Source-linked AI summary
Complex Economic Activities Concentrate in Large Cities
Pierre-Alexandre Balland, Cristian Jara-Figueroa, Sergio Petralia, Mathieu Steijn, David Rigby, Cesar A. Hidalgo
TL;DR
The paper asks why some economic activities agglomerate more than others and why concentration can continue increasing despite advances in communication and transportation. It analyzes patents, research papers, industries, and occupations using urban scaling and complexity measures, finding that more complex activities concentrate more in large cities and that complex patenting has become increasingly concentrated since 1850.
Problem
The paper examines why economic activities differ in spatial concentration and why some activities become more concentrated despite advances in communication and transportation.
Method
The study compares urban scaling across patents, research papers, industries, and occupations using domain-specific measures of knowledge complexity and historical patent data.
Results
More complex activities are more concentrated in large cities across all four domains, with complexity-concentration correlations ranging from Pearson’s r = 0.62 to r = 0.82.
Takeaways & Limitations
The findings suggest that increasing urban concentration of jobs and innovation may be a consequence of the growing complexity of the economy.
Abstract
from arXiv · showhide
Why do some economic activities agglomerate more than others? And, why does the agglomeration of some economic activities continue to increase despite recent developments in communication and transportation technologies? In this paper, we present evidence that complex economic activities concentrate more in large cities. We find this to be true for technologies, scientific publications, industries, and occupations. Using historical patent data, we show that the urban concentration of complex economic activities has been continuously increasing since 1850. These findings suggest that the increasing urban concentration of jobs and innovation might be a consequence of the growing complexity of the economy.
Introduction
The paper asks why economic activities differ in their urban concentration and shows that complexity is associated with stronger concentration in large cities across multiple domains. It also reports that the concentration of complex knowledge production has increased over time.
- Introduction: More complex economic activities tend to concentrate more strongly in large cities across technologies, research papers, industries, and occupations.The paper links this pattern to differences in knowledge complexity and urban scaling.
- Introduction: Complex industries such as biotech and semiconductors are more concentrated in large cities than less complex industries such as apparel and furniture manufacturing.
- Introduction: The concentration of more complex inventions has increased continuously since at least 1850, while that of the least complex technologies has decreased since the 1970s.
- Introduction: Urban concentration varies across activities, but the paper argues that complex processes require deeper knowledge division and specialization.Cities provide multiple interaction opportunities that can help address the resulting coordination costs.
- Introduction: The paper measures urban concentration through scaling exponents and explains differences in those exponents largely through differences in economic complexity.
DATA
The study analyzes patents, scientific publications, industries, and occupations across 353 U.S. metropolitan statistical areas using contemporary and historical data sources.
- DATA: The analysis covers the spatial distribution of patents, research papers, industries, and occupations in 353 U.S. Metropolitan Statistical Areas.
- DATA: Recent patent data provide inventor-address coordinates for USPTO patents granted from 1975 to 2010.
- DATA: Historical patents from 1850 to 1974 come from HistPat, which combines publicly available USPTO documents with text-mining algorithms and statistical models.
- DATA: Scientific publication data come from Elsevier’s Scopus database for 1996–2008 and are divided into 23 major thematic disciplines.
- DATA: Patents are disaggregated into 30 technologies using National Bureau of Economic Research two-digit subcategories.
RESULTS
Across patents, research papers, industries, and occupations, more complex activities concentrate more strongly in large cities. Historical patent data further show that this concentration has increased over time, especially for complex technologies.
- Overall urban concentration: Patents, research papers, industry GDP, and employment all concentrate highly in large U.S. metropolitan areas, with scaling exponents of 1.26, 1.54, 1.11, and 1.04, respectively.These activities follow super-linear relationships between metropolitan population and output or employment.
- Activity-level differences: Computer, hardware, and software patents scale more strongly with city population (E = 1.57) than pipes and joints patents (E = 1.1).Comparable variation appears across research fields, industries, and occupations.
- Complexity measures: Complexity is measured by patent knowledge vintage, publication team size, industry worker education, and occupational worker education.These measures operationalize recency, division of knowledge, and specialization without using spatial information.
- Complexity relationships: Urban concentration rises with complexity across technologies, scientific fields, industries, and occupations.The reported correlations are 0.82, 0.72, 0.70, and 0.62, respectively, all with p<1x10^-3.
- Historical trends: The scaling exponent of the most complex patents increased continuously from 1850 to 2010, reaching almost 1.8 in 2010.After the 1970s, the least complex patents became less concentrated, with their exponent falling below 1.2.
- Historical trends: Electrical and electronic patents drove much of the post-1870 rise in scaling, while computers and communications and drugs and medical patents became most concentrated after 1950.The analysis tracks six NBER technology categories across decades from 1850 to 2010.
DISCUSSION
The paper addresses why some activities agglomerate more than others and why agglomeration persists despite improved communication and transportation. It argues that complexity explains these differences and links stronger concentration of growth-related activities to increasing spatial inequality.
- Interpretation: Complex economic activities tend to concentrate more in large urban areas because they require a deeper division of knowledge and labor.The paper connects this specialization to coordination needs and tacit knowledge embedded in social networks.
- Interpretation: Across patents, research papers, industries, and occupations, more knowledge-intensive activities are more likely to exhibit super-linear scaling.The paper presents complexity as an explanation for variation in the degree of agglomeration across economic processes.
- Historical divergence: During the IT revolution, the concentration of the most and least complex activities diverged.Complex activities became more concentrated while the least complex technologies became less concentrated in cities.
- Implications: The paper argues that continued concentration of complex growth activities may increase spatial inequality between large and small cities.It specifically identifies pharma, artificial intelligence, and data services as activities that may remain concentrated in a few large cities.