Abstract

Agglomeration economies as a kind of effective resource allocation effect in the economic transition and the structural adjustment process, its cultivation has become a strong driving force of urban sustainable development and the development of urban and rural, regional coordination.At theoretical level, agglomeration is a core proposition in the research of urban economics, regional economics and new economic geography.But because of its invisibility, agglomeration economies have been regarded as a"black box".In addition, in the aspects of the degree of economic integration and the liquidity of production factors, China has a big gap and great differences in economic geography, compared with Europe and the United States and other developed countries.Thus the agglomeration economic theory and its relevant experience originated in Europe and the United States and other developed countries, needs to make further serious analysis and research on its application and development in the domestic academic circle of China.

Based on Marshall's classic interpretation in three aspects of agglomeration economies, and with the aim of confirming the existence of agglomeration economies and investigating the internal mechanism of agglomeration economies, this thesis carries on the identification and measurement in agglomeration economies research from multiple dimensions.On research perspectives, on the one hand, given the importance of space factor in identifying and measuring agglomeration economies, space factor is divided into two types—independent space and link space.From the viewpoint of link space, this thesis examines the geographical proximity effect and the proximity effect of organizational culture; on the other hand, the main body of agglomeration economies are divided into three levels—macro or urban level, meso or enterprise level and micro or labor force level.This thesis emphasizes the close links between behavioral agents at the three levels and strives to explore a variety of effects on agglomeration economies caused by behavioral agents'heterogeneity at each level.On technical route, considering the Circular and Cumulative Causation plays a key role in the process of production factors' agglomeration, based on the theory of matrix from the deep excavation of urban spatial effect, heterogeneity and endogeneity, this thesis employs spatial analysis, multi-level analysis and micro econometric tools, to try to solve the problems, such as the effect of spatial effect, heterogeneity and endogeneity, which are contrary to the basic assumptions of classical regression.Among them, this thesis treats identification and effective treatment of endogeneity problem of agglomeration economies as the pinnacle of empirical analysis.

The agglomeration economies'identification and estimation results based on wage premium confirm that there is a noticeable agglomeration economies'effect during the development of urban economy in China.Labor is able to achieve higher abilities in the cities with greater employment density, or we can say that the cities with greater employment density are able to pay higher level of wages to labor.These results reveal the main reason that the labor force flows into big cities and more advanced regions, even if these areas of production and the cost of living is high.If the government that is blind to the fact that does not pay attention to deepen the developed areas of openness, gradually elimi-nate the institutional obstacles factors movement and reduce the transaction costs, but blindly introduce relevant policies to attract capital and enterprises settled in to the weaker region agglomeration economy, so policy is bound to the wasted effort effect.So how to promote the urbanization process of our country is currently a difficult problem in the process of modernization, especially under the background of urban and rural dual structure and regional development imbalance.

This thesis has seven chapters besides introduction part and conclusion part.

In chapter one from agglomeration economies's three aspects—its connotation, its dimension and its function mechanism, this thesis defines the concept of agglomeration economies, and makes a detailed explanation on agglomeration index and econometric estimation method.Furthermore it elaborates the basis of agglomeration economies theory from three angles—the polarization theory and location selection, externality theory, and spatial economic theory.

In chapter two, treating China's urban and rural economic sectors, industry, urban and geographical space as the measurement obj ect, it analyzes the related indicators about the measurement of the agglomeration economies, adopting original geographic concentration index, location quotient and Gini coefficient, location Gini coefficient, EG index, urbanization economy, local economy and the density of employment, market potential and Moran's I index respectively.

In chapter three and chapter four, according to the classic model of urban economics and the new model of economic geography, combined with the micro individual data, using instrumental variable method, it deals effectively with endogenous problems in the process of identification and estimation of the agglomeration economies.The empirical results show that the coefficient on wage elasticity of employment density is about 0.10, and the coefficient on the wage elasticity of market potential is between 0.23 and 0.34.In other words, the results show that cities with a higher degree in employment density or market potential will get a higher capacity to pay labor force.Under the condition of controlling employment density or market potential, labor wages in larger urban land area are higher than that in smaller urban land area.It reveals the deep reason why big cities will be strongly attractive to labor.

In chapter five, this article uses 286 panel data from 1997 to 2010 and also uses urban economics and new economic geography as empirical analysis framework.In addition, this article uses system GMM to conduct the endogeneity in the process of identification of urban agglomeration economics.The empirical result shows that new economic geography model is more appropriate for analyzing the current Chinese economic geography.The elastic coefficient of wages for the market potential is about 0.23, and it appears significant at 1% level, which explains that the labor force is able to achieve higher ability in the cities with higher potential markets, or we can say that the cities with higher potential markets are able to pay higher wage to the labor force.The result reveals the main reason that the labor force flows into big cities and more advanced regions.

In chapter six, centered on the vicinity of organizational culture in agglomeration economies, according to the perspective of regional heterogeneity and the typical data structure that industry nests in city, using the analysis method of the Hierarchical Linear Models（HLM）and indirect estimation strategy of ag-glomeration economies, it attempts to reverify the existence form of agglomeration economies to capture unobservable heterogeneity of agglomeration economies from the industry dimension and reveal the internal close contacts among the dimensions in agglomeration economies.The empirical results not only verify that the localization economy（MAR externality）exists in manufacturing, but also confirm that at industry-level, agglomeration economies are both influenced by industry's own characteristics, and deeply affected by the social environment of industry nested urban economy.

In chapter seven, treating the identification and measurement of agglomeration economic and spatial externalities as the theme. Spatial econometrics which models externality in the form of spatial lags is an effective tool to capture externalities, and the spatial externalities are divided into direct effects and indirect effects, so as to avoid explanations for the model parameter error.The empirical analysis shows that spatial spillover effect is important to the urban economic growth.

Keywords:Identification problem; Instrumental Variables; System Generalized Method of Moments; Hierarchical Linear Models; Spatial Durbin model.