A modularized agent-based framework for causal-based policy-making

Explaining social phenomena from a causal perspective and based on which to design and evaluate policies for improving them is critical. Agent-based Modeling (ABM) is a powerful means for modeling social systems from a bottom-up perspective, simulating counter-factual outcomes, and investigating the impacts of policy alternatives. How to establish a causal understanding of the model behaviors has thus been vital for enabling causal-based policy-making. In this work, we extend, integrate and generalize the existing methods to a systematic modularized framework, which aims to abstract and conceptualize the social phenomenon with cognitive agent-based models, to form a causal explanation of the model behaviors, and based on which to support explainable policy formulation and evaluation for improving the phenomenon. We apply it to a specific long-term care (LTC) model to demonstrate its effectiveness in facilitating LTC design. This modularized framework is expected to be integrally computerized and made publicly available in future to inspire discussions among stakeholders for policy-making.