Abstract: In this paper, we carefully estimate the productive capital stock of network infrastructure and its different types of facilities, construct provincial panel data from 1993 to 2017, and use the spatial Dubin model to test its impact on high-quality regional economic growth. Empirical analysis shows that although the direct effect of the overall network infrastructure capital is significantly positive, it has an inverted "U" shape during the sample period, and it is the strongest in the eastern region and the weakest in the western region. The indirect effect of the overall network infrastructure capital is significantly positive, but it is not significant in the early and middle stages of economic development, and it is the strongest in the eastern region and the weakest in the central region. In terms of differences in specific structural types, the direct effects of transportation facilities are the strongest overall and weakest in the central region, communi- cation facilities are second and continue to decline during the sample period, and energy facilities are the smallest and strongest in the central regions. The overall indirect effects of energy facilities are the largest, transportation facilities are the second and weakest in the western region, and communications infrastructure is the smallest and significantly negative in the early stages. In addition, the direct effect of point infrastructure continues to increase after the economy enters the "new normal" and is relatively greater in the western re- gion. Accordingly, countermeasures and suggestions are put forward for further optimizing various infrastructure investment strategies and promoting high-quality regional economic growth.

Key words: network infrastructure, productive capital stock, high-quality regional economic growth, spatial Dubin model