1# RECORD 37a309783a11809263623703621db3a26e46597f2# TITLE: The Effect of Central Business District on House Prices in Chengdu Metropolitan Area: A Hedonic Approach3# YEAR: 2015.0 VENUE: nan DOI: 10.2991/cas-15.2015.834# AUTHORS: Gao-Lu Zou5# PAGES: 4 CANDIDATE TABLE PAGES (1-based): [2, 3]6# ABSTRACT: Gaolu Zou Chengdu Economic and Development Institute of Conventions and Exhibitions, Chengdu University, Chengdu 610106, Sichuan Province, China Abstract—This paper aims to analyse the effects of the CBD and several selected housing features on house prices in Chengdu. Sampling is focused on ordinary old housing units. Data were collected from metropolitan Chengdu, China. Utilizing log-linear hedonic techniques, the study suggests a small negative price gradient from the CBD and a marginal and nonlinear effect of housing age and underground railway locations on house prices. Therefore, the study supports the classic inference for spatial house prices. Additionally, we suggest higher property tax rates for large-floor space units.789===== PAGE 1 (score 29, numeric-paren 0, table-words 2, var-words 25) =====1011The Effect of Central Business District on House Prices in Chengdu Metropolitan 12Area: A Hedonic Approach 13 14Gaolu Zou 15Chengdu Economic and Development Institute of Conventions and Exhibitions, Chengdu University, Chengdu 610106, 16Sichuan Province, China 17 18 19Abstract—This paper aims to analyse the effects of the CBD 20and several selected housing features on house prices in 21Chengdu. Sampling is focused on ordinary old housing units. 22Data were collected from metropolitan Chengdu, China. 23Utilizing log-linear hedonic techniques, the study suggests a 24small negative price gradient from the CBD and a marginal 25and nonlinear effect of housing age and underground railway 26locations on house prices. Therefore, the study supports the 27classic inference for spatial house prices. Additionally, we 28suggest higher property tax rates for large-floor space units. 29Keywords-central business district; housing feature; price, 30hedonic; log-linear; marginal effect. 31I. INTRODUCTION 32Chengdu metropolis displays a spatially concentric 33pattern. Classic urban housing market modeling starts from 34a monocentric city with an only business or employment 35center (the Central Business District known as CBD), e.g. 36[1, 2]. House prices vary inversely with distance or travel 37time to the CBD, holding other amenities constant. This is 38because the housing spatial equilibrium requires a negative 39price gradient. However, empirical evidence for the 40negative price gradient is inconsistent[3]. For a long period, 41the CBD did not affect house prices. However, house values 42now decline by more than 8% per mile (from the CBD)[4]. 43The CBD in Chengdu is situated in Yan Shi Kou nexus. 44Starting from CBD outwards are four ring expressways, i.e., 45Ring Road No. 1, Ring Road No. 2, Ring Road No. 3 and 46Around-the-city Expressway. Renming Nan Lu Road and 47Shu Du Da Dao Avenue are two main radial avenues 48starting from CBD outwards. Thus, nonlinear and concave 49urban property prices may exist in Chengdu metropolitan 50area[5, 6]. House prices have a tendency to change 51marginally with various features such as the distance and 52age[7]. 53Chengdu 54has 55since 562008 57conducted 58large-scale 59development activities on underground railways. The dual 60effects of urban transit tracks on real estate prices are 61suggested[8-10]. Elasticity of private house price relative to 62walking distance is -0.02[11]. Railway stations[12], light 63rails[13], underground stations[5] increased either land or 64house values. However, negative externalities such as noise 65and congestion reduced house prices[8, 14]. 66This paper mainly aims to examine the nonlinear effects 67of the CBD and several selected housing featureson house 68prices in metropolitan Chengdu. We introduced a hedonic 69technique and thus estimated implicit house prices. We 70expect that the paper would contribute to the house price 71gradient from the CBD, property valuation and property tax 72policy. 73II. METHODOLOGY 74Housing properties are composed of various features[14, 7515]. Hedonic methods could discover the prices of these 76housing features[16]. Hedonic prices are often called 77implicit prices because they are not explicitly ‘viewed’ or 78listed on board. [17] suggested the alternative functional 79forms of hedonic models. Past studies on real estate markets 80widely use hedonic techniques, e.g. [14, 18-28]. 81One commonly used hedonic technique is the log-linear 82equation[29-31]. Right inferences could be reached using 83the log-linear form[32]. It is formulated as 84ε85β86α87+88+89+90=91∑92∑93=94=95n96j97j98i99m100i101i102D103c104X105PRICE1061107j1081109ln110ln111 112Where PRICE is the house price. Xi represents the ith 113feature of m housing features. Dj represents the jth feature of 114n housing features. D stands for dummy variable, e.g. CBD 115location and lake view. Ln denotes the natural logarithm. εis 116an error term. βi is constant elasticity given continual 117variable Xi, or implicit price Pi=118i119i120β121X122given integer variable 123Xi. Existence of Dj implies that price change is 124%125100126)1271128(129×130−131j132ce133[33]. Marginal house price changes 134could be suggested using estimates of the coefficients[7]. 135III. 136DATA AND VARIABLE DEFINITION 137We selected 180 ordinary-type older units in Chengdu 138metropolitan area. Sampling area is bounded within the 139Around-the-city Expressway. Suppose that eleven housing 140features impact house price PRICE. For each housing unit, 141the selected 11 features include floor space AREA, number 142of bedrooms BEDROOM, number of halls HALL, number 143of bathrooms BATHROOM, floor level FLOORLEVEL, 144house age YEAR, green coverage rate GREENCOVERAGE, 145decoration DECORATION, distance to CBDDIST_CBD, 146ring 147location 148RING, 149walking 150distance 151to 152nearest 153underground railway station UNDERGROUND. Data 154contain 2160 observations. Sampling spans the period from 155mid-January to mid-April. Table 1 statistically describes the 156raw data. We defined various details for house price and the 157International Conference on Circuits and Systems (CAS 2015) 158© 2015. The authors - Published by Atlantis Press159349160161162===== PAGE 2 (score 61, numeric-paren 1, table-words 22, var-words 11) =====16316411 features as follows. 165House price is call price listed on websites[34, 35]. The 166call price is a supply-side price. Floor space is limited to 50 167to 200 square meters. Distance to CBD is that to Yan Shi 168Kou (city center). The city government has planned that 169Yan Shi Kou is CBD, where business buildings and 170activities become highly concentrated. Ring location 171(dummy variable) includes core area within Ring Road No. 1721 (=1), ring zone 1 between Ring Road No. 1 and Ring 173Road No. 2 (=2), ring zone 2 between Ring Road No. 2 and 174Ring Road No. 3 (=3), and ring zone 3between Ring Road 175No. 3 and Around-the-city Expressway (=4). 176There are two underground railway lines currently available 177in Chengdu. Walking distance to nearest underground 178railway station is divided into five zones and corresponding 179dummy variables: up to 400 meters (below five minutes) 180denoted by 1401 to 800 meters (five to ten minutes) 181denoted by 2801 to 1200 meters (ten to fifteen minutes) 182denoted by 3, 1201 to 1600 meters (fifteen to twenty 183minutes) denoted by 4, and above 1600 meters (above 184twenty minutes) denoted by 5. 185TABLE I. 186STATISTICAL DESCRIPTION FOR THE RAW DATA 187 188Mean 189Median 190Maximum 191Minimum 192Standard. Deviation 193Jarque-Bera 194P-value195PRICE 19697.8 19788 198365 19943 20041.1 2011354.7 2020.00 203AREA(sq. meters) 20499.7 20591.95 206223 20749 20828.1 20987.2 2100.00 211DECORATION 2123.1 2134 2144 2151 2161.3 21730.1 2180.00 219BEDROOM 2202.5 2212 2225 2231 2240.7 2257.3 2260.03 227HALL 2282.0 2292 2303 2311 2320.3 233368.5 2340.00 235BATHROOM 2361.5 2371 2383 2391 2400.6 24120.5 2420.00 243FLOOR LEVEL 2449.8 2457 24633 2471 2488.1 24946.2 2500.00 251YEAR 2528.6 2538 25429 2551 2564.7 25768.6 2580.00 259GREENCOVERAGE(%) 26029.3 26130 26270 2631.7 26412.5 26519.9 2660.00 267RING 2682.6 2693 2704 2711 2721.1 27313.2 2740.00 275DIST_CBD(meters) 2764939.3 2775270 27812867.8 279201 2802661.3 2810.4 2820.81 283UNDERGROUND 2844.1 2855 2865 2871 2881.3 28935.5 2900.00 291 292IV. 293ESTIMATION AND ANALYSIS 294We took a stepwise regression strategy. Preliminary 295regression (Table II) suggested that five out of 11 variables 296were statistically insignificant. However, we found that 297removing RING and HALL from the regression did not 298increase the values of R-squared and adjusted R-squared, 299but led to worsening heteroskedasticity. Thus, we could not 300remove more features than two from the regressions. We 301decided the estimated results in Table III. Fitness appears to 302be satisfactory, figure. 1. 303TABLE II. RESULTS OF PRELIMINARY REGRESSION. 304Variable 305Estimate 306t-statistic 307P-value 308LOG(AREA) 3090.89 3108.35 3110.00 312LOG(DECORATION) 3130.04 3141.59 3150.11 316LOG(BEDROOM) 3170.07 3180.89 3190.37 320LOG(HALL) 3210.07 3220.80 3230.42 324LOG(BATHROOM) 3250.06 3261.10 3270.27 328LOG(FLOORLEVEL) 329-0.01 330-0.84 3310.40 332LOG(YEAR) 333-0.16 334-5.22 3350.00 336LOG(GREENCOVERAGE) 3370.05 3382.79 3390.01 340LOG(RING) 3410.01 3420.12 3430.91 344LOG(DIST_CBD) 345-0.09 346-2.04 3470.04 348LOG(UNDERGROUND) 349-0.08 350-1.90 3510.06 352Error term 3531.31 3542.30 3550.02 356Notes: R-squared=0.72, Adjusted R-squared=0.70, S.E. of regression=0.19, 357F-statistic=39.2. White Heteroskedasticity(P-value)=35.7(0.03). 358 359 360Figure 1. 361Actual and fitted values for house prices 362Estimates are significant for AREA, DECORATION, 363YEAR, 364GREENCOVERAGE, 365DIST_CBD, 366and 367UNDERGROUND. Hence, given other features constant, 368the growth of 1% in distance from the CBD implies that the 369price reduces by about 0.1%. However, a 1% growth in 370floor area implies that the house price grows by 0.92%. A 371growth of 10% in green coverage rate may provoke a rise of 3720.5% in price. 373Given other features constant, for an average one-year 374‘new’ unit, the implicit price is 1(-0.17)=1 yuan RMB/sqm. 375For an average 5-year unit, the implicit price is 5(-0.17)=0.76 376-.8377-.4378.0379.4380.83813.53824.03834.53845.03855.53866.0387253885038975390100391125392150393175394Residual395Actual396Fitted397350398399400===== PAGE 3 (score 47, numeric-paren 1, table-words 14, var-words 15) =====401402yuan RMB/sqm. For an average 10-year unit, the implicit 403price is 10(-0.17)=0.68 yuan RMB/ sqm, and so on. Hence, 404house prices normally decrease marginally with house ages, 405figure. 2. 406TABLE III. RESULTS OF STEPWISE REGRESSIONS 407Variable 408Estimate 409t-statistic 410P-value411LOG(AREA) 4120.92 4139.06 4140.00 415LOG(DECORATION) 4160.05 4171.68 4180.09 419LOG(BEDROOM) 4200.07 4210.94 4220.35 423LOG(BATHROOM) 4240.06 4251.06 4260.29 427LOG(FLOORLEVEL) 428-0.01 429-0.84 4300.40 431LOG(YEAR) 432-0.17 433-5.38 4340.00 435LOG(GREENCOVERAGE) 4360.05 4372.76 4380.01 439LOG(DIST_CBD) 440-0.09 441-3.83 4420.00 443LOG(UNDERGROUND) 444-0.07 445-1.85 4460.07 447Error term 4481.19 4492.60 4500.01 451Notes: R-squared=0.72, Adjusted R-squared=0.70, S.E. of regression=0.19, 452F-statistic=48.2, White Heteroskedasticity(P-value)=34.0(0.01). 453 454Figure 2. 455House prices decreased marginally with age 456Given other features constant, for an average unit, the 457implicit price for the walking distance of 400 to 800 meters 458to nearest ground station, relative to the walking distance of 459within 400 meters, reduces by 4.7% (2(-0.07)-1). The implicit 460price for the walking distance of 801 to 1200 meters to 461station reduces by 7.4%. The implicit price for the walking 462distance of 1201 to 1600 meters to station reduces by 9.2%. 463The implicit price for the walking distance exceeding 1600 464meters to station reduces by 10.7%. Thus, effects of 465walking distance to rail station on house prices are 466nonlinear and marginal, fig.3. 467 468 469 470Figure 3. 471House prices decreased nonlinearly with walking 472distance to nearest underground stationgiven the price in the 473zone within 400 meters (=1) 474V. 475CONCLUSIONS 476It has been argued that house prices vary inversely with 477distance or travel time to the CBD. Various structural and 478environmental features impact housing prices. Few studies 479analyzed the implicit prices in Chengdu. The Chengdu 480metropolis is characterized by a monocentric pattern and 481has rapidly expanded in terms of economy, population and 482built area over the past decade. Two underground railway 483lines are in effect. Five others are in construction. We 484selected ordinary-type older housing units within the 485Around-the-city Expressway. 486The paper employed a hedonic technique and conducted 487regressions using log-linear equations. We find that house 488prices decrease with distance from CBD and walking 489distance to the nearest underground station. The finding 490supports a negative price gradient from the CBD. In 491addition, larger floor space, good decoration and higher 492green coverage increase but age reduces house prices. 493We suggest that the government could impose higher 494property tax rates on units with a larger floor space than 495those with a smaller floor space. 496ACKNOWLEDGMENTS 497A Project Supported by Scientific Research Fund of 498SiChuan Provincial Education Department (2014-2015). 499REFERENCES 500[1] Alonso, W., Location and land use. Cambridge, MA.: Harvard 501University Press, 1964. 502[2] Fujita, M., Spatial patterns of residential development. Journal of 503Urban Economics, 12(1), pp. 22-52, 1982. 504[3] Bender, B. & Hwang H.-S., Hedonic housing price indices and 505secondary employment centers. Journal of Urban Economics, 17(1), 506pp. 90-107, 1985. 507[4] McMillen, D.P., The return of centralization to chicago: Using repeat 508sales to identify changes in house price distance gradients. Regional 509Science and Urban Economics, 33(3), pp. 287-304, 2003. 510[5] Wieser, R. Hedonic prices on viennás urban residential land markets, 511working paper nr.: 2/2006. IFIP-working paper 2/2006: Centre of 512Public Finance and Infrastructure Policy, Vienna University of 513Technology. 2006. 514[6] Colwell, P.F. & Munneke H.J., The structure of urban land prices. 515Journal of Urban Economics, 41(3), pp. 321-336, 1997. 516[7] DiPasquale, D. & Wheaton W.C., Urban economics and real estate 517‐60%518‐40%519‐20%5200%521052210523205243052540526price change527house age (years)528‐4.7%529‐7.4%530‐9.2%531‐10.7%532400m to 533800m534800m to 5351200m5361200m to 5371600m538exceeding 5391600m540351541542543===== PAGE 4 (score 15, numeric-paren 0, table-words 0, var-words 15) =====544545markets. New Jersey, Englewood Cliffs: Prentice-Hall, 1996. 546[8] Simons, R.A. & El Jaouhari A., The effect of freight railroad tracks 547and train activity on residential property values. Appraisal Journal, 54872(3), pp. 223-233, 2004. 549[9] Smith, J.J. & Gihring T.A., Financing transit systems through value 550capture. American Journal of Economics and Sociology, 65(3), pp. 551751-786, 2006. 552[10] Kilpatrick, J.A., Throupe R.L., Carruthers J.I. & Krause A., The 553impact of transit corridors on residential property values. Journal of 554Real Estate Research, 29(3), pp. 303-320, 2007. 555[11] Mok, H.M., Chan P.P. & Cho Y.-S., A hedonic price model for private 556properties in hong kong. The Journal of Real Estate Finance and 557Economics, 10(1), pp. 37-48, 1995. 558[12] Voith, R., Changing capitalization of cbd-oriented transportation 559systems: Evidence from philadelphia, 1970–1988. Journal of Urban 560Economics, 33(3), pp. 361-376, 1993. 561[13] Knaap, G., Hopkins L. & Pant A. Does transportation planning 562matter? Explorations into the effects of planned transportation 563infrastructure on real estate sales, land values, building permits, and 564development sequence. Research Paper, Lincoln Institute of Land 565Policy. 1996. 566[14] Forrest, D., Glen J. & Ward R., The impact of a light rail system on 567the structure of house prices: A hedonic longitudinal study. Journal of 568Transport Economics and Policy, pp. 15-29, 1996. 569[15] Chau, K., Wong S., Yiu C. & Leung H., Real estate price indices in 570hong kong. Journal of Real Estate Literature, 13(3), pp. 337-356, 5712005. 572[16] Rosen, S., 573Hedonic prices and implicit markets: Product 574differentiation in pure competition. Journal of Political Economy, 57582(1), pp. 34-55, 1974. 576[17] Halvorsen, R., Henry O. Pollakowski. , Choice of functional form for 577hedonic price equations. Journal of Urban Economics, 10(1), pp. 57837-49, 1981. 579[18] Besner, C., A spatial autoregressive specification with a comparable 580sales weighting scheme. Journal of Real Estate Research, 24(2), pp. 581193-211, 2002. 582[19] Blackley, D.M., Follain J.R. & Lee H., An evaluation of hedonic 583price indexes for thirty-four large smsas. AREUEA Journal: Journal 584of the American Real Estate & Urban Economics Association, 14(2), 585pp. 179-205, 1986. 586[20] Bourassa, S., Cantoni E. & Hoesli M., Spatial dependence, housing 587submarkets, and house price prediction. Journal of Real Estate 588Finance & Economics, 35(2), pp. 143-160, 2007. 589[21] Brasington, D.M., Which measures of school quality does the 590housing market value? Journal of Real Estate Research, 18(3), pp. 591395, 1999. 592[22] Butsic, V., Hanak E. & Valletta R.G., Climate change and housing 593prices: Hedonic estimates for ski resorts in western north america. 594Land Economics, 87(1), pp. 75-91, 2011. 595[23] Dale-Johnson, D. & Phillips G.M., Housing attributes associated with 596capital gain. AREUEA Journal: Journal of the American Real Estate 597& Urban Economics Association, 12(2), pp. 162-175, 1984. 598[24] Dinan, T.M. & Miranowski J.A., Estimating the implicit price of 599energy efficiency improvements in the residential housing market: A 600hedonic approach. Journal of Urban Economics, 25(1), pp. 52-67, 6011989. 602[25] Goodman, A.C., Hedonic prices, price indices and housing markets. 603Journal of Urban Economics, 5(4), pp. 471-484, 1978. 604[26] Goodman, A.C. & Thibodeau T.G., The spatial proximity of 605metropolitan area housing submarkets. Real Estate Economics, 35(2), 606pp. 209-232, 2007. 607[27] Harding, J.P., Knight J.R. & Sirmans C.F., Estimating bargaining 608effects in hedonic models: Evidence from the housing market. Real 609Estate Economics, 31(4), pp. 601-622, 2003. 610[28] Jim, C. & Chen W.Y., Value of scenic views: Hedonic assessment of 611private housing in hong kong. Landscape and Urban Planning, 91(4), 612pp. 226-234, 2009. 613[29] Case, B., Pollakowski H.O. & Wachter S.M., On choosing among 614house price index methodologies. Real Estate Economics, 19(3), pp. 615286-307, 1991. 616[30] Gatzlaff, D.H. & Ling D.C., Measuring changes in local house prices: 617An empirical investigation of alternative methodologies. Journal of 618Urban Economics, 35(2), pp. 221-244, 1994. 619[31] Benson, E.D., Hansen J.L., Schwartz Jr A.L. & Smersh G.T., The 620influence of canadian investment on u.S. Residential property values. 621Journal of Real Estate Research, 13(3), pp. 231-249, 1997. 622[32] Kang, H.B. & Reichert A.K., An evaluation of alternative estimation 623techniques and functional forms in developing statistical appraisal 624models. Journal of Real Estate Research, 2(1), pp. 1-29, 1996. 625[33] Wooldridge, J.M., Introductory econometrics: A modern approach: 626South-Western Pub, 2009. 627[34] Sohu. Second-hand real estate. http://cd.esf.focus.cn/shop/. 2014. 628Sina. Second-hand real estate. http://cd.esf.sina.com.cn/. 2014. 629 630352631