Abstract: Quantitative evaluation of the impact of atmospheric regulation on economic development will help to improve the performance of atmospheric regulation. To effectively implement the relevant control requirements of atmospheric environmental quality and achieve the cooperative effect of high-efficient atmospheric environmental protection and high-quality economic development, this paper focused on the verification of the Porter hypothesis to analyze the dynamic impact of atmospheric regulation on the development of regional productivity. Based on the panel data of the '2+26' cities from 2006 to 2017, taking capital stock, labour and energy as input factors, economic development as desirable output, industrial sulphur dioxide and dust discharge as undesirable outputs, data envelopment analysis (DEA)-Malmquist index was used to estimate the total factor productivity (TFP) that could represent the green productivity of the '2+26' cities, and then the key driving factor of TFP was identified by decomposing the TFP dynamic index. On this basis, a vector autoregressive model of atmospheric regulation, TFP and its key driving factor was constructed. Then, the impact of atmospheric regulation in short-term and long-term was explored through impulse response function, which testified the strong and weak Potter hypotheses. The results show that the TFP of the '2+26' cities showed a downward trend from 2006 to 2015, an upward trend from 2015 to 2017, and the TFP decreased by 4.6% during the entire period from 2006 to 2017. Technological innovation was the key factor affecting TFP. From a regional perspective, there were significant regional differences of the impact of atmospheric regulation on TFP and technological innovation, which means that the strong and weak Porter hypotheses were not applicable in some cities. In terms of duration, the impact of atmospheric regulation could last for 15 to 20 years after implementation, while the impact differed in the long-term and short-term. The transition of impact usually occurred within 2 to 4 years, and the establishment of strong and weak Porter hypotheses would change accordingly. The research implies that when making policies and implementing atmospheric regulation governance, targeted control measures should be taken, and urban heterogeneity should be considered from the aspects of atmospheric regulation impact as well as the economic and social development. In addition, in order to design and implement atmospheric regulation comprehensively and accurately, the regulation should be evaluated and adjusted on a regular basis (2 to 4 years after the implementation).