# RECORD 602e1fb4b1e820fd13056357b14a38759c9354ab
# TITLE: Metro station inauguration, housing prices, and transportation accessibility: Tehran case study
# YEAR: 2021.0  VENUE: Journal of Transport and Land Use  DOI: 10.5198/jtlu.2021.1622
# AUTHORS: Yekta Yazdanifard; Hosein Joshaghani; Masoud Talebian
# PAGES: 25  CANDIDATE TABLE PAGES (1-based): [5, 12, 13, 15, 18, 19, 20, 21, 22, 23]
# ABSTRACT: The opening of a new metro station, as a mode of the transportation corridor, potentially could have different effects on housing prices. We have investigated its effect on the value of residential properties around those stations, using data from large expansions of the metro network in Tehran, Iran. In the period of our study (April 2010 to December 2018), forty-five metro stations were inaugurated in Tehran. We use a difference-in-difference regression method to identify the causal effect of interest, where adjacent properties are used as the treatment group and similar but distant properties as the control group. The results indicate that, on average, the adjacent properties are affected by a 3.7 percent increase in price relative to distant properties. We also extend our study by categorizing new metro stations according to the extent of ex-ante access to other modes of public transportation such as bus rapid transit (BRT). We find 2 to 11 percent positive effect of new metro stations in regions with lower public transport, while in regions with ex-ante extensive public transportation system, we find less than 2 percent positive effect.


===== PAGE 1 (score 7, numeric-paren 0, table-words 0, var-words 7) =====


http://jtlu.org
. 14 . 1 [2021] pp. 537–561
Metro station inauguration, housing prices, and transportation
accessibility: Tehran case study
Yekta Yazdanifard
Sharif University of Technology
Hosein Joshaghani
Khatam University
Masoud Talebian
Sharif University of Technology
Abstract: e opening of a new metro station, as a mode of the transporta-
tion corridor, potentially could have different effects on housing prices. We
have investigated its effect on the value of residential properties around those
stations, using data from large expansions of the metro network in Tehran,
Iran. In the period of our study (April 2010 to December 2018), forty-five
metro stations were inaugurated in Tehran. We use a difference-in-difference
regression method to identify the causal effect of interest, where adjacent
properties are used as the treatment group and similar but distant properties
as the control group. e results indicate that, on average, the adjacent prop-
erties are affected by a 3.7 percent increase in price relative to distant proper-
ties. We also extend our study by categorizing new metro stations according
totheextentofex-anteaccesstoothermodesofpublictransportationsuchas
bus rapid transit (BRT). We find 2 to 11 percent positive effect of new metro
stations in regions with lower public transport, while in regions with ex-ante
extensive public transportation system, we find less than 2 percent positive
effect.
Article History:
Received: June5,2019
Accepted: May 27,
2020
Availableonline: April
26, 2021
1
Introduction
Public transportation in large cities and metropolitan regions is of paramount importance, and there-
fore are policies regarding its creation and expansion. While the transportation corridors can bring
some disadvantages to residents adjacent to them, e.g., crowding and pollution, they are mainly per-
ceived as advantageous because they make the region more accessible and easier for daily commutes.
erefore, itisexpectedthatnearbyhousingpricestoincreaseduetomoreandfasteraccesstootherre-
gions of the city. In this study, we not only measure this effect but also examine factors that determine
its magnitude.
OuruniqueadministrativedatacomesfromTehran,andweconsiderthetotalnumberof1,336,266
transactions. Tehran is the capital and biggest city of Iran and home to about 10 million population.
Copyright 2021 Yekta Yazdanifard, Hosein Joshaghani, Masoud Talebian.
doi: 10.5198/jtlu.2021.1622
ISSN: 1938-7849 | Licensed under the Creative Commons Attribution – NonCommercial License 4.0.
e Journal of Transport and Land Use is the official journal of the World Society for Transport and Land Use
(WSTLUR) and is published and sponsored by the University of Minnesota Center for Transportation Studies.


===== PAGE 5 (score 18, numeric-paren 0, table-words 4, var-words 8) =====

Metro station inauguration, housing prices, and transportation accessibility: Tehran case study

3.1
Housing Prices
Data on housing prices are according to the recorded transactions, obtained from the Office of Plan-
ning and Housing Economy (OPHE) of the Ministry of Roads and Urban Development in Iran. is
administrative information includes complete data on all individual housing transactions from April
1, 2010, to December 20, 2018 (nine years) in Iran. We focus on part of data which is for the resi-
dential housing transaction in Tehran, which consists of almost 1,260,000 housing sale transactions.
All real estate transactions in Iran are done by real estate agencies. ese transactions must be fully
integrated into a system connected with the Ministry of Roads and Urban Development. e system
has been in operation throughout the country since 2010. e information recorded within the sys-
tem is provided by real estate managers. OPHE record all the housing transaction information which
are officially registered in real estate agencies since 2010². We refer to Appendix 2 for data cleaning
steps. For each transaction, the OPHE data has the following variables, and the summary statistics is
reported in Table (1). e description of the time categories is presented in the appendix.
1. Transaction code,
2. e transaction dates,
3. Total transaction price,
4. e price per square meter,
5. e surface area of the house,
6. Age of the building,
7. Using type of the building,
8. Frame type of the building skeleton,
9. e first six digits of the ten digits postal code of the housing traded,
10. Percentage of the property traded.
e using type of the building in our administrative data is divided in 4 categories: i) Residen-
tial Buildings ii) Commercial Buildings iii) Institutions iv) Office Buildings. More than 99% of these
transactions are residential houses. In this research, we calculate the impact of metro station inaugu-
ration on residential houses and omit the other types of building which are less than 1%.
Table 1: Summary statistics of residential properties’ valuables
Total
Treatment
Control
mean
sd
mean
sd
mean
sd
Area
84.93
38.30
76.58
33.41
84.89
40.56
Price (Mill. Tomans)
4.24
3.48
3.66
2.74
4.25
3.49
Price (Dollars)
1000.95
613.69
881.19
501.85
1007.79
628.66
Age
8.82
8.59
10.12
9.37
9.08
9.02
Observations
1,267,466
66,324
766,958
Comments: USD/IRR changes in different years. In 2010 each Dollar was averagely
1100 Tomans and in 2019 it grows up to 12000 Tomans in the market. Prices are
nominal and are not adjusted for inflation. Treatment and control groups are intro-
duced later in Section 4.3 of the paper.
As shown in the Table (2), the skeleton is mostly made of concrete, metal or both. Only less than
1% of skeleton properties are different.
Residential properties in Iran and Tehran are one or more units in one building, each unit having
one household and owner. In this situation, several families may live in one apartment. ere are, of
course, single-unit duplex houses that are very few. Each residential property in Iran is divided into 6
² e housing information used in this article is available in the following link: https://bit.ly/2IteOEz


===== PAGE 6 (score 6, numeric-paren 0, table-words 1, var-words 4) =====


.
Table 2: Summary statistics by frame type of residential properties
Count
Col %
Cum %
Frame Type
Concrete
609,141
48.1
48.1
Steel
584,520
46.1
94.2
Steel and Concrete
69,396
5.5
99.7
Without Frame
2,738
0.2
99.9
Brick or Cement Block
1,091
0.1
100.0
Wood
317
0.0
100.0
Adobe or Clay
262
0.0
100.0
Total
1,267,466
100.0
Comments : More than 99 percent of frame types are concrete or/and
metal.
shares, where any real or legal person can own one or more of these shares, but generally one person
owns all 6 shares. At the time of trading, any of these shares can be traded. As a result, 16% ,33% ,50%
,67% ,84% and 100% of a property may be traded on each transaction. In the administrative OPHE
data, the number of transactions that is traded less than 100% of the house is only 1.7%, so they are
omitted from our study.
3.2
Metro Stations
e metro stations data has been obtained from Tehran Metro Operations Company. is informa-
tion includes:
1. Time of the inauguration of the metro station,
2. Geographic coordinates of the metro station,
3. Name of the metro station,
4. e number of metro lines which passes through the station.
e total number of stations operated in Tehran is 105, which is planned to increase up to 150 stations
by 2022. e number of stations opened during the period considered in this study is presented in
Figure (1). e red lines specify the stations opened within our research period.
As will be discussed further, we use difference-in-difference method to investigate the causal effect
of a metro station inauguration on residential housing prices. To implement this method, it is neces-
sary to have access to the data before and aer the inauguration of the metro station. Based on the
availability of housing prices from April 1, 2010, only those stations inaugurated aer this date have
been investigated. erefore, 45 stations are investigated, most of which are related to stations of lines
C and D.
It is important to mention that four of these stations (A1, A8, B9, D1) have been inaugurated two
times because they are the junctions of two different metro lines. ese stations opened for the first
time before the period under investigation, and for the second time within our research period. Since
the second opening of metro station can have a different impact from the first opening, we produce
two different outputs: with and without these stations. Location of these 45 stations investigated is
depicted on Figure (2).
3.3
Postal Codes and Geographic Coordinates
In Iran, each property has a unique 10-digit postal code that not only identifies the city and neigh-
borhood of the property but also its exact geographic coordinates. Postal code in Tehran begins with
numbers 11 to 19 (excluding 12). According to the first three digits of the postal code, the city of


===== PAGE 12 (score 47, numeric-paren 30, table-words 4, var-words 3) =====


.
It can be argued that the increase in housing prices is a combined effect of metro station inaugura-
tion and change in usage of houses around metro station. We believe this is not the case in the city of
Tehran as the stations opened in the period of our analysis are located in regions that have a persistent
structure in type of usage of buildings and this concern cannot significantly affect our results.
Table 4: Regression results for average effect of metro station
(1)
(2)
(3)
(4)
(5)
T=1
0.5941∗∗∗
-0.0123∗∗∗
-0.0172∗∗∗
-0.0202∗∗∗
-0.0248∗∗∗
(0.0015)
(0.0013)
(0.0012)
(0.0012)
(0.0015)
M=1
-0.0457∗∗∗
0.0122∗∗∗
0.0180∗∗∗
0.0228∗∗∗
0.0139∗∗∗
(0.0037)
(0.0022)
(0.0021)
(0.0021)
(0.0021)
T=1 * M=1
0.0105∗
0.0236∗∗∗
0.0247∗∗∗
0.0226∗∗∗
0.0371∗∗∗
(0.0053)
(0.0031)
(0.0030)
(0.0029)
(0.0030)
Age (centuries)
-1.1762∗∗∗
-2.5029∗∗∗
-2.5031∗∗∗
(0.0046)
(0.0125)
(0.0125)
Area (Hectares)
16.9128∗∗∗
10.2480∗∗∗
10.7196∗∗∗
(0.1210)
(0.3469)
(0.3442)
Age2 (centuries)
4.1766∗∗∗
4.1717∗∗∗
(0.0354)
(0.0352)
Area2 (Hectares)
1.279 × 102∗∗∗
1.143 × 102∗∗∗
(12.3798)
(12.2858)
Constant
0.8337∗∗∗
-0.0478
-0.0261
0.0407
-0.0441
(0.0012)
(0.0421)
(0.0401)
(0.0417)
(0.2509)
Region
No
Yes
Yes
Yes
Yes
uarter
No
Yes
Yes
Yes
Yes
Frame
No
No
No
Yes
Yes
N
833,282
833,282
833,282
833,281
833,281
R-squared
.1683
.7155
.742
.7466
.7524
Standard errors in parentheses
∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001
Comments: Regression number (1) has no control variable and regression number (5) uses all
existing control variables and fixed effects. As specified in the table, by adding control variables,
the value of r 2 goes to 75.2%. e post, treat variables are defined as the time and place dummy
variables.
5.2
Inauguration of the Metro Station for the Second Time
To test the second hypothesis, those stations that have been opened for the second time during the
study are examined separately. e last column of Table 7 considers only those four stations and shows
that the coefficient is not statistically different from zero. ese results indicate that those stations


===== PAGE 13 (score 41, numeric-paren 24, table-words 4, var-words 4) =====

Metro station inauguration, housing prices, and transportation accessibility: Tehran case study

opened for the second time due to connection of a new line, do not have a significant impact on the
prices of the properties near them. In other words, these stations have had their impact on the housing
prices aer their initial opening.
We also estimate the primary regression without the four re-opened stations in Table 7. e co-
efficients of the case of considering all 45 stations together and the case of removing of four stations
openedforthesecondtimecanbecompared. Asitcanbeseenincluding/excludingthefourre-opened
stations does not affect the statistical significance of the coefficients. ⁴
Table 5: Regression results to test the effect of opening the station for the
second time
(1)
(2)
(3)
45 Stations
41 Stations
4 Stations
T=1
-0.0248∗∗∗
-0.0233∗∗∗
-0.0243∗∗∗
(0.0015)
(0.0017)
(0.0072)
M=1
0.0139∗∗∗
0.0164∗∗∗
-0.0178∗∗
(0.0021)
(0.0023)
(0.0058)
T=1 * M=1
0.0371∗∗∗
0.0388∗∗∗
-0.0005
(0.0030)
(0.0032)
(0.0083)
Age (centuries)
-2.5031∗∗∗
-2.5244∗∗∗
-2.5746∗∗∗
(0.0125)
(0.0135)
(0.0314)
Area (Hectares)
10.7196∗∗∗
13.8140∗∗∗
-17.5511∗∗∗
(0.3442)
(0.3653)
(1.1811)
Age2 (centuries)
4.1717∗∗∗
4.2409∗∗∗
4.1220∗∗∗
(0.0352)
(0.0381)
(0.0908)
Area2 (Hectares)
1.143 × 102∗∗∗
3.212 × 10∗
9.903 × 102∗∗∗
(12.2858)
(12.8276)
(48.9493)
Constant
-0.0441
-0.0564
0.3082
(0.2509)
(0.2533)
(0.1760)
N
833,281
725,519
107,762
R-squared
.7524
.7542
.7396
Standard errors in parentheses
∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001
Comments: In regression number (1), all stations are considered and
in regression number (2) 4 stations that have been opened for the
second time have been removed. e (post × treat) factor in these
two situations is roughly equal, indicating no appreciable effect on
housing prices at the time of the second opening of the station
⁴ Notice that the diff-in-diffcoefficient (Pos t ×T reat) in the two cases (3.71% and 3.88%) are economically close, but
statistically different, so we make no claim about the equivalence of columns 1 and 2 in Table (5).


===== PAGE 14 (score 14, numeric-paren 0, table-words 1, var-words 12) =====


.
5.3
Clustering Stations
In this section, we divide stations based on their urban differences, and test the third hypothesis. We
will explain each of these cases below.
- North (B1-B3 and C1-C2): Stations located in the North of Tehran are in this category. One
of the characteristics of this category is that household income in these regions is higher than
the average of the community and less public transport than the city center.
- South (B19-B23): Stations located in the south of Tehran are in this category. Regarding the
weak economic conditions in the south of Tehran, it is expected that the property prices near
these metro stations will improve.
- West (A1-A4): e common characteristic of these stations is that they are all on the western
margins of Tehran, and the metro is one of the most important vehicles for transportation of
the residents of this region. erefore, it is anticipated that the opening of the metro station in
this region will have a positive impact on the price of housing around it.
- Sayyad (B4-B10): Due to the lack of a BRT line and the stations of other metro lines, it is
expected that the opening of metro stations in this region will have a positive impact on the
property prices around it.
- Azadi–Enqelab(A5-A14): Again,regardingthepresenceoftheBRTline,whichwasoperated
before2010, itisexpectedthattheopeningofmetrostationsonthesestreetswillnothavemuch
impact on the property prices around it.
- Vali Asr (B11-B18): Regarding the existence of the BRT line, which was implemented before
2010, it is expected that the opening of metro stations on this street will not have much effect
on the property prices around it.
- Navvab (D1-D4): Regarding the presence of the BRT line in this street, which was operated
before 2010, it is predictable that the opening of metro stations on these streets will not have a
positive effect regarding public transport access for the properties around it. It should be noted
that these stations belong to line 7 of the Tehran Metro, which have recently been operated
in 2018. Based on the literature reviewed earlier, we conjecture that the negative effects, e.g.,
disturbance, noise, crime rates, and pollution have contributed to the negative coefficient.
Results of the regression analysis for each of these categories are reported in Table (6) which indi-
cates that those stations opened in regions with access to bus rapid transit (BRT) lines did not have a
significant effect on the prices of the properties around it. We interpret this finding as follows: Advan-
tage of having access to the metro station that causes the prices of the region’s properties to be raised
aer the opening of the metro station is more and faster access to the rest of the city. erefore, prop-
erties with good access to BRT lines and already easy access to city center, we find less than 2 percent
effect on their prices. Categories 5, 6, and 7 have the same status.
Unlike the categories described in the previous paragraph, the properties located in Tehran’s pe-
ripheral regions (north, northeast, west, and south) had a growth of prices from 2 to 11% at the time of
opening of these stations. e most significant price increase is identified for the properties around the
stations in the north of Tehran that includes Tajrish and Mahallati stations. ese stations are in more
affluent part of the city and while there is good access to well developed highways, there is relatively
less public transportation in these regions, therefore the opening of the metro station in these regions
cause the price of the surrounding properties to increase. e least increase is found for properties
around the Sayyad highway (No.4).


===== PAGE 15 (score 76, numeric-paren 56, table-words 2, var-words 5) =====

Metro station inauguration, housing prices, and transportation accessibility: Tehran case study

Table 6: Regression results for 7 defined categories
(1)
(2)
(3)
(4)
(5)
(6)
(7)
North
South
West
Sayyad
Azadi-Enghelab
Valiasr
Navvab
T=1
0.0323∗∗∗
0.0248∗
-0.0326∗∗∗
0.0105
-0.0101∗∗
0.0753∗∗∗
0.0996∗∗∗
(0.0079)
(0.0100)
(0.0059)
(0.0082)
(0.0035)
(0.0056)
(0.0129)
M=1
-0.1403∗∗∗
-0.1193∗∗∗
-0.0323∗∗
-0.0819∗∗∗
0.0569∗∗∗
0.0516∗∗∗
0.0071∗
(0.0392)
(0.0094)
(0.0101)
(0.0075)
(0.0062)
(0.0052)
(0.0029)
T=1 * M=1
0.1138∗∗
0.0688∗∗∗
0.0337∗∗
0.0010
0.0175∗∗
0.0128
-0.0137∗
(0.0415)
(0.0133)
(0.0124)
(0.0122)
(0.0067)
(0.0088)
(0.0064)
Age (centuries)
-3.2597∗∗∗
-2.3733∗∗∗
-3.1817∗∗∗
-2.7542∗∗∗
-2.6335∗∗∗
-2.6704∗∗∗
-2.6744∗∗∗
(0.0485)
(0.0724)
(0.0397)
(0.0420)
(0.0195)
(0.0408)
(0.0364)
Area (Hectares)
24.7873∗∗∗
-1.5277
-30.7669∗∗∗
23.5485∗∗∗
17.7836∗∗∗
7.4146∗∗∗
-8.8337∗∗∗
(1.0445)
(2.8774)
(1.3572)
(1.0859)
(0.7445)
(1.2790)
(1.6689)
Age2 (centuries)
6.8146∗∗∗
4.7623∗∗∗
9.3918∗∗∗
5.1907∗∗∗
4.1099∗∗∗
4.1474∗∗∗
3.9768∗∗∗
(0.1437)
(0.2565)
(0.1322)
(0.1278)
(0.0532)
(0.1043)
(0.1124)
Area2 (Hectares)
-2.19 × 102∗∗∗
2.62 × 102
14.73 × 102∗∗∗
-1.61 × 102∗∗∗
-1.43 × 102∗∗∗
-4.32 × 10
2.81 × 10
(29.6552)
(161.2154)
(56.9606)
(36.8792)
(34.1970)
(47.4865)
(97.8459)
Constant
0.2933
0.6732∗
0.2830
1.0138∗∗∗
-1.2945∗∗∗
0.0475
0.6529∗∗
(0.3666)
(0.3161)
(0.1889)
(0.2605)
(0.2191)
(0.2724)
(0.2147)
N
80,423
55,045
88,720
71,263
284,352
87,913
81,984
R-squared
.641
.5066
.7257
.7436
.7222
.7417
.7136
Comments: Columns 1-4 are the categories without initial bus rapid transit (BRT) lines in these regions. On
the other hand, columns number 5-7 are located near BRT lines, initally operational before metro stations are
opened.


===== PAGE 16 (score 5, numeric-paren 0, table-words 0, var-words 5) =====


.
6
Conclusion
In this paper, using a unique administrative data of transacted housing properties in Tehran, capital of
Iran, we investigate the effect of metro station inauguration on prices of proximate properties consid-
ering the proximity of other public transportation such as Bus Rapid Transit (BRT) to those Metro
stations. In summary, we find that opening of the metro station in Tehran, on average, significantly
has increased the price of housing in the regions around them. Studying the impact of all stations aer
their opening shows a 3.7% increase in property prices at a radius of 400 meters from the metro station
compared with the property at a radius of 400 meters to 3 kilometers. is finding confirms earlier
studies that the advantage of metro station due to ease of access to other city regions overcomes the
possible disadvantages of crowding and pollution, confirming the results of Chen et al. (1998).
Using diversity of various city regions, we find that the lower the initial quality and volume of
public transport near the opened metro station is, the more profound the positive effect on the price of
surrounding properties will be. Our study finds the price increase of 2 to 11% of the nearby properties
with less ex ante access to public modes of transportation. However, if these stations open in regions
with an extensive public transportation system such as BRT lines, then there will be no significant
positive impact on housing prices in those regions or even negative impact.
It is worth noting that it usually takes time for the construction of metro station from the date that
the construction is publicized to the actual date of opening. erefore, in practice, prices can adjust to
the news of metro inauguration before the inauguration takes place. It implies that by comparison of
prices before and aer inauguration, we underestimate the effect of metro stations. Hence we interpret
our findings as minimum effect of metro stations on the house prices. Nevertheless, the result of this
research shows that metro operation in Tehran generally results in at least a 3.7% increase in housing
prices. Finally, further research is needed to study dynamics of house price adjustment which was
beyond scope of this research.
7
Appendix
7.1
Calculate Distance and latitude-longitude
e geographic coordinate system is a system through which the location of each point on the earth
can be determined using several components. In the coordinates, one component represents the verti-
cal location or the height of the desired point, and the second component determines the horizontal
location. An ordinary coordinate system has three components, height (elevation), latitude, and lon-
gitude. Latitude (ϕ) of each point is the northern or southern angle between the equatorial plane and
the line passing through the point and the center of the earth. Points with the same latitude constitute
the circuits, the circles parallel to the Equator that their radiuses vary from the largest (the Equator) to
the smallest (in poles). e longitude (λ) of a point represents the eastern or western angle between
the meridian on which the point is located and the prime meridian, whose longitude is zero degree.
is angle is at most 180◦E (East) or W (West).
e city of Tehran expands between 51◦24′E and 51◦36′E longitude with an approximate length
of50kmandbetween35◦34′E and35◦50′E latitudewithapproximately30kmwidth. egeographic
coordinates of each point as degrees, minutes, and seconds can be converted to the decimal degrees by
the following formula:
Decimal Degrees = Degrees + ( Minutes /60) + ( Seconds /3600)
For example:
35◦40′20′′ = 35 +
40
60 +
20
3600 = 35.672222


===== PAGE 18 (score 47, numeric-paren 30, table-words 4, var-words 3) =====


.
Table 7: Regression results in different ranges of treatment group
(1)
(2)
(3)
(4)
(5)
200m
400m
600m
800m
1000m
T=1 * M1=1
0.0484∗∗∗
(0.0062)
Age (centuries)
-2.4984∗∗∗
-2.5031∗∗∗
-2.5052∗∗∗
-2.5119∗∗∗
-2.5083∗∗∗
(0.0125)
(0.0125)
(0.0125)
(0.0125)
(0.0125)
Area (Hectar2)
10.8345∗∗∗
10.7197∗∗∗
10.6780∗∗∗
10.6476∗∗∗
10.8049∗∗∗
(0.3444)
(0.3443)
(0.3442)
(0.3441)
(0.3441)
Age2 (centuries)
4.1609∗∗∗
4.1717∗∗∗
4.1743∗∗∗
4.1881∗∗∗
4.1844∗∗∗
(0.0352)
(0.0352)
(0.0352)
(0.0351)
(0.0352)
Area2 (Hectar2)
110.7304∗∗∗
114.3933∗∗∗
116.0682∗∗∗
116.9601∗∗∗
113.1338∗∗∗
(12.2888)
(12.2858)
(12.2818)
(12.2776)
(12.2798)
T=1 * M2=1
0.0371∗∗∗
(0.0030)
T=1 * M3=1
0.0271∗∗∗
(0.0022)
T=1 * M4=1
0.0136∗∗∗
(0.0019)
T=1 * M5=1
0.0111∗∗∗
(0.0019)
Constant
-0.0388
-0.0441
-0.0575
-0.0693
-0.0659
(0.2510)
(0.2509)
(0.2509)
(0.2508)
(0.2508)
N
833,281
833,281
833,281
833,281
833,281
R-squared
.7522
.7524
.7526
.7527
.7526
Standard errors in parentheses
∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001
Comments: As shown in the table, the post * treat coefficient is increased by shrinking the
range.
price appears more significantly. More specifically, if the range under control is considered up to 2km,
the effect of the metro opening will be 4.1%, while if we increase the range up to 6km, the impact of
the metro opening will be approximately 3%. As said before, the basic case considered in this research
is up to 3km.
7.5
Robustness check - Stepwise-Linear-Quadratic Model
Table(9)reportstheresultsofmodelsweconsiderhere. Intheprimarymodel,thedistancewasdivided
into two categories coded by 0 (400m to 3km) and 1 (up to 400m). Here, for robustness check we
consider the distance as a linear and a quadratic function, and only transactions within the range of
1kmofthemetrostationareconsidered, thenumberof transactionsis 305,468 inthis case. Also, a case


===== PAGE 19 (score 49, numeric-paren 30, table-words 4, var-words 5) =====

Metro station inauguration, housing prices, and transportation accessibility: Tehran case study

Table 8: Regression results in different ranges of control group
(1)
(2)
(3)
(4)
(5)
2km
3km
4km
5km
6km
T=1 * M1=1
0.0413∗∗∗
(0.0030)
Age (centuries)
-2.5158∗∗∗
-2.5031∗∗∗
-2.4259∗∗∗
-2.3996∗∗∗
-2.3846∗∗∗
(0.0143)
(0.0125)
(0.0117)
(0.0112)
(0.0108)
Area (Hectares)
7.7151∗∗∗
10.7196∗∗∗
12.3332∗∗∗
12.2424∗∗∗
12.4509∗∗∗
(0.3964)
(0.3442)
(0.3236)
(0.3108)
(0.3007)
Age2 (centuries)
4.1837∗∗∗
4.1717∗∗∗
4.0274∗∗∗
4.0404∗∗∗
4.0831∗∗∗
(0.0402)
(0.0352)
(0.0333)
(0.0321)
(0.0312)
Area2 (Hectares)
2.21 × 102∗∗∗
11.14 × 102∗∗∗
6.84 × 10∗∗∗
9 × 10∗∗∗
1.028 × 102∗∗∗
(14.2462)
(12.2858)
(11.6416)
(11.2525)
(10.9528)
T=1 * M2=1
0.0371∗∗∗
(0.0030)
T=1 * M3=1
0.0349∗∗∗
(0.0030)
T=1 * M4=1
0.0342∗∗∗
(0.0030)
T=1 * M5=1
0.0307∗∗∗
(0.0030)
Constant
-0.0341
-0.0441
-0.1212
-0.1341
-0.1510
(0.2495)
(0.2509)
(0.2539)
(0.2534)
(0.2528)
N
628,782
833,281
977,048
1,077,506
1,169,264
R-squared
.7538
.7524
.7478
.7481
.7478
Standard errors in parentheses
∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001
Comments: As shown in the table, the post * treat coefficient is increased by shrinking the range.
with a stepwise function is considered. In this case, we consider distances to four categories 0 to 150m,
150 to 400m and 400m to 1km and 1 to 3km. In the case with stepwise function, the coefficient of
the variable of the difference-in-difference in the first category, which includes properties in the range
of up to 150 meters from the metro station, is 4.6 percent, which is smaller for the other categories,
so that it is negative for the last category (400m to 1 kilometers). In the linear and quadratic case, the
coefficient of the difference-in-difference variable of degree one is -4.4%, meaning that, up to 1km, for
every 100 meters away from the metro station, 0.44% of the price of the property is reduced due to the
opening of the metro station. As summarized in the Table (9), the coefficient for the term of degree
one is equal to -14%.


===== PAGE 20 (score 29, numeric-paren 11, table-words 6, var-words 4) =====


.
Table 9: Comparison of linear, quadratic and stepwise models
(1)
(2)
(3)
(4)
Basic
Stepwise
Linear
uadratic
T=1 * M=1
0.0371∗∗∗
0.0460∗∗∗
(0.0030)
(0.0096)
T=1 * M=2
0.0362∗∗∗
(0.0032)
T=1 * M=3
0.0056∗∗
(0.0020)
T=1 * M
-0.0442∗∗∗
-0.1434∗∗∗
(0.0055)
(0.0276)
T=1 * M2
0.0895∗∗∗
(0.0230)
Constant
-0.0441
-0.0667
0.2523∗∗∗
0.1934∗∗
(0.2509)
(0.2508)
(0.0679)
(0.0681)
N
833,281
833,281
305,468
305,468
R-squared
.7524
.7527
.7365
.7367
Standard errors in parentheses
∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001
Comments: Stepwise, linear and quadratic modes are compared
with the initial state. In stepwise mode, the treatment group is
divided into 3 groups of less than 150 meters, between 150 and
500 meters and 1000 meters. In linear and quadratic mode, the
range is 1 km.
7.6
Robustness check - ln(Age), ln(Area) and categorized age and area
In Table (10) logarithm of variables Age and Area are used instead of their levels. Moreover, in Table
(11) we used 10 decile fixed effects for variables Age and Area. As can be seen, controlled for fixed
effects, the effect of metro station inauguration changes slightly with model specification, but this
change is negligible regarding its amount. erefore, we conclude that our results are robust to model
specification.


===== PAGE 21 (score 41, numeric-paren 28, table-words 2, var-words 4) =====

Metro station inauguration, housing prices, and transportation accessibility: Tehran case study

Table 10: Regression results for log form of variables
(1)
(2)
(3)
(4)
(5)
T=1
0.5941∗∗∗
0.6266∗∗∗
0.0505∗∗∗
-0.0194∗∗∗
-0.0265∗∗∗
(0.0015)
(0.0014)
(0.0014)
(0.0013)
(0.0016)
M=1
-0.0457∗∗∗
0.0513∗∗∗
-0.0413∗∗∗
0.0211∗∗∗
0.0127∗∗∗
(0.0037)
(0.0032)
(0.0027)
(0.0022)
(0.0022)
T=1 * M=1
0.0105∗
0.0157∗∗∗
-0.0063
0.0222∗∗∗
0.0371∗∗∗
(0.0053)
(0.0045)
(0.0039)
(0.0031)
(0.0032)
lnAge
-0.1259∗∗∗
-0.0785∗∗∗
-0.1069∗∗∗
-0.1049∗∗∗
(0.0006)
(0.0005)
(0.0004)
(0.0005)
lnArea
0.0883∗∗∗
0.5667∗∗∗
0.1081∗∗∗
0.1076∗∗∗
(0.0019)
(0.0013)
(0.0013)
(0.0013)
Constant
0.8337∗∗∗
0.3309∗∗∗
-2.0560∗∗∗
-0.2625∗∗∗
-0.4028
(0.0012)
(0.0119)
(0.0507)
(0.0408)
(0.2553)
Region
No
Yes
No
Yes
Yes
uarter
No
No
Yes
Yes
Yes
Region*uarter
No
No
No
No
Yes
Frame
No
No
No
No
Yes
N
833,282
759,057
759,057
759,057
759,056
R-squared
.1683
.4435
.5708
.7296
.736
Standard errors in parentheses
∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001
lnArea and lnAge are exerted instead of Area and Age.


===== PAGE 22 (score 29, numeric-paren 20, table-words 2, var-words 1) =====


.
Table 11: Regression results for categorized form of variables
(1)
(2)
(3)
(4)
(5)
T=1
0.5941∗∗∗
0.5921∗∗∗
0.0522∗∗∗
-0.0198∗∗∗
-0.0244∗∗∗
(0.0015)
(0.0014)
(0.0013)
(0.0012)
(0.0015)
M=1
-0.0457∗∗∗
0.0504∗∗∗
-0.0419∗∗∗
0.0227∗∗∗
0.0141∗∗∗
(0.0037)
(0.0030)
(0.0026)
(0.0021)
(0.0021)
T=1 * M=1
0.0105∗
0.0210∗∗∗
-0.0051
0.0238∗∗∗
0.0380∗∗∗
(0.0053)
(0.0042)
(0.0037)
(0.0029)
(0.0030)
Age Deciles FE
No
Yes
Yes
Yes
Yes
Area Deciles FE
No
Yes
Yes
Yes
Yes
Constant
0.8337∗∗∗
0.8589∗∗∗
0.1502∗∗
0.1756∗∗∗
0.0386
(0.0012)
(0.0086)
(0.0503)
(0.0398)
(0.2510)
Region
No
Yes
No
Yes
Yes
uarter
No
No
Yes
Yes
Yes
Region*uarter
No
No
No
No
Yes
Frame
No
No
No
No
Yes
N
833,282
833,282
833,282
833,282
833,281
R-squared
.1683
.4667
.5759
.7444
.7504
Standard errors in parentheses
∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001
Values for Age and Area are categorized in 10 deciles. First column report re-
gression results for the whole sample without any control and any fixed effect. In
columns 2-5, deciles of age and area are-fixed effects are used to control any nonlin-
ear effect of these variables.


===== PAGE 23 (score 17, numeric-paren 0, table-words 3, var-words 4) =====

Metro station inauguration, housing prices, and transportation accessibility: Tehran case study

Table 12: Summary statistics of variables comparison
2010
2011
2012
2013
2014
mean
sd
mean
sd
mean
sd
mean
sd
mean
sd
Area
84.24
38.28
84.82
38.66
83.22
38.20
83.18
38.66
82.31
37.54
Price (Mill. Tomans)
1.62
.86
1.87
.89
2.57
1.25
3.73
2.03
4.01
2.20
Price (Dollars)
1480.46
785.57
1041.09
499.02
704.50
343.93
1036.44
566.40
1135.68
620.19
Age
11.41
8.19
10.97
8.51
9.59
8.32
8.05
8.31
7.19
8.19
Observations
85964
133167
168097
111369
143292
2015
2016
2017
2018
2019
mean
sd
mean
sd
mean
sd
mean
sd
mean
sd
Area
83.56
37.59
86.04
37.07
88.22
37.35
87.67
40.48
85.94
40.09
Price (Mill. Tomans)
4.07
2.29
4.31
2.24
4.85
2.46
7.19
4.25
12.32
6.76
Price (Dollars)
1131.91
637.83
1151.65
599.18
1143.08
579.34
719.56
425.54
147.39
80.97
Age
7.15
8.17
7.74
8.40
8.57
8.81
9.22
9.07
9.64
9.01
Observations
131341
146927
162445
131727
53137


===== PAGE 24 (score 12, numeric-paren 0, table-words 0, var-words 12) =====


.
References
Badoe, D. A. and E. J. Miller. 2000. Transportation–land-use interaction: Empirical findings in North
America, and their implications for modeling. Transportation Research Part D: Transport and En-
vironment, 5(4):235–263.
Bajic, V. 1983. e effects of a new subway line on housing prices in metropolitan Toronto. Urban
Studies, 20(2):147–158.
Baker, C. 1983. Tracking Washington’s Metro. American Demographics, 5(11):30–35.
Billings, S. B. 2011. Estimating the value of a new transit option. Regional Science and Urban Eco-
nomics, 41(6):525–536.
Bohman, H. and D. Nilsson. 2016. e impact of regional commuter trains on property values: Price
segments and income. Journal of Transport Geography, 56:102–109.
Bowes, D. R. and K. R. Ihlanfeldt. 2001. Identifying the impacts of rail transit stations on residential
property values. Journal of Urban Economics, 50(1):1–25.
Bryan, G., E. Glaeser, and N. Tsivanidis. 2019. Cities in the developing world. Technical report,
National Bureau of Economic Research.
Chen, H., A. Rufolo, and K. J. Dueker. 1998. Measuring the impact of light rail systems on single-
family home values: A hedonic approach with Geographic Information System application. Trans-
portation Research Record, 1617(1):38–43.
Chica-Olmo, J., R. Cano-Guervos, and I. Tamaris-Turizo. 2019. Determination of buffer zone for
negative externalities: Effect on housing prices. e Geographical Journal, 185(2):222–236.
Coffman, C. and M. E. Gregson. 1998. Railroad development and land value. e Journal of Real
Estate Finance and Economics, 16(2):191–204.
Debrezion, G., E. Pels, and P. Rietveld. 2011. e impact of rail transport on real estate prices: An
empirical analysis of the Dutch housing market. Urban Studies, 48(5):997–1015.
Diao, M. 2015. Selectivity, spatial autocorrelation and the valuation of transit accessibility. Urban
Studies, 52(1):159–177.
Diao, M., D. Leonard, and T. F. Sing. 2017. Spatial-difference-in-differences models for impact of new
mass rapid transit line on private housing values. Regional Science and Urban Economics, 67:64–77.
Diao, M., Y.Qin, andT.F.Sing.2016. Negativeexternalitiesofrailnoiseandhousingvalues: Evidence
from the cessation of railway operations in Singapore. Real Estate Economics, 44(4):878–917.
Forouhar, A. and M. Hasankhani. 2018. e effect of Tehran metro rail system on residential prop-
erty values: A comparative analysis between high-income and low-income neighbourhoods. Urban
Studies, 55(16):3503–3524.
Forrest, D., J. Glen, and R. Ward. 1996. e impact of a light rail system on the structure of house
prices: A hedonic longitudinal study. Journal of Transport Economics and Policy, pp. 15–29.
Giuliano, G. 2004. Land use impacts of transportation investments. e Geography of Urban Trans-
portation, 3:237–273.
Hess, D. B. and T. M. Almeida. 2007. Impact of proximity to light rail rapid transit on station-area
property values in Buffalo, New York. Urban Studies, 44(5-6):1041–1068.
Hu, L., S. He, Z. Han, H. Xiao, S. Su, M. Weng, and Z. Cai. 2019. Monitoring housing rental prices
based on social media: An integrated approach of machine-learning algorithms and hedonic mod-
eling to inform equitable housing policies. Land Use Policy, 82:657–673.
McMillen, D.P.andJ.McDonald.2004. Reactionofhousepricestoanewrapidtransitline: Chicago’s
Midway line, 1983–1999. Real Estate Economics, 32(3):463–486.
Mohammad, S. I., D. J. Graham, and P. C. Melo. 2015. e effect of the Dubai Metro on the value of
residential and commercial properties. Journal of Transport and Land Use, 10(1).
Murat Celik, H. and U. Yankaya. 2006.
e impact of rail transit investment on the residential
property values in developing countries: e case of Izmir subway, Turkey. Property Management,
24(4):369–382.