【Abstract】The optimal strategy for the government to supervise low-carbon subsidy is studied in this paper, as well as the problem of supervision efficiency and supervision stability. Firstly, the evolutionary game models are constructed in un-supervision and supervision cases respectively, based on agents of the government and enterprises, and three supervision strategies and related rules are given. Secondly, the supervision efficiency and supervision stability are discussed based on the established small-world network model, then the optimal supervision strategy and the corresponding optimal supervision probability are obtained. Moreover, parameter sensitivity analysis is carried out based on benefit stability. The simulation results show that: (1) the optimal strategy for the government to supervise low-carbon subsidy is random supervision of enterprises who declare high subsidy, with the optimal probability of supervision 0.48, and the overall optimal supervision probability 0.24. (2) the supervision efficiency and supervision stability can not achieve at the same time. However, in order to maximize the revenues of both agents, the measure of efficiency first and stability second can be considered. (3) when there is more noise in the environment, the average revenues of both agents are lower, and the benefit stability is worse. When the supervision cost is higher, the result is the same as the case with more noise in the environment. However, the increase in punishment cost does not necessarily lead to the increase in revenue. When punishment cost increases to a certain extent, and continues to increase, the revenues of both agents reduce first, and then increase, and the benefit stability becomes worse first, and then better. Finally, the related policy recommendations and future work are presented.

　　【Keywords】Low-carbon subsidy;Supervision efficiency;Stability;Small-world network model;Evolutionary game model.

　　本文刊登于《Journal of Cleaner Production》，2017年第168卷。该期刊为SCI收录，在中科院分区中为二区期刊，5年影响因子6.207，为学院B类奖励期刊。