The aim of this thesis is to extend the so-called New Economic Geography (NEG) in two directions. Te first is to advance the extent to which Ricardian insights have been integrated into the NEG theory. Chapter 1 approached from "old" trade theory (Continuum-of-Goods Trade Model) and modelled regional level of knowledge spillovers (Marshallian externalities) and comparative advantage. In contrast to the standard NEG model, agglomeration occurs at the high trade costs while diversification occurs at low trade costs. The second is to integrate some of the insights from the so-called new-new trade theory (with its emphasis on firm-level differences) by Melitz into the NEG theory. Chapter 2 incorporated "new new" trade theory into the Footloose Capital Model and Chapter 3 was integrated into the Vertical Linkage model. As a result, we found that firm heterogeneity works as dispersion force and that delocation subsidy and taxation have "spatial sorting effect"