1# RECORD 461f8c8aa68c6a2b5157bc03f6d689991ee3ac642# TITLE: Estimating the Effects of Light Rail Transit (LRT) System on the Property Values in the Klang Valley, Malaysia: A Hedonic House Price Approach3# YEAR: 2013.0 VENUE: nan DOI: 10.11113/jt.v61.16204# AUTHORS: M. Dziauddin; S. Alvanides; N. Powe5# PAGES: 13 CANDIDATE TABLE PAGES (1-based): [8, 9]6# ABSTRACT: This study investigates the increased value of land in the form of house prices as a result of improved accessibility owing to the construction of Light Rail Transit (LRT) systems. Kelana Jaya Line LRT system is chosen as the case study in this research. Hedonic house price modelling is employed to estimate the effects of the LRT system on the prices of the house that are located within the radius of two kilometres from the Kelana Jaya LRT stations. Selling prices, structural attributes, land use and socio-economic attributes were collected from the database of Department of Valuation and Services of Malaysia, selected maps and reports. Fifty-five factors that are likely to influence a house price were identified and used to estimate the overall effects of the LRT system on it. However, only significant variables were included in the final deliberation and these were identified by using correlation analysis and modified step-wise procedures. The outcome of this study shows a positive relationship between the existence of the LRT system and house prices. In other words, properties that are located within close proximity to the LRT station are valued more than properties that are located further away.789===== PAGE 1 (score 11, numeric-paren 0, table-words 0, var-words 11) =====1011 1261:1 (2013) 35–47 | www.jurnalteknologi.utm.my | eISSN 2180–3722 | ISSN 0127–9696 13 14Full paper 15Jurnal 16Teknologi 17Estimating the Effects of Light Rail Transit (LRT) System on the Property 18Values in the Klang Valley, Malaysia: A Hedonic House Price Approach 19 20Mohd Faris Dziauddina*, Seraphim Alvanidesb, Neil Powec 21 22aDepartment of Geography and Environment, Faculty of Human Sciences, Universiti Pendidikan Sultan Idris, 35900 Tanjong Malim, Perak, Malaysia 23bFaculty of Humanities and Social Sciences, Newcastle University, Newcastle upon Tyne, NE1 7RU, United Kingdom 24cSchool of Architecture Planning and Landscape, Claremont Tower Newcastle University Newcastle upon Tyne NE1 7RU, United Kingdom 25 26*Corresponding author: fdziauddin@yahoo.com 27 28 29Article history 30Received :19 September 2012 31Received in revised form :4 32December 2012 33Accepted :12 February 2013 34Graphical abstract 35 36Abstract 37This study investigates the increased value of land in the form of house prices as a result of improved 38accessibility owing to the construction of Light Rail Transit (LRT) systems. Kelana Jaya Line LRT system 39is chosen as the case study in this research. Hedonic house price modelling is employed to estimate the 40effects of the LRT system on the prices of the house that are located within the radius of two kilometres 41from the Kelana Jaya LRT stations. Selling prices, structural attributes, land use and socio-economic 42attributes were collected from the database of Department of Valuation and Services of Malaysia, selected 43maps and reports. Fifty-five factors that are likely to influence a house price were identified and used to 44estimate the overall effects of the LRT system on it. However, only significant variables were included in 45the final deliberation and these were identified by using correlation analysis and modified step-wise 46procedures. The outcome of this study shows a positive relationship between the existence of the LRT 47system and house prices. In other words, properties that are located within close proximity to the LRT 48station are valued more than properties that are located further away. 49Keywords: Light rail transit systems (LRT); property values; hedonic price model; Klang Valley 50Abstrak 51Kajian ini menyiasat peningkatan nilai tanah dalam bentuk harga rumah hasil daripada peningkatan 52ketersampaian menerusi pembinaan sistem Transit Aliran Ringan (TAR). Sistem TAR laluan Kelana Jaya 53dipilih sebagai kajian kes dalam penyelidikan ini. Model harga rumah hedonik digunakan untuk 54menganggarkan kesan sistem TAR pada harga rumah yang terletak dalam lingkungan dua kilometer dari 55stesen TAR Kelana Jaya. Harga jualan, ciri-ciri fizikal rumah, lokasi dan sosio-ekonomi dikumpul 56daripada pangkalan data Jabatan Penilaian dan Perkhidmatan Malaysia, peta terpilih dan laporan. Lima 57puluh lima faktor yang mungkin mempengaruhi harga rumah telah dikenal pasti dan digunakan bagi 58menganggarkan kesan keseluruhan sistem TAR di atasnya. Walau bagaimanapun, hanya pemboleh ubah 59yang signifikan secara statistik telah dimasukkan dalam perbincangan akhir dan ini telah dikenal pasti 60dengan menggunakan analisis korelasi dan prosedur ‘step-wise’ yang telah diubah suai. Hasil kajian ini 61menunjukkan terdapat hubungan positif antara kewujudan sistem TAR dan harga rumah. Dalam erti kata 62lain, hartanah yang terletak dalam jarak yang hampir dengan stesen TAR mempunyai nilai yang lebih 63tinggi berbanding hartanah yang terletak jauh. 64Kata kunci: Sistem transit aliran ringan (TAR); nilai hartanah; model harga hedonik; Lembah Klang 65© 2012 Penerbit UTM Press. All rights reserved.66 67 68 691.0 INTRODUCTION 70 71Rail transit systems, namely heavy and light transit systems are 72a public good and have been seen as serving a number of 73purposes and producing a number of public benefits, particularly 74to those areas that have been served by high service quality rail 75transit systems. These public benefits can be categorised into 76two; direct and indirect benefits. 77The direct benefits of rail transit systems are defined in terms of 78improved regional mobility, consumer savings, vehicle cost 79savings, energy conservation, improved mobility for non-drivers 80and disadvantaged groups, congestion reduction, roadway cost 81savings, increased traffic safety, and pollution emission 82reductions (see, for example, Litman, 2003, 2004a, 2004b, 2007, 832012; Banister and Banister, 1995; Knowles, 1996; Pucher, 842004). 85 Alongside these direct benefits, the provision of high 86service quality of public transportation such as rail transit 87systems has also potentially influenced local land use and 88increased local property values (indirect benefits), particularly 899091===== PAGE 8 (score 49, numeric-paren 0, table-words 12, var-words 18) =====929342 Mohd Faris Dziauddin, Seraphim Alvanides & Neil Powe / Jurnal Teknologi (Sciences & Engineering) 61:1 (2013) 35–47 94 95 96α10LnSECONDARYSCHi + α11LnPRIMARYSCHi + 97α12LnPARKi 98+ 99α13LnRECREATIONi 100+ 101α14LnHOSPITALi 102+ 103α15LnINDUSTRYi 104+ 105α16LnMALAYi + εi 106 107 108 (4) 109 110 where the actual-distance to a station is given by 111NETDIST: the network-distance from the property to an LRT 112station measured in metres. The results from the hedonic house 113price models are presented in a standard format. This standard 114format is shown in Table 4. Note that the results from straight-115line-distance and network-distance were separately displayed. 116As can be seen in Table 4, the results from the hedonic house 117price models include five attributes which represent the 118predictor, coefficient, standard error, t-value and VIF. The 119predictor is the variable that has been used to measure the house 120prices and it has included focused, structural, locational and 121socio-economic variables. The second column provides 122coefficients which represent the prices of each predictor used in 123the model. The standard errors of the coefficients are presented 124in the third column. The fourth column gives us the results of a 125t-value of the predictor. The greater t-value of the predictors 126implies that the greater its function in determining a house price. 127The multicollinearity level of the predictors used in the final 128model is shown in the final column, namely the variance 129inflation factor (VIF). Meanwhile, the adjusted R2 for the 130number of variables in the model is shown below each model. 131 132Table 2 Definition of variables and data sources 133 134Table 3 Characteristics of dependent and independent variables 135 136Vector 137Predictor 138Mean 139Standard deviation 140Minimum 141Maximum 142Dependent variable 143House price transactions (P) 144 145Independent variables 146Focus variables (R) 147 148Structural variables (S) 149 150 151 152 153Locational variables (L) 154 155 156 157 158 159 160 161 162 163 164 165Socio-economic 166and 167ethnic 168variables (N) 169 170LOGSELLING 171 172 173 174 175LOGSTRDIST 176LOGNETDIST 177LOGTIMESAVINGS 178LOGFLRAREA 179LOGBEDS 180TYPTRRD 181TYPSEMID 182TYPDETCH 183TYPCONDO 184LOGCBD 185LOGPRIMARYSCH 186LOGSECONDARYSCH 187LOGCOMMERCIAL 188LOGPARK 189LOGHOSPITAL 190LOGRECREATION 191LOGINDUSTRY 192LOGMALAY 193 19412.49 195 196 197 198 1996.73 2007.15 2012.15 2026.17 2031.07 2040.46 2050.01 2060.12 2070.19 2089.08 2096.65 2106.65 2116.62 2126.94 2137.87 2148.67 2157.19 2163.73 217 2180.59 219 220 221 222 2230.64 2240.65 2251.14 2260.68 2270.26 2280.49 2290.12 2300.32 2310.39 2320.29 2330.82 2340.64 2350.99 2362.27 2370.89 2380.49 2390.55 2400.24 241 24210.82 243 244 245 246 2474.33 2484.33 249-4.07 2504.79 2510.00 2520.00 2530.00 2540.00 2550.00 2567.91 2571.45 2582.73 2592.27 2600.52 2614.25 2627.09 2634.29 2643.46 265 26614.43 267 268 269 270 2717.64 2728.08 2733.38 2748.54 2751.79 2761.00 2771.00 2781.00 2791.00 2809.43 2817.85 2827.93 2837.94 2849.04 2859.04 2869.44 2877.94 2884.49 289Vector 290Predictor 291Operational definition 292Units 293Data source 294Dependent variable 295House price transactions 296(P) 297 298Independent variables 299Focus variables (R) 300 301Structural variables (S) 302 303 304 305 306 307 308Locational variables (L) 309 310 311 312 313 314 315 316Socio-economic variables 317(N) 318 319SELLING 320 321 322 323STRDIST 324NETDIST 325TIMESAVINGS 326FLRAREA 327BEDS 328TYPTRRD 329TYPSEMID 330TYPDETCH 331TYPCONDO 332CBD 333PRIMARYSCH 334SECONDARYSCH 335COMMERCIAL 336PARK 337HOSPITAL 338RECREATION 339 340INDUSTRY 341MALAY 342 343House price transactions 344 345 346 347Straight-line-distance 348Network-distance 349Travel time savings to CBD 350Floor area 351Number of bedrooms 352Terraced house 353Semi-detached house 354Detached house 355Condominium 356Proximity to CBD 357Proximity to primary schools 358Proximity to secondary schools 359Proximity to commercial areas 360Proximity to parks 361Proximity to hospitals 362Proximity to recreational areas 363Proximity to industrial areas 364Proportion of Malays 365 366Malaysia Ringgits (MYR) 367 368 369 370Metre 371Metre 372Minutes 373Square foot 374Number 375Dummy (0 or 1) 376Dummy (0 or 1) 377Dummy (0 or 1) 378Dummy (0 or 1) 379Metre 380Metre 381Metre 382Metre 383Metre 384Metre 385Metre 386Metre 387Proportion of Malays (for 388each Mukim) 389 390Department of Valuation 391and Property Services of 392Malaysia (DVPA) 393Calculated using GIS 394Calculated using GIS 395Calculated using GIS 396DVPA 397DVPA 398DVPA 399DVPA 400DVPA 401DVPA 402Calculated using GIS 403Calculated using GIS 404Calculated using GIS 405Calculated using GIS 406Calculated using GIS 407Calculated using GIS 408Calculated using GIS 409Calculated using GIS 410 411Malaysia Census 2000 412413414===== PAGE 9 (score 68, numeric-paren 0, table-words 15, var-words 23) =====41541643 Mohd Faris Dziauddin, Seraphim Alvanides & Neil Powe / Jurnal Teknologi (Sciences & Engineering) 61:1 (2013) 35–47 417 418 419Table 4 The traditional hedonic specification of Model 3 and Model 4 420 421Vector 422Predictor 423Straight-line distance (Model 3) 424Network distance (Model 4) 425Coefficient 426Standard 427error 428t-value 429VIF 430Coefficient 431Standard 432error 433t-value 434VIF 435 436Focus variables 437 438 439Structural 440variables 441 442 443 444 445Locational 446variables 447 448 449 450 451 452 453Socio-economic 454and 455ethnic 456variables 457Constant 458STRDIST 459NETDIST 460TIMESAVINGS 461FLRAREA 462BEDS 463TYPTRRD 464TYPSEMID 465TYPDETCH 466TYPCONDO 467CBD 468PRIMARYSCH 469SECONDARYSCH 470COMMERCIAL 471PARK 472HOSPITAL 473RECREATION 474INDUSTRY 475MALAY 47635667.67 477-10.56 478 4792019.25 480310.63 48131485.34 48271160.14 483139696.87 484173145.42 485184622.86 486-17.96 48726.25 488-11.11 489-10.13 490-1.13 4913.40 492-4.81 49328.22 4943673.45 4951.74 4964.95 497 498296.16 49913.09 5003922.05 5019509.88 50222626.95 50317052.20 50410165.74 5051.52 5063.94 5074.66 5083.28 5090.31 5101.29 5111.70 5123.72 513613.52 51418.85 515-2.16 516 5176.82 51824.07 5198.01 5207.44 5216.19 52210.16 52318.23 524-11.75 5256.75 526-2.40 527-3.17 528-3.33 5292.64 530-2.84 5317.74 5325.99 533 5341.85 535 5363.27 5376.23 5381.87 5394.33 5401.40 5416.02 5423.02 5433.18 5441.93 5451.41 5462.41 5471.24 5483.27 5495.87 5501.70 5517.61 55243739.2 553 554-6.61 5552046.17 556312.65 55731594.29 55872143.93 559138713.09 560170849.93 561184950.79 562-18.61 56326.57 564-11.11 565-9.54 566-1.13 5673.58 568-4.65 56926.91 5703520.07 571 5721.76 573 5743.03 575296.15 57613.09 5773922.04 5789509.88 57922626.95 58017052.20 58110165.73 5821.55 5833.93 5844.66 5853.28 5860.31 5871.28 5881.70 5893.71 590605.85 59118.89 592 593-2.31 5946.87 59523.99 5968.06 5977.61 5986.15 59910.02 60018.25 601-11.92 6026.79 603-2.36 604-2.94 605-3.36 6062.73 607-2.76 6087.23 6095.83 610 611 6121.70 6133.30 6146.35 6151.87 6164.29 6171.40 6186.03 6193.03 6203.33 6211.93 6221.41 6232.50 6241.24 6253.32 6265.80 6271.80 6287.37 629 630 631R2 (adj.) = 78.2 per cent 632R2 (adj.) = 78.2 per cent 633 634 6355.0 RESULTS: HEDONIC PRICE MODELS 636 637Table 4 presents the summary of the parameter estimates of 638Models 3 and 4 – the basic model for straight-line-distance and 639network-distance in estimating the effect of the LRT system on 640house prices. To reduce the complication of the interpretation 641process, continuous independent variables are deviated around 642their means. In other words, the models are estimated with 643respect to the average property size of 651.35 square feet. The 644results of both models show that most of predictor variables that 645have been used to estimate the LRT-house prices relationship 646produce correct signs (positive or negative) as theoretically 647expected, except for primary school and significance level of 6480.01 and 0.05. In terms of the R2 statistic, the model explains the 649variation of the house price within a two-kilometre radius from 650an LRT station in the Klang Valley reasonably well for both 651straight-line and network-distance model with 78.2 per cent. 652The result of hedonic model also shows that the parameter 653estimates of several independent variables are found to be 654slightly different between these two models. 655 656(1) Value of the LRT system; An examination of Table 4 shows 657that the parameter estimates of focus variables are found to 658be statistically significant with the correct sign. Evidently, 659it can be seen that the house price decreases as we move 660further away from the LRT station – for every metre away 661from an LRT station, the value of a residential property 662decreases by MYR10.56 (straight-line-distance model) and 663by MYR6.61 (network-distance model). This implies that a 664residential property located anywhere within 1,000 metres 665of an LRT station would generally be valued at an average 666rate between MYR10,560 (straight-line-distance model) 667and MYR6,610 (network-distance model) more than a 668residential property located outside of this distance. As 669for the magnitude of effect, the LRT system has 670significantly contributed at -2.16 (t-value of straight-line-671distance) and -2.31 (t-value of network-distance) in 672determining the house price in the study area. It is notable 673that the t-value of the parameter estimate of straight-line-674distance measures is slightly higher than that of network-675distance measures. 676(2) Value of structural attributes; The result of the estimation 677indicates that the most influential factor in determining the 678house price is floor area. For every square-foot increase in 679the floor area, the house price increases by an average of 680around MYR310.63 for straight-line-distance model. 681Similarly, an increase around MYR312.65 is deduced for 682network-distance model. The greater magnitude of the 683effect of floor area was expected since floor areas are 684always associated with the size of the property – this is 685consistent with the most of the hedonic house price 686literature. In the case of number of bedrooms, a potential 687buyer has to pay on average MYR31,000 for one additional 688bedroom of a property for both models. As for property-689type attribute, its role is only to indicate the price for 690different types of housing in the study area. As to be 691expected, the price for a detached or semi-detached house 692would be higher than a terraced house. After examining the 693results for both models, the conclusion that can be made is; 694there are significant differences in price between different 695types of housing. In other words, implicit prices of 696detached (MYR170,000), semi-detached (MYR139,000), 697condominium 698(MYR184,000) 699and 700terraced 701houses 702(MYR71,00) for both models should all have reflected 703some value-added by the attributes that they possess. In the 704case a condominium property, the explanation that could be 705given is; since a condominium property has various 706facilities such as swimming pool, sauna, club house, sport 707facilities, landscape and security, it can be expected that a 708condominium property has a positive effect on house price. 709The same thing can be expected from a detached property, 710since it stands exclusively on its own and normally it 711comes with a bigger plot of land with lower density – these 712attributes will definitely contribute to the positive effect on 713house price. 714715716===== PAGE 10 (score 15, numeric-paren 0, table-words 0, var-words 15) =====71771844 Mohd Faris Dziauddin, Seraphim Alvanides & Neil Powe / Jurnal Teknologi (Sciences & Engineering) 61:1 (2013) 35–47 719 720 721(3) Value of locational attributes; The results of the estimation 722have also confirmed the importance of locational attributes 723in determining house prices. Distance from the CBD is 724significant with the correct sign of the estimated parameter 725for both models. The model suggests that a rent gradient 726for the Klang Valley of around MYR17.96 per metre for 727the straight-line-distance model and MYR18.61 per metre 728for the network-distance model. This can be interpreted as 729a distance decay relationship between land rent and the 730distance from the CBD. For every metre away from the 731CBD (Kuala Lumpur city centre), the property value 732decreases around MYR17.96 and MYR18.61 for straight-733line 734and 735network-distance 736respectively. 737The 738other 739locational attributes that also show significant contribution 740to the house price is the proximity to commercial areas. 741The interpretation that can be made is that for every metre 742away from the commercial areas, house price decreases at 743the rate of MYR10.15 (straight-line-distance model) and 744MYR9.54 (network-distance model). The parameter 745estimates for proximity to a secondary school also shows 746statistical significance with the anticipated sign. The 747implicit price for proximity to secondary schools suggests 748that for every metre away from secondary schools, house 749price would generally decrease by about MYR11 for both 750models. With regards to the proximity to primary schools, 751the result is statistically significant, however, with 752unexpected signs. The model suggests that for every metre 753away from primary schools, there is a degree of increment 754in housing value. It clearly shows that on average, for 755every metre away from primary schools, the house price 756experiences an increase by about MYR26 for both straight-757line and network-distance models. This supports the study 758on the impact of school on house prices carried out by 759Cheshire and Sheppard (2004). The logical reason for the 760decreasing house value by being located too close to the 761primary schools is due to negative externalities such as 762noise and traffic congestion that can be associated with the 763existence of primary schools. The other possible reason is 764that house price would normally respond positively 765(increase) when located in the neighbourhood where the 766school is regarded to perform well in major examinations 767(in practice most of the major examinations take place in 768secondary schools). In other words, the quality of schools 769assessed by their performance in major examinations is 770found to be more important than just being located near to 771an ordinary school. Thus, it is reasonable to expect house 772prices to increase for every metre away from primary 773schools after considering the explanation that has been 774given above. Proximity to parks is also found to be 775statistically significant in determining house prices with the 776expected signs. The parameter estimates for proximity to 777parks is MYR1.13 per metre (rent gradient) for both 778models. Similarly, proximity to recreational areas is also 779found to be significant in determining house prices. For the 780proximity to recreational areas, the parameter estimates 781indicate that for every metre away from recreational areas, 782the house price reduces at the rate of MYR4 for both 783models. In terms of proximity to health centres that is 784hospitals, the parameter estimates suggest that house price 785increases by about MYR3 for straight-line and network-786distance models of every metre away from this amenity. 787Continuing the observation of the study, the price-distance 788function for proximity to industrial areas shows positive 789signs. The estimated parameter indicates that there is an 790increase in house price at the rate of MYR28.22 (straight-791line-distance model) and MYR26.91 (network-distance 792model) for every metre away from industrial areas. The 793rationale behind this observation is that being located 794adjacent to industrial areas suggests that residential 795property is prone to suffer from traffic congestion and air 796and noise pollution. Additionally, the norm of industry 797workers to populate surrounding areas close to factories for 798example, result in social problems and hence, brings the 799perceived value of properties in the area down. Therefore, 800it is very understandable at least in the context of this study 801(where house price increases for every metre away from 802industrial areas) since people tend to avoid negative 803externalities caused by the industry, the associated 804community and its activities. 805(4) Neighbourhood variables; The only significant factor of 806socio-economic and ethnic attributes in determining house 807prices in this study is the percentage of Malay ethnic 808(Malay proportion). The model suggests that if a property 809is located in the areas with a higher percentage of Malay 810ethnics, it would result in an increase in housing price with 811an average of MYR3,600 (straight-line-distance model) 812and 813MYR3,500 814(network-distance 815model). 816The 817contribution of the Malay ethnic towards higher house 818prices in these areas can be reasoned out by examining the 819history of these affected areas. Unlike today’s well spread-820out city of Kuala Lumpur or CBD of the Klang Valley, 821over 40 years ago these areas were considered to be 822suburban areas for the Klang Valley. Those days, these 823areas were considered to be affordable among the ethnic 824Malays while the actual downtown of the city centre was 825dominated by high-income dwellers. However, due to the 826rapid urbanisation in the Klang Valley these areas that were 827once considered as suburban areas have inevitably become 828an integral part of the city of Kuala Lumpur that are well 829connected by the roads and highways with no distinction of 830differences. These newly integrated demographically 831Malay dominated areas have quickly emerged as attractive 832housing estates among middle and high income Malays. 833The house price difference can easily be explained through 834the gain of premium value for the mentioned areas. 835Consequently, observation can be made today that the more 836affluent second and third generation of the Malays would 837have the preference to purchase a house in these Malay 838dominated areas since they are willing to pay a higher price 839in order to be located in these areas with perceived 840advantages of becoming an integral part of the city and 841belonging to a community of the same race that they are 842familiar with. 843 844 8456.0 CONCLUSIONS AND POLICY IMPLICATIONS 846 847The study in the Klang Valley was carried out in order to 848provide empirical evidence for the estimation of the negative 849rent gradient with respect to the trade-off between the 850improvement of accessibility to the CBD and house prices. 851Hedonic house price models was employed – the role of hedonic 852house price models is to provide empirical evidence of a 853negative land rent gradient for a unit of house from an LRT 854station. In other words, it has been assumed that house prices 855would decrease for every metre further away from an LRT 856station. 857 The estimating of the effects of the LRT system on house 858prices using hedonic house price models in this study reveals a 859number of key findings. Firstly, the hedonic house price models 860estimate that houses located within two kilometres of an LRT 861station in the Klang Valley decrease in price as the distance 862