Abstract
Households are crucial in the energy transition, accounting for over 25% of the European Union's energy consumption. To design effective policy measures that motivate households to change their behavior in favor of the energy transition, agent-based models (ABMs) are vital. For ABMs to reach their full potential in policy design, they must appropriately represent behavioral dynamics. One way to accomplish this is by strengthening the fit in ABMs between behavioral determinants (e.g., trust in energy companies) and the behavior of interest (e.g., adopting tariff structures). This study investigates whether a structured behavioral analysis improves this "determinants-behavior-fit." A systematic review of 71 ABMs addressing household energy decisions reveals that models incorporating a behavioral analysis formalize nearly twice as many behavioral determinants, indicating a more systematic uptake. Subsequently, we find a difference between models focusing on investment-related behaviors (e.g., households buying solar panels) and those examining daily energy practices (e.g., households adjusting charging habits). Models in the first category integrate more social factors when incorporating behavioral analyses, corresponding with the influence of networks and peer effects on investment behaviors. Models in the second category emphasize individual and external factors in response to behavioral analyses, corresponding with the energy practices' habitual and contextual nature. Despite the benefits of a behavioral analysis for improving the determinants-behavior fit in ABMs, only one-third of the studies apply it partially. On top of that, almost half of the studies do not report a rationale for their choice of behavioral determinants. This suggests that many models may not fully capture the behavioral mechanisms underlying household energy decisions, limiting ABMs' potential to inform policymakers. Our findings highlight the need for systematic behavioral assessments in model development. We conclude that collaboration between behavioral scientists and modelers is crucial to accomplish such integration, and we emphasize the importance of allowing sufficient time and resources for meaningful exchange. Future research could further investigate empirical validation of behavioral insights in ABMs and explore how ABM results improve with a better determinants-behavior fit. By bridging behavioral science with computational modeling, ABMs' decision-support power to policymakers can be improved, ultimately accelerating the energy transition.