Sustainable development through green innovation and resource allocation in cities: Evidence from machine learning

China has promoted innovation-driven and green development to unprecedented strategic heights. However, compared to the large and rapid innovation investment, total factor productivity's (TFP) growth rate has shown a downward trend. Consequently, this study assesses the inefficiency caused by resource mismatch and discusses the impact of green innovation activities on green total factor productivity (GTFP). We use a causal forest-based machine learning method to solve the endogenous problem. The empirically analyzes the observation samples of 272 prefecture-level cities in China from 2008 to 2018 and obtains the asymptotic normality estimation on the average treatment effect (ATE). Simultaneously, clustering causal forest and ridge expressions, discusses the robustness of related problems. According to the results, (1) the effect of China's green innovation on GTFP is negative for a short time and positive for a long time; (2) the impact of green innovation activities on GTFP is subject to capital mismatch, while the statistical law of the impact of labor mismatch is not obvious but the adverse impact of resource mismatch is gradually improving; and (3), Green innovation has significantly improved China's GTFP, but it did not lead to ideal Growth rate of GTC.