Green Transformation and New Quality Productivity: A Triple Difference Approach with Double Machine Learning

This paper proposes an integrated causal inference framework combining triple difference (DDD) and double machine learning (DML) to evaluate the impact of green technological innovation on new quality productivity growth. Grounded in Porter's hypothesis and technological innovation theory, the study addresses key challenges in environmental policy evaluation by merging quasi-experimental designs with machine learning techniques. The methodology employs propensity score matching to construct comparable treatment and control groups, followed by DDD estimation to identify causal effects across multiple dimensions, while leveraging DML to control for high-dimensional co-founders and uncover heterogeneous treatment effects. Focusing on China's manufacturing sector, the analysis demonstrates that green technology adoption significantly enhances productivity, with substantial variation across firm ownership structures and regional characteristics. The methodological contribution lies in demonstrating how machine learning augments traditional difference-in-differences approaches through improved covariate balancing, detection of nonlinear treatment effects, and refined subgroup analysis. The findings offer theoretical insights into the mechanisms of green productivity growth and provide empirical evidence for policymakers to design targeted environmental innovation strategies.