he regional innovation process can be decomposed into interrelated two stages, namely research process and commercial process. By constructing the Malmquist productivity index of DEA, this paper evaluates regional two-stage innovation productivities in China and finds obviously spatial correlation between them which are distributed in multi-growth-pole as the center of the group. In the east, innovative advantage regions are relatively concentrated, but in the Midwest, the innovative growth poles are relatively dispersed, and there is difference on different extents in the two-stage innovation performance change between different provinces. On this basis, the paper further incorporates the effects of value chain outflow and spatial correlation into two dynamic spatial panel models to separately estimate these effects on both regional research and commercial productivities changes. The analysis shows that there is significantly positive correlation between the two-stage innovation performances, that the coordinated innovation chain structure with interactive correlation is basically formed, among which the knowledge spillover effect of “learning and research” on “production” is significant, however, the feedback impact of research result commercialization on research and development is weak. In different regions, the east regions basically form the innovation advantage zones with the coordination of the strong with the strong, but the chain innovation structure in the Midwest is not perfect. China should sufficiently make use of the regional innovation spillover and spatial correlation effects to enhance the feedback effect of research results commercialization on research and development, break regional innovation barriers, strengthen regional innovation coordination, and further improve regional innovation ability and performance.