A method, data, and quality note introducing the Dynamic Global Microgeographic Patent Database

Abstract

The spatial concentration of innovation is increasingly understood as a driver of economic growth, yet its microgeography remains poorly understood. Inventor address-level patent records are valuable for studying the geography of innovation, but they remain underused because raw records require substantial retrieval, cleaning, harmonisation, deduplication, and geocoding. Using improvements in programmatic bulk retrieval, Dynamic Global Microgeographic Patent Database (DGMPD) addresses this gap. As a 'living' dataset, continuously updated from patents available through trusted international and domestic patent office APIs, the DGMPD provides daily, technology-specific (IPC/CPC) data on the microgeography of innovation. Given this unique position, my note is primarily methods focused, explaining the creation of the DGMPD. However, I evidence preliminary external validation tests and results for a fixed subsample of the DGMPD. This includes 281,115 deduplicated and cleaned raw patents priority filed between 01-01-2022 to 31-12-2022, reflecting patenting activity from 117 countries by 630,299 inventors postcode-matched with a c.86% success rate to 199,201 unique matched addresses (anonymised to 40,143 unique 1km 2 grids). Back-and forward-filling of the DGMPD is ongoing for 1980 to April/May 2026. The DGMPD demonstrates how new data infrastructures can refine understanding of evolving technological and spatial patterns of innovation. 

JEL Classification: O31, O34, R12, C81, O33