To test the driving effect of China's tax and fee reduction policies on independent innovation, we established a model of Dynamic Spatial Durbin (SDM) and introduced DMSP/OLS night lighting data and Malmquist productivity index for partial differential decomposition. We found that: (1) Affected by the tax and fee reduction policies, the local province tends to increase the level of independent innovation in the short term, while neighboring provinces tend to purchase and rely on foreign technology; (2) In the long term, the tax and fee reduction policies do not significantly increase the level of independent innovation in local and neighboring regions; (3) There is a strategic choice behavior of local government between political promotion incentives and promoting independent innovation; (4) The policy externality of tax reduction and fee reduction has a two-way feedback effect. We conclude that: (1) The spatial agglomeration characteristics of tax and fee reduction policies require the government to fully consider the local innovation and economic foundation, and break the resource endowment of administrative divisions; (2) The spatial feedback feature of the tax and fee reduction policies requires the government to focus on the two-way interaction of independent innovation in the adjacent regions, rather than just one-way assistance, imitation and learning; (3) The spatial lag characteristics of tax and fee reduction policies require the government to establish a accountability system or life-long system for innovative performance evaluation. Moreover, the study fails to provide causality evidence from the spatial agglomeration and spatial time-delay.