1986 年 17 巻 p. 127-143
There have been many studies about estimation of residential bid rent with some different approaches. They are traditional cross section analyses, hedonic price approach methods and disaggrigate models newly developed.
Those studies commonly assume the monocentric city hypothesis. It guarantees that we can estimate the unique bid rent curve along the distance from city center for the households who have uniform income and utility function. But actual cities where we try to estimate bid rent functions have fairly wide job distribution. As different job sites yield the different bid rent curves, we need more information on households' job location and commuting time adding to house location and land price there for the estimation. And it is very hard to gather such collective information.
However, we can get rid of the trouble under certain conditions of job and house distribution. We discuss the condition in the case of a simple one dimention urban model with job sites. It is found that if the number of households working at job sites distributed from city center to a point in the city exceeds the number of houses supplied for them locating in the area from city center to the point, each bid rent curve of household at different job site lies on the unique bid rent curve.
The range of the area is examined using actual job house distribution data in Tokyo. Then we can estimate the bid land price function for the households living in the area using only the data in the Land nautification book published by National land agency. The estimation is executed by three building type, high-rise apartment, low-rise wooden apartment and detached house. The random bid rent model which is developed by Lerman and Kern and has never actually computed is used. This model has the both characteristics of regression analysis and discrimination analysis. The computed results show fairly good levels of these two way of analysis.
The most important variable of bid land price function estimated is the time distance from the city center to the location of a house in the three estimation cases. As for the absolute number of parameter value estimated for the variable the biggest one is of high-rise apartment, the second low-rise wooden apartment and the third detached house. It is found too that the width of street in front of a building is important for high-rise apartment and the distance to the nearest railway station from a house location is important for low-rise wooden apartment. These characteristics of bid land price functions estimated have good correspondings to our observations of actual housing market. Compared with the parameter value estimated by regression analysis, the difference of the parameter values estimated by this model among building types is much bigger than the one by regression analysis.