An analysis on differences in spatial computable general equilibrium models by market structure assumption : -a comparison of perfect competition modeling and monopolistic competition modeling-

Spatial Computable General Equilibrium (SCGE) models are convenient methods of the analysis of the change of inter-regional economic interaction or regional benefit by policy shocks. Recent SCGE models have two main streams in terms of the assumption of market structure; perfect competition models and monopolistic competition models. Benefit measured by perfect competition based models is usually independent of economy of scale and therefore the policy assessment result is consistent with normal cost-benefit analysis. It is an important factor for practical welfare analysis when validity of policy implementation is discussed from a point of view of efficiency. On the other hand, monopolistic competition based models is suitable to theoretical framework of new economic geography field which highlights the economic agglomeration. Agglomeration effect is also an important factor from a point of view of regional economic development effects. Thus the both of two types of models have theoretical and practical merits respectively. However, the results of the model analyses of course depend on the model formulations and can be different in not only detail but also feature of benefit distribution. Understanding the difference of the model outputs by theoretical assumption is crucial theme of practical policy assessment. This paper attempts to compare the economic effects of a road transport development project estimated by a perfect competition based SCGE model and a monopolistic competition based SCGE model quantitatively. Our analysis emphasizes especially the differences in the magnitude of benefit and the regional distribution pattern of benefit because they are usually the largest interests of actual policy assessments. The results show that elasticity of substitution, which is a dominant parameter of monopolistic competition models as a key factor of markup, sensitively affects to benefit and its distribution. It mainly causes the difference of the outputs of the perfect competition based SCGE model and the monopolistic competition model, which implies that the elasticity parameter should be chosen carefully. We furthermore analyze the relationship between size of analysis target region and benefit as well as sensitivity analysis of model parameters. The analysis shows that the regional scale also influences to the benefit estimation in particular by monopolistic competition model. Finally, we summarize the tendency of model outputs of the two types of the models and points to keep in mind for the practical policy analysis by SCGE models.