Sorting garbage and recycling renewable resources is a new and effective means to address the “garbage siege.” To examine residents' classification behaviors and scientifically assess effects of policies, this study comprehensively analyzes factors influencing residents' classification behavior at the government, resident and societal dimensions and construct multiagent decision-making functions and an interaction model by Anylogic platform. The empirical results show publicity and education have the most significant impacts on classification behavior followed by classification infrastructure and subjective norms. The simulation results show publicity and education, charging policies and government subsidies significantly transform unclassified residents into classified residents and all three approaches double rates of classification behavior. From perspectives of multisubject interaction and multicenter governance, this study analyzes motivations behind classification behavior from a micro level and quantifies effects of policies from a macro level, providing a common research method and paradigm that can be used in behavior research and policy formulation.

分类垃圾和回收可再生资源是解决“垃圾围困”的一种新的有效手段。为了研究居民分类行为并科学评估政策效果，本研究全面分析了影响政府，居民和社会层面居民分类行为的因素，并通过Anylogic平台构建了多主体决策功能和互动模型。实证结果表明，宣传和教育对分类行为的影响最大，其次是分类基础设施和主观规范。模拟结果显示了宣传和教育，收费政策和政府补贴将未分类居民大大转变为已分类居民，并且这三种方法均使分类行为的比率提高了一倍。从多主体互动和多中心治理的角度出发，本研究从微观层面分析了分类行为背后的动机，并从宏观层面量化了政策的效果，提供了可用于行为研究和政策制定的通用研究方法和范式。