Abstract:From the perspective of consumer utility, we develop a computing experimental model to study the outsourcing strategies of manufacturers under the energy saving and emission reduction. We study the impact of carbon reduction policies and consumer behavior on manufacturers' outsourcing decisions. The reinforcement learning algorithm is applied into the dynamic decision making process of manufacturers. The aim of this paper is to provide theoretical basis for the manufacturer to develop a reasonable strategy in multi-cycle decision-making. The results show that the manufacturer should produce low-carbon component in house when low-carbon preference of the consumer is higher, otherwise, the manufacturer should outsource. The manufacturer should produce low-carbon product in house when the rewards and punishments of carbon emissions is high, otherwise, the manufacturer should outsource. In addition, when the manufacturer’s difficulties of reducing carbon emissions are easy, the manufacturer can produce component in house, otherwise, the manufacturer should outsource.