The government's unitary management of marine eco-economic systems has several drawbacks, underscoring the urgency of exploring the effectiveness and feasibility of a multi-agent collaborative governance model. In this study, we developed a simulation model based on statistical and survey data to examine the effectiveness of multi-agent collaborative governance in marine eco-economic systems. Using the data simulation function of system dynamics (SD) Vensim PLE, we verified the input parameters, income variables, behavior probability, and governance outcomes for five types of stakeholders in marine governance, finding that: (1) In a multi-agent the governance framework, arbitrarily increasing the participation probability of one stakeholder type tends to lead to the involvement of two or more types. This leads to the establishment of a compact multi-agent collaborative governance model, significantly enhancing both participation and fairness of governance compared to the traditional government-centric model. (2) Regardless of which parameter is optimized, the multi-agent collaborative governance model can improve the performance of marine eco-economic systems. This governance model is more conducive to the coordinated, healthy, and sustainable development of marine eco-economic system. (3) By calculating the degree of alignment between stakeholders (central government, local government, sea-related producers, sea-related consumers, and coastal communities) across 11 coastal provinces and cities in China, this study identified one or more types of multi-agent collaborative governance models, proving the feasibility of this approach. (4) While collaborative governance models led by five different entities have their unique advantages and disadvantages, further strengthening relationships and cooperation among these stakeholders is essential for advancing the maturity and effectiveness of the multi-agent governance model.