Double/debiased machine learning for semiparametric synthetic difference-in-differences models
提出一种结合半参数双重差分和合成双重差分的因果效应估计器，通过重新加权单位来放松平行趋势假设，并利用机器学习进行第一步估计，在模拟和绿色金融对绿色创新的影响评估中表现良好。
We propose a new estimator for causal effects when the data do not satisfy the parallel trends assumption in high-dimensional settings. Our estimator combines insights from the semiparametric difference-in-differences and synthetic difference-in-differences methods and compensates for the lack of parallel trends by re-weighting units. The score/moment function used for estimating effects satisfies Neyman orthogonality, which allows for the application of machine learning in the first-step estimation. Additionally, we establish the consistency and asymptotic normality of the estimator. In simulation experiments, our estimator performs well compared to other estimators. Finally, we apply this estimator to evaluate the effects of green finance on green innovation, and the results show that green finance has a significant positive effect on green innovation.