The effects of high-density suburban development on commuter mode choices in Seoul, Korea

Abstract

This study empirically assesses the effects of high-density suburban development on commuter mode choices in Seoul, Korea. We separated the impacts of densification and new suburban development on mode choice through an examination of interaction effects between population density and the sizes of newly developed suburban areas in logistic regression models. As expected, population and employment densities are positively related to public transit use, while new suburban residential and nonresidential developments are positively related to automobile use. However, the interactive logistic model empirically confirms that a one-unit increase in development density of the new suburban residential increased automobile usage by 27% in 1996 and 17% in 2006 (13% in 1996 and 16% in 2006 for new nonresidential developments), indicating that suburbanization and density are not at odds, but rather are corroborative in encouraging automobile use in the Seoul metropolitan area (SMA). One of the primary explanations for this finding is the strong housing preference for suburban high-rise apartment buildings among SMA residents.

Introduction

Urban planners are concerned about the undesirable consequences of urban sprawl, such as high levels of automobile-associated congestion and air pollution. As a reaction to unfettered outward low-density development, Smart Growth advocates suggest increasing development densities, encouraging mixed land use and investing in public transit (Downs, 2005). The underlying assumption of Smart Growth is that a positive relationship exists between dense mixed land uses and public transit ridership.

Urban sprawl contributes to automobile dependence because the separation between land uses in low-density development makes driving inevitable, while mixed land use and densification reduce automobile dependence and promote transit and use of non-motorized modes (Cervero, 2002, Cervero and Gorham, 1995, Frank and Pivo, 1994, Friedman et al., 1994, Handy, 2005, Kitamura et al., 1997, Nelson and Nygaard, 1995, Newman and Kenworthy, 1989, Newman and Kenworthy, 1999, Pushkarev and Zupan, 1977, Seskin and Cervero, 1996). Unlike existing studies, which have focused primarily on cities in North America and Europe, in this study we empirically analyze the relationships between land use patterns and commuter mode choices in Seoul, the capital of South Korea. Development density, levels of mixed land use and public transit share are much higher in the Seoul metropolitan area (SMA) than in Western cities, already reaching beyond the levels suggested by Smart Growth proponents. In addition, Seoul has experienced high-density suburban development in the last several decades, which provides a good test bed for investigating how the two contradictory factors, one encouraging automobile use (suburbanization) and the other promoting transit ridership (density), affect commuter mode choice.

Seoul¹ is unique in terms of development density, mixed land use patterns and provision of public transit compared to Western cities. First, the population of Seoul, known as one of the fastest growing metropolises in the world, has almost quintupled over the last half-century, from 5.2 million in 1960 to 24.5 million in 2007 (Korean Statistical Information Service: http://kosis.kr). Thus, the core city of Seoul is one of the most densely urbanized areas in the world with 52,500 inhabitants per square mile, a figure 8.5 times higher than the Los Angeles metropolitan area as of 2000.² In addition, Seoul is noted for its rapid suburbanization. Massive suburban residential development, including five new towns, has attracted approximately 3.3 million residents from the core city of Seoul, as well as from other parts of Korea, to the Seoul suburbs in the last 20 years (Korean Statistical Information Service: http://kosis.kr).

Second, although Seoul first introduced zoning in 1934, Seoul’s land-use pattern is a mixture of residential and nonresidential uses, in part because high-density residential development makes it possible to attract small- and medium-size businesses, mostly within walking distance, and in part because of flexible zoning laws that allow commercial buildings to be located within residential areas. In addition, Seoul’s zoning laws do not include minimum lot size, which is believed to be partly responsible for low-density suburban development in the US. The separation between large-scale low-density residential development and shopping malls in suburban Seoul, which are popular elements in major US metropolitan suburban areas, is difficult to detect.

Third, massive investments in public transit have been made to connect newly developed suburban areas and downtown Seoul, focusing on a heavy rail system. The SMA subway system is one of the most extensive and heavily used transit systems in the world. The SMA subway system has 11 lines and handles more than 10 million daily trips (Ministry of Land, Transport, and Maritime Affairs: http://www.mltm.go.kr). Beginning with the operation of Line 1 in 1974, the SMA subway system has been extended significantly by the construction of new lines between 1995 and 2005. The total length of the subway system increased by 101% from 222 km in 1995 to 447 km in 2005 (Ministry of Land, Transport, and Maritime Affairs: http://www.mltm.go.kr). In addition, in 2004 the Seoul metropolitan government finished constructing a Bus Rapid Transit (BRT) system that also links suburban residential areas to urban employment centers. As of 2010, 81 km of the BRT network covers seven different corridors, with plans to extend the BRT to a length of 191.2 km over 16 routes (Seoul Metropolitan Government, 2009).

Due to Seoul’s unique features in terms of development density, mixture of land uses and public transit provision, we sought to empirically examine how these land use features affect commuter mode choices using logistic regression models, focusing on the impacts of recent high-density suburban development on commuter mode choice. We include development density, land sizes of new developments and levels of mixed use in logistic regression analyses in order to measure the effects of these variables on commuter mode choice. Moreover, we include two interaction variables for investigating the net effects of suburban development and density on mode choice³: product terms of (1) newly developed residential areas and population densities of residential areas, and of (2) newly developed nonresidential areas and employment densities at workplaces.

This paper consists of four sections. The first reviews the existing literature, focusing on theoretical and empirical studies pertaining to factors affecting commuter mode choice. The second introduces the characteristics of Seoul’s suburban development patterns and commuter mode choices. The third section describes binary logistic regression models and explanatory variables. The fourth section presents model results and interpretations, followed by conclusions and policy suggestions.