The impact of carbon trading policy on the total factor productivity of enterprises: An Innovative Approach Using PSO and Machine Learning

With the development and improvement of the national carbon emission trading market, more industries and enterprises are participating in the carbon emission trading policy. Based on the current status of the single empirical research method and the three doubts about the traditional SCM method, this study chooses to use the PSO algorithm and machine learning model to innovate the traditional SCM method, and uses the production and operation data of all listed companies from 2011 to 2022 for training and verification. The improved SCM method shows good fitting results. This study uses this method to examine the impact of the policy on the total factor productivity of regulated enterprises since the establishment of the carbon emission trading market in 2011. The results show that the negatively affected enterprises are mainly concentrated in technology-intensive, high-innovation-cost, equipment-upgrading and other enterprises that are difficult to adapt to the policy in the short term; while for high-energy-consuming industries, enterprises have strong policy adaptability and carbon emission cost transfer capabilities, so the policy impact on enterprises in these industries is improved; In addition, although the impact of policies on the total factor productivity of regulated enterprises varies, overall the negative impact is slightly greater than the positive impact, and this phenomenon is more obvious as carbon prices rise. This conclusion at the enterprise level should be consistent with the conclusions at the industry level and the overall level, that is, high carbon prices will have a suppressive effect on the total factor productivity of regulated enterprises or industries.