DEMAND FOR HOUSING IN TOKYO METROPOLITAN

REGION: A DISCRETE CHOICE ANALYSIS

Housing is a commodity, which is a bundle of characteristics specific to a housing market,

tenure, income and other household characteristics. Voluminous literature exists on

housing demand in which a quantitative measure defined as housing services is used to

measure housing demand. Housing or, more precise, the service stream from a housing

unit, is a heterogeneous commodity. Some dimensions, such as age or size of structure,

are measured on continuous scale, others, such as tenure or type of structure, are discrete

properties. Measuring the volume of housing services as housing expenditure, as is done

in previous research, essentially ignores the heterogeneity, and for large number of policy

purposes like impact of tax on tenure choice, choice between owning and renting etc., the

distribution of housing consumption into qualitatively different categories is of more

interest than an aggregate qualitative measure of housing expenditure alone. Earlier

research has addressed the issues of methodologies in housing demand estimation and

different market and related differences in demand elasticities. The econometric theory of

joint discrete/continuous models is well studied, and there exist a variety of applications.

However, there is paucity of research applications to analyze housing demand using

discrete models. The limited research in this area has focused only on American or

German housing markets. There is no research on housing markets, which treats housing

demand as discrete choices, for Japan, despite the economic importance of Japanese

economy. We model housing demand in Japan using a discrete choice model. A nested

multinomial logit model (NMNL) as the basic analytical tool for our analysis. The

microeconomic and econometric foundations of NMNL models encompass the elegant

theory of housing economics of a utility maximizing household. NMNL models impose a

hierarchical structure on the choice set that can be visualized in the form of a decision

tree. Three dimensions of choice, tenure, dwelling size (as number of rooms) and

structure type (as type of unit) generate these steps of clustering. This paper estimates the

choice probabilities and demand elasticities of various housing alternatives for Tokyo

using 1993 housing survey data for 23 wards. The paper concludes with policy

prescriptions.