1# RECORD c15ba8a0a6017224eaf4694459bf9621790746622# TITLE: A Study of the Metro and Housing Value Interaction with Hedonic Price Model: Comparison of Ankara Batıkent and Koru Metro Stations3# YEAR: 2023.0 VENUE: Mimarlık Bilimleri ve Uygulamaları Dergisi (MBUD) DOI: 10.30785/mbud.12848434# AUTHORS: Fatma Erdoğanaras; Kübra Cihangir Çamur; Gülsel Satoğlu5# PAGES: 17 CANDIDATE TABLE PAGES (1-based): [10, 11, 12]6# ABSTRACT: The effects of urban rail systems on the real estate market were studied by analysing housing prices around Ankara Koru and Batıkent Metro Stations using the Hedonic Price Model. Quantitative data on residential sales were used. The findings reveal that sale values of residences are more elastic in term of metro effect in Batıkent, where middle-lower income groups live, compared to Koru Metro Station, where middle-upper income groups live. As the distance to the metro station decreases in Batıkent, prices increase. These results suggest that subways can help reduce polarization between income groups in cities while increasing accessibility. Providing access to neighbourhoods with both low and high-income individuals by metro increases housing prices more in favour of low-income groups and reduces income polarization between income groups.789===== PAGE 1 (score 10, numeric-paren 0, table-words 0, var-words 10) =====1011Journal of Architectural Sciences and Applications 12JASA 2023, 8 (1), 161-177. 13Research article 14e-ISSN: 2548-0170 15https://dergipark.org.tr/en/pub/mbud 16 17Citation: Erdoğanaras, F., Cihangir Çamur, K. & Satoğlu, G. (2023). A study of the metro and housing value 18interaction with Hedonic Price Model: Comparison of Ankara Batıkent and Koru Metro Stations. Journal of 19Architectural Sciences and Applications, 8 (1), 161-177. 20DOI: https://doi.org/10.30785/mbud.1284843 21 22 23 24Received: 19/04/2023 – Accepted: 26/05/2023 161 25A Study of the Metro and Housing Value Interaction with Hedonic 26Price Model: Comparison of Ankara Batıkent and Koru Metro 27Stations 28Fatma ERDOĞANARAS 1*29, Kübra CİHANGİR ÇAMUR 230 Gülsel SATOĞLU 331 32ORCID 1: 0000-0002-0892-8431 33ORCID 2: 0000-0003-0343-3306 34ORCID 3: 0000-0003-4601-2034 351,2 Gazi University, Faculty of Architecture, Department of City and Regional Planning, 06570, Ankara, Türkiye. 363 Master's Degree, Gazi University, Faculty of Architecture, Department of City and Regional Planning, 06570, 37Ankara, Türkiye. 38* e-mail: fatmaerdoganaras1862@gmail.com 39Abstract 40The effects of urban rail systems on the real estate market were studied by analysing housing prices around 41Ankara Koru and Batıkent Metro Stations using the Hedonic Price Model. Quantitative data on residential sales 42were used. The findings reveal that sale values of residences are more elastic in term of metro effect in Batıkent, 43where middle-lower income groups live, compared to Koru Metro Station, where middle-upper income groups 44live. As the distance to the metro station decreases in Batıkent, prices increase. These results suggest that 45subways can help reduce polarization between income groups in cities while increasing accessibility. Providing 46access to neighbourhoods with both low and high-income individuals by metro increases housing prices more in 47favour of low-income groups and reduces income polarization between income groups. 48Keywords: Metro stations, housing prices, hedonic price model, accessibility, transportation planning, urban 49planning. 50Metro ve Konut Değeri Etkileşiminin Hedonik Fiyat Modeli ile 51İncelenmesi: Ankara Batıkent ve Koru Metro İstasyonlarının 52Karşılaştırılması 53Öz 54Kentsel raylı sistemlerin emlak piyasası üzerindeki etkileri, Hedonik Fiyat Modeli kullanılarak Ankara Koru ve 55Batıkent Metro istasyonları çevresindeki konut fiyatları analiz edilerek incelenmiştir. Konut satış değerlerine 56ilişkin nicel veriler kullanılmıştır. Bulgular, orta-alt gelir grubunun yaşadığı Batıkent'teki konutların satış 57değerlerinin, orta-üst gelir grubunun yaşadığı Koru Metro İstasyonu'na kıyasla metro etkisi açısından daha 58esnek olduğunu ortaya koymaktadır. Batıkent'te, metro istasyonuna yaklaştıkça fiyatlar artmaktadır. Bu 59sonuçlar, metroların erişilebilirliği artırırken şehirlerdeki gelir grupları arasındaki kutuplaşmayı azaltmaya 60yardımcı olabileceğini göstermektedir. Hem düşük hem de yüksek gelirli bireylerin yaşadığı mahallelere metro ile 61erişim sağlanması, konut fiyatlarını düşük gelirli gruplar lehine artırmakta ve gelir grupları arasındaki 62kutuplaşmayı azaltmaktadır. 63Anahtar kelimeler: Metro istasyonları, konut fiyatları, hedonik fiyat modeli, erişilebilirlik, ulaşım planlaması. 646566===== PAGE 10 (score 27, numeric-paren 10, table-words 4, var-words 8) =====6768Journal of Architectural Sciences and Applications, 2023, 8 (1), 161-177. 69 70170 71 72Table 2. Correlation relationships of Batıkent Metro Station for housing sale prices 73 74Distance 75to Metro 76Station 77(m) 78Site 79/Apartment 80 81Price 82(TL) 83Square 84Meter 85 (m2) 86Number 87of 88Rooms 89Floor 90Level 91 92Buildin93g Age 94(year) 95Distance 96to Bus 97Stop 98 (m) 99Distance 100to 101School 102 (m) 103Distance to Metro 104Station (m) 1051 106.142 107-.162 108-.161 109-.330 110.020 111.133 112.532*113* 114-.160 115Site/Apartment 116.142 1171 118.296 119-.108 120-.046 121.226 122.274 123.156 124.041 125Price (TL) 126-.162 127.296 1281 129.609** 130.421* 131.007 132-.345 133.173 134-.103 135Square meter(m2) 136-.161 137-.108 138.609** 1391 140.689** 141-.307 142-.582** 143-.111 144-.274 145Number of Rooms 146-.330 147-.046 148.421* 149.689** 1501 151-.067 152-.188 153-.260 154-.080 155Floor Level 156.020 157.226 158.007 159-.307 160-.067 1611 162.363 163-.233 164.251 165Building Age 166 (year) 167.133 168.274 169-.345 170-.582** 171-.188 172.363 1731 174.134 175-.194 176Distance to Bus 177Stop (m) 178.532** 179.156 180.173 181-.111 182-.260 183-.233 184.134 1851 186-.239 187Distance to School 188(m) 189-.160 190.041 191-.103 192-.274 193-.080 194.251 195-.194 196-.239 1971 198 199According to correlation matrix (Table 2) there is a very strong positive relation between the 200variables price and square meter (0.609), a strong positive relation between price and the number of 201rooms/sections (0.421), positive relation between price and whether it is located in a site or an 202apartment building (0.296), price and the distance to the bus stop (0.173) and a negative relation 203between the variables price and the age of the building (0.345), price and the distance to the metro 204station (0.162), price and the distance to the school (0.103). According to the results of the Hedonic 205Regression Model, the variables of distance to the school, floor level, and square meter size, location 206in a site or apartment building, distance to the metro stop, distance to the bus stop, age, and number 207of rooms explain 53.5% of the variation in the housing price. For the One-Way Analysis of Variance 208(ANOVA) to determine whether multiple independent variables are significant or not in the created 209model, the significant value of the F-test is 0.005, which is less than 0.05. Therefore, a statistically 210significant relationship between the variables in the regression model has been determined. 211The significant value (B) of the T-test is greater than 0.05 for variables such as location in a 212site/apartment building, housing size, number of rooms, floor level, and housing age, indicating that 213there is no statistically significant relationship between these variables and the housing price. 214However, the significant value (B) of the T-test is less than 0.05 for the variables of distance to the 215metro station and distance to the bus stop, indicating that there is a statistically significant 216relationship between these variables and the housing price. 217The table shows the correlation coefficients between the price and different variables of residential 218properties located near the Batıkent Metro Station. The variables include square meter size, number 219of rooms, building age, distance to the bus stop, distance to the metro station, distance to the 220school, floor level, and whether the property is located in a site or apartment building. 221The correlation coefficients range from -1 to 1, where a value of -1 indicates a perfect negative 222correlation, 0 indicates no correlation, and 1 indicates a perfect positive correlation. The table shows 223that the variables with the strongest positive correlation to price are the square meter size (0.609), 224number of rooms (0.421), and whether the property is located in a site or apartment building (0.296). 225The variables with the strongest negative correlation to price are age (-0.345) and distance to the 226metro station (-0.162). 227228229===== PAGE 11 (score 23, numeric-paren 7, table-words 4, var-words 6) =====230231Journal of Architectural Sciences and Applications, 2023, 8 (1), 161-177. 232 233171 234 235Table 3. Independent variable relationships of Batıkent Metro Station for housing sale prices 236Coefficients 237Model 238Unstandardized 239Coefficients 240Standardized 241Coefficients 242T 243Sig. 244Collinearity Statistics 245B 246Std. 247Error 248Beta 249 250 251Tolerance 252VIF 2531 254(Constant) 25513.205 256.394 257 25833.482 259.000 260 261 262Distance to Metro 263Station (m) 264.000 265.000 266-.379 267-2.152 268.047 269.651 2701.537 271Site/ 272Apartment 273.140 274.072 275.297 2761.945 277.070 278.867 2791.153 280Square Meter (m2) 281.008 282.004 283.556 2842.099 285.052 286.288 2873.467 288Number of Rooms 289.018 290.099 291.042 292.184 293.856 294.389 2952.571 296Floor 297.021 298.010 299.338 3002.015 301.061 302.720 3031.390 304Building Age (years 305) 306-.050 307.043 308-.232 309-1.153 310.266 311.501 3121.996 313Distance to Bus 314Stop (m) 315.002 316.001 317.520 3182.854 319.011 320.609 3211.643 3224.2. The Effect of Koru Metro Station on Housing Sale Prices 323The distance of the house for sale to the metro station is an independent variable that affects the 324housing price in fourth place. As the distance increases, there is an increase in the sale price of the 325house, so the relationship between distance and price is proportional. 326Table 4. Correlation relationship for Koru Station for housing sale prices 327Correlations 328Pearson 329Correlation 330Coefficient 331DISTANCE to 332METRO 333STATION(m) 334 SITE/ 335APARTMENT 336PRICE 337(TL) 338 SQUARE 339METER (m2) 340NUMBER of 341ROOMS 342FLO343OR 344BUILDING 345AGE (yıl) 346DISTANCE 347to BUS 348STOP (m) 349DISTANCE to 350SCHOOL(m) 351Distance to 352Metro 353Station (m) 3541 355.274 356.351 357.298 358.146 359.263 360-.368 361.284 362-.077 363Site/ 364Apartment 365.274 3661 367-.037 368.023 369.258 370.066 371.000 372-.218 373-.281 374Price (Tl) 375.351 376-.037 3771 378.937** 379.695** 380-381.027 382-.563* 383.242 384-.143 385Square 386Meter 387.298 388.023 389.937*390* 3911 392.684** 393-394.077 395-.536* 396.381 397-.140 398Number of 399Rooms 400.146 401.258 402.695*403* 404.684** 4051 406-407.280 408-.409 409-.072 410-.165 411Floor 412.263 413.066 414-.027 415-.077 416-.280 4171 418-.440 419.168 420.046 421Building Age 422(Years) 423-.368 424.000 425-426.563* 427-.536* 428-.409 429-430.440 4311 432-.319 433.268 434Distance to 435Bus Stop (M) 436.284 437-.218 438.242 439.381 440-.072 441.168 442-.319 4431 444-.174 445Distance to 446School (m) 447-.077 448-.281 449-.143 450-.140 451-.165 452.046 453.268 454-.174 4551 456According to correlation matrix (Table 4) there is a very strong positive relation between the 457variables price and square meter (0.937), price and the number of rooms/sections (0.695), positive 458relation between price and distance to metro stop (0.351) and distance to bus stop (0.242) and a 459negative relation between price and building age (0.563), whether it is located in a site or an 460apartment building (0.037) and the floor (0.027). 461According to the summary of the Hedonic Regression Model, variables such as distance from the 462school, the floor it is located on, square meter size, whether it is located in a site/apartment, distance 463to the metro station, distance to the bus stop, age, and number of rooms explain 93.2% of the price 464variable of the houses for sale in Korukent. For one-way analysis of variance (ANOVA) to determine 465whether a model with multiple independent variables is significant or not, the null (H0) and 466alternative (H1) hypotheses are tested. Since the F-test's significant value is less than 0.05, it has 467been determined that the regression model is significant and there is a statistically significant 468469470===== PAGE 12 (score 23, numeric-paren 2, table-words 3, var-words 13) =====471472Journal of Architectural Sciences and Applications, 2023, 8 (1), 161-177. 473 474172 475 476relationship between the variables. The significant value (B) of the t-test is greater than 0.05, 477indicating that there is no statistically significant relationship between the distance of the house to 478the metro, whether it is located in a site/apartment, the age of the house, the distance to the bus 479stop or school, and the price of the house. However, since the significant value (B) of the t-test is less 480than 0.05, there is a statistically significant relationship between the square meter, the number of 481rooms, and the floor number of the house and the price of the house. When the square meter value 482of the house increases by one unit, the price of the house increases by 0.004 units, when the number 483of rooms increases by one unit, the price of the house increases by 0.344 units, and when the floor 484number of the house increases by one unit, there is a change of 0.028 units in the price. The Table 5 485shows the correlation coefficients between the price and different variables of residential properties 486located near the Koru Metro Station. The variables include square meter size, number of rooms, 487building age, distance to the bus stop, distance to the metro station, distance to the school, floor 488level, and whether the property is located in a site or apartment building. The correlation coefficients 489of the model range from -1 to 1, where a value of -1 indicates a perfect negative correlation, 0 490indicates no correlation, and 1 indicates a perfect positive correlation. The table shows that the 491variables with the strongest positive correlation to price are the square meter size (0.677), number of 492rooms (0.414), and the floor (0,291). The variable distance school has a negative correlation (-0.120). 493Table 5. Independent variable relationships of Koru Metro Station for housing sale prices 494Coefficients 495 496Unstandardized 497Coefficients 498 499Standardized 500Coefficients 501T 502Sig. 503 504 505B 506Std. Error 507Beta 508 509 510VIF 511Constant 51213.629 513.380 514 51535.882 516.000 517 518Distance to Metro 519Station(M) 520-1.345E-5 521.000 522-.007 523-.090 524.930 5251.439 526Site/ Apartment 527-.168 528.164 529-.084 530-1.027 531.331 5321.646 533Square Meter (M2) 534.004 535.001 536.677 5376.015 538.000 5393.155 540Number of Rooms 541.344 542.099 543.414 5443.484 545.007 5463.508 547Floor 548.028 549.009 550.291 5513.105 552.013 5532.179 554Building Age (Years) 555.029 556.055 557.057 558.526 559.612 5602.963 561Distance to Bus Stop 562(M) 563.000 564.000 565.027 566.319 567.757 5681.777 569Distance to School(M) 570-.001 571.000 572-.120 573-1.613 574.141 5751.374 576 5774.3 Evaluation of the Housing Sale Prices around the Batıkent and Koru Metro Stations Compared 578to the Cases 579The residential areas around the Batıkent Metro Station hosts both developing and established 580housing areas subject to planning implementations related to middle and low-income households. 581On the other hand, the residential areas around the Koru Metro Station is characterized by low-rise 582housing units as well as high-rise apartments for upper-middle and high-income households. The 583hedonic regression analysis of the housing sale prices presents the differences and similarities of the 584both locations, Batıkent and Koru, in terms of variables affecting the housing sale prices. The most 585important variables affecting the housing sale prices in Batıkent case are the size of the house, the 586distance to bus stop, the number of the rooms, the age of the building, whether it is located in a site 587or an apartment building and distance to metro station. On the other hand it is seen for the Koru 588case that only the variables related to housing characteristics are affecting the housing sale prices 589are; the size of the house, the number of the rooms, the floor number and distance to school. Some 590of the findings of the two cases investigated in this article are parallel to the cases found in the 591literature. In a middle-low-income residential area (Batıkent), the distance to the bus stop and metro 592station affects the housing sale prices, whereas in a high-income residential area (Koru), there is no 593such effect. Studies in the literature indicate that rail system investments affect housing prices, and 594the accessibility to metro stations also contributes positively to the increase in housing prices, just as 595the proximity to the city center does (Bajic, 1983; Cervero, 2003; Pan & Zhang, 2008). 596597598===== PAGE 13 (score 14, numeric-paren 0, table-words 0, var-words 14) =====599600Journal of Architectural Sciences and Applications, 2023, 8 (1), 161-177. 601 602173 603 604Research conducted in Istanbul shows that being close to a metro line has a positive effect on the 605surrounding housing prices (Şahin, 2019). Similarly, another study examining the effect of Istanbul 606metro stations on housing prices found that the closer the distance to the station, the higher the 607housing prices (Cengiz, 2020). 608In a thesis study that examined the effects of metro stations and metro lines on the housing sub-609market, it was found that there is no significant difference in the prices of the housing around the 610station areas and metro lines within 250 meters. However, it was determined that the values of the 611for-sale and rental housing around the station areas were higher than those in areas with similar 612proximity to the metro line (Alas, 2021). Another study found that proximity to transportation points 613such as public transportation stations positively affect housing prices, in addition to variables such as 614size, age, floor number, garden view, sea view, etc. In this study, it was found that housing prices are 615high around public transportation stations (Choy et al., 2007). Another study conducted in Istanbul 616states that the housing prices around the station areas increase in areas where housing prices are 617already high and similar results were obtained for rental values (Alkay, 2011). However, there are 618also studies that demonstrate the opposite of this general trend, such as the effect of metro stations 619decreasing housing prices in areas where high-income groups live (Nelson, 1999). 6205. Conclusion and Suggestions 621In the relevant literature, it is emphasized that metro stations within walking distance cause an 622increase in housing prices and accessibility is an important criterion in people's transportation mode 623choices. The results of the Hedonic Price Model applied in this study reveal that the impact of 624Batıkent and Koru Metro Stations, with different socio-economic characteristics, on housing sales 625prices varies depending on accessibility within the 1500-meter radius defined as a walkable distance. 626Batıkent with a mixed-use land pattern and is inhabited by the middle-lower income group, as the 627distance of the housing to the station decreases, the sale prices of the houses increase. In this case, it 628can be said that users' lower income levels make them more sensitive to travel costs and less 629expected to own a car. On the other hand, in Koru, where high-income group lives, the increase in 630housing sales prices within the 1500-meter radius of the Koru Metro Station as one moves away from 631the station indicates that housing-related features such as the size, number of rooms, and number of 632floors are more important in determining housing prices. This situation is also supported by the fact 633that high-income individuals use private cars more when compared to public transportation and their 634time is more valuable. 635In other words, the relationship between accessibility and housing sales values is lower near Koru 636Metro Station, where users with a high socio-economic level, while it is higher near Batıkent Metro 637Station, where users with a low socio-economic level. This situation shows that metro transportation 638systems are effective in reducing polarization between different socio-economic groups in the city. In 639the future, conducting similar studies in high and lower socio-economic housing areas using the same 640method and applying this model at regular intervals will contribute to monitoring socio-economic 641problems such as polarization in the city and overcoming these problems with transportation 642policies. Additionally, this study can be expanded to include rental housing prices around metro 643stations and compare sales and rental values. Similarly, sales and rental values in areas located along 644the metro line and those far away from it can be compared, taking into account socio-economic 645differences, to obtain more general results. 646Acknowledgements and Information Note 647We thank the anonymous reviewers and our colleagues for their comments and guidance on the 648development of the article. The article complies with national and international research and 649publication ethics. Ethics Committee approval was not required for the study. 650Author Contribution and Conflict of Interest Declaration Information 651All authors have contributed equally. The authors declared no potential conflicts of interest with 652respect to the research, authorship, and/or publication of this article. 653