1# RECORD 7670a6d2dc2183fc23719395205238727f1783172# TITLE: Spatial Estimation of the Impact of Airport Noise on Residential Housing Prices3# YEAR: 2008.0 VENUE: nan DOI: 10.1007/bf033992674# AUTHORS: M. Salvi5# PAGES: 30 CANDIDATE TABLE PAGES (1-based): [11, 18, 22, 23]6# ABSTRACT: SummaryThe paper applies spatial econometric techniques to measure the impact of airport noise on the price of single-family homes in the Zurich Airport area. We exploit a large database of geo-referenced noise measurements to investigate the reaction of prices to different noise metrics. The particular institutional setting of Zurich Airport, with changing patterns of runways configurations allows to distinguish the impact of noise at different times of the day. The use of neighborhood fixed-effects is compared to the results given by a costlier modelling strategy involving a rich set of location descriptors. The paper documents the impact of airport noise on housing prices. In the base model specification the Noise Discount Index is 0.97% with typical discounts in the range of −2% to −8%. Accounting for the spatiality of the data has little effect on the results.789===== PAGE 1 (score 6, numeric-paren 0, table-words 0, var-words 6) =====1011© Swiss Journal of Economics and Statistics 122008, Vol. 144 (4) 577–60613a 14Ecole Polytechnique Fédérale, Lausanne, and Zürcher Kantonalbank, Zürich. Email: marco.15salvi@epfl.ch. I am grateful to Patrik Schellenbauer, Philippe Thalmann, Adrian Lüscher, 16Gion-Reto Hassler, Alice Hollenstein and the editors for their helpful comments and sug-17gestions. I thank the ZKB GIS-team for invaluable help with the GIS. The views expressed 18herein are my own and do not necessarily represent the views of the ZKB.19Spatial Estimation of the Impact of Airport Noise on 20Residential Housing Prices21Marco Salvia22JEL-Classification: Q53, Q51, R31, L93, C2123Keywords: noise, airports, housing prices, hedonic pricing, spatial econometrics241. Introduction25The impact of airport proximity on neighboring communities is a hotly debated 26topic. The positive aspects of proximity are related to the provision of communi-27cation links and to the direct economic importance of large airports. The down-28side of proximity are the adverse environmental effects primarily associated with 29aircraft noise. The growing interdependence and global connectivity of modern 30economies is bound to increase the economic importance of air travel. On the 31other hand, environmental quality tends to be a superior good – as per capita 32income rises, the social demand for environmental quality increases more than 33proportionally. It is fair to say that airport noise is likely to fuel ongoing public 34debate.35The recent story of Zurich Airport is in this regard exemplary. Until the year 362000 the use of German airspace for approaching and leaving Zurich airport had 37been governed by a bilateral agreement between Switzerland and Germany. The 38agreement was terminated by Germany in 2000, forcing the Swiss authorities to 39set out new landing procedures. As a result, the number of residents subject to 40aircraft noise increased significantly. Under the Swiss noise protection law these 41residents may be entitled to compensation and over 19 000 have filed a claim. 42This has drastically increased the public awareness for the evaluation of the cost 43of noise. Specifically, the use of the hedonic method for the scope of evaluation 44and compensation has been proposed. Recently, the Federal Supreme Court of 454647===== PAGE 11 (score 34, numeric-paren 0, table-words 4, var-words 22) =====4849Spatial Estimation of the Impact of Airport Noise on Residential Housing Prices 5058751Table 1: Descriptive Statistics of the House Characteristics52Variable 53Mean 54Median 55Min 56Max 57Std. Dev 58House price [CHF 1000] 59772.16 60727.00 61155.00 622 600.00 63262.94 64Continuous variables) 65Lot size [m2] 66521.62 67447.00 68150.00 692 430.00 70315.61 71Size, volume of building [m3] 72723.12 73691.00 74350.00 751 995.00 76220.67 77Number of rooms 785.49 795.00 802.00 8111.00 821.15 83Number of bathrooms 842.27 852.00 861.00 876.00 880.73 89Age of building [years] 9030.76 9121.00 920.00 93156.00 9432.67 95Categorical variables 96Condition of the dwelling 97 New 980.36 99 Renovated 1000.13 101 Well maintained 1020.41 103 Needs renovation 1040.10 105Brick building 1060.95 107Detached 1080.46 109Cellar 1100.95 111Double glazing 1120.66 113Floor heating 1140.45 115Garage 1160.49 117Underground garage 1180.25 119Modern kitchen 1200.33 121Swimming pool 1220.02 123Sauna 1240.04 125The table reports the descriptive statistics of the 3 737 observations included in the final sample. 126The original data set contained 3 967 transactions single-family homes sold at arm’s length in the 127Canton of Zurich between 1995 and 2005. 230 observations (5.8%) were excluded because of 128likely data entry mistakes, of which 178 transactions with very small lot sizes under 150 m2. We 129suspect that for these transactions, the house living area was erroneously entered.130131132===== PAGE 12 (score 24, numeric-paren 0, table-words 1, var-words 15) =====133134588 135Marco Salvi136Table 2: Descriptive Statistics of the Location Characteristics137Variable 138Mean 139Median 140Min 141Max 142Std. Dev 143Aircraft noise (in excess of 50 dB) 144Leq 16h [dB] 1451.11 1460.00 1470.00 14820.40 1492.84 150Leq Evening [dB] 1510.24 1520.00 1530.00 15416.10 1551.25 156Leq 1h Peak [dB] 1570.67 1580.00 1590.00 1608.20 1611.50 162Location variables 163Road traffic noise, Lr 16h [dB] 1640.78 1650.00 1660.00 16718.50 1682.55 169View on lakes [ha] 170512.00 1710.00 1720.00 1736 971.00 1741 236.00 175View on Alps [km2] 176187.83 177168.83 1780.00 179986.48 180132.56 181Travel distance to CBD [min] 18232.9 18333.00 18412.00 18556.00 1868.01 187Near power line ( < 200m) 1880.03 1890.00 1900.00 1911.00 1920.16 193Near railway ( < 100m) 1940.09 1950.00 1960.00 1971.00 1980.28 199Slope terrain [%] 2004.59 2013.83 2020.13 20321.33 2043.49 205Aspect: East 2060.28 2070.00 2080.00 2091.00 2100.45 211Aspect: West 2120.44 2130.00 2140.00 2151.00 2160.50 217Share of Swiss residents 2180.84 2191.00 2200.00 2211.00 2220.36 223Maximum built density [%] 22443.48 22540.00 22615.00 227333.00 22818.89 229Community tax revenue [Index] 230106.78 23190.93 23242.34 233511.42 23451.95 235The location variables were matched at the hectare level with each of the 3 737 single-family homes, 236sold in the Canton of Zurich between 1995 and 2005. The aircraft noise exposure and, where pos-237sible, the other location variables are those of the transaction year. All noise variables are reported 238as the values in excess of 50 dB(A). Daytime car travel noise is measured with the Lr metric (rating 239level). The distance to CBD is measured as the car travel time to Zurich Main station. Aspect 240(orientation) dummies are computed only for terrains with a slope greater than 1%. The share of 241Swiss residents is computed in a neighborhood of 300m around the transaction. View variables 242are computed at 4 meters above ground. The density variables report the maximum allowed built 243density on the lot expressed as the ratio of the living space to the lot size. The community tax is 244the communal per capita tax revenue. See the main text for further explanations.245246247===== PAGE 18 (score 27, numeric-paren 0, table-words 2, var-words 13) =====248249594 250Marco Salvi25111 ML estimation requires the repeated evaluation of the information matrix, which is computa-252tionally demanding in this context (Pace and Gilley, 1998). Therefore, we use a sparse matrix 253implementation.254of the spatial model in Table 5 (Model 1 and 2) for different specifications of 255the aircraft noise, alongside the results of the OLS estimation of a non-spatial 256specification (Model 3).11257Table 5: Estimation Results: SAR Models and Non-Spatial Hedonic Regression258Model 1 (ML SAR) 259Model 2 (ML SAR) 260Model 3 (OLS) 261Variable 262Estimate 263SE 264Estimate 265SE 266Estimate 267SE 268Intercept 2699.2801 2700.1337 2719.3899 2720.1083 2739.2325 2740.1253 275Leq 16h 276−0.0098 2770.0010 278−0.0070 2790.0013 280−0.0098 2810.0009 282Leq Evening 283−0.0064 2840.0019 285Leq 1h Peak 286−0.0063 2870.0023 288Age of building 289−0.0028 2900.000 291−0.0029 2920.0003 293−0.0026 2940.0003 295Age squared 2960.0007 2970.000 2980.0007 2990.0002 3000.0006 3010.0002 302New 3030.1377 3040.012 3050.1373 3060.0116 3070.1358 3080.0119 309Renovated 3100.1451 3110.009 3120.1452 3130.0093 3140.1419 3150.0096 316Well maintained 3170.1126 3180.008 3190.1125 3200.0080 3210.1096 3220.0082 323Brick building 3240.0284 3250.009 3260.0290 3270.0092 3280.0298 3290.0094 330Detached 3310.0256 3320.006 3330.0264 3340.0056 3350.0253 3360.0057 337Cellar 3380.0476 3390.010 3400.0478 3410.0103 3420.0487 3430.0106 344Double glazing 3450.0312 3460.005 3470.0317 3480.0055 3490.0322 3500.0056 351Floor heating 3520.0356 3530.006 3540.0359 3550.0064 3560.0357 3570.0066 358Garage 3590.0253 3600.005 3610.0257 3620.0055 3630.0253 3640.0056 365Underground garage 3660.0290 3670.007 3680.0293 3690.0071 3700.0264 3710.0072 372Modern kitchen 3730.0589 3740.005 3750.0586 3760.0052 3770.0592 3780.0054 379Swimming pool 3800.0637 3810.015 3820.0641 3830.0147 3840.0653 3850.0152 386Sauna 3870.0396 3880.011 3890.0396 3900.0109 3910.0379 3920.0113 393394395===== PAGE 19 (score 32, numeric-paren 0, table-words 2, var-words 14) =====396397Spatial Estimation of the Impact of Airport Noise on Residential Housing Prices 398595399Model 1 (ML SAR) 400Model 2 (ML SAR) 401Model 3 (OLS) 402Variable 403Estimate 404SE 405Estimate 406SE 407Estimate 408SE 409Log lot size 4100.1691 4110.006 4120.1692 4130.0059 4140.1646 4150.0060 416Log size 4170.4163 4180.011 4190.4155 4200.0111 4210.4206 4220.0113 423Log rooms 4240.1243 4250.013 4260.1246 4270.0130 4280.1223 4290.0134 430Log bathrooms 4310.1005 4320.008 4330.0998 4340.0078 4350.1018 4360.0080 437Share Swiss residents 4380.0149 4390.006 4400.0157 4410.0064 4420.0159 4430.0063 444Road traffic noise 445−0.0050 4460.001 447−0.0051 4480.0009 449−0.0053 4500.0009 451Log travel dist. to CBD 452−0.2310 4530.015 454−0.2354 4550.0151 456−0.2290 4570.0133 458Low density (< 40%) 459−0.0034 4600.006 461−0.0040 4620.0059 463−0.0044 4640.0056 465Middle density (40–50%) 466−0.0024 4670.007 468−0.0029 4690.0068 470−0.0034 4710.0065 472Higher density (> 50%) 473−0.0117 4740.007 475−0.0126 4760.0072 477−0.0132 4780.0070 479Lake view > 0–20 km24800.0552 4810.007 4820.0514 4830.0066 4840.0563 4850.0059 486Lake view 20–40 km2 4870.1390 4880.012 4890.1346 4900.0118 4910.1411 4920.0107 493Lake view > 40 km2 4940.1121 4950.014 4960.1074 4970.0138 4980.1103 4990.0120 500View 50–100 km2 5010.0359 5020.009 5030.0357 5040.0091 5050.0406 5060.0084 507View 100–250 km2 5080.0236 5090.008 5100.0244 5110.0078 5120.0270 5130.0071 514View > 250 km2 5150.0163 5160.009 5170.0183 5180.0090 5190.0159 5200.0081 521Slope 0–9% 5220.0194 5230.006 5240.0191 5250.0061 5260.0191 5270.0060 528Slope > 9% 5290.0357 5300.008 5310.0360 5320.0080 5330.0353 5340.0076 535Near power line 536−0.0103 5370.015 538−0.0133 5390.0150 540−0.0118 5410.0140 542Near railway 543−0.0243 5440.008 545−0.0237 5460.0082 547−0.0247 5480.0078 549Aspect: East 550−0.0032 5510.007 552−0.0034 5530.0068 554−0.0002 5550.0064 556Aspect: West 5570.0102 5580.006 5590.0093 5600.0063 5610.0139 5620.0060 563Log tax revenue 5640.1559 5650.010 5660.1568 5670.0099 5680.1549 5690.0089 570ρ5710.1852 5720.1853 573– 574LR test (p-value) 575< 0.001 576< 0.001 577– 578Residual standard error 5790.1279 5800.1275 5810.1309 582N 5833737 5843737 5853737 586Log likelihood 5872364 5882374 589AIC 590−4625591−4640 592For reasons of space, yearly time dummies are not reported here. They are available from the author 593upon request.594595596===== PAGE 22 (score 27, numeric-paren 0, table-words 2, var-words 12) =====597598598 599Marco Salvi600In spatial econometric applications, the choice of spatial matrix may cause 601greater differences in parameter estimates than the choice of estimation tech-602nique. This is unsatisfactory as this choice remains largely untested and many 603possible matrices are compatible with the underlying economic story. We thus 604conclude the spatial econometric section by reporting the results of the estima-605tion of Model 1 given alternative choices for the spatial matrices. The results are 606reported in Table 6, in the columns labelled Model 4 (for the weighting matrix 607with cut-off at 600m) and Model 5 (for the weighting matrix based on nearest-608neighbor contiguity).609Table 6: ML SAR Estimation Results with Alternative Weighting Schemes610Model 4 (c = 600m) 611Model 5 (NN) 612Model 6 (No GIS) 613Variable 614Estimate 615SE 616Estimate 617SE 618Estimate 619SE 620Intercept 6219.3800 6220.1137 6239.3701 6240.1124 6259.3226 6260.0840 627Leq 16h 628−0.0099 6290.0011 630−0.0099 6310.0010 632−0.0086 6330.0023 634Age of building 635−0.0030 6360.0003 637−0.0030 6380.0003 639−0.0034 6400.0003 641Age squared 6420.0008 6430.0002 6440.0008 6450.0002 6460.0010 6470.0002 648New 6490.1359 6500.0116 6510.1367 6520.0116 6530.1271 6540.0117 655Renovated 6560.1443 6570.0093 6580.1448 6590.0092 6600.1491 6610.0093 662Well maintained 6630.1119 6640.0080 6650.1119 6660.0080 6670.1111 6680.0080 669Brick building 6700.0290 6710.0093 6720.0278 6730.0093 6740.0280 6750.0093 676Detached 6770.0277 6780.0056 6790.0271 6800.0056 6810.0267 6820.0057 683Cellar 6840.0468 6850.0103 6860.0458 6870.0103 6880.0441 6890.0104 690Double glazing 6910.0340 6920.0055 6930.0326 6940.0055 6950.0291 6960.0055 697Floor heating 6980.0333 6990.0064 7000.0349 7010.0064 7020.0326 7030.0065 704Garage 7050.0254 7060.0055 7070.0259 7080.0055 7090.0278 7100.0056 711Underground garage 7120.0270 7130.0071 7140.0289 7150.0072 7160.0287 7170.0073 718Modern kitchen 7190.0587 7200.0053 7210.0585 7220.0052 7230.0634 7240.0053 725Swimming pool 7260.0598 7270.0147 7280.0624 7290.0147 7300.0587 7310.0149 732Sauna 7330.0391 7340.0109 7350.0378 7360.0109 7370.0526 7380.0110 739Log lot size 7400.1700 7410.0059 7420.1693 7430.0060 7440.1786 7450.0059 746Log size 7470.4165 7480.0111 7490.4158 7500.0112 7510.4223 7520.0113 753754755===== PAGE 23 (score 29, numeric-paren 0, table-words 3, var-words 14) =====756757Spatial Estimation of the Impact of Airport Noise on Residential Housing Prices 758599759Model 4 (c = 600m) 760Model 5 (NN) 761Model 6 (No GIS) 762Variable 763Estimate 764SE 765Estimate 766SE 767Estimate 768SE 769Log rooms 7700.1199 7710.0130 7720.1207 7730.0130 7740.1246 7750.0131 776Log bathrooms 7770.0960 7780.0077 7790.0969 7800.0077 7810.0969 7820.0078 783Share Swiss residents 7840.0160 7850.0064 7860.0169 7870.0066 788Road traffic noise 789−0.0053 7900.0009 791−0.0054 7920.0009 793Log travel dist. to CBD 794−0.2335 7950.0162 796−0.2307 7970.0159 798Low density (< 40%) 799−0.0077 8000.0059 801−0.0045 8020.0060 803Middle density (40–50%) 804−0.0065 8050.0068 806−0.0027 8070.0069 808Higher density (> 50%) 809−0.0145 8100.0072 811−0.0120 8120.0073 813Lake view > 0–20 km28140.0545 8150.0067 8160.0557 8170.0067 818Lake view 20–40 km2 8190.1318 8200.0123 8210.1357 8220.0122 823Lake view > 40 km2 8240.1080 8250.0146 8260.1119 8270.0144 828View 50–100 km2 8290.0319 8300.0092 8310.0344 8320.0093 833View 100–250 km2 8340.0213 8350.0081 8360.0216 8370.0080 838View > 250 km2 8390.0135 8400.0093 8410.0137 8420.0092 843Slope 0–9% 8440.0171 8450.0060 8460.0177 8470.0061 848Slope > 9% 8490.0357 8500.0080 8510.0345 8520.0081 853Near power line 854−0.0090 8550.0151 856−0.0064 8570.0154 858Near railway 859−0.0244 8600.0080 861−0.0228 8620.0083 863Aspect: East 864−0.0032 8650.0068 866−0.0031 8670.0069 868Aspect: West 8690.0093 8700.0063 8710.0106 8720.0065 873Log tax revenue 8740.1588 8750.0107 8760.1599 8770.0105 878ρ8790.2579 8800.2233 8810.1453 882LR test (p-value) 883< 0.001884< 0.001 885< 0.001 886Residual standard error 8870.1276 8880.1274 8890.1263 890N 8913737 8923737 8933737 894AIC 895−4640 896−4645 897−4439 898Model 4 uses a contiguity matrix with a cut-off value set at 600m, while in Model 5 it is con-899strained to the four nearest neighbors. For Model 6 the same weighting matrix as in Model 1 of 900Table 5 is used, with GIS-matched location variables replaced by community indicator variables. 901For reasons of space, both the time and the community dummies are not reported.902903904===== PAGE 24 (score 4, numeric-paren 0, table-words 1, var-words 2) =====905906600 907Marco Salvi908Both alternative weighting matrices are associated with marginally higher, posi-909tive spatial autocorrelation parameters ρ. Accordingly, the standard error of the 910estimated coefficients are somewhat higher. Nevertheless, as far as the added 911value of the spatial econometric specification is concerned, this analysis confirms 912the previous negative results.9134.2.3 Robustness of Results914Higher noise levels may exert a disproportionate impact on house prices – an 915effect not captured by a crude linear specification. In order to examine this 916issue we reformulate Model 1 as a generalized additive model. This nonpara-917metric estimation method relaxes the assumption of log-linearity of the mean 918noise exposure, allowing for a more flexible functional form (Hastie and Tib-919shirani, 1990). The nonparametric component is modelled as a third-order 920spline. Figure 3 illustrates the result of the estimation where we report the non-921parametric fit with the respective 95%-confidence interval.922Figure 3. Non-Parametric Estimation of the Impact of Mean Noise Exposure.923Leq 16h over 50 dB924Effect on log house price9250.00926−0.05927−0.10928−0.15929−0.20930−0.259310 9325 93310 93415 93520936The nonparametric component is modelled as a third-order spline. 937The dotted lines represent the 95% confidence interval.938