The rapid growth of digital platforms (DPs) driven by technologies like big data, cloud computing, and AI, has profoundly reshaped global industries. Countries worldwide have embraced digital development strategies, leading to a surge in the platform economy. While China has made significant strides in scientific and technological innovation, achieving high rankings in global innovation indices, it faces the challenge of low application rates of scientific achievements and a low proportion of basic R&D investment. This "high quantity, low quality" problem necessitates a reevaluation of China's science and technology management system. DPs, as key enablers of the digital economy, offer unique advantages such as open sharing, information matching, and platform spillover, fostering a favorable environment for innovation. However, regional development biases and institutional obstacles lead to resource mismatches (Misa), potentially hindering the positive impacts of DPs on INN. This research addresses the question of how DPs affect INN, focusing on the mechanisms involved and the role of Misa, particularly considering China's regional heterogeneity. The study employs a theoretical framework incorporating DPs, INN, and Misa to analyze the function of DPs in improving INN and provide policy recommendations for China to become a global innovation leader.

Existing literature on DPs focuses on their definition, measurement, and impact. While the connotation and measurement of DPs are still under development, studies highlight their positive impacts on technology management, industrial organization, and strategic management, alongside potential risks such as market monopolies. Research on digitization's effect on INN examines the impact of digital services, the digital economy, and digital technologies, revealing both positive impacts and the limitations of focusing solely on quantity rather than quality of innovation. Studies on Misa's impact on INN investigate its formation mechanisms, measurement methods, and consequences, indicating it hinders total factor productivity and innovation effectiveness. This study addresses gaps in the literature by combining DP and INN research, using a more robust measure of INN (patent citations recognized by foreign academic circles), and incorporating Misa as both an intermediary and threshold variable.

To investigate the direct, indirect, and nonlinear impacts of DPs on INN, the study uses a panel data set covering 30 Chinese provinces from 2011 to 2021 (excluding Tibet, Taiwan, Hong Kong, and Macao due to data unavailability). Three models are employed: (1) A dynamic panel GMM model to assess the direct effect of DP on INN, incorporating lagged INN and control variables such as government intervention (gov), urbanization (urb), international competition (com), and technology market development (tech). (2) An intermediary effect model to examine the mediating role of Misa between DP and INN. (3) A dynamic threshold regression model to explore the nonlinear relationship between DP and INN, using Misa as the threshold variable. The dependent variable, INN, is measured using the number of patent citations recognized by foreign academic circles. The core explanatory variable, DP, is proxied using the natural logarithm of the interaction term between broadband internet users and online retail sales. Misa is calculated based on capital and labor factor mismatches. Descriptive statistics are presented, and linear interpolation is used to handle missing data. Robustness tests are conducted by altering explanatory variables, reducing control variables, and adjusting the sample regions.

The baseline regression results (Table 2) reveal a significant positive direct effect of DP development on INN across all regions, with stronger effects in the East compared to the Central and West. The intermediary effect analysis (Table 3) shows that Misa plays a significant mediating role, with DP indirectly promoting INN by improving resource allocation efficiency. A negative relationship is observed between DP and Misa, indicating that DPs contribute to better resource allocation. The dynamic threshold regression analysis (Table 4 and 6) confirms a significant single-threshold effect of Misa. The threshold value of 0.399 signifies that the positive impact of DP on INN is considerably higher when Misa is below this level (weak mismatch), while the effect is considerably reduced when Misa exceeds the threshold (strong mismatch). Robustness tests (Table 8) validate these findings across different model specifications and sample selections.

The findings support the hypotheses that DPs have a direct positive effect on INN, Misa mediates this relationship, and the effect is nonlinear and contingent on the level of Misa. The significant positive direct effect of DP underscores the potential of digital platforms to enhance innovation quality through open sharing, information matching, and platform spillovers. The mediating role of Misa highlights the importance of addressing resource mismatches to fully realize the innovation-enhancing potential of DPs. The nonlinear relationship, characterized by a threshold effect, emphasizes that the impact of DPs is particularly strong under conditions of relatively efficient resource allocation. The regional variations in the impact of DPs reflect the diverse resource endowments and development levels of different regions in China.

This study demonstrates the significant positive impact of digital platforms on innovation quality in China, mediated by resource mismatches and subject to a threshold effect. Policy recommendations include promoting the development of digital technology-based enterprises, implementing dynamic resource allocation policies to mitigate resource mismatches, and encouraging enterprises to focus on both quantity and quality of innovation. Future research should explore the impact of digital monopolies and data protection, and delve deeper into the mechanisms through which digital platforms empower innovation, potentially utilizing frameworks like TOE.

The study acknowledges limitations, including the potential impact of digital monopolies and data protection issues on both innovation and consumer welfare. The current analysis focuses on the aggregate level of innovation quality and doesn't explore variations within specific sectors or types of innovation. Additionally, while the robustness tests enhance confidence, the specific methods employed for handling missing data and the limitations of proxy variables should be considered.