1986 年 17 巻 p. 33-53
Many authors have been working on land use models focused on metropolitan areas. However, those models vary largely in their perceptions of regions and activities, combinations of endogenous and exogenous variables, and the ways to mediate locational competitions. Since many of those models are so-called engineering models motivated to evaluate the impacts of certain regional policies or projects, they are not necessarily consistent with the macro-econometric models which are to provide their control totals.
The primal purpose of this study is to propose a metropolitan land use simulation model which has a direct link with the system of regional accounts. The model is composed of three major model blocks; regional frame, activity, and location models, among which, the last model is the principal subject of this article. The role of location model is to estimate the regional allocation of stock variables, including land areas, number of employees, industrial assets, number of housing and population.
The model consists of five submodels, which can be summarized as follows. (a) Basic location submodel: to determine increments of land areas occupied by each activity, along with investments and housing construction accompany them, based on Ellickson's random bidding model. (b) Stock accounting submodel: to calculate number of employees and existing amounts of stock variables. (c) Auxiliary location submodel: to distribute investment items and employees of “non-basic” activities. (d) Communting submodel: to determine commuting patterns and population allocation based on a constrained gravity model. (e) Demolition submodel: to estimate the amounts of land released by each activity.
The study area is Kanto Region consists of 1 metropolis and 6 prefectures, which are subdivided into 51 zones. Activities are classified into 35 sectors and 6 housing types. Data are collected and processed for every year between two census years, 1975 and 1980.