From Elevated Freeway to Linear Park: Land Price Impacts of
Seoul, Korea’s CGC Project
Chang Deok Kang and Robert Cervero
WORKING PAPER
UCB- ITS- VWP- 2008- 7
August 2008
From Elevated Freeway to Linear Park:
Land Price Impacts of Seoul, Korea’s CGC Project
Chang Deok Kang and Robert Cervero
Track One
Berkeley Center for the Future of Urban Transport
A Volvo Center of Excellence
August 2008
1
From Elevated Freeway to Linear Park:
Land Price Impacts of Seoul, Korea’s CGC Project
1. Int roduct ion
Freeways and other high- performance roadway investments have long been considered
vital to the economic well- being of metropolitan areas. Empirical research shows that
limited- access, grade- separated freeway systems increase a region’s economic productivity
by lowering transportation costs, a factor input to economic production ( Aschauer, 1990;
Boarnet, 1997). Past studies also reveal that urban land markets capitalize the benefit of
proximity to freeway interchanges, especially for non- residential uses and in areas
experiencing worsening traffic conditions ( Gillen, 1996; Boarnet, 1997; Bhatta and
Drennan, 2003).
Increasingly, city leaders are turning to a different kind of public asset to grow local
economies, notably public amenities, urban parks, and other civic functions that enhance
quality of central- city living. In an increasingly competitive, knowledge- based global
marketplace, improved civic spaces and expansion of the arts and cultural- entertainment
offerings, proponents contend, will appeal to highly sought professional- class workers,
Richard Florida’s ( 2002) so- called “ creative class”.
The desire for high- speed mobility, and freeways in particular, and quality of place pose a
dilemma. On the one hand, access- controlled, grade- separated freeways provide vital
2
mobility, funneling suburbanites to good- paying white- collar office jobs and providing
connectivity to the region at large. On the other hand, they sever longstanding
neighborhoods, form barriers and visual blight, cast shadows, and spray noise, fumes, and
vibrations on surrounding areas. Do the mobility benefits of freeways offset their
nuisance effects, particularly in today’s amenity- conscious workplace? Some city leaders
have decided no, opting to tear down long- standing elevated freeway structures and
replace them with linear parkways or less obtrusive, more human- scale surface boulevards.
In a sense, this represents a re- ordering of public priorities and perhaps even a paradigm
shift -- from a focus on “ automobility” to a focus on “ liveability”, from an emphasis on
expediting the movement of professional- class suburbanites to central cities to one of
attracting professional- class workers to reside in central cities.
Freeway deconstruction is not new. More than 30 years ago, residents of Portland,
Oregon voted to bulldoze the Harbor Drive freeway and replace it with a 37- acre
waterfront park on the edge of downtown. Following the 1989 Loma Prieta earthquake,
San Francisco razed double- deck freeways along its waterfront and several miles inland in
favor of attractively landscaped, multi- way boulevards. Milwaukee recently tore down its
Park East Freeway, opting to use the vacated land for housing, shops, and offices.
Hoping to reverse the flight of households and businesses from the central city, then-
Mayor John Norquist spearheaded a community- based effort to transform 26 acres of
prime urban real estate to a New Urbanism- type “ new town/ in town”. A ground- level
six- lane boulevard, McKinley Avenue, has since been constructed, adorned with tree- lined
medians, granite pavers, and wide sidewalks. At present, freeway demolition is planned
3
for the Innerloop in Rochester, NY, Route 29 in Trenton, NJ, and the Whitehurst freeway
in Washington, D. C. and serious discussions are presently under way to remove the Jones
Falls Expressway in Baltimore, Seattle’s Alaska Way Viaduct, Buffalo’s Skyway, the
Sheridan Expressway in the Bronx, Robert Moses Parkway in Niagra Falls, sections of
Interstate- 5 in Portland, and segments of Paris’s Pompiedou Expressway ( Preservation
Institute, 2007).
Without question, the boldest and most dramatic freeway removal to date has been Seoul,
Korea’s Cheong Gye Cheon ( CGC) project. Under the leadership of then- Mayor ( and
recently elected president of South Korea) Myung- Bak Lee, Seoul’s CGC elevated
expressway was torn down in 2003. A stream buried beneath the freeway was soon
thereafter brought back to the surface as a linear park and bike path. The mayor staked
his 2002 mayoral election campaign on this $ 313 million stream- restoration project, calling
it “ a new paradigm for urban management in the new century” ( Seoul Metropolitan
Government, 2003). Echoing the sentiments of urban visionaries like Jaime Lerner of
Curitiba, Brazil and Enrique Penalosa of Bogotá, Colombia, Mayor Lee defended the
project on the grounds: “ we want to make a city where people come first, not cars”.
The CGC project, perhaps more than any freeway removal to date, represents a recasting
of public priorities.. Specifically, it marks a shift from infrastructure that enhances
“ automobility” to infrastructure that enhances public amenities and quality of urban living.
In this paper, we investigate the land- price impacts of replacing the CGC freeway with a
linear park/ waterway on commercial and residential properties. First, the literature is
reviewed. Next, we present background information on the project, discuss methodology
4
and data sources, and then present a hedonic price model that isolates the influences of
proximity to the freeway and its successor, the urban- stream/ linear- park, on property
values. Multilevel modeling is used to estimate land- value impacts during the 2001- 2002
and 2005- 2006 periods. The article concludes with discussions on the public policy
implications of the research findings.
2. Lite rature Re view
The impact of high- capacity urban highways on neighborhoods has been studied in depth
since freeways first appeared on the urban scene over 50 years ago. Past studies suggest
that under receptive conditions, urban highway networks significantly influence the
location choices of firms and households ( Boarnet, 1998; Boarnet and Chalermpong,
2001). Rapid population and employment growth, matched by worsening traffic
congestion, must be present if nearby properties are to capitalize the accessibility benefits
conferred by highway and freeway interchanges. Even then, other preconditions, like
supportive zoning and complementary infrastructure, must also be in place if road
investments are to have significant and lasting land- use impacts ( Giuliano, 2004). One
study in California concluded that land- value appreciation attributable to highway
investments depend on the network structure and the composition of economic growth
( Boarnet and Haughwout, 2000). The marginal accessibility benefits of an improvement
to an already extensive roadway network might be relatively small, thus bringing about
modest land- value gains. Site features also matter. For commercial activities that rely on
visibility, exposure, and ease of site access, land appreciation is often limited to parcels
close to an interchange ( Voith, 1993).
5
The impacts of highways on development and land prices also vary by length of time and
geographical scale of analysis ( Bhatta and Drennan, 2003). Impacts on urban structure
might be immeasurable in the near term yet appreciable over a number of years due to
time- lags in land development ( Giuliano, 2004). Furthermore, a macro- scale study may
reveal modest land- use effects while a micro- scale study, such as at the neighborhood
level, might find a significant degree of land- use adjustments.
Valuation impacts have also been carried out in the past on a number of different site
amenities as well as dis- amenities, including open space, waste facilities, building designs,
streetscapes, and waterfronts. As an externality, a site amenity or dis- amenity typically
exerts its price influences on a specific aspect of a property, such as a view or proximity
( Kain and Quigley, 1970; Cheshire and Sheppard, 1995). In middle and upper income
settings, property owners are often willing to pay for aesthetics and architectural design to
increase property premium ( Asabere, Hachey, and Grubaugh, 1989; Vandell and Lane,
1989). The impacts of nature, however, are not always as clear. In the case of open space,
studies show the land price impacts vary tremendously. Open space can increase land
prices by its intrinsic qualities ( e. g., greenery, spaciousness) but also by reducing the
amount of developable land available. However, the noise and foot traffic generated by
nearby popular parks and open areas might be viewed as a nuisance by residential
property owners ( Frech and Lafferty, 1984). Proximity to nuisances like toxic waste sites
or airport flight paths universally lowers property values, with residential parcels losing the
most ( Kohlhase, 1991). Other forms of neighborhood amenities rarely overcome such
nuisances. One study, for example, showed that average land prices were no different in
6
nicely landscaped districts than in less lavished ones with similar levels of traffic
congestion ( Polinksy and Shavel, 1976).
Research also shows that the land- price effects of open space vary by size and type. One
study found that small neighborhood parks increased residential property values the most
( Espey and Owusu- Edusei, 2001). Another study found that natural area parks were more
highly valued than urban parks ( Lutzenhiser and Netusil, 2001). A Portland, Oregon
study found proximity to private parks and golf courses increased residential sales prices
( Bilitzer and Netusil, 2000). In general, research shows, the benefits of open space are
most capitalized in residential property values in areas that are denser, have higher
household incomes, and are closer central business areas ( Anderson and West, 2006;
Dehring and Dunse, 2006). The impacts of open space and public amenities on non-residential
properties are less clear. Urban economists like Glaeser et al. ( 2001) contend
that urban parks, open space, and waterfront improvements can help cities attract skilled
workers and knowledge- based industries in addition to stabilizing declining
neighborhoods. One study in greater Los Angeles found that public amenities like parks
did influence the location patterns of firms ( Sivitanidou, 1995).
Far less is known about the price impacts of converting from one form of public amenity
to another, such as replacing an elevated freeway with a linear parkway. A study of
Boston’s notorious “ Big Dig” project ( wherein an elevated freeway was replaced by a
tunnel and a linear park) found that proximity to open space had a positive impact on
residential and commercial property values ( Tajima, 2003).. This, however, was a cross-
7
sectional study that looked only at post- freeway impacts. Our study aims to extend such
work by examining impacts on residential and commercial properties both before and
after the replacement of an elevated freeway with a linear park.
3. Pro ject Des cr ipt ion
A combination of factors weighed in on the decision to tear down Seoul’s CGC elevated
freeway and restore the urban stream – flanked by a linear park-- that pre- dated the
freeway. Cheong Gye Cheon, which means “ clear valley stream,” has long been a source
of fresh water and centerpiece of urban life in Seoul, going back to the 14th century.
During the Chosun dynasty ( 1392- 1910), city dwellers did their laundry in the stream and
frequently socialized on its banks. Following the Korean War ( 1950- 1953), the stream’s
character quickly changed when temporary refugee housing was built along its banks.
Untreated waste was dumped directly into the waterway, turning it into a veritable
cesspool and eventually prompting city officials to cover the stream with an elevated
freeway.
The Cheong Gye Cheon freeway, 50 to 80 meters in width and 6 kms in length, opened in
1971.. Below the road was stream and a sewer trunk- line. The freeway quickly became an
important conduit for movement within central- city Seoul. A 1992 study by the Korean
Society of Civil Engineers, however, found that more than 20% of the freeways steel
beams were seriously corroded and were in need of urgent repair ( CGC Restoration
Project Headquarters, 2004). The Seoul Metropolitan Government immediately began
8
repairing the road’s understructure however due to concerns about the road’s long- term
safety and stability, this was seen as a stopgap to either totally reconstructing the freeway
or tearing it down altogether.
At roughly the same time the CGC freeway was being repaired and under increased
scrutiny, an urban regeneration movement was taking place in Seoul. In the 1980s,
numerous new towns were built on the periphery of Seoul in hopes of relieving central-city
congestion and population growth pressures. Rising automobile travel and the
convergence of more and more radial commutes from the periphery into the urban center
led to steadily worsening traffic conditions. By the late 1990s, some critics charged that
new towns were a failed experiment, exacerbating traffic congestion and environmental
quality due to lengthier, more car- dependent commutes. The idea of re- urbanizing Seoul’s
central areas through “ new towns/ in town” began to surface.
The person who led the charge of re- generating Seoul city was Myung- bak Lee. In 2001,
Lee ran for mayor of Seoul, largely on a platform of reinvigorating Seoul’s central city as
means of creating a more sustainable yet productive city. Lee campaigned on the premise
that Seoul could achieve a better balance between function and the environment. Prior to
becoming major, Lee found and led the Hyundai Group for three decades, Korea’s largest
builder of public works and infrastructure projects. During this time, he earned the
nickname “ Bulldozer” Lee, partly due to the company’s legacy of constructing massive
roadways throughout the country but also because he reputedly once took apart a
bulldozer to study its mechanism and figure out why it kept breaking down..
9
Once taking office in early 2002, Mayor Lee quickly began acting upon his campaign
promises. His vision for future urban transportation called for not only expanding public
transit services but also reducing the ecological footprint of the private car by reclaiming
urban space consumed by roads and highways. The Cheong Gye Cheon freeway- to- linear
park conversion was a natural choice to launch Lee’s vision of a more sustainable urban
landscape for Seoul. Change was swift. By February 2003, a plan for the freeway removal
was completed and five months later, the freeway had been completed disassembled.
Some two years later, in October 2005, the restored CGC stream and linear parky was
opened to the public following a high- profile public celebration and ribbon cutting by
Mayor Lee. The entire cost of the freeway demolition and stream restoration was US$ 313
million.
10
Figure 1.. Location of Case Areas in Seoul
Source: Seoul Metropolitan Government
Today, the restored CGC stream flows east to west in central Seoul ( Figure 1), passing
through 13 districts ( the smallest municipal administrative unit) in four different wards.
In 2003, the land- use composition of building floorspace in these 13 districts was: offices
( 29%), commercial- retail ( 49%), residential ( 13%), and other miscellaneous uses ( 9%).
The CGC project has not been without controversy. One obvious concern would be the
effects of taking out a 6 km freeway on traffic congestion. To help relieve possible
congestion impacts, the Seoul Metropolitan Government opened 68 kms of rapid- bus-only
median lanes along major arterials in 2004 and reconfigured bus routes to better fed
11
into the city’s extensive subway system. Additionally, many small shopkeepers and
merchants were against the project out of fear of losing business. Alongside the former
elevated freeway was an assembly of small- scale shops and markets selling shoes, apparel,
tools, and appliances. Through intensive negotiations, the Seoul Metropolitan
Government was able to head off this opposition by financially compensating merchants
and relocating a number of shops to a newly constructed market center south of the Han
River that was easily accessible by highways and public transit.
4. Study Are a and Ca ses
In order to study the capitalization effects of converting the CGC freeway to a restored
stream and linear park, data were compiled on land prices and site as well as
neighborhood attributes of individual land parcels in the four wards ( of Seoul’s 25 total
wards) that contain stretches of the CGC corridor. Figure 2 shows the CGC corridor
along with plots of the following points in these four wards: land parcels; five freeway
interchange ramps ( when the freeway existed); and 29 pedestrian entrances to the linear
park ( that currently exist).
12
Figure 2. Spatial Distribution of Land Parcels and Access Points
Source: Authors’ work and Seoul Metropolitan Government
Land prices in Seouls’ four central wards vacillated during the 1990s but have trended
upwards since 2000. As shown in Figure 3, land prices plummeted by about 20% in 1998
in the wake of Asia’s financial crisis. Prices rebounded in 2001 and 2002, and in more
recent years have appreciated at an annual rate of 5% to 10%.
13
Figure 3. Annual Rate of Land Appreciation ( 1990~ 2006)
- 30.00
- 20.00
- 10.00
0.00
10.00
20.00
30.00
40.00
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
Year
Rate of Land Appreciation (%)
Seoul
Jongro
Joong
Sungdong
Dongdaemun
Source: Seoul Statistics Office
5. Dat a and Me t hodolo gy
The principal database used in this research was the Publicly Announced Land Price data
maintained by South Korea’s central government, obtained for the periods of 2001- 2002
to 2005- 2006. The database contains records on property address, land use, assessed land
price, and ordinal values for property’s shape, slope, and level of road access. The South
Korean government assesses the price of each land parcel annually for taxation purposes.
Once the central government estimates the value of base parcels, the local government
assesses individual parcel values based on location and site attributes.
Supplementing parcel data were obtained from other government sources on the
economic, demographic, location, land use and public finance attributes for the wards and
districts surrounding land parcels that were studied. Tables 1 and 2 list these additional
14
variables in addition to presenting descriptive statistics for both residential and non-residential
( i. e., office and retail- commercial) parcels. Of note are location variables that
measure the straight- line and network distances of each parcel to the CGC corridor
( specifically to the former freeway ramps and current pedestrian entrances) as well as City
Hall and major transportation infrastructure ( using the proximity function of ArcMap 9.2
and network distance measurement of Arc Workstation).
Multilevel modeling ( MLM) was used in estimating the price effects of proximity to the
former freeway and present- day linear park to account for the hierarchical structure of the
data – i. e., parcels nested within neighborhoods. MLM also allowed for the statistical
control of various covariates, such as attributes of sites ( e. g., parcel shape) and
neighborhoods ( e. g. population density). MLM improves the quality of model estimates
since it accounts for the fact that nearby land parcels share neighborhood characteristics.
Figure 4 diagrams the nature of this interdependence among variables. The failure of the
more traditional approach to hedonic price modeling using ordinal least squares ( OLS) to
account for this violation of independence assumption can result in biased parameter
estimates. Our use of MLM, we believe, represents an important methodological
improvement over past hedonic price studies.
Additionally, we estimated MLM in log- log form for two reasons: one, it provided better
statistical fits than linear formulations and two, it moderated the effects of
heteroschedastic error terms and variables with non- normal distributions. In the models
that follow, all continuous- scale dependent and independent variables were converted to
natural log form. A side benefit of log- log formulations is that estimated coefficients
15
represent elasticities, revealing the relative sensitivity of land prices to changes in the right-hand
side predictor variables.
The multilevel hedonic models we explicitly accounted for the non- independence of error
terms of parcels that share the same wards. Our MLMs took the following random
intercept form:
ij 00 1 ij 2 ij 3 ij 4 ij 0 j ij P = ! + " L + " S + " D + " N + μ + #
Where:
ij P = the CPI- adjusted assessed land price ( per square meter) of parcel i ( Level 1)
in ward j ( Level 2);
Lij = a vector of land attributes such as land use, shape, slope, and road
accessibility in parcel i
( Level 1) in ward j;
Sij = a vector of neighborhood socio- demographic characteristics, such as
population density, employment density, percentage of age class, percentage of
college graduation from Seoul Statistics and Population and Housing Census in
Korea in ward j ( Level 2) that is assigned to each parcel i in that ward;
Dij = a vector of distance to Cheong Gye Cheon, CBD, subway stations, and
roads from each parcel i ( Level 1) in ward j; and
Nij = a vector of neighborhood land use and public finance such as park density,
developable land ratio, road ratio, retail ratio, permit ratio of residential and
16
commercial, and local tax per households in ward j ( Level 2) that is assigned to
each parcel i in the ward.
The statistic used to justify MLM is the intraclass correlation ( ! ), representing the
proportion of the variance explained by the grouping structure in the population. The
intraclass correlation also reveals the expected correlation between two randomly chosen
units that are in the same group. By convention, a intraclass correlation of 0.05 or more
indicates MLM should be used to account for the non- independence of error terms.
Figure 4. Data Structure by Level 1 and 2
Figure 5 reveals the temporal nature of analysis that was carried out. During the period of
study, there were three notable time periods: 2001- 2002, when the CGC freeway was still
Level 1
Land Parcels
Level 2
Wards or Districts
• Land Attributes and Use
• Distance to Cheonggyecheon , CBD,
Subway stations, Roads ( Straight - line)
• Distance or Network Distance to Freeway
Ramps and Pedestrian Entrances
• Economic and Demographic Attributes
• Neighborhood Land Use and Public Finance
17
in operations; 2003- 2004, a transitional period when the freeway was demolished and the
stream restoration/ linear park construction began; and 2005- 2006, presence of central
Seoul’s new urban amenity: the CGC stream and linear park. For purposes of examining
the price effects of the freeway and its urban stream/ linear park replacement, MLMs were
estimated for two periods: 2001- 2002 ( freeway period) and 2005- 2006 ( urban stream/ liner
park period). The 2003- 2004 transitional period was excluded because of the disruptive
effects of the massive construction activities underway at that time.
Figure 5. Time Periods of Projects and Model Design
6. De sc rip ti ve Sta tis tics
Demolition of Freeway Opening Urban Stream
Year
2001 2002 2003 2004 2005 2006
July October
Non Residential and
Residential Models
for Freeway Project
Non Residential and
Residential Models
for Urban Stream Project
18
Table 1 presents descriptive statistics for non- residential parcels in 2001- 2002 ( the
freeway period) and 2005- 2006 ( the urban stream/ linear park period). A total of 6,483
and 7,235 land- parcel observations were available for these two periods, respectively. For
variables related land attributes and use, specifically a parcel’s shape, slope, and road
accessibility, the ordinal scores were assigned by property assessors. Lower values are
desirable– e. g., square parcels on flat slopes with superior road accessibility receive scores
of 1.
For both time periods, the distribution of land uses among non- residential parcels was as
follows: commercial- retail ( 65%), undeveloped land zoned for commercial- retail ( 9%),
office ( 6%), mixed- use ( 19%), and undeveloped land zoned for mixed- use ( 1%). Table 1
also shows that employment densities tended to be higher than population densities in
areas surrounding the sampled non- residential parcels. Further, residents in areas
surrounding the surveyed non- residential parcels tended to be in the mid life- cycle stages
( 20 to 40 years of age).
For the residential parcels surveyed, there were 3,769 and 4,244 observations for the two
time points ( 2001- 2001 and 2005- 2006, respectively). Table 2 shows the surveyed
residential land parcels were primarily used for single family housing ( 77%), followed by
multi- family housing ( 15%), non- developed or open land available for residential use
( 4%), row housing ( 2%), and condominiums ( 2%). For the residential parcels studied,
surrounding population densities were higher than surrounding employment densities.
19
The two tables also shed light on proximity of various activities to the CGC corridor.
Non- residential parcels tend to be closer to CGC than residential lands owing to the fact
that the stream flows through the heart of Seoul where historically offices and retail shops
had predominated. Similarly, non- residential parcels tend to be closer to City Hall
( generally considered to be Seoul’s 100% corner, or heart of the CBD), subway stations,
and major roads. In terms of land- use composition, public expenditures, park land, and
the ratio of parcels that are developed, areas surrounding the surveyed residential and
non- residential parcels are quite similar. However, non- residential parcels tended to have
larger shares of surrounding land devoted to retail activities and high rates of local taxes
( on a per household basis).
20
Table 1. Descriptive Statistics for Non Residential Models
Variable Obs Mean Std. Dev. Min Max Obs Mean Std. Dev. Min Max
CPI- adjusted Land Price ( Korea Won per Square Meter) 6,483 4,496,620 4,161,874 936,567 39,700,000 7,235 6,425,122 5,933,833 1,400,000 58,100,000
Distance to CGC Corridor
Distance to Cheonggyecheon ( straight ine distance, m) 6,483 1,069.2 975.0 20.5 5,2809 7,235 1,0782 978.1 20.5 5,280.9
dummy ( distance ! 100) 6,483 0064 0.245 0000 1000 7,235 0.062 0241 0000 1000
dummy ( 100 < distance ! 200) 6,483 0076 0.265 0000 1000 7,235 0.073 0260 0000 1000
dummy ( 200 < distance ! 300) 6,483 0069 0.254 0000 1000 7,235 0.068 0252 0000 1000
dummy ( 300 < distance ! 400) 6,483 0067 0.249 0000 1000 7,235 0.065 0247 0000 1000
dummy ( 400 < distance ! 500) 6,483 0067 0.249 0000 1000 7,235 0.067 0251 0000 1000
Other Location Factors
Distance to CBD: City Hal ( s raight line distance, m) 6,483 3,360.9 2,347.7 130.4 8,9896 7,235 3,371.4 2,3468 130.4 8,989.6
Distance to Subway Stations ( s raight line distance, m) 6,483 354.1 295.4 48 3,599.1 7,235 356.9 301.1 4.8 3,683.1
Distance to Arterial Roads ( s raight line distance, m) 6,483 36.2 272 00 152.2 7,235 365 27.2 0.0 152.2
Land Attributes and Use
Shape ( 1~ 8)** 6,483 3801 1.794 1000 8000 7,235 3.775 1.782 1000 8000
Slope ( 1~ 5)*** 6,483 2039 0.219 1000 5000 7,235 2.035 0208 1000 5000
Road Accessibi ity ( 1~ 12)**** 6,483 5683 2.967 1000 11.000 7,235 5.765 2910 1000 11.000
Commercial ( 0/ 1) 6,483 0655 0.475 0000 1000 7,235 0.639 0.480 0000 1000
Commercial Raw Lands and Others ( 0/ 1) 6,483 0037 0.188 0000 1000 7,235 0.048 0213 0000 1000
Office ( 01) 6,483 0076 0.265 0000 1000 7,235 0.080 0271 0000 1000
Mixed- Use ( 0/ 1) 6,483 0215 0.411 0000 1000 7,235 0.212 0.409 0000 1000
Mixed- Use Raw Lands and Others ( 01) 6,483 0018 0.132 0000 1000 7,235 0.022 0.146 0000 1000
Neighborhood Economic and Demographic Attributes
Populaion Density ( Persons per Gross Square Km) 6,483 16,1233 11,610.2 1,908.4 44,023.3 7,235 15,5166 11,557.1 1,354.7 42,253.3
Employment Density( Employee per Gross Square Km) 6,483 27,890.4 22,585.0 712.2 91,617.2 7,235 22,4845 18,050.8 638.6 77,221.2
Percentage of College Degree (%) 6,483 0238 0.067 0.100 0350 7,235 0.136 0034 0080 0265
Percentage of 20 to 40 years old (%) 6,483 0366 0.036 0285 0.451 7,235 0.331 0040 0275 0.440
Percenatge of 40 to 60 years old (%) 6,483 0282 0.035 0228 0343 7,235 0.274 0042 0229 0366
Percentage of more than 60 years old (%) 6,483 0.136 0.036 0076 0218 7,235 0.133 0047 0084 0242
Other Neighborhood Attributes
Park Density Ratio ( Park Area per Ward Area) 6,483 0221 0.150 0061 0.441 7,235 0.129 0.146 0008 0.445
Developable Land Ratio ( Developable Land per Ward Area) 6,483 0673 0.138 0.477 0800 7,235 0.674 0.140 0.477 0820
Road Area Raio ( Road Area per Ward Area) 6,483 0.154 0.048 0094 0207 7,235 0.159 0047 0096 0210
Retal Area Raio ( Retail Area per Ward Area) 6,483 0.109 0.119 0006 0298 7,235 0.134 0.144 0012 0341
Percentage of Residential Permit per Total Permit (%) 6,483 0266 0.115 0052 0.478 7,235 0.157 0.114 0041 0.410
Percentage of Commercial Permit per Total Permit (%) 6,483 0557 0.175 0223 0.767 7,235 0.600 0207 0.142 0927
* Shape: Vaue for assessed shape of parcels
** Slope : Value for assessed slope of parcels
*** Road Accessibility: Value for assessed access bility to roads
( Detal Description in Appendix)
2001~ 2002 2005~ 2006
21
Table 2.. Descriptive Statistics for Residential Models
Variable Obs Mean Std. Dev. Min Max Obs Mean Std. Dev. Min Max
CPI- adjusted Land Price ( Korean Won per Square Meter) 3,769 1,218,846 282,184 75,863 2,699,516 4,244 1,858,086 437,631 210,000 5,600,000
Distance to Ramps or Pedestrian Entrances
Network D stance to Freeway Ramps ( m) 3,769 3,2852 1,662.7 1968 7,9629
Network Distance to Cheonggyecheon Pedestrian Enrances ( m) 4,244 2,425.2 1,366.1 193.2 7,254.1
dummy ( distance ! 500) 3,769 0.023 0.150 0000 1.000 4,244 0052 0221 0.000 1.000
dummy ( 500 < distance ! 1,0 00) 3,769 0.136 0.343 0000 1.000 4,244 0.178 0382 0.000 1.000
dummy ( 1,0 00 < distance ! 1,5 00) 3,769 0.134 0.341 0000 1.000 4,244 0205 0.403 0.000 1.000
dummy ( 1,5 00 < distance ! 2,000 ) 3,769 0.125 0.331 0000 1.000 4,244 0.194 0396 0.000 1.000
dummy ( 2,000 < distance ! 2,5 00) 3,769 0.168 0.374 0000 1.000 4,244 0.115 0320 0.000 1.000
dummy ( 2,500 < distance ! 3,0 00) 3,769 0.107 0.310 0000 1.000 4,244 0068 0251 0.000 1.000
Other Locaion Factors
Distance to CBD: C ty Hal ( straight ine distance, m) 3,769 4,5728 2,248.5 7798 9,1149 4,244 4,609.5 2,264.9 779.8 9,114.9
Distance to Subway Stations ( straight ine distance, m) 3,769 627.1 608.1 41.3 3,919.4 4,244 6395 6239 41.3 3,919.4
Distance to Arterial Roads ( s raight line distance, m) 3,769 47.5 33.3 0.0 253.6 4,244 47.2 33.0 0.1 2536
Land Attributes, Type, and Neighborhood Attributes
Shape ( 1~ 8)* 3,769 3.578 1.707 1000 8.000 4,244 3587 1688 1.000 8.000
Slope ( 1~ 5)** 3,769 2.544 0.798 1000 5.000 4,244 2528 0.779 1.000 5.000
Road Accessiblity ( 1~ 12)*** 3,769 8.586 1.380 1000 12000 4,244 8501 1388 1.000 11.000
Single Family Housing ( 01) 3,769 0.814 0.389 0000 1.000 4,244 0.765 0.424 0.000 1.000
Row Housing ( 0/ 1) 3,769 0.021 0.143 0000 1.000 4,244 0023 0.151 0.000 1.000
Multi Famly Housing ( 0/ 1) 3,769 0.127 0.333 0000 1.000 4,244 0.150 0357 0.000 1.000
Condominium ( 01) 3,769 0.013 0.114 0000 1.000 4,244 0023 0.149 0.000 1.000
Raw Land in Residenial ( 0/ 1) 3,769 0.012 0.110 0000 1.000 4,244 0014 0.118 0.000 1.000
Other Lands in Residential ( 01) 3,769 0.012 0.110 0000 1.000 4,244 0024 0.153 0.000 1.000
Population Density ( Persons per Gross Square Km) 3,769 21,710.2 10,7675 3,178.4 44,023.3 4,244 21,185.3 11,087.6 2,888.0 42,2533
Employment Density( Employee per Gross Square Km) 3,769 11,338.9 12,334.1 7122 77,971.7 4,244 12,112.3 10,813.8 638.6 71,2100
Percentage of College Degree (%) 3,769 0.239 0.069 0.105 0.350 4,244 0.148 0039 0.094 0.265
Percentage of 20 to 40 years old (%) 3,769 0.386 0.032 0312 0.451 4,244 0347 0042 0.283 0.440
Percenatge of 40 to 60 years old (%) 3,769 0.260 0.020 0228 0.340 4,244 0260 0028 0.229 0.366
Percentage of more han 60 years old (%) 3,769 0.114 0.021 0076 0.196 4,244 0.119 0030 0.084 0.242
Other Neighborhood Attributes
Park Dens ty Raio ( Park Area per Ward Area) 3,769 0.215 0.171 0061 0.441 4,244 0.120 0.153 0.008 0.445
Developable Land Ratio ( Developable Land per Ward Area) 3,769 0.645 0.144 0.477 0.800 4,244 0648 0.148 0.477 0.820
Road Area Ratio ( Road Area per Ward Area) 3,769 0.142 0.042 0094 0.207 4,244 0.146 0044 0.096 0.210
Retal Area Ratio ( Retail Area per Ward Area) 3,769 0.059 0.095 0006 0.298 4,244 0072 0.115 0.012 0.341
Percentage of Residential Permit per Total Permit (%) 3,769 0.308 0.099 0052 0.478 4,244 0.199 0.113 0.041 0.410
Percentage of Commercial Permit per Total Permit (%) 3,769 0.489 0.167 0223 0.767 4,244 0528 0.198 0.142 0.927
Local Tax per Households ( Korean Won) 3,769 4,370,875 3,406,631 1,259,339 11,900,000 4,244 5,108,723 4,231,003 1,311,776 13,800,000
* Shape: Value for assessed shape of parcels
** Slope : Value for assessed slope of parcels
*** Road Access bi ity: Value for assessed access bility to roads
( Detail Description in Appendix)
2001~ 2002 2005~ 2006
7. Multile vel Line ar He donic Mo dels: Non Resi dential Pa rcels, 2001~ 2002 and
2005~ 2006
For modeling purposes, all of the variables shown in Table 1 were candidates for inclusion
in multi- level hedonic price models. Predictor variables were retained if their coefficients
were statistically significant and matched a priori expectations.
The MLM results for non- residential parcels are presented in Table 3. The intraclass
correlations ( ! ), representing the share of variation explained by the grouping structure
22
( i. e., the wards) , were quite high, justifying the use of MLM. Specifically, 57.8% and
60.2% of the variation of land price is explained by between- group variation among 182
and 185 districts over the two time periods respectively. In both models shown in Table
3, all predictor variables were statistically significant at the 5 percent probability level..
7.1 Effects of Distance to the CGC Corridor
The findings related to the core research question – the relative influences of proximity to
the CGC freeway ( in 2001- 2002) and the CGC urban stream/ linear park ( in 2005- 2006) –
are revealed by the coefficients on the dummy variables for different straight- line distance
intervals, in 100 meter bands ( up to 500 meters). Lying in the 500 meters of the CGC
corridor increased non- residential property values, although more for the linear- park
amenity than the former freeway, for each distance ring. Controlling for other variables in
the model, Figure 6 reveals the marginal distance effects of the urban stream/ linear- park
amenity versus the freeway for the five 100m bands. For non- residential uses, the
marginal effects of the stream/ linear- park amenity within 100 meters are 13% point
higher than that of the freeway. The marginal advantage of the stream/ linear- park,
however, erodes with distance from the corridor, but still exists ½ kilometer away.
23
Figure 6. Marginal Effects of Freeway and Amenity by Distance Intervals
0%
5%
10%
15%
20%
25%
30%
35%
0~ 100 100~ 200 200~ 300 300~ 400 400~ 500
Distance Interval to CGC Corridor ( m)
Marginal Effects
Freeway
Amenity
7.. 2 Effects of Other Factors
The results in Table 3 reveal that other features of a site and surrounding neighborhood
significantly influenced non- residential land prices. This subsection summarizes the
hedonic price influences of the control variables used in the analysis.
Other Location Factors
In general, non- residential land prices fell with distance to Seoul’s City Hall, the nearest
subway station, and arterial roadways in both time periods. The eroding effects of
distance to City Hall on land prices increased in more recent years.
24
Land Attributes and Land Use
In both time periods, high ordinal values assigned to a parcel’s shape, slope, and road
accessibility ( denoting odd shapes, steep slopes, and poor road access) lowered non-residential
land prices. In terms of the type of non- residential use, in the 2001- 2002
period when the CGC freeway was in place, offices enjoyed a 6% land- price premium
while mixed- use parcels were discounted by 25% ( in comparison to the reference land use
category, retail- commercial parcels). These effects were somewhat moderated during the
2005- 2006 era of the stream/ linear- park.
Neighborhood Economic and Demographic Attributes
In 2001- 2002, higher population densities in the four wards traversed by the CGC freeway
were associated with lower non- residential land prices while the opposite held for
employment densities. Office, retail, and other non- residential parcels that are inter- mixed
with similar activities tend to average higher prices because of agglomeration economies as
well as greater opportunities for comparative shopping. The only neighborhood
demographic variable with high enough statistical significance to appear in either model
was the percentage of residents more than 60 years old during the 2005- 2006 period. The
results suggest that the presence of residential areas with older residents are associated
with lower non- residential property prices.
25
Other Neighborhood Attributes
In general, non- residential property values declined as the amount of land devoted to
parks and that was fully developed increased. The availability of developable land appears
to lower non- residential land prices, however retail space and residential building activities
tend to increase prices due to increased competition for land. This was so, however, only
in the 2001- 2002 freeway era.
26
Table 3. Multilevel Linear Hedonic Model for Predicting Non Residential Land Prices per Square
Meter ( 􀀀 )
Variable Coefficient t p Coefficient t p
Fixed Effects
Distance to CGC Corridor
dummy ( 1, if distance ! 100 , o herwise 0 ) 0 203 8.48 0 000 0.330 14 53 0 000
dummy ( 1, if 100 < distance ! 200 , o herwise 0 ) 0.187 8.79 0 000 0 238 11 88 0 000
dummy ( 1, if 200 < distance ! 300 , o herwise 0 ) 0.167 8.69 0 000 0.189 10 50 0 000
dummy ( 1, if 300 < distance ! 400 , o herwise 0 ) 0.114 6.31 0 000 0.133 7.76 0 000
dummy ( 1, if 400 < distance ! 500 , o herwise 0 ) 0 057 3.34 0 001 0 073 4 56 0 000
Other Location Factors
ln( Distance to CBD: City Hall) - 0.469 - 13 87 0 000 - 0 520 - 17.64 0 000
ln( Distance to Subway Stations) - 0.139 - 19.39 0 000 - 0.136 - 20 23 0 000
ln( Distance to Arterial Roads) - 0 016 - 4.63 0 000 - 0 016 - 4 82 0 000
Land Attributes and Use
Shape ( 1~ 8) - 0 008 - 3.99 0 000 - 0 005 - 2.45 0 014
Slope ( 1~ 5) - 0.131 - 7.78 0 000 - 0.146 - 8 86 0 000
Road Accessibility ( 1~ 12) - 0 072 - 53.31 0 000 - 0 072 - 55 50 0 000
Commercial Raw Lands and Others ( 0/ 1) - 0.149 - 8.01 0 000 - 0.177 - 11.30 0 000
Office ( 0/ 1) 0 060 4 28 0 000 0 027 2.14 0 033
Mixed- Use ( 0/ 1) - 0 249 - 24.15 0 000 - 0 239 - 24.43 0 000
Mixed- Use Raw Lands and O hers ( 0/ 1) - 0.332 - 12.71 0 000 - 0.332 - 14.72 0 000
Neighborhood Economic and Demographic Attributes
ln( Popula ion Density) - 0.134 - 2.24 0 025
ln( Employment Density) 0.105 3.46 0 001
ln( Percentage of more han 60 years old) - 0.124 - 2.30 0 021
Other Neighborhood Attributes
ln( Park Density Ra io) - 0.695 - 9.03 0 000 - 0 024 - 2 88 0 004
ln( Developable Land Ra io) - 2.157 - 6.23 0 000
ln( Retail Area Ra io) 0 265 10.62 0 000
ln( Percentage of Residen ial Permit per Total Permit) 0 051 5 97 0 000
Constant 18 591 27.92 0 000 25 553 21.10 0 000
Random Effects
Sigma_ u 0.305 0.321
Sigma_ e 0 261 0 261
Rho 0 578 0.602
Summary Statistics
Number of Observa ions 6483 7235
Number of Groups 182 185
2001~ 2002 2005~ 2006
27
8. Multile vel Line ar Hedo nic Mo dels: Resi dent ial Pa rcels, 2001~ 2002 and
2005~ 2006
Table 4 presents the multilevel hedonic price model results for residential parcels, for each
of the two time periods.. The intraclass correlation ( ! ) confirms that multi- level
modeling is appropriate: 29.8% and 38.7% of the variation in land values is explained by
between- group variation across the 109 and 107 districts for the 2001- 2002 and 2005- 2006
time periods, respectively.
8.. 1 Effects of Distance to Freeway Ramps or Pedestrian Entrances
We found that the CGC freeway generally conferred accessibility benefits among
residential properties surveyed throughout the 109 ward districts however the freeway’s
nuisance effect often detracted from land values within a ½ to 3 km buffer of the freeway.
Consistent with expectations, we found that by 2005- 2006, conversion to an urban
stream/ linear- park increased the value of nearby residential properties, particularly within
½ kilometers of a pedestrian entryway.
The elasticity coefficient on the “ network distance” variable indicates that a doubling of
distance to the freeway ramps was associated with a 15.1% decrease in residential property
values in 2001- 2002, all else being equal. The coefficients on the fixed- effect dummy
variables, however, reveal that residential land values got discounted within a ½ to 3 km
straight line distance of a freeway ramps– generally within earshot and eyeshot of the
28
elevated structure, indicating visual blight and noise depressed values within this zone.
We note that no discount was recorded for the 0 to ½ km buffer due to statistical
insignificance within our sample for the 2001- 2002 period. In general, most residential
properties that were studied were located farther from both the freeway and urban
stream/ linear park than non- residential properties.
By 2005- 2006 when the urban stream/ linear- park had replaced the freeway, the slope
coefficient of the variable measuring network distance to the access points ( which was
now the pedestrian entrances) was flatter than for the 2001- 2002 period. All else being
equal, a doubling of network distance to the pedestrian entrances was associated with a
5.4% decline in residential property values. However there was also a clear amenity
benefit of being situated 2 km of the urban stream/ linear- park – what might be
considered a 5 to 10 minute walkshed. Based on the coefficients on the fixed- effect
dummy variables, the greatest value- added was for residential properties within ½ km of a
pedestrian entranceway; the smaller coefficients on the other dummies indicate this
amenity- based premium declined as one got farther away, up to 2 km. We expect the
premium was associated mainly with the benefit of having a nearby recreational outlet
rather than a view since 77 percent of the residential parcels surveyed were low- rise single-family
homes. For most, their view of the urban stream/ linear- park is obstructed by high-rise
office towers and apartment buildings.
Figure 7 summarize the results by plotting the network marginal effects of proximity to
both the CGC freeway ( in 2001- 2002) and urban stream/ linear- park ( in 2005- 2006). The
29
plots, derived from coefficients on the distance to CGC ramps or pedestrian entrances,
clearly show that the nuisance effects of proximity to the freeway and premium effects of
proximity to the urban stream/ linear- park eroded with distance, up to the 3 km mark.
Thus the freeway was a significant dis- amenity to residential parcels that were the closest
to it whereas the urban stream/ linear- park was just as clearly an amenity.
Figure 7. Marginal Effects of Freeway and Amenity by Distance Intervals
- 20%
- 15%
- 10%
- 5%
0%
5%
10%
0~ 500 500~ 1000 1000~ 1500 1500~ 2000 2000~ 2500 2500~ 3000
Distance Interval to Cheonggyecheon Ramps or Entrances ( m)
Marginal Effects
Freeway
Amenity
8.. 2 Effects of Other Factors
Below, the influences of control variables used in the modeling hedonic price effects of
residential properties are summarized.
30
Other Location Factors
Distance to Seoul City Hall and subway stations only had a significant eroding effect on
residential property values in 2005- 2006 – possibly in the latter case due to the loss of
freeway capacity and thus a higher value placed on being near Seoul’s subway and 100%
corner ( i. e., City Hall). Table 4 reveals that proximity to arterial roadways was also valued
slightly more in the 2005- 2006 post- freeway era.
Land Attributes, Type, and Neighborhood Attributes
Similar to the finding for non- residential properties, odd parcel shapes, steep slopes, and
poorer road access were associated with lower residential land values, in both time
periods. In terms of the type of residential property, Table 4 clearly shows that
condominiums – a popular housing choice throughout Korea -- commanded the highest
premium: 19.8% compared to single- family housing which served as the reference group.
In 2001- 2002, the model shows residential property values tended to be higher in settings
with higher employment density and more highly educated residents. Higher employment
densities increase residential values by exerting competitive pressures on bid rents and also
because retail densities also rise which means closer proximity to consumer goods and
services.
Other Neighborhood Attributes
31
Table 4 shows that during both time periods, residential property values were positively
associated retail activities in the neighborhood ( suggesting the benefits of mixed uses) but
negatively associated with development intensities. A high density of nearby parks also
generally had a depressing effect on residential property values. This could reflect
peculiarities of downtown Seoul wherein housing near parks only appreciate in value
when the parks are of significant size and substantially developed. Lastly, the negative
sign on the coefficient for the variable denoting local tax per household likely reflects the
negative capitalization of tax burdens once factors like quality of nearby public
infrastructure ( e. g., proximity to subway) are accounted for.
32
Table 4. Multilevel Linear Hedonic Model for Predicting Residential Land Prices per Square Meter
( 􀀀 )
Variable Coefficient t p Coefficient t p
Fixed Effects
Distance to Freeway Ramps or Pedestrian Entrances
ln( Network Distance to Freeway Ramps or Pedestrian Entrances ) - 0.151 - 7 54 0 000 - 0 054 - 3 57 0 000
dummy ( 1, if distance ! 500, o herwise 0 ) 0 083 2.35 0 019
dummy ( 1, if 5 00 < distance ! 1,0 00 , o herwise 0 ) - 0.154 - 5 24 0 000 0 060 2.19 0 029
dummy ( 1, if 1,0 00 < distance ! 1,5 00, o herwise 0 ) - 0 088 - 3 51 0 000 0 057 2.34 0 019
dummy ( 1, if 1,5 00 < distance ! 2,000, otherwise 0 ) - 0 041 - 1 83 0 068 0 052 2.37 0 018
dummy ( 1, if 2,000 < distance ! 2,5 00, o herwise 0 ) - 0 059 - 3 26 0 001
dummy ( 1, if 2,500 < distance ! 3,0 00, o herwise 0 ) - 0 039 - 2 81 0 005
Other Location Factors
ln( Distance to CBD: City Hall) - 0.137 - 4 98 0 000
ln( Distance to Subway Stations) - 0 027 - 4 51 0 000
ln( Distance to Arterial Roads) - 0 010 - 4 05 0 000 - 0 014 - 5.67 0 000
Land Attributes, Type, and Neighborhood Attributes
Shape ( 1~ 8) - 0 010 - 6.63 0 000 - 0 007 - 5 08 0 000
Slope ( 1~ 5) - 0 099 - 25 92 0 000 - 0 094 - 23.52 0 000
Road Accessibility ( 1~ 12) - 0 054 - 28 23 0 000 - 0 053 - 28.44 0 000
Row Housing ( 0/ 1) 0 033 1 97 0 049 0 040 2 57 0 010
Mul ifamily Housing ( 0/ 1) 0 023 2 94 0 003 0 015 2 05 0 040
Condominium ( 0/ 1) 0.198 8 96 0 000 0.198 11.57 0 000
Raw Land in Residen ial ( 0/ 1) - 0.695 - 30.38 0 000 - 0 573 - 27.28 0 000
ln( Employment Density) 0 062 4 96 0 000
ln( Percentage of College Degree) 0 208 4.45 0 000
Other Neighborhood Attributes
ln( Park Density Ra io) - 0 273 - 3 51 0 000 - 0 021 - 3 5 0 000
ln( Developable Land Ra io) - 1.735 - 6 59 0 000 - 0 599 - 3 2 0 001
ln( Retail Area Ra io) 0 213 13 24 0 000 0.123 3 84 0 000
ln( Percentage of Residen ial Permit per Total Permit) 0 033 2.47 0 014
ln( Percentage of Commercial Permit per Total Permit) 0.106 5 08 0 000
ln( Local Tax per Households ($)) - 0.155 - 3.68 0 000
Constant 17 094 9.62 0 000 18 913 22.6 0 000
Random Effects
Sigma_ u 0 091 0.115
Sigma_ e 0.140 0.145
Rho 0 298 0.387
Summary Statistics
Number of Observa ions 3769 4244
Number of Groups 109 107
2001~ 2002 2005~ 2006
33
9. Conclusio n
These empirical results support the core hypotheses of our research. In general, the CGC
corridor conferred benefits mainly to offices, retail shops, and other non- residential uses
when it was a freeway. However within a 500 meter buffer, non- residential properties
were generally worth more when the freeway was replaced by an urban stream/ linear-park.
This could reflect a number of influences, such as the successful substitution of a
massively expanded busway network for the lost freeway capacity and the presence of a
major public amenity in influencing office locations ( e. g., attraction of high- skilled white
collar workers to location near an attractive recreational outlet). Similarly, retail land
prices might have gone up near the urban stream/ linear- park in anticipation of more high-salaried,
professional class workers and residents being attracted to the area. While the
statistical results do not exactly tell us why non- residential properties capitalized even
greater benefits from an urban stream and linear park than a freeway, we believe these to
be plausible explanations. Follow- up research using qualitative methods, including
informant interviews, will be carried out to address this matter further.
The impacts of the freeway- to- amenity conversion on residential property values are
unequivocal. Residences within an nuisance zone of the elevated freeway structure sold at
a discount whereas a few years later when the urban stream/ linear- park was in place, they
sold at a premium. While proximity to freeway on- ramps was valued by residential
properties, this benefit was offset by nuisance effects of noise, dust, fumes, and visual
blight for residences within several kms of the structure.
34
Our research findings have clear policy implications as related to public investment and
dis- investment. For one, replacing a freeway with a restored urban stream and nicely
landscaped linear park clearly changed the spatial distribution of capitalized benefits and
dis- benefits. As with most public resource allocation choices, there were winners and
losers, however since the net marginal effects were highest following the freeway removal,
we believe this massive change in public works in central Seoul resulted in winners far
outnumbering losers.
Additionally, the magnitude of land- value impacts from the freeway and urban stream
were sensitive to the location of land parcels. Offices, retail stores, and other non-residential
activities were closest to the elevated freeway and generally enjoyed the
enhanced regional accessibility benefits that were conferred. Most residences were
situated well beyond ½ kilometer of the freeway and while they enjoyed some regional
accessibility advantages, these were eclipsed by disamenities for parcels within an view-shed,
noise- shed, and “ odor”- shed of the elevated structure. By the time the urban
stream/ linear- park was in place, land- use conversions had occurred. Not only were
nearby offices and retail shops commanding higher rents, but new residences in high- rise
apartments were appearing, taking advantage of this public amenity. The spillover benefit
of the urban stream and linear- park is revealed by a heat- island effect study that found
ambient temperatures along the inner- city stream were 3.3O to 5.9O Centigrade lower than
along a parallel road 4 to 7 blocks away ( Seoul Development Institute, 2006)
In close, Seoul Korea’s bold experiment with replacing a heavily- trafficked inner- city
freeway with a restored stream and linear park has been an unqualified success based on
35
land- market performance. Are these lessons transferable to other places? Perhaps.
Ideally, research will be performed on other freeway- to- amenity conversions that have
occurred or are now under way to probe the generalizability of our research findings. It is
important to note that the Cheong Gye Cheon conversion has not been Seoul’s only
massive public- works transformation. At roughly the same time the freeway- to- stream
conversion was taking place, Mayor Lee converted a massive 1.32 ha- sized roadway
interchange near Seoul’s City Hall to an oval inner- city park. Also, 74 linear kms of
roadway space has been expropriated and given over to the city’s massive expanded
dedicated bus- lane network. As part of a larger urban experiment aimed at reinvigorating
the city, the benefits conferred by the Cheong Gye Cheon conversion could reflect the
influences of a larger urban renaissance that is now underway. Clearly, the dynamic
transformation of Seoul’s urban landscape in recent years calls for more research to enrich
our collective understanding of the welfare implications of designing a city less for
automobiles and more for people.
36
Refer ence s
Anderson, S. T. and West, S. E. ( 2006) Open Space, Residential Property Values, and
Spatial Context, Regional Science and Urban Economics, 36, pp. 773- 789.
Asabere, P. K., Hachey, G., and Grubaugh, S. ( 1989) Architecture, Historic Zone, and the
Value of Homes, Journal of Real Estate Finance and Economics, 2, pp. 181- 195.
Aschauer, D. A. ( 1990) Highway Capacity and Economic Growth, Economic
Perspectives, pp. 14- 24.
Bhatta, S. D. and Drennan, M. P. ( 2003) The Economic Benefits of Public Investment in
Transportation: A Review of Recent Literature, Journal of Planning Education
and Research, 22, pp. 288- 296.
Boarnet, M. ( 1997) Highways and Economic Productivity: Interpreting Recent Evidence,
Journal of Planning Literature, 11( 4), pp. 476- 486.
Boarnet, M. ( 1998) Spillovers and the Locational Effects of Public Infrastructure, Journal
of Regional Science, 38( 3), pp. 381- 400.
Boarnet, M. and Chalermpong, S. ( 2001) New Highways, House Prices, and Urban
Development: A Case Study of Toll Roads in Orange County, CA, Housing
Policy Debate, 12( 3), pp. 575- 605.
Boarnet, M. and Haughwout, A. F. ( 2000) Do Highways Matter? Evidence and Policy
Implications of Highway's Influence on Metropolitan Development..
Washington, DC: Brookings Institution
Bolitzer, B. and Netusil, N. R. ( 2000) The Impact of Open Spaces on Property Values in
Portland, Oregon, Journal of Environmental Management, 59, pp. 185- 193.
37
CGC Restoration Project Headquarters ( 2004) CGC Restoration.. Seoul: Seoul
Metropolitan Government.
Cheshire, P., and Sheppard, S. ( 1995) On the Price of Land and the Value of Amenities,
Economica, 62, pp. 247- 267.
Dehring, C. N. D. ( 2006) Housing Density and the Effect of Proximity to Public Open
Space in Aberdeen, Scotland, Real Estate Economics, 34( 4), pp. 553- 566.
Dipasquale, D., and Wheaton, W. C. ( 1996) Urban Economics and Real Estate Markets..
Englewood Cliffs, NJ: Prentice Hall.
Espey, M. and Owusu- Edusei, K. ( 2001) Neighborhood Parks and Residential Property
Values in Greenville, South Carolina, Journal of Agricultural and Applied
Economics, 33( 3), pp. 487- 492.
Florida, R. ( 2002) The Rise of Creative Class, Jackson: Basic Books.
Frech III, H. E. and Lafferty, R. N. ( 1984) The Effect of the California Coastal
Commission on Housing Prices, Journal of Urban Economics, 16, pp. 105- 123.
Gillen, D. ( 1996) Transportation Infrastructure and Economic Development: A Review
of Recent Literature, Logistics and Transportation Review, 32( 1), pp. 39- 62.
Giuliano, G. ( 2004) Land Use Impacts of Transportation Investments: Highway and
Transit, in: S. Hanson and G. Giuliano ( Ed.), The Geography of Urban
Transportation, pp. 237- 273. New York: Guilford Press..
Glaeser, E. L., Kolko, J., and Saiz, A. ( 2001) Consumer City, Journal of Economic
Geography, 1, pp. 27- 50.
Hamilton, L. C. ( 2006) Statistics with STATA. Belmont: Thomson Higher Education.
38
Kain, J. F. and Quigley, J. M. ( 1970) Measuring the Value of Housing Quality, Journal of
the American Statistical Association, pp. 532- 548.
Kohlhase, J. E. ( 1991) The Impact of Toxic Waste Sites on Housing Values, Journal of
Urban Economics, 30, pp. 1- 26.
Lutzenhiser, M. and Netusil, N. R. ( 2001) The Effect of Open Spaces on A Home's Sale
Price, Contemporary Economic Policy, 19( 3), pp. 291- 298.
Ministry of Construction and Transportation, Publicly Announced Land Price data ( 2001
~ 2006). Seoul, Korea.
National Statistical Office, Population and Housing Census, 2000 and 2005. Dae- jeon,
Korea.
Polinsky, M. and Shavell, S. ( 1976) Amenities and Property Values in a Model of an Urban
Area, Journal of Public Economics, 5, pp. 119- 129.
Rabe- Hesketh, S. and Skrondal, A. ( 2005) Multilevel and Longitudinal Modeling Using
Stata.. College Station , TX: STATA Press.
Rosen, S. ( 1974) Hedonic Prices and Implicit Markets: Product Differenciation in Pure
Competition, Journal of Political Economy, 82( 1), pp. 34- 55.
Seoul Development Institute ( 2006) Study on Urban Structure and Form in CGC Project.
Seoul, Korea.
Seoul Metropolitan Government ( 2003) CGC Restoration Project. Seoul, Korea.
Seoul Statistics Home Page,
http:// www. seoul. go. kr/ v2007/ publicinfo/ statistics/ data/ 4 02. html.
39
Sivitanidou, R. ( 1995) Urban Spatial Variation in Office- Commercial Rents: The Role of
Spatial Amenities and Commercial Zoning, Journal of Urban Economics, 38, pp.
23- 49.
Tajima, K. ( 2003) New Estimates of the Demand for Urban Green Space: Implication for
Valuing the Environmental Benefits of Boston's Big Dig Project, Journal of
Urban Affairs 25( 5), pp. 641- 655.
The Preservation Institute Home Page, http:// www. preservenet. com/.
Vandell, K. D. and S. J. Lane ( 1989) The Economics of Architecture and Urban Design:
Some Preliminary Findings, AREUEA Journal, 17, pp. 235- 260.
Voith, R. ( 1993). Changing Capitalization of CBD- Oriented Transportation Systems-
Evidence from Philadelphia, 1970- 1988, Journal of Urban Economics, 33( 3), pp.
361- 376.
Woodridge, J. ( 2006) Introductory Econometrics.. Mason, OH: Thompson South
Western.
40
Ap pendix : De sc rip tion on Shape, Slop e, and Roa d Acce ssib ility Var ia bles
1. Shape ( Number means the value of assessed land shape)
2. Slope
Value Description
1 Land remarkably lower than roads or nearby landscape
2 Flat land
3 Land with slope less than 15 ˚
4 Land with slope more than 15 ˚
5 Land remarkably higher than roads or nearby landscape
3. Road Accessibility
Value Description
1 Land facing to more than 25meter- wide roads
2 Land facing to 8~ 12 meter- wide and more than 25meter- wide roads
3 Land facing to less than 8 meter- wide and more than 25meter- wide roads
4 Land facing to 12~ 25 meter- wide roads
5 Land facing to 12~ 25 meter- wide and other roads
6 Land facing to 8~ 12 meter- wide roads
7 Land facing to 8~ 12 meter- wide and other roads
8 Land facing to less than 8 meter- wide roads
9 Land facing to auto accessible roads
10 Land facing to small tractor accessible roads
11 Land facing to two small tractor accessible roads
12 Land without accessible roads
1
2
Road
3
4
Road
5
6 7
8