Evolution of retailer's competitive performance considering price and service combination strategies: an agent-based simulation

This paper explores an agent-based model that incorporates the Q-learning algorithm, and this model includes a competitive multi-agent retail-consumer interaction network. In the network model, various retail agents are constructed to compete for consumer groups under different network features (consumer neighbour nodes, consumer network reconnection probability, and consumer herding psychology intensity) with different pricing and service level combinations. All retail agent agents adjust their product prices and service levels under the Q-learning mechanism to maximise their expected sales and profits. Compared to previous studies, we make contributions that include, but are not limited to, constructing consumer networks with nodes of new network characteristics, as well as designing individual consumer characteristics with more complexity, including heterogeneous attributes such as the consumer's income level and the consumer's expectation level. The purpose of this paper is to provide recommendations for selecting the appropriate combination strategies for retailers in a complex market environment.