- Email: [email protected]
Hedonic pricing model of assessed and market land values: A case study in Bangkok metropolitan area, Thailand
Accepted Manuscript Hedonic pricing model of assessed and market land values: A case study in Bangkok metropolitan area, Thailand Sathita Malaitham, Atsushi Fukuda, Varameth Vichiensan, Vasinee Wasuntarasook PII: DOI: Reference:
S2213-624X(17)30055-X https://doi.org/10.1016/j.cstp.2018.09.008 CSTP 291
To appear in:
Case Studies on Transport Policy
Received Date: Revised Date: Accepted Date:
17 February 2017 12 April 2018 17 September 2018
Please cite this article as: S. Malaitham, A. Fukuda, V. Vichiensan, V. Wasuntarasook, Hedonic pricing model of assessed and market land values: A case study in Bangkok metropolitan area, Thailand, Case Studies on Transport Policy (2018), doi: https://doi.org/10.1016/j.cstp.2018.09.008
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Title page 1.
Title: Hedonic pricing model of assessed and market land values: A case study in Bangkok Metropolitan Area, Thailand
2.
Authors: Sathita Malaitham, Corresponding Author Nihon University Department of Transportation System Engineering 7-24-1 Narashinodai, Funabashi, Chiba, 274-8501, Japan Tel: (+81)-47-469-5355; Fax: (+81)-47-469-5355 Email: [email protected] Atsushi Fukuda Nihon University Department of Transportation System Engineering 7-24-1 Narashinodai, Funabashi, Chiba, 274-8501, Japan Tel: (+81)-47-469-5355; Fax: (+81)-47-469-5355 Email: [email protected] Varameth Vichiensan Kasetsart University Department of Civil Engineering 50 Phahonyothin Rd, Ladyao, Jatujak, Bangkok, 10900, Thailand Tel: (+66)-2-942-8555; Fax: (+66)-2-579-4575 Email: [email protected] Vasinee Wasuntarasook Nagoya University Department of Environmental Engineering C1-2(651), Furo, Chikusa, Nagoya, 464-8603, Japan Tel: (+81)-52-789-2773; Fax: (+81)-52-789-2773 Email: [email protected]
Abstract Bangkok Metropolitan Area, the capital of Thailand, is known as one of the world’s most traffic-congested cities. Lots of transport-related projects to alleviate traffic congestion especially rapid transit system always require an amount of land. Therefore, several privately owned lands have to be acquired by the government agencies due to the lack of available spaces in an urban area. However, the assessment of compensation for a compulsory land acquisition is determined on the basis of the assessor’s database that often values each of properties lower than expected. In this paper, the assessed prices data and the offering prices for sale data of Bangkok are analyzed with the use of regression framework through the hedonic pricing model. The spatial non-stationarity to examine the variations of the implicit effects to property value is also included bases on the geographically weighted regression (GWR) technique. The results show that the determinants of property value are myriad and varied over space, i.e. spatial non-stationarity exists in the study area. The obtained results indicate that the difference between two data is extremely large. These results provide the basic information to compute the fair compensation for the landowners and allow the government agencies to tax the direct beneficiaries of their investments in advance so as to finance infrastructure projects. © 2017 The Authors. Published by Elsevier B.V. Peer-review under responsibility of WORLD CONFERENCE ON TRANSPORT RESEARCH SOCIETY.
2214-241X © 2017 The Authors. Published by Elsevier B.V. Peer-review under responsibility of WORLD CONFERENCE ON TRANSPORT RESEARCH SOCIETY.
2
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
Keywords: Assessed land value, Actual land value, Hedonic study, Spatial effect, Developing countries
1. Introduction Over the past several decades there has been a dramatic increase in the discussion of the capitalization effects of investments in transportation infrastructure (e.g., transits) in many ways. It is well acknowledged that changes in accessibility to areas of influence are considered as one of benefits of such investments (1). The accessibility, as defined by Hansen (2) and Ben-Akiva and Lerman (3), provides the potential of opportunities to participate in activities in different locations. For example, people who live nearby transit stations can reduce travel time and costs to other destinations. On the other hand, business activities such as shops and jobs can also be reached from any destinations. Therefore, households and businesses are willing to pay for higher accessibility locations. Such accessibility-improved areas have unavoidably become more attractive to investors for residential and commercial land development that pushing up their values, further owners of properties rightly obtain these benefits(4). Bangkok Metropolitan Area, the capital city of Thailand, is known as one of the world’s most traffic-congested cities due to the growing number of registered vehicles. Encouraging people to use public transport modes instead of their private vehicles in order to decrease the number of vehicles on the roads have been being argued as the solution to address this issue. A rail-based transit system, which is the affordable and comfortable transport system that run efficiently to meet the needs of the people, has been introduced and proposed by the policy-makers and planners. High capacity, performance, and full grade-separated system compared to other transport modes including buses and express boats causes an increase in quantity demand for land in locations with good accessibility to this system. Vichiensan et al. (5) indicated that an opening of the first transit system of Bangkok (i.e., BTS Skytrain) significantly influences assessed land values of station or corridor. Furthermore, they claimed that rail transit system has played significant role in accelerating urban development in Bangkok. From a policy perspective, it is important to understand how benefit of transit investment is capitalized into nearby land under market conditions. In this paper, the purpose is to examine the effects of transits on two types of values (i.e., assessed and market land values) by applying the hedonic pricing model. The relationship between assessed and market values may be quite complex, but there is no literature that compares whether the effects are the same for these two types or not in Bangkok studies. In fact, the assessed values assign to each parcel by government agencies for the purposes of taxation. Furthermore, the compensation for the compulsory land acquisition before the construction of transit infrastructure is evaluated based on land assessment database that is somewhat higher or lower than the market values. Hedonic model will help to explore the relative contribution of each of these attributes on the assessed and market land values. As known, each property has a unique bundle of attributes such as home features, its accessibility to transport, work, etc. (6, 7, 8, 9). As transit investment influences each area differently, the spatial function, i.e., spatial nonstationarity is accommodated with the hedonic pricing model to capture land value appreciation across the study area. The obtained results will indicate the difference between assessed and market land values in terms of various functional characteristics. This information can help to estimate how much beneficiaries should return their benefit to the public. 2. Literature reviews Assessed value corresponded by governmental agencies is used to calculate the value of the property for tax purposes. Conversely, market value refers to price of asset on the open market. Hedonic price model, which was firstly introduced by Rosen (10), has been applied to examine the relationships between property value and its characteristics for many years. A market value of properties was used to examine as a dependent variable in many hedonic studies (11, 12, 13, 14, 15, 16, 17, 18), whereas there are a small number of studies using assessed land value (4, 19, 20, 21). Either assessed or Market values were used to capture the determinants of single-family homes (22), industrial land (23), and farmland (24). Although assessed values might be imperfectly correlated with market values (25), there might have some errors in market values as a result of the distorted market (15). Kowalski and Colwell (23) investigated the factors that influence the assessed and market values of parcels zoned for industrial uses in western Wayne County, Michigan (i.e., west of Detroit). The implicit prices exhibited by market for industrial parcels in the west suburbs of Detroit are somewhat different than those which are organized in the assessment process in the same area, so industrial parcels are systematically or structurally under-and over- assessed. Ma and Swinton (24) captured the determinants of appraised value and sale price for farmland uses in southwestern Michigan. Results suggest that appraised values are a poor substitute for sale prices if the research goal is to understand dynamically evolving determinants of land value in exurbanizing regions, especially the value of natural amenities. Over the past decades, it has become increasingly clear that the presence of rail transit system can increase property values due to the accessibility improvement. Bajic (11) performed one of the earliest who attempted to identify the effects of a subway line in Toronto on the values of housing. Gatzlaff and Smith (7) examined the impact of the development of the Miami Metrorail system on property values proximate to its station locations. Some studies performed sought to distinguish between the accessibility benefits of rail transit and other transportation systems (6, 26). Forrest et al. (27) examined the relationship between the availability of commuter rail services and the pattern of house prices in an urban area of Manchester, England. Armstrong and Rodriguez (28) found that properties located in municipalities with commuter rail stations exhibit values that are between 9.6 percent and 10.1 percent higher than properties in municipalities without a commuter rail station in Eastern Massachusetts. Premium of light rail
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
3
transit accessibility is approximately $2.31 (using geographical straight-line distance) and $0.99 (using network distance) for every foot closer to a station in Buffalo, New York (4), while the impact of rail transit has increased around 7.814 billion yuan on the surrounding residential values of Chengdu Metro Line 1 (29). Although, a great deal of ascending property value produced by transits and their facilities has been captured in developed countries for many years, few studies have investigated in developing countries. One reason for little research is that transit system has been introduced to their countries during the past 20 years. Lots of studies have been carried out on the significance of accessibility to employment that mostly refers to the distance to CBD (16, 27, 30). Land values increase 16 percent for every mile closer to the CBD (30). Likewise, longer distances to the city’s sub-centers are associated with lower prices, for every kilometer increases in the distance to XuJiaHui, the price would drop by 4.0 percent (31). On the other hand, non-work activities such as shopping center, bank branch offices, green space or park, hospital, airport, etc. were found to be the factors that impact on the property values. For example, the presence of shopping centers and sports facilities, are important factors in determining prices of properties (9). The shopping mall, one of the attractiveness of the location that found the highly significant and uplift the office rents (32). Among variables describing neighborhood quality, median income level was statistically significant and it had the positive impact on property values in Queen (21). Cervero and Duncan (33) revealed that every $10,000 increase in median household income associated with a $1.67 per square foot increase in multi-family parcels in Santa Clara County. While every $100 increase in the median annual household income associated with a $36 increase in property value in Buffalo, suggesting that houses were more likely to sell for a higher price in affluent neighborhoods (4). Furthermore, population and employment density were both significant; prices negatively associated with population density but positively associated with employment density (34). The school quality, flooding, level of security, percent of population black, racial composition, crime rate, the noise level, and number of fireplaces in the neighborhood were also used to examine the effects on the property values of their methods. Previous research has provided mixed evidence including large positive, small positive as well as negative effects. 3. Methodology 3.1. Model specifications Hedonic pricing model is used to model the relationship between one (or more) dependent or response variables and independent or predictor variables. The equation of the hedonic pricing model can be written in the basis of the classical linear regression specification as follow:
Yi 0 1 X i1 ... k X ik i
(1)
An equation(1) can be written more compactly for n observations as
Y = Xβ ε
(2)
where Y X β
= = = =
a vector (n x 1) of observations corresponding to a dependent variable, a matrix (n x k) of observations of k independent variables, a vector (k x 1) of regression parameters, and a vector (n x 1) of random errors, assumed to be normally distributed
To solve the equation (2), the ordinary least square (OLS) is a simply fitting mechanism based on minimizing the sum of squared residuals or residual sum of squares (RSS). The vectors of regression parameters that minimizes are given by the usual expression:
β = (X'X)-1 X'Y
(3)
Also, the equation (1) transformed into natural logarithms as expressed in equation (4) is common in hedonic studies (35, 36).
ln(Yi ) 0 1 ln( X i1 ) ... k ln( X ik )
(4)
Spatial regression model applied to tease out the spatial effect, i.e., non-stationarity in this study is geographically weighted regression (GWR) model. The model takes on the following form (37):
Yi 0 (ui , vi ) 1 (ui , vi ) X1 ... k (ui , vi ) X k i
(5)
4
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
where (ui,vi) denotes the latitude and longitude coordinates of observation i, and k(ui,vi) represent the k regression parameters at estimation location of observation i. The equation (5) can be written more compactly as:
Y(i) = Xβ(i) + ε(i)
(6)
The estimation of regression parameters specific t every location is accomplished by performing local regressions, each using a sub-sample data in the vicinity of estimation’s location. Thus, a total of n observations can perform n local regression models. The estimator of GWR model is written as follows:
β(i) = (X'W(i)X)-1 X'W(i)Y
(7) where W(i) is the geographical weight function between the estimation location of observation i and the observation j,…,n located within the bandwidth of observation i (hi ). Within the bandwidth, the observations spatially closer to estimation location of the observation i have more influence in the local regression estimation compared with the data located farther away. In this study, the weight function is defined as w(d)=exp(-d2/h), where h is bandwidth and d is the distance between the focus point, i, and other points. 3.2. Study area The study area compasses the Bangkok Metropolitan Area. The region, shown in Fig. 1, covers an area of 1,568.7 km2 with 50 districts. There is the Chao Phraya River meandering through the city in the southward direction. This river divides Bangkok into two sides: the eastern bank and the western bank of the Chao Phraya. On the west side, it is known as the Bangkok’s old town where the original CBD situated. As the city expanded, parts of the areas along Silom Rd. and Sukhumvit Rd. located on the east side of the Chao Phraya River has been the CBD of Bangkok. There are lots of premium office buildings, five-star hotels, luxury condominiums, entertainment buildings, shopping malls and so on throughout the areas. By contrast, the Bangkok’s old town has become the top tourist destination and led them be the heart of Bangkok’s tourist and commercial center where many unique retail shops are concentrated.
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
5
Fig. 1. Map of Bangkok Metropolitan Area, Existing Rail Transit Network and Sample Data
Bangkok has a registered population of 5.69 million while a total number of registered vehicles have hit 8.92 million in 2015 which was two times higher than the total number of registered vehicles in 2004 (National Statistical Office). Undoubtedly, the growth of motorization is remarked as one of the critical causes that contributes to traffic congestion. The rail transit system to alleviate the traffic congestion was firstly introduced at the end of the 1970s. Although it has been known more than 30 years, the first rail transit of Bangkok, namely BTS Skytrain, has started its service along the heavily congested arterial roads such as Phaholyothin Rd., Sukhumvit Rd., Silom Rd., Sathorn Rd., etc. since December 1999. Five year later, the second rail transit line or so-called MRT Blue Line was started its operation from Bang Sue station to Hua Lumphong Station by running in an underground level of other crowded roads namely Ratchadapisek Rd., Asok Rd., Rama 4 Rd., etc. Then, Airport Rail Link was constructed to serve between the city area and the new international airport of Thailand. In March 2016, the new route, i.e., MRT Purple Line opened the service from Bang Sue station to Bang Yai station. Also, the extensions of BTS Skytrain and MRT Blue Line are now under construction whereas other new lines are now planning to start their construction. 3.3. Descriptive statistics
140000
140000
120000
120000
Price (baht/sq.m)
Price (baht/sq.m)
The data used for modeling involve two types of land values data, i.e., assessed and market land values. Both are assigned as the dependent variable (Y) in the equation (2). The assessment data obtaining from the reports of Treasury Department, Thailand are evaluated based on type of land use categories, proximity from main roads, size of lands, etc *. This report continuously uses to levy the property’s taxes over the period of 4 years without major change or update the values, while land values in the market are annually changing. The period of land value report employs to capture taxes during the year 2012 to 2015 were used in this study ; however, it was generally evaluated before published around 2 years, i.e., this assessed value had started evaluated since 2010 and published in the year 2012. The market values refer to the prices at which land are offered for sale on the market. The offering prices data to estimate were collected from the for-sale advertisements by landowners on the internet from August 2014 to February 2015. Advertisements normally include location, price, and size of land lot. Furthermore, convenient access such as to the nearest shopping malls, expressway ramps and also rail transit stations are almost mentioned in the advertisements. In the hedonic studies, the value of property located within the immediate area of rail transit stations is higher compared with the property from stations farther away. In case of Bangkok, the statistics of the assessed prices data and the offering prices data by the distance intervals of each operating line are presented in Fig. 2. First of all, the differences between the prices per square meter of assessed and market data are noticeably large. Land parcels being located closer to the stations tend to add more valuable than those being located farther away as expected. Surprisingly, the averages of market value of land parcel being located within 1,500 meters of BTS Skytrain, MRT Blue Line and Airport Rail Link were approximately three times higher than the assessor’s database.
100000 80000 60000 40000 20000
100000 80000 60000 40000 20000
0
0 0-500
500-1000 1000-1500
>1500
0-500
Distance to station (m.) BTS Skytrain
MRT Blue Line
Airport Rail Link
(a)
500-1000 1000-1500
>1500
Distance to station (m.) BTS Skytrain
MRT Blue Line
Airport Rail Link
(b)
Fig. 2. Property values by station’s distance intervals: (a) Assessed value and (b) Market value
To include the effects of mentioned attributes on the advertisements and the reports as the independent variables (X), GIS was used to measure the straight-line distances such as the distance from the land parcels to the nearest rail transit stations, to school, to the central business district, and so on that mentioned. Also, GIS was used to measure how far the parcels are located from the
*
http://www.treasury.go.th/main.php?filename=index
6
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
main roads and expressway entrances/exit. As indicated in the previous study, not only are the proxy attributes influenced on the property values, but also zonal attributes such as employment density, household density, median income, etc. Furthermore, the land use proportions in each zone were included in the models. All the zonal variables in this study were collected from National Statistical Office (NSO) and Bangkok Metropolitan Administration (BMA). The descriptive statistics (i.e. mean values) of the dependent variables and their attributes are summarized in Table 1.
Table 1. Descriptive statistics Assessed value (1,871 data)
Market value (823 data)
Property value (baht/m2)
13,736
29,202
Size of land Parcel (m2)
3,704
4,073
Independent variable Property attributes DIST_MAINRD DIST_STA
Distance to the nearest main road (km) Distance to the nearest station (km)
0.49 4.29
0.58 4.58
TIME_CBD DIST_SHOP
Time to CBD (min) Distance to the nearest shopping mall (km)
61.05 3.09
71.39 2.82
DIST_EXP JOB_ACCESS
Distance to expressway (km) Job accessibility (jobs)
5.31 2,298,110
4.97 2,250,783
Zonal attributes HH_DENS EMP_DENS
Household density (households/sq.km) Employment density (jobs/sq.km)
3,675 9,160
3,116 5,599
MED_INC %L_RESIDENCE %L_COMMERC
Median income (baht) Percentage of residence area Percentage of commercial area
30,787 33.21 4.51
29,388 34.37 4.61
%L_INDUST %L_EDUCA
Percentage of industrial area Percentage of education area
11.36 2.19
11.44 1.92
%L_VACANT
Percentage of vacant area
20.31
22.31
Variables
Description
Dependent variable P_VALUE LOT_SIZE
4. Results 4.1. OLS models The results of OLS models are shown in Table 2 which contains the estimate of parameters, the t-statistic, and p-value. The goodness-of-fit is evaluated by the coefficient of determination (R2), residual sum of squares (RSS) and Akaike information criterion (AIC). First of all, the elasticities with respect to the proxy variables including distance to main road, the nearest station, time to the CBD, distance to the nearest shopping mall and the nearest expressway ramps have the negative sign as expected. For example, the estimation results of both data reveal that the elasticities with respect to the distance to the main road are approximately -0.777 (using the assessment data) and -0.060 (using the offering prices data) while the elasticities of being used the assessed prices data and the offering prices data with respect the distance to the nearest station are -0.044 and -0.263, respectively. These results suggest a property value discount of 0.777% and 0.060% for a percentage increase in the distance to main road. Similarly, a 1-percent decrease in rail transit station proximity was estimated to increase the assessed value and offering price for sale by 0.044% and 0.263%, respectively, meaning that a-100,000 baht/m2 property located 100 m from the station will be priced at 44 baht/m2 (the assessed prices data) and 263 baht/m2 (the offering prices for sale data) more than an identical property that located 101 m from the station and around 440 baht/m2 (the assessed prices data) and 2,630 baht/m2 (the offering prices for sale data) more than another identical property that is located 110 m from the station. Obviously, there are somewhat different between the estimated parameters
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
7
from the assessed prices data and the offering prices for sale. For instance, holding everything constant, the identical property located near the shopping mall can sell more expensive than the assessor’s database about 45.39 %. And, that property will be worth about 5.5 times higher than the assessment if it located near the expressway ramps. Besides, the employment accessibility variable coefficient indicates that the property with greater accessibility is more expensive either the assessment data or the selling prices data.
Table 2. Estimation results for OLS models Assessed prices data (1,871 data)
Variables Parameter
P-Value
ln (DIST_MAINRD)
-0.777
0.000
ln (DIST_STA)
-0.044
ln (TIME_CBD)
-0.241
ln (DIST_SHOP) ln (DIST_EXP)
Offering prices data (823 data) Parameter
P-Value
***
-0.060
0.005
***
0.055
*
-0.263
0.000
***
0.000
***
-0.394
0.000
***
-0.077
0.000
***
-0.141
0.000
***
-0.033
0.028
n/s
-0.006
0.912
n/s
0.281
0.059
*
0.580
0.028
**
ln (HH_DENS)
0.089
0.000
***
0.039
0.310
n/s
ln (MED_INC)
0.245
0.004
***
0.036
0.792
n/s
-0.329
0.000
***
0.016
0.865
n/s
Property attributes
ln (JOB_ACCESS) Zonal attributes
ln (%L_RESIDENCE) ln (%L_COMMERC)
0.444
0.000
***
0.183
0.005
***
ln (%L_INDUST)
-0.077
0.000
***
-0.029
0.365
n/s
ln (%L_EDUCA)
0.079
0.005
***
-0.041
0.401
n/s
-0.137
0.000
***
-0.110
0.030
**
3.156
0.000
***
2.413
0.564
n/s
ln (%L_VACANT) Constant Goodness-of-fit R2
0.641
0.604
RSS
973.962
429.613
AIC
4,118.432
1,792.245
*** = significant at .1% level ** = significant at 1% level *
= significant at 5% level
n/s = insignificant at 5% level
For zonal attributes, the results obtained by using the assessed price data yields the statistically significant estimate. However, the estimated parameter of %L_RESIDENCE variable, and %L_EDUCA variable yield the different sign between two models. Regarding the OLS model of the offering prices for sale, the estimated parameters only, i.e., percentage of lands for commercial usages and the vacant lands are significant at 95% confidence level. Higher household density is associated with a higher property value in the assessment database. And every 1% increase in median household income is associated with a 0.245% increase in the assessed prices data. Either higher or lower percentage of land usage are associated with property values. For example, the higher percentage of land for industrial usages or the vacant land tend to decrease property value in both data, all else being equal. On the other hand, having the significant percentage of land for commercial usages remarkably increase values. 4.2. GWR models The estimation results of GWR models, shown in Table 3, contain the estimate of the parameters (i.e. mean value), and p-value. The goodness-of-fit is evaluated similar to the OLS models. From the table, the GWR models have significant improvement over
8
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
the OLS models for this study. In addition to the coefficient of determination (R2), residual sum of squares (RSS) and Akaike information criterion (AIC), the GWR models give the better explanation as expected. As stated earlier, the GWR model gives the ability to examine the spatial effect, i.e., non-stationarity hidden in the study. Therefore, the GWR models estimate the parameters at each observed point, i.e., a total 1,817 sets and 823 sets of estimation parameter for the assessed prices data and the offering prices for sale data, respectively are obtained. For ease of interpretations, the local estimated parameters were illustrated in the contour maps where the inverse distance weighting method is employed to interpolate as shown Fig. 3 to Fig. 8. Only accessibility variables are concentrated even if all the parameter estimates can be mapped.
Table 3. Estimation results for GWR models Assessed prices data (1,871 data)
Variables Parameter
P-Value
ln (DIST_MAINRD)
-0.063
0.000
ln (DIST_STA)
-0.141
ln (TIME_CBD)
-0.325
ln (DIST_SHOP) ln (DIST_EXP)
Offering prices data (823 data) Parameter
P-Value
***
-0.061
0.100
0.000
***
-0.239
0.030
*
0.000
***
-0.459
0.000
***
-0.145
0.000
***
-0.139
0.020
*
-0.092
0.000
***
0.011
0.030
*
0.289
0.000
***
0.634
0.000
***
ln (HH_DENS)
0.022
0.000
***
0.041
0.040
*
ln (MED_INC)
0.099
0.000
***
-0.027
0.450
n/s
-0.673
0.000
***
0.035
0.070
n/s
Property attributes
ln (JOB_ACCESS)
n/s
Zonal attributes
ln (%L_RESIDENCE) ln (%L_COMMERC)
0.057
0.000
***
0.106
0.000
***
ln (%L_INDUST)
-0.146
0.000
***
0.014
0.060
n/s
ln (%L_EDUCA)
0.353
0.000
***
-0.011
0.680
n/s
-0.154
0.000
***
-0.066
0.060
n/s
5.255
0.000
***
2.419
0.000
***
ln (%L_VACANT) Constant Goodness-of-fit R2
0.771
0.660
RSS
619.734
374.931
AIC
3,713.007
1,792.245
*** = significant at .1% level ** = significant at 1% level *
= significant at 5% level
n/s = insignificant at 5% level
Obviously, the estimated parameters vary substantially within the study area; indicating that there is a varying spatial relationship, i.e., non-stationarity in the model parameters. As stated, the results from the OLS models suggest that closer to the major roads can add significant value to land prices either the assessed data or the actual data. Such relationships seem to be appeared in the outer areas where there are a few services of public transport system and private vehicles are necessary for commuting as illustrated in Fig. 3 and Fig. 4. These areas can add value at the highest rates, that is, about 0.16% (using the assessed prices data) and 0.35% (using the offering prices for sale data) for every 1% closer to the major roads. However, Table 3 reveals that better accessibility to major roads adds an average premium value to the property about 0.06%. On the other hand, on average, property located closer to the station were sold around 40% higher than the assessment by the government. Both Fig. 5 and Fig. 6 indicate that being near the stations can significant add worth to the property. However, the premium benefits accrue to the property that are located fairly close to the BTS Skytrain, MRT Blue Line and Airport Rail Link as shown in Fig. 6. Unfortunately, such variations of capitalized effects clearly disappear on the Fig. 5.
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
Fig. 3. The local estimated parameters of the ln (DIST_MAINRD) variable using the assessed prices data
Fig. 4. The local estimated parameters of the ln (DIST_MAINRD) variable using the offering prices for sale data
9
10
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
Fig. 5. The local estimated parameters of the n (DIST_STA) variable using the assessed prices data
Fig. 6. The local estimated parameters of the ln (DIST_STA) variable using the offering prices for sale data
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
11
Fig. 7. The local estimated parameters of the ln (DIST_EXP) variable using the assessed prices data
Fig. 8. The local estimated parameters of the ln (DIST_EXP) variable the offering prices for sale data
Although globally shorter distance to expressway ramps can add value, from Fig. 7, it can be seen that in most of areas, such relationships appear to be unrelated. On the other hand, Fig. 8 presents the variations of proximity to expressway vary substantially between the left-hand side and right-hand side. The right-hand side, where the distance to the nearest expressway ramp has a significant negative effect on property value for sale, relate to the property located in the areas of expressway network, which may acquire a negative effect from this proximity. Another area on the left-hand side exhibit a significant positive effect. As seen, this area is not located in any catchment areas of the expressway network. The additional closeness to expressway ramps every 1% raise the value of the property by more than 0.1%. 5. Conclusion In this paper, the assessed prices data and the offering prices for sale data of Bangkok are analyzed with the use of regression framework through the hedonic pricing model. The spatial effect is also included in the model in the basis of the geographically
12
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
weighted regression (GWR) technique. The results reveal that the implications between two data are large. Furthermore, the determinants of property value are myriad. Besides, the impacts of each factor are quite complicated and varied over space, i.e. spatial non-stationarity exists in the given area. Also, the GWR models give the better explanation. Firstly, the descriptive statistics claim the same thing as many previous studies, that is, being near the station can add value to the property. The offering prices of property that located within 1,500 meters of BTS Skytrain, MRT Blue Line, and Airport Rail Link are nearly three times higher than the assessor’s database. As stated earlier, the hedonic pricing model are used to examine the implicit prices data based on their various characteristics. The estimated results indicated that there are somewhat different between the capitalized effects from the two OLS models in absolute terms, around 90% (distance to the main road), 80% (distance to the station), 40% (time to CBD and distance to the shopping mall), etc. For example, every 1% closer to the main road can add significant value to the property about 0.777% (using the assessed prices data) and 0.060% (using the offering prices data). Likewise, a 1-percent decrease in rail transit station proximity is estimated to increase the assessed value and offering price for sale by 0.044% and 0.263%, respectively. Other factors, e.g., closeness to the city center, shopping malls, expressway ramps, etc. also reveal the varying relationships to land price uplifts. The contour maps present the results of the GWL models. As stated earlier, the GWR model gives the ability to examine the spatial effect, i.e., non-stationarity hidden in the study. Therefore, the GWR models estimate the parameters at each observed point, i.e., a total 1,817 sets and 823 sets of estimation parameter for the assessed prices data and the offering prices for sale data, respectively. Each contour map reveals the spatial variations vary substantially. For example, being near the stations can significant add worth to the property. However, the premium benefits accrue to the property that are located fairly close to the BTS Skytrain, MRT Blue Line, and Airport Rail Link. Similarly, the distance to the nearest expressway ramp has a significant negative effect on property value for sale, relate to the property located in the areas of expressway network, which may acquire a negative effect from this proximity. This may be a usual case in many cities in the developing cities. Finally, the fair compensation for the landowners can computed based on this study. Also, the basic formulation of value capture policies to tax the direct beneficiaries in the affected districts in advance so as to finance the rail transit infrastructure projects, in particular the more than 10 transit lines that are planned for construction in the future can be formed. References 1. Banister, D., and J. Berechman. Transport Investment and Economic Development. UCL Press, London, 2000. 2. Hansen, W.G. How Accessibility Shapes Land Use. Journal of the American Institute of Planners, Vol. 25 (2), 1959, pp. 73-76. 3. Ben-Akiva, M., and S. Lerman. Disaggregate Travel and Mobility Choice Models and Mesures of Accessibility. Croom Helm, Andover, Hants, 1979, pp. 654-679. 4. Hess, D.B., and T.M. Almeida. Impact of Proximity to Light Rail Rapid Transit on Station-Area Property Values in Buffalo, New York. Urban Studies, Vol. 44 (5-6), 2007, pp. 1041-1068. 5. Vichiensan, V., S. Malaitham, and K. Miyamoto. Hedonic Analysis of Residential Property Values in Bangkok: Spatial Dependence and Nonstationarity Effects. Journal of the Eastern Asia Society for Transportation Studies, Vol. 8, 2011, pp. 886-899. 6. Cervero, R., and J. Landis. Assessing the Impacts of Urban Rail Transit on Local Real Estate Markets using Quasi-Experimental Comparisons. Transportation Research Part A: Policy and Practice, Vol. 27 (1), 1993, pp. 13-22. 7. Gatzlaff, D.H., and M.T. Smith. The Impact of the Miami Metrorail on the Value of Residences near Station Locations. Land Economics, Vol. 69 (1), 1993, pp. 54-66. 8. Sherry, R. Property Values and Transportation Facilities: Finding the Transportation-Land Use Connection. Journal of Planning Literature, Vol. 13 (4), 1997, pp. 412-427. 9. So, H.M., R.Y.C. Tse, and S. Ganesan. Estimating the Influence of Transport on House Prices: Evidence from Hong Kong. Journal of Property Valuation and Investment, Vol. 15 (1), 1997, pp. 40-47. 10. Rosen, S. Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition. The Journal of Political Economy, Vol. 82 (1), 1974, pp. 34-55. 11. Bajic, V. The Effects of a New Subway Line on Housing Prices in Metropolitan Toronto. Urban Studies, Vol. 20 (2), 1983, pp. 147-158. 12. Bowes, D.R., and K.R. Ihlanfeldt. Identifying the Impacts of Rail Transit Stations on Residential Property Values. Journal of Urban Economics, Vol. 50 (1), 2001, pp. 1-25. 13. Debrezion, G., E. Pels, and P. Rietveld. The Impact of Railway Stations on Residential and Commercial Property Value: A Meta -analysis. The Journal of Real Estate Finance and Economics, Vol. 35 (2), 2007, pp. 161-180. 14. Diao, M., and J. Ferreira Jr. Residential Property Values and the Built Environment: Empirical Study in the Boston, Massachusetts, Metropolitan Area. In Transportation Research Record: Journal of the Transportation Research Board, No. 2174, Transportation Research Board of the National Academies, Washington, D.C., 2010, pp. 138-147. 15. Doss, C.R., and S.J. Taff. The Influence of Wetland Type and Wetland Proximity on Residential Property Values. Journal of Agricultural and Resource Economics Vol. 21 (1), 1996, pp. 120-129. 16. Haider, M., and E.J. Miller. Effects of Transportation Infrastructure and Location on Residential Real Estate Values: Application of Spatial Autoregressive Techniques. In Transportation Research Record: Journal of the Transportation Research Board, No. 1722, Transportation Research Board of the National Academies, Washington, D.C., 2000, pp. 1-8. 17. Li, W., K. Joh, C. Lee, H. Park, and A. Woo. Assessing Benefits of Neighborhood Walkability to Single-Family Property Values: A Spatial Hedonic Study in Austin, Texas. Journal of Planning Education and Research, Vol., 2015, pp. 1-18. 18. Sunding, D.L., and A.M. Swoboda. Hedonic Analysis with Locally Weighted Regression: An Application to the Shadow Cost of Housing Regulation in Southern California. Regional Science and Urban Economics, Vol. 40 (6), 2010, pp. 550-573. 19. Cervero, R., and C.D. Kang. Bus Rapid Transit Impacts on Land Uses and Land Values in Seoul, Korea. Transport Policy, Vol. 18 (1), 2011, pp. 102-116. 20. Clower, T.L., and B.L. Weinstein. The Impact of Dallas (Texas) Area Rapid Transit Light Rail Stations on Taxable Property Valuations. Australasian Journal of Regional Studies, Vol. 8 (3), 2002, pp. 389-400. 21. Lewis-Workman, S., and D. Brod. Measuring the Neighborhood Benefits of Rail Transit Accessibility In transportation Research Record: Journal of the Transportation Research Board, No. 1576, Transportation Research Board of the National Academies, Washington, D.C., 1997, pp. 147-153. 22. Berry, B.J., and R.S. Bednarz. A Hedonic Model of Prices and Assessments for Single-Family Homes: Does the Assessor Follow the Market or the Market Follow the Assessor? Land Economics, Vol. 51 (1), 1975, pp. 21-40. 23. Kowalski, J.G., and P.F. Colwell. Market Versus Assessed Values of Industrial Land. Real Estate Economics, Vol. 14 (2), 1986, pp. 361-373.
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000 24. 25. 26. 27. 28. 29. 30. 31.
32. 33. 34. 35.
36.
37.
13
Ma, S., and S.M. Swinton. Hedonic Valuation of Farmland Using Sale Prices Versus Appraised Values. Land Economics, Vol. 88 (1), 2012, pp. 1-15. Freeman, M. The Measurement of Environmental and Resource Values: Theory and Methods. Resources for the Future, Washington, D.C., 2003. Ryan, S. The Value of Access to Highways and Light Rail Transit: Evidence for Industrial and Office Firms. Urban Studies, Vol. 42 (4), 2005, pp. 751-764. Forrest, D., J. Glen, and R. Ward. The Impact of a Light Rail System on the Structure of House Prices: A Hedonic Logitudinal Study. Journal of Transport Economics and Policy, Vol. 30 (1), 1996, pp. 15-29. Armstrong, R., and D. Rodriguez. An Evaluation of the Accessibility Benefits of Commuter Rail in Eastern Massachusetts using Spatial Hedonic Price Functions. Transportation, Vol. 33 (1), 2006, pp. 21-43. Wei, X., W.K. Zhang, C. QWang, and G.J.X. Xu. The Impact of Urban Rail Transit on Surrounding Residential Prices: Line 1 of Chengdu Metro as an Example. Modern Applied Science, Vol. 6 (3), 2012, pp. 58-64. McDonald, J.F., and D.P. McMillen. Employment Subcenters and Land Values in a Polycentric Urban Area: The Case of Chicago. Environment and Planning A, Vol. 22 (12), 1990, pp. 1561-1574. Pan, Q., H. Pan, M. Zhang, and B. Zhong. Effects of Rail Transit on Residential Property Values Comparison Study on the Rail Transit Lines in Houston, Texas, and Shanghai, China. In Transportation Research Record: Journal of the Transportation Research Board, No. 2453, Transportation Research Board of the National Academies, Washington, D.C., 2014, pp. 118-127. Bollinger, C.R., K.R. Ihlanfeldt, and D.R. Bowes. Spatial Variation in Office Rents within the Atlanta Region. Urban Studies, Vol. 35 (7), 1998, pp. 10971118. Cervero, R., and M. Duncan. Neighbourhood Composition and Residential Land Prices: Does Exclusion Raise or Lower Values? Urban Studies, Vol. 41 (2), 2004, pp. 299-315. Bae, C.-H.C., M.-J. Jun, and H. Park. The Impact of Seoul's Subway Line 5 on Residential Property Values. Transport Policy, Vol. 10 (2), 2003, pp. 85-94. Gingerich, K., H. Moaoh, and W. Anderson. Location and Transportation Effects on Onresidential Real Estate Price Regressions in Windsor, Ontario, Canada. In Transportation Research Record: Journal of the Transportation Research Board, No. 2397, Transportation Research Board of the National Academies, Washington, D.C., 2013. Chalermpong, S. Rail Transit and Residential Land Use in Developing Countries: Hedonic Study of Residential Property Prices in Bangkok, Thailand. In Transportation Research Record: Journal of the Transportation Research Board, No. 2038, Transportation Research Board of the National Academies, Washington, D.C., 2007, pp. 111-119. Fotheringham, S.A., C. Brunsdon, and M. Charlton. Geographically Weighted Regression: The Analysis of Spatially Varying Relationships. John Wiley & Sons, West Sussex, 2002.
14
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
Market value of land lots within 1.5 km of stations are expensive nearly 3 times higher than the assessor’s database. Spatial non-stationarity exists in the given area. Contour map reveals the spatial variations vary substantially. A short walk to stations can significantly add worth to the land value.
S2213-624X(17)30055-X https://doi.org/10.1016/j.cstp.2018.09.008 CSTP 291
To appear in:
Case Studies on Transport Policy
Received Date: Revised Date: Accepted Date:
17 February 2017 12 April 2018 17 September 2018
Please cite this article as: S. Malaitham, A. Fukuda, V. Vichiensan, V. Wasuntarasook, Hedonic pricing model of assessed and market land values: A case study in Bangkok metropolitan area, Thailand, Case Studies on Transport Policy (2018), doi: https://doi.org/10.1016/j.cstp.2018.09.008
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Title page 1.
Title: Hedonic pricing model of assessed and market land values: A case study in Bangkok Metropolitan Area, Thailand
2.
Authors: Sathita Malaitham, Corresponding Author Nihon University Department of Transportation System Engineering 7-24-1 Narashinodai, Funabashi, Chiba, 274-8501, Japan Tel: (+81)-47-469-5355; Fax: (+81)-47-469-5355 Email: [email protected] Atsushi Fukuda Nihon University Department of Transportation System Engineering 7-24-1 Narashinodai, Funabashi, Chiba, 274-8501, Japan Tel: (+81)-47-469-5355; Fax: (+81)-47-469-5355 Email: [email protected] Varameth Vichiensan Kasetsart University Department of Civil Engineering 50 Phahonyothin Rd, Ladyao, Jatujak, Bangkok, 10900, Thailand Tel: (+66)-2-942-8555; Fax: (+66)-2-579-4575 Email: [email protected] Vasinee Wasuntarasook Nagoya University Department of Environmental Engineering C1-2(651), Furo, Chikusa, Nagoya, 464-8603, Japan Tel: (+81)-52-789-2773; Fax: (+81)-52-789-2773 Email: [email protected]
Abstract Bangkok Metropolitan Area, the capital of Thailand, is known as one of the world’s most traffic-congested cities. Lots of transport-related projects to alleviate traffic congestion especially rapid transit system always require an amount of land. Therefore, several privately owned lands have to be acquired by the government agencies due to the lack of available spaces in an urban area. However, the assessment of compensation for a compulsory land acquisition is determined on the basis of the assessor’s database that often values each of properties lower than expected. In this paper, the assessed prices data and the offering prices for sale data of Bangkok are analyzed with the use of regression framework through the hedonic pricing model. The spatial non-stationarity to examine the variations of the implicit effects to property value is also included bases on the geographically weighted regression (GWR) technique. The results show that the determinants of property value are myriad and varied over space, i.e. spatial non-stationarity exists in the study area. The obtained results indicate that the difference between two data is extremely large. These results provide the basic information to compute the fair compensation for the landowners and allow the government agencies to tax the direct beneficiaries of their investments in advance so as to finance infrastructure projects. © 2017 The Authors. Published by Elsevier B.V. Peer-review under responsibility of WORLD CONFERENCE ON TRANSPORT RESEARCH SOCIETY.
2214-241X © 2017 The Authors. Published by Elsevier B.V. Peer-review under responsibility of WORLD CONFERENCE ON TRANSPORT RESEARCH SOCIETY.
2
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
Keywords: Assessed land value, Actual land value, Hedonic study, Spatial effect, Developing countries
1. Introduction Over the past several decades there has been a dramatic increase in the discussion of the capitalization effects of investments in transportation infrastructure (e.g., transits) in many ways. It is well acknowledged that changes in accessibility to areas of influence are considered as one of benefits of such investments (1). The accessibility, as defined by Hansen (2) and Ben-Akiva and Lerman (3), provides the potential of opportunities to participate in activities in different locations. For example, people who live nearby transit stations can reduce travel time and costs to other destinations. On the other hand, business activities such as shops and jobs can also be reached from any destinations. Therefore, households and businesses are willing to pay for higher accessibility locations. Such accessibility-improved areas have unavoidably become more attractive to investors for residential and commercial land development that pushing up their values, further owners of properties rightly obtain these benefits(4). Bangkok Metropolitan Area, the capital city of Thailand, is known as one of the world’s most traffic-congested cities due to the growing number of registered vehicles. Encouraging people to use public transport modes instead of their private vehicles in order to decrease the number of vehicles on the roads have been being argued as the solution to address this issue. A rail-based transit system, which is the affordable and comfortable transport system that run efficiently to meet the needs of the people, has been introduced and proposed by the policy-makers and planners. High capacity, performance, and full grade-separated system compared to other transport modes including buses and express boats causes an increase in quantity demand for land in locations with good accessibility to this system. Vichiensan et al. (5) indicated that an opening of the first transit system of Bangkok (i.e., BTS Skytrain) significantly influences assessed land values of station or corridor. Furthermore, they claimed that rail transit system has played significant role in accelerating urban development in Bangkok. From a policy perspective, it is important to understand how benefit of transit investment is capitalized into nearby land under market conditions. In this paper, the purpose is to examine the effects of transits on two types of values (i.e., assessed and market land values) by applying the hedonic pricing model. The relationship between assessed and market values may be quite complex, but there is no literature that compares whether the effects are the same for these two types or not in Bangkok studies. In fact, the assessed values assign to each parcel by government agencies for the purposes of taxation. Furthermore, the compensation for the compulsory land acquisition before the construction of transit infrastructure is evaluated based on land assessment database that is somewhat higher or lower than the market values. Hedonic model will help to explore the relative contribution of each of these attributes on the assessed and market land values. As known, each property has a unique bundle of attributes such as home features, its accessibility to transport, work, etc. (6, 7, 8, 9). As transit investment influences each area differently, the spatial function, i.e., spatial nonstationarity is accommodated with the hedonic pricing model to capture land value appreciation across the study area. The obtained results will indicate the difference between assessed and market land values in terms of various functional characteristics. This information can help to estimate how much beneficiaries should return their benefit to the public. 2. Literature reviews Assessed value corresponded by governmental agencies is used to calculate the value of the property for tax purposes. Conversely, market value refers to price of asset on the open market. Hedonic price model, which was firstly introduced by Rosen (10), has been applied to examine the relationships between property value and its characteristics for many years. A market value of properties was used to examine as a dependent variable in many hedonic studies (11, 12, 13, 14, 15, 16, 17, 18), whereas there are a small number of studies using assessed land value (4, 19, 20, 21). Either assessed or Market values were used to capture the determinants of single-family homes (22), industrial land (23), and farmland (24). Although assessed values might be imperfectly correlated with market values (25), there might have some errors in market values as a result of the distorted market (15). Kowalski and Colwell (23) investigated the factors that influence the assessed and market values of parcels zoned for industrial uses in western Wayne County, Michigan (i.e., west of Detroit). The implicit prices exhibited by market for industrial parcels in the west suburbs of Detroit are somewhat different than those which are organized in the assessment process in the same area, so industrial parcels are systematically or structurally under-and over- assessed. Ma and Swinton (24) captured the determinants of appraised value and sale price for farmland uses in southwestern Michigan. Results suggest that appraised values are a poor substitute for sale prices if the research goal is to understand dynamically evolving determinants of land value in exurbanizing regions, especially the value of natural amenities. Over the past decades, it has become increasingly clear that the presence of rail transit system can increase property values due to the accessibility improvement. Bajic (11) performed one of the earliest who attempted to identify the effects of a subway line in Toronto on the values of housing. Gatzlaff and Smith (7) examined the impact of the development of the Miami Metrorail system on property values proximate to its station locations. Some studies performed sought to distinguish between the accessibility benefits of rail transit and other transportation systems (6, 26). Forrest et al. (27) examined the relationship between the availability of commuter rail services and the pattern of house prices in an urban area of Manchester, England. Armstrong and Rodriguez (28) found that properties located in municipalities with commuter rail stations exhibit values that are between 9.6 percent and 10.1 percent higher than properties in municipalities without a commuter rail station in Eastern Massachusetts. Premium of light rail
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
3
transit accessibility is approximately $2.31 (using geographical straight-line distance) and $0.99 (using network distance) for every foot closer to a station in Buffalo, New York (4), while the impact of rail transit has increased around 7.814 billion yuan on the surrounding residential values of Chengdu Metro Line 1 (29). Although, a great deal of ascending property value produced by transits and their facilities has been captured in developed countries for many years, few studies have investigated in developing countries. One reason for little research is that transit system has been introduced to their countries during the past 20 years. Lots of studies have been carried out on the significance of accessibility to employment that mostly refers to the distance to CBD (16, 27, 30). Land values increase 16 percent for every mile closer to the CBD (30). Likewise, longer distances to the city’s sub-centers are associated with lower prices, for every kilometer increases in the distance to XuJiaHui, the price would drop by 4.0 percent (31). On the other hand, non-work activities such as shopping center, bank branch offices, green space or park, hospital, airport, etc. were found to be the factors that impact on the property values. For example, the presence of shopping centers and sports facilities, are important factors in determining prices of properties (9). The shopping mall, one of the attractiveness of the location that found the highly significant and uplift the office rents (32). Among variables describing neighborhood quality, median income level was statistically significant and it had the positive impact on property values in Queen (21). Cervero and Duncan (33) revealed that every $10,000 increase in median household income associated with a $1.67 per square foot increase in multi-family parcels in Santa Clara County. While every $100 increase in the median annual household income associated with a $36 increase in property value in Buffalo, suggesting that houses were more likely to sell for a higher price in affluent neighborhoods (4). Furthermore, population and employment density were both significant; prices negatively associated with population density but positively associated with employment density (34). The school quality, flooding, level of security, percent of population black, racial composition, crime rate, the noise level, and number of fireplaces in the neighborhood were also used to examine the effects on the property values of their methods. Previous research has provided mixed evidence including large positive, small positive as well as negative effects. 3. Methodology 3.1. Model specifications Hedonic pricing model is used to model the relationship between one (or more) dependent or response variables and independent or predictor variables. The equation of the hedonic pricing model can be written in the basis of the classical linear regression specification as follow:
Yi 0 1 X i1 ... k X ik i
(1)
An equation(1) can be written more compactly for n observations as
Y = Xβ ε
(2)
where Y X β
= = = =
a vector (n x 1) of observations corresponding to a dependent variable, a matrix (n x k) of observations of k independent variables, a vector (k x 1) of regression parameters, and a vector (n x 1) of random errors, assumed to be normally distributed
To solve the equation (2), the ordinary least square (OLS) is a simply fitting mechanism based on minimizing the sum of squared residuals or residual sum of squares (RSS). The vectors of regression parameters that minimizes are given by the usual expression:
β = (X'X)-1 X'Y
(3)
Also, the equation (1) transformed into natural logarithms as expressed in equation (4) is common in hedonic studies (35, 36).
ln(Yi ) 0 1 ln( X i1 ) ... k ln( X ik )
(4)
Spatial regression model applied to tease out the spatial effect, i.e., non-stationarity in this study is geographically weighted regression (GWR) model. The model takes on the following form (37):
Yi 0 (ui , vi ) 1 (ui , vi ) X1 ... k (ui , vi ) X k i
(5)
4
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
where (ui,vi) denotes the latitude and longitude coordinates of observation i, and k(ui,vi) represent the k regression parameters at estimation location of observation i. The equation (5) can be written more compactly as:
Y(i) = Xβ(i) + ε(i)
(6)
The estimation of regression parameters specific t every location is accomplished by performing local regressions, each using a sub-sample data in the vicinity of estimation’s location. Thus, a total of n observations can perform n local regression models. The estimator of GWR model is written as follows:
β(i) = (X'W(i)X)-1 X'W(i)Y
(7) where W(i) is the geographical weight function between the estimation location of observation i and the observation j,…,n located within the bandwidth of observation i (hi ). Within the bandwidth, the observations spatially closer to estimation location of the observation i have more influence in the local regression estimation compared with the data located farther away. In this study, the weight function is defined as w(d)=exp(-d2/h), where h is bandwidth and d is the distance between the focus point, i, and other points. 3.2. Study area The study area compasses the Bangkok Metropolitan Area. The region, shown in Fig. 1, covers an area of 1,568.7 km2 with 50 districts. There is the Chao Phraya River meandering through the city in the southward direction. This river divides Bangkok into two sides: the eastern bank and the western bank of the Chao Phraya. On the west side, it is known as the Bangkok’s old town where the original CBD situated. As the city expanded, parts of the areas along Silom Rd. and Sukhumvit Rd. located on the east side of the Chao Phraya River has been the CBD of Bangkok. There are lots of premium office buildings, five-star hotels, luxury condominiums, entertainment buildings, shopping malls and so on throughout the areas. By contrast, the Bangkok’s old town has become the top tourist destination and led them be the heart of Bangkok’s tourist and commercial center where many unique retail shops are concentrated.
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
5
Fig. 1. Map of Bangkok Metropolitan Area, Existing Rail Transit Network and Sample Data
Bangkok has a registered population of 5.69 million while a total number of registered vehicles have hit 8.92 million in 2015 which was two times higher than the total number of registered vehicles in 2004 (National Statistical Office). Undoubtedly, the growth of motorization is remarked as one of the critical causes that contributes to traffic congestion. The rail transit system to alleviate the traffic congestion was firstly introduced at the end of the 1970s. Although it has been known more than 30 years, the first rail transit of Bangkok, namely BTS Skytrain, has started its service along the heavily congested arterial roads such as Phaholyothin Rd., Sukhumvit Rd., Silom Rd., Sathorn Rd., etc. since December 1999. Five year later, the second rail transit line or so-called MRT Blue Line was started its operation from Bang Sue station to Hua Lumphong Station by running in an underground level of other crowded roads namely Ratchadapisek Rd., Asok Rd., Rama 4 Rd., etc. Then, Airport Rail Link was constructed to serve between the city area and the new international airport of Thailand. In March 2016, the new route, i.e., MRT Purple Line opened the service from Bang Sue station to Bang Yai station. Also, the extensions of BTS Skytrain and MRT Blue Line are now under construction whereas other new lines are now planning to start their construction. 3.3. Descriptive statistics
140000
140000
120000
120000
Price (baht/sq.m)
Price (baht/sq.m)
The data used for modeling involve two types of land values data, i.e., assessed and market land values. Both are assigned as the dependent variable (Y) in the equation (2). The assessment data obtaining from the reports of Treasury Department, Thailand are evaluated based on type of land use categories, proximity from main roads, size of lands, etc *. This report continuously uses to levy the property’s taxes over the period of 4 years without major change or update the values, while land values in the market are annually changing. The period of land value report employs to capture taxes during the year 2012 to 2015 were used in this study ; however, it was generally evaluated before published around 2 years, i.e., this assessed value had started evaluated since 2010 and published in the year 2012. The market values refer to the prices at which land are offered for sale on the market. The offering prices data to estimate were collected from the for-sale advertisements by landowners on the internet from August 2014 to February 2015. Advertisements normally include location, price, and size of land lot. Furthermore, convenient access such as to the nearest shopping malls, expressway ramps and also rail transit stations are almost mentioned in the advertisements. In the hedonic studies, the value of property located within the immediate area of rail transit stations is higher compared with the property from stations farther away. In case of Bangkok, the statistics of the assessed prices data and the offering prices data by the distance intervals of each operating line are presented in Fig. 2. First of all, the differences between the prices per square meter of assessed and market data are noticeably large. Land parcels being located closer to the stations tend to add more valuable than those being located farther away as expected. Surprisingly, the averages of market value of land parcel being located within 1,500 meters of BTS Skytrain, MRT Blue Line and Airport Rail Link were approximately three times higher than the assessor’s database.
100000 80000 60000 40000 20000
100000 80000 60000 40000 20000
0
0 0-500
500-1000 1000-1500
>1500
0-500
Distance to station (m.) BTS Skytrain
MRT Blue Line
Airport Rail Link
(a)
500-1000 1000-1500
>1500
Distance to station (m.) BTS Skytrain
MRT Blue Line
Airport Rail Link
(b)
Fig. 2. Property values by station’s distance intervals: (a) Assessed value and (b) Market value
To include the effects of mentioned attributes on the advertisements and the reports as the independent variables (X), GIS was used to measure the straight-line distances such as the distance from the land parcels to the nearest rail transit stations, to school, to the central business district, and so on that mentioned. Also, GIS was used to measure how far the parcels are located from the
*
http://www.treasury.go.th/main.php?filename=index
6
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
main roads and expressway entrances/exit. As indicated in the previous study, not only are the proxy attributes influenced on the property values, but also zonal attributes such as employment density, household density, median income, etc. Furthermore, the land use proportions in each zone were included in the models. All the zonal variables in this study were collected from National Statistical Office (NSO) and Bangkok Metropolitan Administration (BMA). The descriptive statistics (i.e. mean values) of the dependent variables and their attributes are summarized in Table 1.
Table 1. Descriptive statistics Assessed value (1,871 data)
Market value (823 data)
Property value (baht/m2)
13,736
29,202
Size of land Parcel (m2)
3,704
4,073
Independent variable Property attributes DIST_MAINRD DIST_STA
Distance to the nearest main road (km) Distance to the nearest station (km)
0.49 4.29
0.58 4.58
TIME_CBD DIST_SHOP
Time to CBD (min) Distance to the nearest shopping mall (km)
61.05 3.09
71.39 2.82
DIST_EXP JOB_ACCESS
Distance to expressway (km) Job accessibility (jobs)
5.31 2,298,110
4.97 2,250,783
Zonal attributes HH_DENS EMP_DENS
Household density (households/sq.km) Employment density (jobs/sq.km)
3,675 9,160
3,116 5,599
MED_INC %L_RESIDENCE %L_COMMERC
Median income (baht) Percentage of residence area Percentage of commercial area
30,787 33.21 4.51
29,388 34.37 4.61
%L_INDUST %L_EDUCA
Percentage of industrial area Percentage of education area
11.36 2.19
11.44 1.92
%L_VACANT
Percentage of vacant area
20.31
22.31
Variables
Description
Dependent variable P_VALUE LOT_SIZE
4. Results 4.1. OLS models The results of OLS models are shown in Table 2 which contains the estimate of parameters, the t-statistic, and p-value. The goodness-of-fit is evaluated by the coefficient of determination (R2), residual sum of squares (RSS) and Akaike information criterion (AIC). First of all, the elasticities with respect to the proxy variables including distance to main road, the nearest station, time to the CBD, distance to the nearest shopping mall and the nearest expressway ramps have the negative sign as expected. For example, the estimation results of both data reveal that the elasticities with respect to the distance to the main road are approximately -0.777 (using the assessment data) and -0.060 (using the offering prices data) while the elasticities of being used the assessed prices data and the offering prices data with respect the distance to the nearest station are -0.044 and -0.263, respectively. These results suggest a property value discount of 0.777% and 0.060% for a percentage increase in the distance to main road. Similarly, a 1-percent decrease in rail transit station proximity was estimated to increase the assessed value and offering price for sale by 0.044% and 0.263%, respectively, meaning that a-100,000 baht/m2 property located 100 m from the station will be priced at 44 baht/m2 (the assessed prices data) and 263 baht/m2 (the offering prices for sale data) more than an identical property that located 101 m from the station and around 440 baht/m2 (the assessed prices data) and 2,630 baht/m2 (the offering prices for sale data) more than another identical property that is located 110 m from the station. Obviously, there are somewhat different between the estimated parameters
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
7
from the assessed prices data and the offering prices for sale. For instance, holding everything constant, the identical property located near the shopping mall can sell more expensive than the assessor’s database about 45.39 %. And, that property will be worth about 5.5 times higher than the assessment if it located near the expressway ramps. Besides, the employment accessibility variable coefficient indicates that the property with greater accessibility is more expensive either the assessment data or the selling prices data.
Table 2. Estimation results for OLS models Assessed prices data (1,871 data)
Variables Parameter
P-Value
ln (DIST_MAINRD)
-0.777
0.000
ln (DIST_STA)
-0.044
ln (TIME_CBD)
-0.241
ln (DIST_SHOP) ln (DIST_EXP)
Offering prices data (823 data) Parameter
P-Value
***
-0.060
0.005
***
0.055
*
-0.263
0.000
***
0.000
***
-0.394
0.000
***
-0.077
0.000
***
-0.141
0.000
***
-0.033
0.028
n/s
-0.006
0.912
n/s
0.281
0.059
*
0.580
0.028
**
ln (HH_DENS)
0.089
0.000
***
0.039
0.310
n/s
ln (MED_INC)
0.245
0.004
***
0.036
0.792
n/s
-0.329
0.000
***
0.016
0.865
n/s
Property attributes
ln (JOB_ACCESS) Zonal attributes
ln (%L_RESIDENCE) ln (%L_COMMERC)
0.444
0.000
***
0.183
0.005
***
ln (%L_INDUST)
-0.077
0.000
***
-0.029
0.365
n/s
ln (%L_EDUCA)
0.079
0.005
***
-0.041
0.401
n/s
-0.137
0.000
***
-0.110
0.030
**
3.156
0.000
***
2.413
0.564
n/s
ln (%L_VACANT) Constant Goodness-of-fit R2
0.641
0.604
RSS
973.962
429.613
AIC
4,118.432
1,792.245
*** = significant at .1% level ** = significant at 1% level *
= significant at 5% level
n/s = insignificant at 5% level
For zonal attributes, the results obtained by using the assessed price data yields the statistically significant estimate. However, the estimated parameter of %L_RESIDENCE variable, and %L_EDUCA variable yield the different sign between two models. Regarding the OLS model of the offering prices for sale, the estimated parameters only, i.e., percentage of lands for commercial usages and the vacant lands are significant at 95% confidence level. Higher household density is associated with a higher property value in the assessment database. And every 1% increase in median household income is associated with a 0.245% increase in the assessed prices data. Either higher or lower percentage of land usage are associated with property values. For example, the higher percentage of land for industrial usages or the vacant land tend to decrease property value in both data, all else being equal. On the other hand, having the significant percentage of land for commercial usages remarkably increase values. 4.2. GWR models The estimation results of GWR models, shown in Table 3, contain the estimate of the parameters (i.e. mean value), and p-value. The goodness-of-fit is evaluated similar to the OLS models. From the table, the GWR models have significant improvement over
8
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
the OLS models for this study. In addition to the coefficient of determination (R2), residual sum of squares (RSS) and Akaike information criterion (AIC), the GWR models give the better explanation as expected. As stated earlier, the GWR model gives the ability to examine the spatial effect, i.e., non-stationarity hidden in the study. Therefore, the GWR models estimate the parameters at each observed point, i.e., a total 1,817 sets and 823 sets of estimation parameter for the assessed prices data and the offering prices for sale data, respectively are obtained. For ease of interpretations, the local estimated parameters were illustrated in the contour maps where the inverse distance weighting method is employed to interpolate as shown Fig. 3 to Fig. 8. Only accessibility variables are concentrated even if all the parameter estimates can be mapped.
Table 3. Estimation results for GWR models Assessed prices data (1,871 data)
Variables Parameter
P-Value
ln (DIST_MAINRD)
-0.063
0.000
ln (DIST_STA)
-0.141
ln (TIME_CBD)
-0.325
ln (DIST_SHOP) ln (DIST_EXP)
Offering prices data (823 data) Parameter
P-Value
***
-0.061
0.100
0.000
***
-0.239
0.030
*
0.000
***
-0.459
0.000
***
-0.145
0.000
***
-0.139
0.020
*
-0.092
0.000
***
0.011
0.030
*
0.289
0.000
***
0.634
0.000
***
ln (HH_DENS)
0.022
0.000
***
0.041
0.040
*
ln (MED_INC)
0.099
0.000
***
-0.027
0.450
n/s
-0.673
0.000
***
0.035
0.070
n/s
Property attributes
ln (JOB_ACCESS)
n/s
Zonal attributes
ln (%L_RESIDENCE) ln (%L_COMMERC)
0.057
0.000
***
0.106
0.000
***
ln (%L_INDUST)
-0.146
0.000
***
0.014
0.060
n/s
ln (%L_EDUCA)
0.353
0.000
***
-0.011
0.680
n/s
-0.154
0.000
***
-0.066
0.060
n/s
5.255
0.000
***
2.419
0.000
***
ln (%L_VACANT) Constant Goodness-of-fit R2
0.771
0.660
RSS
619.734
374.931
AIC
3,713.007
1,792.245
*** = significant at .1% level ** = significant at 1% level *
= significant at 5% level
n/s = insignificant at 5% level
Obviously, the estimated parameters vary substantially within the study area; indicating that there is a varying spatial relationship, i.e., non-stationarity in the model parameters. As stated, the results from the OLS models suggest that closer to the major roads can add significant value to land prices either the assessed data or the actual data. Such relationships seem to be appeared in the outer areas where there are a few services of public transport system and private vehicles are necessary for commuting as illustrated in Fig. 3 and Fig. 4. These areas can add value at the highest rates, that is, about 0.16% (using the assessed prices data) and 0.35% (using the offering prices for sale data) for every 1% closer to the major roads. However, Table 3 reveals that better accessibility to major roads adds an average premium value to the property about 0.06%. On the other hand, on average, property located closer to the station were sold around 40% higher than the assessment by the government. Both Fig. 5 and Fig. 6 indicate that being near the stations can significant add worth to the property. However, the premium benefits accrue to the property that are located fairly close to the BTS Skytrain, MRT Blue Line and Airport Rail Link as shown in Fig. 6. Unfortunately, such variations of capitalized effects clearly disappear on the Fig. 5.
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
Fig. 3. The local estimated parameters of the ln (DIST_MAINRD) variable using the assessed prices data
Fig. 4. The local estimated parameters of the ln (DIST_MAINRD) variable using the offering prices for sale data
9
10
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
Fig. 5. The local estimated parameters of the n (DIST_STA) variable using the assessed prices data
Fig. 6. The local estimated parameters of the ln (DIST_STA) variable using the offering prices for sale data
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
11
Fig. 7. The local estimated parameters of the ln (DIST_EXP) variable using the assessed prices data
Fig. 8. The local estimated parameters of the ln (DIST_EXP) variable the offering prices for sale data
Although globally shorter distance to expressway ramps can add value, from Fig. 7, it can be seen that in most of areas, such relationships appear to be unrelated. On the other hand, Fig. 8 presents the variations of proximity to expressway vary substantially between the left-hand side and right-hand side. The right-hand side, where the distance to the nearest expressway ramp has a significant negative effect on property value for sale, relate to the property located in the areas of expressway network, which may acquire a negative effect from this proximity. Another area on the left-hand side exhibit a significant positive effect. As seen, this area is not located in any catchment areas of the expressway network. The additional closeness to expressway ramps every 1% raise the value of the property by more than 0.1%. 5. Conclusion In this paper, the assessed prices data and the offering prices for sale data of Bangkok are analyzed with the use of regression framework through the hedonic pricing model. The spatial effect is also included in the model in the basis of the geographically
12
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
weighted regression (GWR) technique. The results reveal that the implications between two data are large. Furthermore, the determinants of property value are myriad. Besides, the impacts of each factor are quite complicated and varied over space, i.e. spatial non-stationarity exists in the given area. Also, the GWR models give the better explanation. Firstly, the descriptive statistics claim the same thing as many previous studies, that is, being near the station can add value to the property. The offering prices of property that located within 1,500 meters of BTS Skytrain, MRT Blue Line, and Airport Rail Link are nearly three times higher than the assessor’s database. As stated earlier, the hedonic pricing model are used to examine the implicit prices data based on their various characteristics. The estimated results indicated that there are somewhat different between the capitalized effects from the two OLS models in absolute terms, around 90% (distance to the main road), 80% (distance to the station), 40% (time to CBD and distance to the shopping mall), etc. For example, every 1% closer to the main road can add significant value to the property about 0.777% (using the assessed prices data) and 0.060% (using the offering prices data). Likewise, a 1-percent decrease in rail transit station proximity is estimated to increase the assessed value and offering price for sale by 0.044% and 0.263%, respectively. Other factors, e.g., closeness to the city center, shopping malls, expressway ramps, etc. also reveal the varying relationships to land price uplifts. The contour maps present the results of the GWL models. As stated earlier, the GWR model gives the ability to examine the spatial effect, i.e., non-stationarity hidden in the study. Therefore, the GWR models estimate the parameters at each observed point, i.e., a total 1,817 sets and 823 sets of estimation parameter for the assessed prices data and the offering prices for sale data, respectively. Each contour map reveals the spatial variations vary substantially. For example, being near the stations can significant add worth to the property. However, the premium benefits accrue to the property that are located fairly close to the BTS Skytrain, MRT Blue Line, and Airport Rail Link. Similarly, the distance to the nearest expressway ramp has a significant negative effect on property value for sale, relate to the property located in the areas of expressway network, which may acquire a negative effect from this proximity. This may be a usual case in many cities in the developing cities. Finally, the fair compensation for the landowners can computed based on this study. Also, the basic formulation of value capture policies to tax the direct beneficiaries in the affected districts in advance so as to finance the rail transit infrastructure projects, in particular the more than 10 transit lines that are planned for construction in the future can be formed. References 1. Banister, D., and J. Berechman. Transport Investment and Economic Development. UCL Press, London, 2000. 2. Hansen, W.G. How Accessibility Shapes Land Use. Journal of the American Institute of Planners, Vol. 25 (2), 1959, pp. 73-76. 3. Ben-Akiva, M., and S. Lerman. Disaggregate Travel and Mobility Choice Models and Mesures of Accessibility. Croom Helm, Andover, Hants, 1979, pp. 654-679. 4. Hess, D.B., and T.M. Almeida. Impact of Proximity to Light Rail Rapid Transit on Station-Area Property Values in Buffalo, New York. Urban Studies, Vol. 44 (5-6), 2007, pp. 1041-1068. 5. Vichiensan, V., S. Malaitham, and K. Miyamoto. Hedonic Analysis of Residential Property Values in Bangkok: Spatial Dependence and Nonstationarity Effects. Journal of the Eastern Asia Society for Transportation Studies, Vol. 8, 2011, pp. 886-899. 6. Cervero, R., and J. Landis. Assessing the Impacts of Urban Rail Transit on Local Real Estate Markets using Quasi-Experimental Comparisons. Transportation Research Part A: Policy and Practice, Vol. 27 (1), 1993, pp. 13-22. 7. Gatzlaff, D.H., and M.T. Smith. The Impact of the Miami Metrorail on the Value of Residences near Station Locations. Land Economics, Vol. 69 (1), 1993, pp. 54-66. 8. Sherry, R. Property Values and Transportation Facilities: Finding the Transportation-Land Use Connection. Journal of Planning Literature, Vol. 13 (4), 1997, pp. 412-427. 9. So, H.M., R.Y.C. Tse, and S. Ganesan. Estimating the Influence of Transport on House Prices: Evidence from Hong Kong. Journal of Property Valuation and Investment, Vol. 15 (1), 1997, pp. 40-47. 10. Rosen, S. Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition. The Journal of Political Economy, Vol. 82 (1), 1974, pp. 34-55. 11. Bajic, V. The Effects of a New Subway Line on Housing Prices in Metropolitan Toronto. Urban Studies, Vol. 20 (2), 1983, pp. 147-158. 12. Bowes, D.R., and K.R. Ihlanfeldt. Identifying the Impacts of Rail Transit Stations on Residential Property Values. Journal of Urban Economics, Vol. 50 (1), 2001, pp. 1-25. 13. Debrezion, G., E. Pels, and P. Rietveld. The Impact of Railway Stations on Residential and Commercial Property Value: A Meta -analysis. The Journal of Real Estate Finance and Economics, Vol. 35 (2), 2007, pp. 161-180. 14. Diao, M., and J. Ferreira Jr. Residential Property Values and the Built Environment: Empirical Study in the Boston, Massachusetts, Metropolitan Area. In Transportation Research Record: Journal of the Transportation Research Board, No. 2174, Transportation Research Board of the National Academies, Washington, D.C., 2010, pp. 138-147. 15. Doss, C.R., and S.J. Taff. The Influence of Wetland Type and Wetland Proximity on Residential Property Values. Journal of Agricultural and Resource Economics Vol. 21 (1), 1996, pp. 120-129. 16. Haider, M., and E.J. Miller. Effects of Transportation Infrastructure and Location on Residential Real Estate Values: Application of Spatial Autoregressive Techniques. In Transportation Research Record: Journal of the Transportation Research Board, No. 1722, Transportation Research Board of the National Academies, Washington, D.C., 2000, pp. 1-8. 17. Li, W., K. Joh, C. Lee, H. Park, and A. Woo. Assessing Benefits of Neighborhood Walkability to Single-Family Property Values: A Spatial Hedonic Study in Austin, Texas. Journal of Planning Education and Research, Vol., 2015, pp. 1-18. 18. Sunding, D.L., and A.M. Swoboda. Hedonic Analysis with Locally Weighted Regression: An Application to the Shadow Cost of Housing Regulation in Southern California. Regional Science and Urban Economics, Vol. 40 (6), 2010, pp. 550-573. 19. Cervero, R., and C.D. Kang. Bus Rapid Transit Impacts on Land Uses and Land Values in Seoul, Korea. Transport Policy, Vol. 18 (1), 2011, pp. 102-116. 20. Clower, T.L., and B.L. Weinstein. The Impact of Dallas (Texas) Area Rapid Transit Light Rail Stations on Taxable Property Valuations. Australasian Journal of Regional Studies, Vol. 8 (3), 2002, pp. 389-400. 21. Lewis-Workman, S., and D. Brod. Measuring the Neighborhood Benefits of Rail Transit Accessibility In transportation Research Record: Journal of the Transportation Research Board, No. 1576, Transportation Research Board of the National Academies, Washington, D.C., 1997, pp. 147-153. 22. Berry, B.J., and R.S. Bednarz. A Hedonic Model of Prices and Assessments for Single-Family Homes: Does the Assessor Follow the Market or the Market Follow the Assessor? Land Economics, Vol. 51 (1), 1975, pp. 21-40. 23. Kowalski, J.G., and P.F. Colwell. Market Versus Assessed Values of Industrial Land. Real Estate Economics, Vol. 14 (2), 1986, pp. 361-373.
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000 24. 25. 26. 27. 28. 29. 30. 31.
32. 33. 34. 35.
36.
37.
13
Ma, S., and S.M. Swinton. Hedonic Valuation of Farmland Using Sale Prices Versus Appraised Values. Land Economics, Vol. 88 (1), 2012, pp. 1-15. Freeman, M. The Measurement of Environmental and Resource Values: Theory and Methods. Resources for the Future, Washington, D.C., 2003. Ryan, S. The Value of Access to Highways and Light Rail Transit: Evidence for Industrial and Office Firms. Urban Studies, Vol. 42 (4), 2005, pp. 751-764. Forrest, D., J. Glen, and R. Ward. The Impact of a Light Rail System on the Structure of House Prices: A Hedonic Logitudinal Study. Journal of Transport Economics and Policy, Vol. 30 (1), 1996, pp. 15-29. Armstrong, R., and D. Rodriguez. An Evaluation of the Accessibility Benefits of Commuter Rail in Eastern Massachusetts using Spatial Hedonic Price Functions. Transportation, Vol. 33 (1), 2006, pp. 21-43. Wei, X., W.K. Zhang, C. QWang, and G.J.X. Xu. The Impact of Urban Rail Transit on Surrounding Residential Prices: Line 1 of Chengdu Metro as an Example. Modern Applied Science, Vol. 6 (3), 2012, pp. 58-64. McDonald, J.F., and D.P. McMillen. Employment Subcenters and Land Values in a Polycentric Urban Area: The Case of Chicago. Environment and Planning A, Vol. 22 (12), 1990, pp. 1561-1574. Pan, Q., H. Pan, M. Zhang, and B. Zhong. Effects of Rail Transit on Residential Property Values Comparison Study on the Rail Transit Lines in Houston, Texas, and Shanghai, China. In Transportation Research Record: Journal of the Transportation Research Board, No. 2453, Transportation Research Board of the National Academies, Washington, D.C., 2014, pp. 118-127. Bollinger, C.R., K.R. Ihlanfeldt, and D.R. Bowes. Spatial Variation in Office Rents within the Atlanta Region. Urban Studies, Vol. 35 (7), 1998, pp. 10971118. Cervero, R., and M. Duncan. Neighbourhood Composition and Residential Land Prices: Does Exclusion Raise or Lower Values? Urban Studies, Vol. 41 (2), 2004, pp. 299-315. Bae, C.-H.C., M.-J. Jun, and H. Park. The Impact of Seoul's Subway Line 5 on Residential Property Values. Transport Policy, Vol. 10 (2), 2003, pp. 85-94. Gingerich, K., H. Moaoh, and W. Anderson. Location and Transportation Effects on Onresidential Real Estate Price Regressions in Windsor, Ontario, Canada. In Transportation Research Record: Journal of the Transportation Research Board, No. 2397, Transportation Research Board of the National Academies, Washington, D.C., 2013. Chalermpong, S. Rail Transit and Residential Land Use in Developing Countries: Hedonic Study of Residential Property Prices in Bangkok, Thailand. In Transportation Research Record: Journal of the Transportation Research Board, No. 2038, Transportation Research Board of the National Academies, Washington, D.C., 2007, pp. 111-119. Fotheringham, S.A., C. Brunsdon, and M. Charlton. Geographically Weighted Regression: The Analysis of Spatially Varying Relationships. John Wiley & Sons, West Sussex, 2002.
14
S.Malaitham/ Transportation Research Procedia 00 (2017) 000–000
Market value of land lots within 1.5 km of stations are expensive nearly 3 times higher than the assessor’s database. Spatial non-stationarity exists in the given area. Contour map reveals the spatial variations vary substantially. A short walk to stations can significantly add worth to the land value.