摘要:AbstractWorld cities (or global cities) are defined by various aspects such as socioeconomic, political and cultural characteristics by several organizations. Some world cities have been growing rapidly, whereas others have not. The purpose of this paper is to investigate economic reasons for the variations in the growth of world cities. According to the literature, there are roughly two possible reasons for the variations: costs of transport and trade and spillover effects of knowledge and technologies.First, Martin and Ottaviano (1999) have developed a dynamic model of new economic geography, where accumulation of knowledge and technologies by research and development is the engine of economic growth. The accumulation of knowledge and technologies often spills over geographically limited regions. Although it is not easy to test the spillover effect of knowledge and technologies, we explore the effects of the economic growth (the GDP growth rate) and the trade dependency (the degree of dependence on foreign trade) as surrogate for knowledge spillover on urban agglomeration (the population share of the largest city).Second, Behrens, Gaigne, Ottaviano and Thisse (2007) have shown that the degree of spatial agglomeration depends not only on the level of transport costs within a country, but also trade costs between countries. More specifically, they have shown that agglomeration is sustained when the domestic transport costs are low and the international trade costs are high. This would be true in Japan because Japan is geographically small with improved transport systems and is imposing high import tariffs especially for agricultural products.On the other hand, this is not true in the United States because the United States is geographically large and is imposing low tariffs. We examine the effect of the trade openness on the degree of agglomeration to the largest city in order to see if the results by Behrens, Gaigne, Ottaviano and Thisse (2007) hold.We explore the reasons for agglomeration and dispersion in the world cities by regression analysis using the international data. We find that the factors affecting the level of agglomeration in the world cities differ from those affecting the change in agglomeration. We show that the level of agglomeration in the world city is negatively affected by the trade openness, while positively affected by the secondary and tertiary shares, and being a national capital. We also show that the change in agglomeration to the world city is strongly affected by economic development, weakly affected by the trade openness, and unaffected by being a national capital. Finally, we confirm that the country fixed effects are important factors of the changes in agglomeration to the world city.