Smart city policies and corporate renewable energy technology innovation: Insights from patent text and machine learning

Abstract

Smart cities have emerged as a key strategy to balance economic growth with carbon emission reduction. This study uses a difference-in-differences (DID) model, supplemented by a double machine learning approach (DML), to examine the impact of China's smart city policies on corporate renewable energy technology innovation (RETI). We further integrate a supervised machine learning bag-of-words (BoW) approach enhanced with TF-IDF weighting and cross-validation to convert patent texts into robust quantitative RETI metrics. Results show that smart city policies significantly enhance RETI, primarily by alleviating financial constraints and improving human capital. These effects are further amplified by well-developed institutional environments and executive's environmental protection background. Additionally, the main effect is more pronounced for nonstate-owned corporate and those in eastern China. These findings offer valuable insights for fostering RETI and advancing sustainable development, with implications for achieving carbon neutrality goals.