The “innovation championship” model has been instrumental in explaining policy innovations in China’s local governments, particularly at the provincial level. However, discrepancies between this model and real-world cases raise questions about its broader applicability. To address this, we employ a dichotomous framework (innovation generation/borrowing) and conduct multi-level quantitative analyses of government work reports. Our analysis suggests that between 2003 and 2022, most provincial innovations were driven by the championship model, which relies on central government recognition, while others were shaped by peer recognition mechanisms. Together, these form a “central and peer” (CP) model that prioritizes innovation generation while incorporating a degree of innovation borrowing. This CP model differentiates the innovation functions among provincial governments, which have formed a collective innovation network: pioneering provinces generate model policies, while others capitalize on these opportunities. Moreover, the extent of the central authority’s influence determines the relative importance of these two mechanisms.

The governance of China under Xi Jinping functions mainly through leading groups. Using the case of water governance, this article examines the interaction between these groups at three levels: the top-level design group, the riverine macroregion groups and the implementation groups. This governance model is designed to avoid nomenklatura failure, restrict fiscal federalism and reduce the agency problem between the centre and local leaders. For the purpose of water governance, China has been divided into five blocs based on river basins, which we call riverine macroregions. Using this approach, the Chinese Communist Party is combining water governance with regional development and enhancing local governments’ collective implementation of central policy.

Many studies put forward the argument that local policy experimentation, a key feature of China's policy process in the Hu Jintao era, has been paralysed by Xi Jinping's (re)centralization of political power – otherwise known as “top-level design.” This narrative suggests that local policymakers have become increasingly risk-averse owing to the anti-corruption campaign and are therefore unwilling to experiment. This article, however, argues that local governments are still expected to innovate with new policy solutions and now will be punished if they do not. By introducing the analytical framework of “experimentation under pressure” and drawing on an analysis of over 3,000 local government regulations and fieldwork data related to foreign investment attraction policies in two localities, Foshan and Ganzhou, the authors highlight new features developing within current experimental policy cycles. Local cadres now have no choice but to experiment as the political risk of shirking the direct command to experiment may be higher than the inherent risk of experimentation itself.