Abstract: Spatial structure is a basic component of regional science, and plays an important role in spatial interactions and regional economic development. The primary purpose of this paper is to examine the development and evolution of China's spatial structure under different traffic systems and its influence on the regional economy and interactions, such as technology diffusion and capital flow. This information is important because we need to explore various measures to improve the imbalanced economic development in China. In this study, unlike in previous structure researches, an agent-based dynamic regional economic model is built from the perspective of bottom-up modeling. In this model, each prefecture-level city in China has been defined as a firm agent, and the Cobb-Douglas production function is adopted. Heterogeneous laborer agents with different technology levels move between firm agents. The laborer migration behavior is affected by Wilson spatial attenuation. The technologies diffuse with the process of laborer migration. Capital also moves between agents as an investment behavior of firms. Different types of roads are endowed with different weights to form a heterogeneous traffic network in this model. A simulation system is built to combine GIS components. Three scenarios are identified to account for the effects of the hierarchical traffic network and preferential tax policies on the evolution of Chinese regional structure. The simulated results indicate that the traditional center-hinterland diffusion mode is no longer appropriate under the impact of heterogeneous traffic network in reality. The technologies firstly diffuse from center cities to sub-hubs along the major traffic lines, following a hub-net diffusion structure, and later from sub-hubs to ordinary nodes located at the branch lines. This phenomenon can also be observed in the capital flow process, in which the capital hubs first interconnect and form a capital flow network along the major traffic roads. After that, the network extends to adjacent ordinary nodes. It is also demonstrated that traffic conditions are important for regional development and that the basic spatial lock-in created by the natural endowments still exists. A preferential tax policy can break such a spatial lock-in to a certain extent and will be conductive to attracting the immigration of laborers and improving the local technology level for less developed regions.

Key words: spatial structure, technology diffusion, capital flow, agent-based simulation, traffic network, China