Impact of a new subway line on housing values in Daegu,Korea: Distance from existing lines

Abstract

Chinese

This study investigates the ex ante as well as ex post impact of a new subway line on housing values in Daegu, Korea where two lines already exist. Housing units are divided into two groups: treatment group versus comparison group based on the distance from the nearest station of the new line. Our results based on Hedonic models in difference-in-differences framework suggest that homes within 500 m from the nearest station of the new line can earn a premium of 99.7 thousand Korean Won (equivalent approximately to 96.3 US dollars) per square metre. However, such a premium from improved accessibility to public transit seems to appear mainly for the homes which are within 500 m from the new line and beyond 5 km from the nearest station of the existing lines. For the homes which are close to existing lines, proximity to the new line stations does not increase residents’ convenience remarkably because people are supposed to easily transfer between the new line and existing lines.

Keywords

accessibility capitalisation difference-in-differences estimation hedonic model

Introduction

The rate of urbanisation in Korea has risen sharply from 40.7% in 1970 to 82.2% in 2013 and traffic congestion in large cities has grown worse. As a result, transportation planning has been emphasised in recent urban planning, together with general land use.¹ In fact, efforts to create systematic and effective urban structure through enhancement of connection between land use and transportation in actual construction of a city continue. Particularly, the importance of public transportation is highlighted in perspective of the burden of traffic congestion costs, consideration of the vulnerability of transportation, and transportation welfare. In addition, its importance has increased with regard to sustainable development, formation of energy efficient urban structure, and expansion of green transportation.

Currently, a subway system is regarded as an essential mode of transport a modern city has to have because of its high level of on-time operation, safety, modal split, and low fares. Another important domain in urban development, with regards to construction of a subway system, is the development of station influence areas. Basically, the development of station influence areas attracts local government’s concerns as it promotes management stability of a subway system by increasing modal split of a subway system, and eventually, improves efficiency of land use and sophistication of urban structure.²

A study of the change in surrounding land and house values resulting from the construction of a subway system needs to be accompanied with estimations of development gains and a feasibility study of the development of station influence areas. This is also an important issue from the policy point of view to the extent that urban planning is established according to such studies (Choi and Yoon, 2004).

This study estimates the willingness to pay for the benefits of improved accessibility to public transportation with empirical analysis of the change in apartment values according to construction of Daegu Subway Line 3 at the time the construction decision was made as well as at the time the operation started. Many previous studies on Korean cases mainly use data after the opening of a subway system. However, this study has advantages from quantitative point of view by applying a difference-in-differences framework utilising panel data before and after subway construction decision was made.³ Particularly, this study is differentiated from previous studies to the extent that it classifies individual apartments into the treatment and comparison group by using distance from existing lines as well as the new line, sets a variety of conditions and analyses them under such conditions. Most of the previous studies about the change in land values or housing prices by the construction of a new subway line only consider accessibility to a new subway line and neglect the effects of accessibility to existing lines even if transfers between a new line and existing lines are available. The purpose of this study is to present the time when development gains of station influence areas are realised and a relevant geographical range by investigating changes in surrounding apartment value. In the model that extends analyses to the time points of operation of the new line, it is observed that two-thirds of the benefits of improved accessibility to the public transport were realised when the construction decision was made and the remaining one-third when the new line started its operation.

Literature review and analytic model

Since a theoretic background for the hedonic approach to house pricing was suggested by Rosen (1974), a lot of theoretic and empirical studies have been conducted. Recently, many variables including accessibility to surrounding physical environment and convenience facilities, educational amenities, and policy as well as characteristics of a house are considered as determinants of housing price.

Empirical works on capitalisation of benefits of accessibility to public transport, particularly a subway system, into land rent or housing price have been done in several previous studies. Among others, recent works (e.g. Andersson et al., 2010; Debrezion et al., 2011; Diao, 2015; Duncan, 2011; Yiu and Wong, 2005) find the positive effect of improvement of rail transport facilities on housing value. Bowes and Ihlanfeldt (2001) on the case of MARTA in Atlanta show that property value between one and three miles of stations increased relative to comparable properties located more than three miles, while properties within one-quarter mile decreased by 19% compared with properties beyond three miles. Weinstein and Clower (2002) for DART in Dallas also find that the assessed value of properties near DART stations increased 32% compared with 20% in control group areas not served by rail. On the other hand, Landis et al. (1995) find no statistically significant effect on home prices on the case studies of Sacramento Light Rail. There are also some studies (e.g. Armstrong and Rodriguez, 2006; Strand and Vagnes, 2001) that show the existence of a negative influence of rail accessibility on property values. They argue that it is due to noise, crime, and environmental concerns around railway stations.

Studies for Korean cases could be summarised as follow. First, Kwon et al. (2001) demonstrate that distance from a subway station has effects on increase of land value through analysis of changes in land value before and after opening of a subway system. Similarly, Choi et al. (2008) propose that there are significant changes in a land near a subway station with analysis of changes in land value by a section from a subway station. In addition, Bae et al. (2003) and Choi and Yoon (2004) estimate effects of Seoul Subway Line 5 and Line 7, respectively, on the price of surrounding apartments. Choi and Seong (2011) estimate the effects of construction of Seoul Subway Line 9 on the price of surrounding apartments by stage of the construction and proposes that such effects exist before approval of the subway construction master plan. However, most previous studies do not consider existing lines although a new line is connected to existing lines in terms of transfer.

The main topic of this study is to estimate changes in surrounding apartment’s value, which was caused by the decision of constructing Daegu Subway Line 3. Generally, matching, double difference, instrumental variable or regression discontinuity design is used as an empirical method to analyse ex-post effect of a policy (Khandker et al., 2009). In this study, hypotheses are tested by estimation of the hedonic model for apartment price as shown below, which uses the double difference framework.

y_{jt} {= b}_{0} + b_{1} {Treat}_{j} + b_{2} {Post}_{t} + b_{3} {(Treat * Post)}_{jt} + C_{j} b_{4} + \sum b_{k} {year}_{k} + e_{jt}

where, y_jt represents the average sale price in log of apartments in the complex j for the year t. Treat_j is a dummy variable indicating the treatment group affected by construction (decision) of the Subway Line 3, which are the apartment complexes located within a certain distance from the nearest station of Line 3. Post_t refers to a dummy variable denoting the period after the policy decision was made. (Treat*Post)_jt indicates the interaction term between the two variables.

Eventually, willingness to pay for the benefits of improved accessibility to public transportation brought by construction (decision) of the Subway Line 3, which is the main topic of this study, is estimated by b ₃, a coefficient of the interaction term. In addition, variables (C_j ) for the age of the apartment in question and its squared term, area (in square metres), number of floors, number of units in the complex, constructor, management entity, and ward or county where the apartment in question locates are included in the regression equation to control for difference in price caused by characteristics of individual apartments. Furthermore, dummy variables for individual years (year_k) are included in the regression equation to control for trend in a business cycle. e_jt represents the usual error term.

While policy decision to construct Daegu Subway Line 3 was made around 2006, completion of the construction and commercial operation of the new subway line were realised in April 2015. Therefore, there is the time difference between decision of policy and operation of the new line. However, it is expected that binding policy decision capitalises future policy effect, i.e. benefits from improvement of accessibility to the public transport, into sales price of an apartment at the time when policy decision was made.⁴ In addition, willingness to pay for the benefits from improved accessibility to public transport could be estimated under various circumstances by applying distance from a station of the existing lines 1 and 2 as well as the new line in difference-in-differences framework.

Data and variables

Figure 1 summarises a timeline of policy-making processes with regard to construction of Daegu Subway Line 3.⁵ The line was included in the Urban Railways Master Plan established by the Ministry of Construction and Transportation (currently, the Ministry of Land, Infrastructure and Transport) in January 1991 and ‘2020 Daegu Urban Master Plan’ prepared in October 2005. The ‘Urban Railways Master Plan of Daegu Subway Line 3’ was approved by the Central Government in 2006. Then, the groundbreaking ceremony was held in June 2009 after the public hearing in May 2007. The line started operation from April 2015.⁶

Figure 1.

Timeline of policy decision-making process for construction of Daegu Subway Line 3.

This study uses apartment sales price provided by Kookmin bank (KB Bank hereafter) and housing statistics of Daegu city. The real estate sales data provided by the KB bank reflects actual sales price as the price has been prepared by sales price investigation companies (two sales price investigation companies were selected from each large apartment complex) to support mortgage loan since January 2004 in accordance with the actual sales price preparation method or sales (rent) case comparison method. This study uses the price as of December in a relevant year. However, the data used in the study provide average sales price for each complex rather than for individual apartments. Thus, we can run regressions only at the complex level but not at the individual apartment level. For housing statistics of Daegu, this study establishes the data base of basic complex information from ‘2013 Daegu Housing Statistical Year Book’ and the KB bank real estate sales price data. In addition, the distance from an individual apartment complex to the nearest subway station is calculated by GIS (see Figure 2). The distance represents Euclidean distance, not the length of a relevant road. Furthermore, apartments reconstructed during the analysis period from 2004 to 2016 are excluded. If an apartment complex has multiple apartment sizes, the price data for apartments whose size is the closest to 85m² of exclusive use area are used in order to control for heterogeneity in size.

Figure 2.

Classification by subway line, distribution of apartment complexes and distance from the nearest subway station.

Two important factors are observed in Figure 3. First, it is found that the sales price of apartments near Line 3 rapidly increased from 2004 to 2006 after the announcement of construction plan of Line 3, compared with that of apartment located in other areas. However, the gap in Chonsei price (one type of housing rent in which a renter makes a lump-sum deposit to the landlord when a lease is signed) remains almost the same during the period. This is because tenants did not have any benefits from the announcement as only policy decision of construction of Line 3 was made and Line 3 was not actually completed and operated during the period. On the other hand, the owner of an apartment will have increased benefits because of improved accessibility to the public transportation in the future when the construction of Line 3 is completed. Therefore, willingness to pay for future benefits appears through capitalisation into housing price at the time when the policy decision was made. Empirical results using single family housing values in Vancouver by Ferguson et al. (1988) show that the urban transit had an impact of the housing market even before the system operates.

Figure 3.

Average sales price and Chonsei price by year.

Second, the change of price gap between two groups only occurred during the period when policy decision was made from 2004 to 2006 and the changed price gap has been maintained afterward. If the significant change in sales price of apartments near Line 3 happened owing to the normal trend of price change, rather than from the influence of policy decision to construct Line 3, the price gap between two groups should keep growing after 2006. However, the fact that the sales price gap between two groups was maintained after 2006 indicates that the change of sales price gap happened between 2004 and 2006 was likely to be affected by policy decision about the construction of Line 3.

Table 1 presents the summary statistics of variables included in the empirical analysis. Such variables should include elements that may have effects on the price of apartments such as a size, as well as characteristics of apartments. However, only some characteristics are considered in this study because of limitations of data. For this study using panel data, estimation of policy effects is not affected if willingness to pay for the characteristics that are not included in the analysis does not change differently between the treatment group and the comparison group before and after implementation of the policy. In addition, as this study targets apartments whose area of exclusive use is most close to 85 m² in each apartment complex, other characteristics (e.g. the number of rooms and the number of bathrooms) are similar to each other. The sales price of the unit area continuously increased from KRW (the currency of Korea, Won) 1,494,300 in 2004 (1,035.1 KRW per US dollar in 2004) to KRW 1,669,400 in 2006 and decreased to KRW 1,532.600 in 2009. After that, the price reached KRW 2,112,600 in 2013 (1,055.4 KRW per US dollar in 2014). The age of apartments ranges from 0 to 35, and its mean value is 11.36. The mean value of the area is 78.68 m², and it ranges from 32.99 m² to 131.59 m². With regard to the number of floors, the mean value is 14 floors. The lowest apartment has 3 floors while the highest apartment has 30 floors. The mean number of households in a given apartment complex is 440. The smallest apartment complex includes 20 households while the largest apartment complex contains 2160 households. Considering the types of constructors, the private apartments and the private operation apartments account for 48.67% and 40.37%, respectively. For the classification by a management entity, it is shown that more than half of the total apartments are autonomously managed by residents with 54.49%.

Table 1.

Summary statistics.

Variables	Mean	Std. dev.	Min.	Max.
Sales price (KRW 10,000 per m²)
2004	149.43	47.41	23.13	319.51
2005	164.07	53.15	18.02	365.15
2006	166.94	50.78	34.38	395.58
2007	162.97	47.62	34.38	395.58
2008	159.05	46.21	11.77	395.58
2009	153.26	44.22	17.65	375.29
2010	155.25	44.00	14.13	375.29
2011	174.48	47.36	15.30	385.43
2012	186.79	47.50	35.47	405.72
2013	211.26	52.88	18.83	418.40
2014	247.38	61.77	23.57	588.30
2015	288.00	81.10	25.08	613.65
2016	279.85	78.27	24.32	684.65
Age of apartment as of 2004 (year)	11.36	6.96	0	35
Area for exclusive use (m²)	78.68	12.64	32.99	131.59
Number of floors	13.71	5.98	3	30
Number of households	440.00	346.45	20	2,160
Constructor (%)
LH	5.15		0	1
Urban Corp.	5.81		0	1
Private	48.67		0	1
Private Operation	40.37		0	1
Management entity (%)
Commissioned	24.09		0	1
Voluntary	21.43		0	1
Autonomous	54.49		0	1

Source: KB Bank Real Estate Sales Price (www.nland.kbstar.com), Each Year.

Empirical results

Distance from the nearest station of the new line

Table 2 shows the baseline regression results in which apartments that had direct influence from construction decision of the new subway line, i.e. Line 3 in Daegu are identified through difference-in-differences framework. First, the first three columns define apartment complexes within 200 m from the nearest station of the new line as the treatment group and other apartment complexes as the comparison group.⁷ The first column does not control for characteristics of individual apartment complexes while the second column does control for them.⁸ In this case, only 11 apartment complexes are included in the treatment group among total 602 apartment complexes. According to the results, construction decision of Subway Line 3 did not increase the price of apartments within 200 m from a station higher than the other apartments. This result shows that benefits from improved accessibility to public transit are not limited to the apartments located within 200 m from a station, which is similar with that of previous studies.⁹

Table 2.

Results where the treatment group is defined by the distance from the nearest station of the new subway line (i.e. Line 3).

	Distance from the nearest station of the new subway line is at most . . .
	200 m			500 m		1 km		2 km		3 km
	2004–2013		2004–2016	2004–2013	2004–2016	2004–2013	2004–2016	2004–2013	2004–2016	2004–2013	2004–2016
Apartment complex within the relevant distance	0.1538⁺	0.0668	0.0659	−0.0961^**	−0.1034^**	−0.0366^*	−0.0441^*	−0.0143	−0.0216	0.0399^**	0.0378^**
Apartment complex within the relevant distance	(1.78)	(1.19)	(1.16)	(−4.71)	(−5.06)	(−2.03)	(−2.47)	(−0.92)	(−1.41)	(2.77)	(2.65)
From 2006	0.0475^**	0.5019^**	0.1584^**	0.4928^**	0.1495^**	0.4908^**	0.1471^**	0.4946^**	0.1507^**	0.5060^**	0.1571^**
	(2.61)	(33.36)	(12.59)	(32.62)	(11.77)	(32.71)	(11.60)	(32.10)	(11.41)	(33.23)	(12.22)
Apartment complex within the relevant distance	0.0072	0.0168	0.0089	0.0624^**	0.0623^**	0.0472^*	0.0465^*	0.0212	0.0206	0.0044	0.0050
* From 2006	(0.07)	(0.26)	(0.14)	(2.73)	(2.77)	(2.47)	(2.46)	(1.32)	(1.29)	(0.30)	(0.34)
From 2015			0.6783^**		0.6733^**		0.6727^**		0.6729^**		0.6788^**
			(52.01)		(51.93)		(51.96)		(51.75)		(51.18)
Apartment complex within the relevant distance			−0.0126		0.0302⁺		0.0223		0.0135		0.0114
* From 2015			(−0.23)		(1.73)		(1.42)		(1.02)		(0.96)
Age		−0.0388^**	−0.0350^**	−0.0381^**	−0.0345^**	−0.0389^**	−0.0351^**	−0.0388^**	−0.0349^**	−0.0388^**	−0.0350^**
		(−23.37)	(−24.48)	(−22.79)	(−23.97)	(−23.25)	(−24.43)	(−23.56)	(−24.61)	(−23.44)	(−24.70)
Square of age		0.0007^**	0.0006^**	0.0007^**	0.0006^**	0.0007^**	0.0006^**	0.0007^**	0.0006^**	0.0007^**	0.0006^**
		(15.79)	(17.03)	(14.98)	(16.31)	(15.52)	(16.82)	(15.77)	(16.96)	(15.28)	(16.54)
Area		0.0024^**	0.0021^**	0.0024^**	0.0022^**	0.0024^**	0.0021^**	0.0024^**	0.0021^**	0.0022^**	0.0019^**
		(7.19)	(7.59)	(7.55)	(8.00)	(7.32)	(7.74)	(7.27)	(7.73)	(6.75)	(7.05)
Number of floors		0.0003	0.0006	0.0006	0.0010	0.0004	0.0008	0.0004	0.0008	−0.0002	0.0001
		(0.30)	(0.86)	(0.72)	(1.33)	(0.48)	(1.10)	(0.47)	(1.10)	(−0.24)	(0.21)
Number of households		0.0001^**	0.0001^**	0.0001^**	0.0001^**	0.0001^**	0.0001^**	0.0001^**	0.0001^**	0.0002^**	0.0002^**
		(18.44)	(21.75)	(18.45)	(21.76)	(18.43)	(21.73)	(18.44)	(21.68)	(18.81)	(22.25)
Constant	4.9529^**	4.9690^**	4.9641^**	4.9748^**	4.9714^**	4.9775^**	4.9738^**	4.9738^**	4.9702^**	4.9456^**	4.9415^**
	(369.91)	(130.92)	(151.89)	(130.70)	(151.32)	(131.07)	(151.49)	(128.87)	(148.95)	(128.04)	(148.30)
Dummy variables for
Year	10	10	13	10	13	10	13	10	13	10	13
Adj. R ²	0.1085	0.5355	0.6867	0.5372	0.6882	0.5351	0.6866	0.5346	0.6862	0.5367	0.6878
No. of apartment complexes within the relevant distance	11			83		148		226		281

Note: Numbers in parentheses are White-Huber’s robust t-values. ^**, ^*, and ⁺ represent the statistical significance at 1%, 5%, and 10% level, respectively. Dummy variables for constructor, management entity, and ward/county are included in the regression, but their results are suppressed because of space limitations. The number of apartment complexes included in the analysis is 602.

Source: KB Bank Real Estate Sales Price (www.nland.kbstar.com), Each Year.

However, the fourth column where the treatment group is defined as apartments within 500 m from the nearest station of the new line shows that construction decision of that line has significant impacts on the sales price. Especially, it is found that the increase in sales price of the treatment group was KRW 99,665 per m² higher than that of the comparison group after the policy decision.¹⁰ This result indicates that expectation of improved accessibility to the public transport formed by the construction decision of the new line is capitalised into the sales price of apartments. The fifth column extends the period of analysis to the time points after operation of Line 3.¹¹ The coefficient of the interaction term between the dummy variable indicating the treatment group and the dummy variable indicating the time points (i.e. December in 2015 and 2016) after the operation (i.e. 23 April 2015) is statistically significant but much smaller both in the magnitude and in the statistical significance. It implies that most of the capitalisation of improved transit accessibility occurs when the decision to construct a subway line is made rather than when a subway line starts operation.

The remaining columns define the treatment group as apartments located within 1 km, 2 km, and 3 km from the nearest station of the new line, respectively. When the treatment group includes apartments with longer distance from the nearest station of the new line, the difference in a price increase between the two groups becomes smaller and statistical significance of the difference also declines. The results of this study indicating that benefits of accessibility to the public transport are capitalised into a housing price in an area within a certain distance from a station confirm the findings of previous studies.¹² However, this study is different from previous studies to the extent that only the policy decision was made and the subway construction was not actually completed around 2006. In addition, it is possible that relative difference in the price increase range between two groups is caused by speculative elements as well as capitalisation.¹³

Table 3 presents analysis of Chonsei price conducted with the same method applied to Table 2, which deals with sales price of apartments. If Daegu Subway Line 3 was completed in its construction and operated around 2006, actual residents’ willingness to pay would increase by improved accessibility to the public transport, irrespective of ownership of apartments and, accordingly, the Chonsei price would also show a higher rise. However, as actual operation of the new line did not begin during 2004–2013, benefits of improved accessibility to the public transportation was not realised to a tenant of an apartment. Therefore, it could be expected that additional Chonsei price increase would not happen to the treatment group contrary to Table 2. Table 3 demonstrates that the theoretical expectation matches the reality, which is also illustrated by the facts proposed in Figure 2.

Table 3.

Results on Chonsei price where the treatment group is defined by the distance from the nearest station of the new subway line (i.e. Line 3).

	Distance from the nearest station of the new subway line is at most . . .
	200 m		500 m	1 km	2 km	3 km
Apartment complex within the relevant distance	0.1442^*	0.0371	−0.0542^**	0.0114	0.0256^*	0.0718^**
	(2.17)	(0.95)	(−3.32)	(0.83)	(2.15)	(6.50)
From 2006	0.0952^**	0.1344^**	0.1305^**	0.1336^**	0.1388^**	0.1470^**
	(5.56)	(13.10)	(12.58)	(12.82)	(13.00)	(13.44)
Apartment complex within the relevant distance	0.0295	0.0321	0.0304⁺	0.0070	−0.0091	−0.0230*
* From 2006	(0.41)	(0.74)	(1.65)	(0.47)	(−0.72)	(−1.99)
Age		−0.0239^**	−0.0234^**	−0.0243^**	−0.0242^**	−0.0241^**
		(−18.10)	(−17.62)	(−18.31)	(−18.33)	(−18.43)
Square of age		0.0002^**	0.0002^**	0.0002^**	0.0002^**	0.0002^**
		(4.52)	(4.06)	(4.65)	(4.61)	(4.03)
Area		0.0016^**	0.0016^**	0.0016^**	0.0015^**	0.0014^**
		(5.36)	(5.66)	(5.45)	(5.26)	(4.77)
Number of floors		0.0083^**	0.0086^**	0.0083^**	0.0083^**	0.0077^**
		(12.28)	(12.85)	(12.42)	(12.38)	(11.46)
Number of households		0.0001^**	0.0001^**	0.0001^**	0.0001^**	0.0001^**
		(17.68)	(17.58)	(17.71)	(18.04)	(18.44)
Constant	4.5380^**	4.5777^**	4.5798^**	4.5795^**	4.5724^**	4.5395^**
	(362.17)	(136.09)	(136.70)	(135.70)	(134.91)	(133.11)
Dummy variables for
Year	10	10	10	10	10	10
Constructor/Management Entity		4/3	4/3	4/3	4/3	4/3
Ward/County		8	8	8	8	8
Adj. R ²	0.1739	0.7206	0.7208	0.7203	0.7203	0.7232
Number of apartment complexes within the relevant distance	11		83	148	226	281

Note: Numbers in parentheses are White-Huber’s robust t-values. ^**, ^*, and ⁺ represent the statistical significance at 1%, 5%, and 10% level, respectively. The number of apartment complexes included in the analysis is 602.

Source: KB Bank Real Estate Sales Price (www.nland.kbstar.com), Each Year.

Spurious results of the empirical analysis using the difference-in-differences framework are generated as it could not distinguish difference in trend, which is completely not relevant with the policy in question, from policy effects. To identify the possibility of spurious results, it is necessary to apply the same analysis framework to the time after the policy decision is made. Table 4 presents results of a kind of placebo test. If the results in Table 2 are caused by difference in trend between the two groups, rather than effects of the policy decision, analysis for years 2006–2013 shall show similar results with those of Table 2. However, in Columns (1)–(3) that set the analyses period as the period from 2006 to 2013 and the policy decision time as a certain time during the period, it is found that the apartments located within 1 km from the nearest station of the new line do not show the significantly larger increase in sales price as a whole. This finding can be interpreted that the results shown in Table 2 are likely to reflect policy effects, rather than the difference in trend between the two groups.

Table 4.

Results for the period after the policy decision made (i.e. 2006-2013) as a placebo test.

	Arbitrary year
	2008	2010	2012
Apartment complex within 1 km from the nearest station of the new subway line	0.0024	0.0062	0.0159
	(0.14)	(0.47)	(1.41)
From an arbitrary year	−0.0511^**	−0.0769^**	0.2400^**
	(−4.35)	(−6.11)	(20.64)
Apartment complex within 1 km from the nearest station of the new subway line	0.0230	0.0268	0.0152
* From an arbitrary year	(1.24)	(1.63)	(0.81)
Constant	4488.07^**	4487.59^**	4487.79^**
	(19.14)	(19.14)	(19.13)
Dummy variables for
Year	8	8	8
Constructor/Management Entity	4/3	4/3	4/3
Ward/County	8	8	8
Characteristics	√	√	√
Adj. R ²	0.5466	0.5467	0.5464

Source: KB Bank Real Estate Sales Price (www.nland.kbstar.com), Each Year.

Distance from existing subway lines

Division of the treatment group

Table 2 shows the difference in willingness to pay for housing with regard to accessibility to the public transport between the two groups defined by the distance from the nearest station of the new line, i.e. Line 3. However, as the new line is designed to allow transfer to existing lines, i.e. Line 1 or 2, geographical proximity to the existing lines is likely to affect willingness to pay and the results shown in Table 2.

To incorporate the influence of proximity to existing lines, apartments locating within 1 km from the nearest station of the new line (i.e. the treatment group in Table 2) are classified into three different subgroups by distance from the nearest station of existing lines; within 2 km, between 2 and 5 km, and over 5 km.¹⁴ The results are presented in Table 5.¹⁵ The number of apartments located within 2 km is 56. This group does not show statistically significant difference in the price increase range around 2006, compared with that of the comparison group, which includes apartments located over 1 km from the nearest station of the new line. The number of apartments located between 2 and 5 km is 40 and the price increase range of this group also does not significantly differ from that of the comparison group. On the other hand, it is found that the price of 52 apartments located over 5 km was KRW 11,600 per m² (or KRW 910,200 for an apartment with average area, i.e. 78.68 m²) higher than the comparison group around 2006, which is statistically significant. This finding indicates that benefits of improved accessibility to the public transportation due to the new subway line construction are limited to households that have limited benefits from existing lines.

Table 5.

Results where the treatment group is divided by the distance from the nearest station of the existing subway lines (i.e. Line 1 or 2).

	The distance from the nearest station of the new subway line is at most 1 km and the distance from the nearest station of the existing lines is . . .
	Less than 2 km	Between 2 km and 5 km	More than 5 km
Apartment complex within the relevant distance	0.1796^**	−0.1436^**	−0.1832^**
	(8.51)	(−5.33)	(−7.33)
From 2006	0.4442^**	0.1378^**	0.1482^**
	(31.14)	(11.14)	(11.73)
Apartment complex within the relevant distance	−0.0077	−0.0036	0.1457^**
* From 2006	(−0.33)	(−0.13)	(5.23)
Constant	4.8430^**	4.8595^**	5.0418^**
	(136.18)	(128.18)	(120.49)
Dummy variables for
Year	10	10	10
Constructor/Management Entity	4/3	4/3	4/3
Ward/County	8	8	8
Characteristics	√	√	√
Adj. R ²	0.5873	0.5749	0.5585
Number of apartment complexes within the relevant distance	56	40	52

Source: KB Bank Real Estate Sales Price (www.nland.kbstar.com), Each Year.

Division of the comparison group

Table 6 defines the treatment group as all apartments located within 1 km from the nearest station of the new line. However, it classifies the comparison group into two subgroups by distance from the nearest station of existing lines, while Table 5 divides the treatment group. It is expected that the difference in sales price increase between the treatment group and the comparison group would be bigger when apartments with longer distance from the nearest station of existing lines are defined as the comparison group.

Table 6.

Results where the comparison group is divided by the distance from the nearest station of the existing subway lines (i.e., Line 1 or 2).

	The distance from the nearest station of the new subway line is more than 1 km and the distance from the nearest station of the existing subway lines is . . .
	At most 800 m	More than 800 m
Apartment complex within 1 km from the nearest station of the new subway line	−0.0348⁺	−0.0319
	(−1.71)	(−1.62)
From 2006	0.1587^**	0.1466^**
	(8.53)	(9.53)
Apartment complex within 1 km from the nearest station of the new subway line	0.0335	0.0605^**
* From 2006	(1.58)	(3.12)
Constant	5.0581^**	5.1483^**
	(104.75)	(96.49)
Dummy variables for
Year	10	10
Constructor/Management Entity	4/3	4/3
Ward/County	8	8
Characteristics	√	√
Adj. R ²	0.5423	0.5175
Number of apartment complexes within the relevant distance	225	229

Source: KB Bank Real Estate Sales Price (www.nland.kbstar.com), Each Year.

According to results in Table 6, apartments that locate beyond 1 km from the nearest station of the new line and within 800 m from the nearest station of existing lines do not show a significantly smaller increase in sales price than those in the treatment group.¹⁶ It implies that such apartments can enjoy benefits of improved accessibility to the public transport because they locate within 800 m from the nearest station of existing lines and it is transferrable between the new line and existing lines although they locate in a certain distance, i.e. beyond 1 km in this case, from a station of the new line. On the other hand, as apartments locate over 800 m from the nearest station of existing lines are geographically limited to gain such benefits, the degree to which their sales price increases is smaller than that of apartments in the treatment group.

Table A2 in the Appendix considers the effect of proximity to existing lines on both the treatment group and the comparison group by dividing the treatment group as in Table 5 and the comparison group as in Table 6. The results for the two extreme cases are reported. As expected from the results in Tables 5 and 6, the largest impact of the policy decision to construct the new line appears when the treatment group is refined by apartments beyond 5 km from the nearest station of existing lines and the comparison group by those beyond 800 m.

Conclusions

These days, opportunity cost of commute and transit time is gradually increasing as the income level rises. As a result, importance of public transportation in a congested large city is emphasised. Analyses of spatial effects of new subway line construction are important resources in several policy decision-making such as estimating development gains by a subway construction and validating feasibility of development of station influence areas.

This study empirically analyses willingness to pay for benefits of improved accessibility to public transportation brought by construction (decision) of Daegu Subway Line 3 using the difference-in-differences framework. According to the results, apartments located within 500 m from the nearest station of the new subway line experienced a relatively larger increase in sales price over two time points; one when the decision to construct the new line was made and the other when it started its operation. However, the degree to which those apartments were relatively more valued was twice as great at the time of the decision-making as at the time of its operation. Also, it is found that the degree of capitalisation of improved accessibility to the public transport into housing prices decreases as the distance from the nearest station of the new line increases. Additional analyses on Chonsei price (i.e. upfront lump-sum deposit) and timing of policy decision made demonstrate robustness of the baseline results of this study. Moreover, the results of the analyses considering accessibility to existing lines, which has been neglected in most of the previous Korean studies, imply that proximity to existing lines has to be taken into account in evaluating the effect of the new line.

This study makes some contributions to the literature on this topic by looking at long enough periods to include the two time points, one when the policy decision to construct a new subway line was made and the other when the new line started its operation. Moreover, proximity to existing lines as well as the new line is considered in evaluating the impact of the new line on housing price.

The evaluation of the feasibility of constructing a transport infrastructure including a subway line has to take into account both the direct benefit (e.g. the number of users) and the indirect benefit (e.g. the increase in sales price of housing units near the infrastructure). The results of this paper suggest that the estimation of the increase in housing sales price has to consider the timing issue and the possibility to transfer to existing lines.

Nevertheless, partly because of the limit on data, the possible bias from residential sorting or move of the city centre could not be taken into account. Maybe, by incorporating socio-economic characteristics of residents in a given location, these two issues could be tackled in the next study.

Footnotes

Appendix

Table A2.

Results where the samples are divided by the distance from the nearest station of the existing subway lines (i.e. Line 1 or 2).

	The distances from the nearest station of the existing subway line are . . .
	Within 2 km for the treatment group; within 800 m for the comparison group	Beyond 5 km for the treatment group; beyond 800 m for the comparison group
Apartment complex within 1 km from the nearest station of the new subway line and in a given distance from the nearest station of the existing lines	0.1868^**	−0.2693^**
	(7.96)	(−8.81)
From 2006	0.1515^**	0.5508^**
	(8.23)	(28.05)
Apartment complex within 1 km from the nearest station of the new subway line and in a given distance from the nearest station of the existing lines	−0.0202	0.1590^**
* From 2006	(−0.81)	(6.05)
Constant	4.8290^**	5.3673^**
	(103.26)	(79.96)
Dummy variables for
Year	10	10
Constructor/Management Entity	4/3	4/3
Ward/County	8	8
Characteristics	√	√
Adj. R ²	0.6223	0.5794
Number of apartment complexes
Treatment group	56	52
Comparison group	225	229

Note: Numbers in parentheses are White-Huber’s robust t-values. ^**, ^*, and ⁺ represent the statistical significance at 1%, 5%, and 10% level, respectively.

Source: KB Bank Real Estate Sales Price (www.nland.kbstar.com), Each Year.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Agostini

Palmucci

(2008) The anticipated capitalization effect of a new metro line on housing prices. Fiscal Studies 29(2): 233–256.

Andersson

Shyr

(2010) Does high-speed rail accessibility influence residential property prices? Hedonic estimates from southern Taiwan. Journal of Transport Geography 18(1): 166–174.

Armstrong

Rodriguez

(2006) An evaluation of the accessibility benefits of commuter rail in eastern Massachusetts using spatial hedonic price functions. Transportation 33(1): 21–43.

Bae

Jun

Park

(2003) The impact of Seoul’s subway line 5 on residential property values. Transport Policy 10: 85–94.

Bowes

Ihlanfeldt

(2001) Identifying the impacts of rail transit station on property values. Journal of Urban Economics 50: 1–25.

Choi

Yoon

(2004) Analysis of spatial influence of subway construction on price of apartments – Focusing on Seoul subway line 7. Seoul Urban Study 5(4): 1–12 [in Korean].

Choi

Seong

(2011) A study of effects of construction of subway line 9 on price of surrounding apartments – Focusing on effects by stage of construction project. Land Planning 46(3): 169–177 [in Korean].

Choi

Kim

Park

(2008) Development of land value estimation model of Seoul subway station influence areas using CHAID analysis. Journal of the Korean Society for Railroad 11(5): 504–512 [in Korean].

Debrezion

Pels

Rietveld

(2011) The impact of rail transport on real estate prices: An empirical analysis of the Dutch housing market. Urban Studies 48(5): 997–1015.

10.

Diao

(2015) Selectivity, spatial autocorrelation and the valuation of transit accessibility. Urban Studies 52(1): 159–177.

11.

Diao

Qin

Sing

(2016) Negative externalities of rail noise and housing values: Evidence from the cessation of railway operations in Singapore. Real Estate Economics 44(4): 878–917.

12.

Duncan

(2011) The impact of transit-oriented development on housing prices in San Diego, CA. Urban Studies 48(1): 101–127.

13.

Ferguson

Goldberg

Mark

(1988) The pre-service impact of the Vancouver Advanced Light Rail Transit System on Sing family property values. In: Clapp

Messner

(eds) Real Estate Market Analysis: Methods and Applications. New York: Greenwood Publishing Group, pp. 78–110.

14.

Henneberry

(1998) Transport investment and house prices. Journal of Property Valuation & Investment 16(2): 144–158.

15.

Khandker

Koolwal

Samad

(2009) Handbook on Impact Evaluation: Quantitative Methods and Practices. Washington, DC: World Bank.

16.

Kim

Lee

(2010) A study on the integrated law-system for revitalization on the station’s area. Urban Design Institute of Korea 2010–10: 515–524 [in Korean].

17.

Korea Transport Institute (2010) 2010 Survey of National Passengers OD. Available at: www.ktdb.go.kr/www/contents.do?key=60 (accessed 2 November 2011).

18.

Korea Transport Institute (2014) Estimation of traffic congestion cost in 2015. Press Release, 23 April.

19.

Kwon

Kim

(2001) Factors influencing change in land value of station influence areas according to opening of subway line. Presented at Autumn Conference of Korea Planner’s Association, 19 October 2001, Seoul [in Korean].

20.

Landis

Cervero

Guhathukurta

et al . (1995) Rail Transit Investments, Real Estate Values, and Land Use Change: A Comparative Analysis of Five California Rail Transit Systems. Research Report No. 48. Institute of Urban and Regional Studies, University of California, Berkeley, CA.

21.

Lee

Kim

(2014) Negative effects of subway stations on price of surrounding apartments – Focusing on different effects of structure and function of station building. Housing Study 22(2): 53–75 [in Korean].

22.

McDonald

Osuji

(1995) The effect of anticipated transportation improvement on residential land values. Regional Science and Urban Economics 25: 261–278.

23.

Rosen

(1974) Hedonic prices and implicit markets: Product differentiation in pure competition. Journal of Political Economy 82(1): 34–55.

24.

Strand

Vagnes

(2001) The relationship between property values and railroad proximity: A study based on hedonic prices and real estate brokers’ appraisals. Transportation 28(2): 137–156.

25.

Weinstein

Clower

(2002) An Assessment of the DART Light Rail Transit on Taxable Property Valuations and Transit-Oriented Development. Center for Economic Development and Research, University of North Texas, Denton.

26.

Yiu

Wong

(2005) The effects of expected transport improvements on housing prices. Urban Studies 42(1): 113–125.

27.

Yoo

Gang

(2012) Analysis of effects of spatial structure of station influence areas on land value. Journal of Korea Society of Civil Engineers 32(1): 61–69 [in Korean].