Understanding residential location choices: an application of the UrbanSim residential location model on Suwon, Korea

ABSTRACT

The residential location choice model is an effective tool to analyze the actual household demand for housing and better living environments, and many researchers have developed various residential location choice models. In this study, a residential location choice model using a discrete choice modelling framework within UrbanSim is applied to Suwon, Korea with the following aims: (1) to investigate factors affecting residential location choice in Suwon, (2) to forecast changes in household residential locations, and (3) to derive policy implications for the local housing market. An extensive database of parcels, households, jobs, land prices, and transportation networks is geocoded on the basis of grid cells that measure 150 × 150 metres. The estimation results show that access to employment opportunities, the ratio of housing cost to income, mixed land use, and the year that housing was built are important factors in determining household residential locations in Suwon. In addition, different age and income groups have different residential location preferences. UrbanSim, a highly disaggregated microsimulation model, is employed to forecast changes in household residential locations using the estimation results of the residential location choice model. These suggest that different income groups show different migration patterns.

KEYWORDS:

• Residential location choice
• land use and transportation
• discrete choice model
• UrbanSim
• Suwon

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 Travel cost by each mode was calculated with consideration of fuel cost (1600 won), base fares of subways and buses plus extra charges per extra distance, and base fares of taxi (1900 won) plus extra charges per extra distance (100 won per 144 m).

2 The parcel-version is much more detailed and sophisticated to forecast urban land use changes than the zone- and grid-version. However, allocation of the population and jobs is not easy because our ‘jipgegu’ data are spatially larger than parcel data. Hence, we apply a grid-version of UrbanSim.

3 HIES and the Korean Census have the same categories of socioeconomic characteristics: housing type, household size, housing ownership, and education status. Therefore, we used these four categories to estimate household income, and then the estimated household income was inflation-adjusted to reflect calendar year 2005. More detailed information is available upon request.

4 Jipgegu is a level of micro census in Korea, and its median area is approximately 0.02 km². Each jipgegu has the numbers of households, employments, housing buildings, and firms (https://sgis.kostat.go.kr).

5 , where N is the number of land use types under consideration and is the fraction of the neighborhood that is of land use type j.