Compared to the burgeoning literature discussing the importance of agglomeration externalities for development, limited attention has been given to network externalities. This is largely due to limited data availability. We propose a general measure to proxy city network externalities based on toponym co-occurrences that indicate the relatedness between cities. This paper extracts intercity relationships based on the co-occurrence of Chinese place names on 2.5 billion webpages. We calculate and map absolute and relative network positions, which we use to explain urban labour productivity. We found that a stronger embeddedness in networks of cities is significantly and positively associated with urban productivity. Smaller cities benefit comparatively more from being well embedded in city networks, suggesting that these relations can compensate for a lack of agglomeration externalities. We also compare the importance for urban performance of city network externalities vis-à-vis agglomeration externalities. City network externalities turn out to be more important in explaining urban performance than agglomeration externalities. This calls for new theorizing on a relational approach to urban and regional development. Rather than stimulating further concentration of urbanization, our findings suggest that fostering relationships between cities is a viable alternative urban development strategy. We conclude with suggestions for a research agenda that delves deeper into city network externalities.

Cities relate to other cities in many ways, and much scholarly effort goes into uncovering those relationships. Building on the principle that strongly related cities will co-occur frequently in texts, we propose a novel method to classify those toponym co-occurrences using a lexicon-based text-mining method. Millions of webpages are analysed to retrieve how 293 Chinese cities are related in terms of six types: industry, information technology, finance, research, culture and government. Each class displays different network patterns, and this multiplexity is mapped and analysed. Further refinement of this lexicon-based approach can revolutionize the study of inter-urban relationships.