Change and driving forces of intercity collaborative innovation networks in the Yangtze River Delta region

[Objective] In the new economic development era, actively integrating into regional innovation networks has become an important way for cities to obtain external innovation resources and enhance their independent innovation capabilities. This study aimed to reveal the spatiotemporal patterns and mechanisms of change of collaborative innovation in the Yangtze River Delta region from a network perspective, and put forward policy recommendations for regional innovation space optimization and high-quality development. [Methods] Based on the co-application data of invention patents, the intercity collaborative innovation networks of 41 cities in the Yangtze River Delta region were constructed for six periods from 2010 to 2020. The characteristics of change of the regional collaborative innovation networks were further investigated through network descriptive analysis and the underlying endogenous and exogenous driving forces were explored through the temporal exponential random graph models. [Results] (1) The intercity collaborative innovation networks in the Yangtze River Delta presented the co-existence of hierarchical structures and clustering structures, as well as remote trans-local linkages and proximate local linkages. Shanghai, Hangzhou, and Nanjing were always the top three in terms of innovation collaboration, and the innovation corridor along Shanghai, Nanjing, Hangzhou, Hefei, and Ningbo became evident. Regional collaboration clusters were highly coupled with administrative divisions and gradually evolved into three communities bounded with provincial boundaries and centered on the cities with high administrative levels, i.e., Jiangsu-Shanghai community, Zhejiang local community, and Anhui local community. Except for Hefei, the development of cities in Anhui Province was more localized and marginalized. (2) The scale and strength of networks increased gradually while the overall change of the networks was stable. The overall structural similarity between the current network and its previous network exceeded 80%. The hierarchical structure tended to converge whereas clustering structures were continuously strengthened, thus the networks were developing towards a balanced and clustering pattern with dense collaboration among multiple cores. (3) In addition to the urban innovation endowments, network change was mainly driven by time dependence, structure dependence, provincial boundary effect, and geographical proximity. [Conclusion] There is still muc