Answers are grounded in this paper and cite their sources.
Can a national innovation-platform policy raise the quality of firms’ green innovation, and if so, through adding resources or by reallocating existing ones? We study China’s Demonstration Bases for Entrepreneurship and Innovation (DBEI), a multi-actor platform policy rolled out in three staggered batches in 2016, 2017, and 2020. Using 41,600 firm-year observations from 4613 Chinese listed firms over 2010–2023, we estimate a staggered difference-in-differences design and corroborate the estimates with heterogeneity-robust estimators (Sun–Abraham and Callaway–Sant’Anna) and a randomization inference permutation test. We find that DBEI designation raises green invention patent grants by about 4.6%, and patent applications by about 3.6%; the effect emerges only from the fourth year after designation and accumulates thereafter, and it is larger for granted and forward-cited patents than for applications—indicating an improvement in innovation quality and direction rather than a mere increase in patent quantity. A distinctive pattern in the channels—digital transformation is significant, whereas human-capital structure and R&D investment are not—supports a reallocation mechanism: the platform lowers coordination and information costs and reorganizes existing factors rather than inducing additional factor inputs. Enterprise-affiliated bases are markedly more effective than university-affiliated and regional bases (an effect roughly eight times larger), which we interpret as evidence that ecosystem performance depends on a firm-centered orchestration mode rather than on the mere presence of complementary actors. The findings contribute to innovation ecosystem and policy-mix theory by distinguishing a reallocation from a resource-addition channel and by specifying the governance boundary conditions under which platform policy works, and they offer guidance for designing quality-oriented, enterprise-centered innovation policy.
No takes yet. Share an insight, caveat, or question.
Zhu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: