基于实验经济学仿真构建碳排放交易的多代理模型

The market efficiency and regulation effects of carbon emission trading strongly depend on decisions made by different kinds of participants in the carbon emission market. Therefore, the participants, especially the policy designers and regulators, should understand the participants' trading behaviors and their underlying mechanisms. Experimental economics can reflect participants' "bounded rationality" characteristic through humans' participating in experiments. However, the scale of experiment is limited due to the difficulty to have enough qualified human participants, and the comparability among the multi-scenario experiments is also limited due to the variations of human participants' concentration level. This paper adopts a hybrid interactive simulation methodology. Human-subjected experiments with a small group of human participants are conducted. Major driving factors and their influencing rules on human participants' uncertain behaviors are extracted, which can aid the construction of multi-agent stochastic models with the same distribution characteristics. In this way, only a small group of humans are needed to represent the special decision makers, who can interact with large numbers of multi-agents constructed by the aforementioned method. The methodology discussed can overcome the scale limitations and meanwhile maintain the merits of experimental economics.