An Agent-Based Modeling Approach for the Diffusion Analysis of Electric Vehicles With Two-Stage Purchase Choice Modeling

Abstract

Abstract Diffusion research of innovative technologies is crucial for new product positioning and strategic planning in product design. As a versatile system simulation method, agent-based modeling (ABM) has been used in many previous studies on the diffusion analysis of electric vehicles (EVs). In these simulations, modeling consumers' purchase decisions is a significant step. Previous studies often adopt simple rule-based decision criteria in this step, while an accurate purchase decision model can contribute to a more reasonable diffusion analysis of EVs. To fill this gap, this brief presents an agent-based modeling approach for the diffusion analysis of electric vehicles with two-stage choice modeling. The core idea is to separate consumers' decision-making process for purchasing cars into two stages. Consumers first form a small choice set from the whole auto market. Then, they make the final choice from the choice set built in the first stage. In addition, the word-of-mouth (WOM) effect and consumers' social networks are also considered in the ABM, which can further improve the accuracy of the diffusion analysis. A case study using data collected from Shanghai, China, is presented to demonstrate the proposed approach. Our approach outperforms other ablation models as well as traditional statistical models in the prediction accuracy of EV's market share. The influence of factors such as government policy and technological improvement on the diffusion of EVs is also discussed. These insights can assist automakers in improving their product design and enhancing their market competitiveness.