Since the 1990s, cities in China have experienced a period of rapid growth. In face of the increasing complexity of the urban sprawl, many big cities are planning to build new towns to avoid over-intensive population and infrastructure construction and to ease the pressure on urban centers. In fact, commerce is one of the most important functions in a city. There is a both-way choice process between the commerce and residence, and the interaction of the two is the foundation of the multi-center structure in metropolitan areas. This paper builds a deductive economic model of spatial structure of metropolitan commerce in two-region scenario based on monopolistic competition, scale economy, spatial cost, preference for variety, and product/service differentiations. However, this kind of traditional mathematic model is based on over-simplifying assumptions, such as heterogeneity irrelevant, fully-rational agents, no interactions among economic agents, unable to adapt and evolve. In addition, the equilibrium analysis is not adapted for dynamic research. To overcome the weakness of traditional deductive model and equilibrium analysis, the paper puts forward a research approach to integrate economic deductive model with agent-based computational experiments for better understanding the evolutional process of metropolitan commerce. By using agent-based modeling and simulation, the spatial structure of metropolitan commerce can be observed dynamically in different scenarios. Therefore, instead of making nonlinear systems tractable by modeling complex building blocks with few interactions, we can make them understandable by modeling simple building blocks with many interactions among different agents in different districts based on our deductive economic model. Dynamic simulations show that (1) With fixed substitution of goods and services and the shopping transport cost, the greater the gap of commercial fixed input between the new and old urban areas is, the more imbalanced commercial space distribution would be, and more easily to form the core-periphery structure. (2) Because of the diversified consumption preferences, to strengthen contribution of inter-regional differences to changes in business market share. (3) Commerce tends to gather in the place with location, population and fixed cost advantages, and improvements in traffic condition will accelerate the commercial spatial concentration. The related methodological issue such as integrating traditional economic model with agent-based geographical computation as well as empirical analysis and econometric test is also discussed.