 4,980 European names to JPO records, establishing a high-fidelity cross-border dataset.
Conclusion and Future Directions
While the methodology provides a powerful tool for cross-border patent analysis, it relies on the assumption that ownership remains constant across offices. Future research could address potential ownership transfers post-filing. Additionally, addressing "truncation lag"—the delay in patent processing that underrepresents recent grants—remains a challenge. Integrating this consolidated data with actual R&D investment figures will be essential for calculating a true "propensity to patent" across global industries.
Summary updated