1# RECORD de48aa9a311d709ef0a33453d7042001c2e721a62# TITLE: Public transport, noise complaints, and housing: Evidence from sentiment analysis in Singapore3# YEAR: 2021.0 VENUE: Journal of Regional Science DOI: 10.1111/jors.125244# AUTHORS: Yi Fan; Ho-Pin Teo; Wayne Xinwei Wan5# PAGES: 27 CANDIDATE TABLE PAGES (1-based): [12, 14, 15, 19, 20, 21]6# ABSTRACT: This paper investigates the effect of a new bus route on noise complaints of residents and the influence of noise on housing price. To overcome the challenge of mapping noise data with subjective emotion, we use a novel data source — text ‐ based noise complaint records from residents in a town in Singapore — and apply natural language processing tools to conduct sentiment analysis. To address the endogeneity concern regarding the bus route, we use a hypothetical least cost path as an instrument for the existing bus route. We find that living closer to the bus route for every 100 m increases noise complaints by around 10 percentage points, and the effect is more severe on medium floor levels (5th – 8th floors) and near bus stops (within 100 m). We further link noise with housing price and discover a price reduction of 3% with a 1 ‐ scale ‐ point increase in noise complaints. This implies that bus noise offsets 17.8% of the benefit from convenience, which sheds light on the importance of noise insulation in transit ‐ oriented developments. propose a novel methodology to measure the quasi real ‐ time noise sentiment at finer detailed level using the noise complaints data, which naturally exists with the789===== PAGE 1 (score 10, numeric-paren 0, table-words 0, var-words 10) =====1011J Regional Sci. 2021;61:570–596.12570 |13wileyonlinelibrary.com/journal/jors14Received: 22 January 2020 | Revised: 25 November 2020 | Accepted: 19 December 202015DOI: 10.1111/jors.1252416R E S E A R C H A R T I C L E17Public transport, noise complaints, and housing:18Evidence from sentiment analysis in Singapore19Yi Fan120|21Ho Pin Teo222|23Wayne X. Wan3241Department of Real Estate, National25University of Singapore, Singapore, Singapore262Department of Building, National University27of Singapore, Singapore, Singapore283Department of Land Economy, The29University of Cambridge, Cambridge, UK30Correspondence31Wayne X. Wan, Department of Land32Economy, The University of Cambridge.3319 Silver St, Cambridge CB3 9EP, UK.34Email: xw357@cam.ac.uk35Funding information36National University of Singapore,37Grant/Award Number: R‐297‐000‐134‐13338Abstract39This paper investigates the effect of a new bus route on40noise complaints of residents and the influence of noise41on housing price. To overcome the challenge of mapping42noise data with subjective emotion, we use a novel data43source—text‐based noise complaint records from re-44sidents in a town in Singapore—and apply natural45language processing tools to conduct sentiment analysis.46To address the endogeneity concern regarding the bus47route, we use a hypothetical least cost path as an in-48strument for the existing bus route. We find that living49closer to the bus route for every 100 m increases noise50complaints by around 10 percentage points, and the51effect52is53more54severe55on56medium57floor58levels59(5th–8th floors) and near bus stops (within 100 m).60We further link noise with housing price and discover a61price reduction of 3% with a 1‐scale‐point increase in62noise complaints. This implies that bus noise offsets6317.8% of the benefit from convenience, which sheds64light on the importance of noise insulation in transit‐65oriented developments.66K E Y W O R D S67housing, noise, public transport, sentiment analysis68This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and69reproduction in any medium, provided the original work is properly cited.70© 2021 The Authors. Journal of Regional Science published by Wiley Periodicals LLC.717273===== PAGE 12 (score 36, numeric-paren 12, table-words 1, var-words 10) =====7475TABLE 176Summary statistics of noise complaints and housing transactions77Total78Near = 0 (100–200 m)79Near = 1 (within 100 m)80Diff ((5)–(8))81N82Mean83SD84N85Mean86SD87N88Mean89SD90t‐test91(1)92(2)93(3)94(4)95(5)96(6)97(7)98(8)99(9)100(10)101Noise sentiment1025931030.4091040.1651053281060.4161070.1701082651090.3991100.1591110.017112Distance (100 m)1135931141.1111150.4931163281171.4831180.3171192651200.6501210.1871220.833***123LCP (100 m)1245931251.7371261.3231273281282.1361291.3491302651311.2431321.1081330.893***134ECP (100 m)1355931361.0681370.7941383281391.5161400.6491412651420.5141430.5781441.001***145Launch (after = 1)1465931470.5751480.4951493281500.5121510.5011522651530.6531540.477155−0.141***156Near (within 100 m = 1)1575931580.4471590.498160RC center (with = 1)1615931620.0691630.2541643281650.06711660.2511672651680.0721690.258170−0.005171Building age17259317329.0221742.26717532817628.9021772.58417826517929.1701801.794181−0.267182Floor level1835931846.4081853.9601863281876.4791883.9761892651906.3211913.9461920.158193Morphology1945931952.9261960.2931973281982.8781990.3712002652012.9852020.122203−0.107***204Gender (Male = 1)2055272060.4672070.4992083002090.5032100.5012112272120.4192130.4942140.085*215To train station (100 m)2165932172.1252181.1552193282202.4352211.2242222652231.7412240.9332250.693***226To LRT line (100 m)2275932281.5312291.1542303282311.8842321.1552332652341.0932350.9942360.790***237To bus stop (100 m)2385932391.2022400.5012413282421.3352430.5482442652451.0382460.3772470.233***248To expressway (100 m)2495932504.2812512.0302523282534.5532542.3572552652563.9452571.4682580.608***259To major road (100 m)2605932611.2172620.8162633282641.3592650.7582662652671.0412680.8512690.318***270To shopping center (100 m)2715932725.3022732.8632743282755.1812763,1542772652785.4522792.467280−0.271281Price (thousand SGD)2821450283375.20728473.5642851003286364.92428769.114288447289398.28229077.968291−33.359***292(Continues)293FAN ET AL.294|295581296297298===== PAGE 13 (score 6, numeric-paren 1, table-words 0, var-words 4) =====299300Total301Near = 0 (100–200 m)302Near = 1 (within 100 m)303Diff ((5)–(8))304N305Mean306SD307N308Mean309SD310N311Mean312SD313t‐test314(1)315(2)316(3)317(4)318(5)319(6)320(7)321(8)322(9)323(10)324Floor32514503266.7103273.63032810033296.6713303.5883314473326.7993333.724334−0.128335Area (m2)3361450337103.59533820.399339100334099.98434119.56342447343111.69634419.928345−11.712***346Note: Morphology equals to 1 if the block follows the new generation building layout, 2 if it follows the point block layout, or 3 if it follows the slab block layout.347Abbreviations: ECP, Euclidean cost path; LCP, least cost path; LRT, light rail train.348*p < 0.1.349**p < 0.05.350***p < 0.01.351582352|353FAN ET AL.354355356===== PAGE 14 (score 77, numeric-paren 56, table-words 2, var-words 8) =====357358TABLE 2359First‐stage IV estimation results of the effects of bus route on noise complaints at individual level360(1)361(2)362(3)363(4)364Launch365× Near366Launch367× Near368Launch ×369Distance370Launch ×371Distance372Launch × ECP373−0.3653***374−0.3306***3750.4184***3760.4025***377(0.0546)378(0.0615)379(0.0520)380(0.0602)381ECP3820.01903830.0367384−0.0070385−0.0491386(0.0419)387(0.0451)388(0.0364)389(0.0395)390Launch × LCP391−0.0787**3920.0491393(0.0360)394(0.0352)395LCP396−0.07903970.1558**398(0.0651)399(0.0644)400Launch4010.7781***4020.8410***4030.8278***4040.7932***405(0.2049)406(0.2067)407(0.2051)408(0.2057)409Building age410−0.0040411−0.0023412−0.0117413−0.0158414(0.0119)415(0.0119)416(0.0104)417(0.0102)418Floor4190.0007420−0.00004210.00294220.0033423(0.0123)424(0.0126)425(0.0102)426(0.0105)427Floor squared4280.00054290.0005430−0.0008431−0.0008432(0.0007)433(0.0007)434(0.0006)435(0.0006)436Point block4370.04014380.06054390.3776**4400.3339**441(0.1951)442(0.1908)443(0.1789)444(0.1692)445Slab block4460.09434470.13154480.14924490.0789450(0.1448)451(0.1457)452(0.1397)453(0.1330)454Male455−0.0402456−0.04054570.01564580.0149459(0.0299)460(0.0286)461(0.0239)462(0.0228)463RC center464−0.0094465−0.0105466−0.0135467−0.0083468(0.0796)469(0.0803)470(0.0684)471(0.0692)472Block and time fixed effect473Y474Y475Y476Y477First‐stage F‐Stats47822.6947926.8148032.6748138.07482J‐Statistics (p value)4830.27874840.2235485Observations486527487527488527489527490Note: Columns (1) and (2) report results for the binary closeness indicator. Columns (3) and (4) report results for the491continuous distance variable. Columns (1) and (3) use the ECP, the Euclidean straight lines connecting bus stops, as the492instrument. Columns (2) and (4) use both ECP and LCP, the least travel time path, as the instrument. Unreported control493variables include distance to train stations, LRT viaduct line, bus stops, expressways, major roads, and shopping centers.494Standard errors are clustered by building blocks. Robust standard errors in parentheses.495Abbreviations: ECP, Euclidean cost path; LCP, least cost path.496*p < 0.1.497**p < 0.05.498***p < 0.01.499FAN ET AL.500|501583502503504===== PAGE 15 (score 109, numeric-paren 82, table-words 3, var-words 9) =====505506TABLE 3507OLS and second‐stage IV estimation results of the effects of bus route on noise complaints at508individual level509Noise sentiment510(1)511(2)512(3)513(4)514(5)515(6)516OLS517ECP IV518Both IVs519OLS520ECP IV521Both IVs522Launch × Near5230.04965240.1090**5250.1118**526(0.0324)527(0.0459)528(0.0437)529Near530−0.0136531−0.0105532−0.0300533(0.0303)534(0.0529)535(0.0476)536Launch × Distance537−0.0405538−0.0953**539−0.0965**540(0.0338)541(0.0430)542(0.0422)543Distance5440.00745450.01255460.0301547(0.0328)548(0.0462)549(0.0423)550Launch551−0.1123552−0.1330553−0.1365554−0.04005550.03195560.0333557(0.1217)558(0.1069)559(0.1054)560(0.1167)561(0.1140)562(0.1114)563Building age564−0.0016565−0.0011566−0.0015567−0.0019568−0.0024569−0.0022570(0.0048)571(0.0044)572(0.0044)573(0.0048)574(0.0044)575(0.0043)576Floor577−0.0130*578−0.0126**579−0.0128**580−0.0128*581−0.0122*582−0.0125**583(0.0070)584(0.0063)585(0.0063)586(0.0071)587(0.0063)588(0.0063)589Floor squared5900.0008*5910.0007*5920.0007*5930.0008*5940.0007*5950.0007*596(0.0004)597(0.0004)598(0.0004)599(0.0004)600(0.0004)601(0.0004)602Point block603−0.0487604−0.0764605−0.0593606−0.0423607−0.0457608−0.0415609(0.0811)610(0.0786)611(0.0786)612(0.0792)613(0.0739)614(0.0728)615Slab block616−0.0429617−0.0675618−0.0505619−0.0409620−0.0484621−0.0393622(0.0543)623(0.0549)624(0.0554)625(0.0508)626(0.0475)627(0.0480)628Male6290.01816300.01956310.01866320.01716330.01666340.0156635(0.0201)636(0.0188)637(0.0186)638(0.0199)639(0.0187)640(0.0186)641RC center6420.0484*6430.0448*6440.0501**6450.0453*6460.0420*6470.0463*648(0.0269)649(0.0251)650(0.0251)651(0.0260)652(0.0236)653(0.0239)654Block and time fixed effect655Y656Y657Y658Y659Y660Y661Observations662527663527664527665527666527667527668R26690.2146700.2036710.2076720.2136730.2066740.208675Note: Columns (1)–(3) report results for the binary closeness indicator and Columns (4)–(6) report results for the676continuous distance variable. Columns (1) and (4) are OLS estimation results. Columns (2) and (5) use the ECP, the677Euclidean straight lines connecting bus stops, as the instrument. Columns (3) and (6) use both ECP and LCP, the least travel678time path, as the instrument. Unreported control variables include distance to train stations, LRT viaduct line, bus stops,679expressways, major roads, and shopping centers. Standard errors are clustered by building blocks. Robust standard errors680in parentheses.681*p < 0.1.682**p < 0.05.683***p < 0.01.684584685|686FAN ET AL.687688689===== PAGE 16 (score 14, numeric-paren 0, table-words 3, var-words 8) =====690691instrument(s), respectively. First‐stage results reveal that both ECP and the combination of ECP and LCP are strongly and692statistically significantly correlated with the actual bus service route, controlling for the physical features, gender of the693complainants, and the fixed effect from time and building. The F‐statistics in all specifications are around 20–30, miti-694gating concerns about weak instruments. For estimations with more instrumental variables than endogenous variables695(Columns (2) and (4)), the J‐statistics for over‐identification test are 0.2787 and 0.2235, respectively.696Table 3 presents OLS and second‐stage IV estimation results. Column (1) presents OLS estimates from697Equation (1), while Columns (2) and (3) display IV estimates using ECP and the combination of ECP and LCP,698respectively. Columns (4)–(6) show corresponding estimates from Equation (2). The OLS estimate is positive,699though with no statistical significance, in Column (1), and it is likely contaminated by unobserved factors. For700example, residents who are less sensitive to noise will choose to live in units closer to roads and will also make less701noise complaints. Therefore, the OLS estimate of the noise effect will be downward biased in magnitude. Using ECP702as an instrument, being close to the bus service route (within 100 m) results in a higher negative sentiment score of7030.109 (Column (2)) and 0.112 (Column (3)) using both ECP and LCP as IVs. Both of the two IV estimates are704statistically significant at the 5% level. Since the sentiment score is normalized to the range of 0 to 1, this indicates705that intensification of public transport services worsens the noise sentiment in surrounding public housing by706approximately 11 percentage points. The continuous distance effect is estimated to be −0.095 (Column (5)) using707ECP as the IV, or −0.097 (Column (6)) using the combined IVs. Both estimates are statistically significant at the7085% level. Consistent with previous findings using a binary indicator of closeness, the distance estimates reveal that709by living closer to road traffic by 100 m, the negative noise sentiment will increase by around 10 percentage points710on average.711One assumption in the interpretations above is that the intensification of the bus service is the only source of712changing noise levels along the bus route. A possible concern about the validity of this assumption is that with the713newly opened bus route, increased bus traffic will change the traffic patterns of other vehicles on the route. For714instance, commuters using private vehicles instead of public buses may choose other routes to avoid congestion.715However, under this scenario, the traffic volume from other vehicles is expected to decrease after the treatment,716and the measured increase in noise sentiment in buildings near the bus route will be smaller than the real increase717due to the new bus. In other words, the real impact of closeness to the new bus service on noise sentiment is likely718larger than our estimates.7196.2720|721Robustness checks722We first examine the robustness of our baseline results by setting different cut‐off distances for the treatment and723control groups in the DID estimations. Since the maximum distance to the bus route for buildings in our sample is724198.5 m, we set the cut‐off distance to be 100 m in the baseline estimation to balance sample size in each group. This725100 m boundary for road traffic noise insulation is also supported by the acoustics literature (Avsar & Gonullu, 2005;726Martin & Hothersall, 2002). As robustness checks, we follow the same empirical strategy as Equation (1) but change the727cut‐off distances to 110 or 90 m. The corresponding estimation results are reported in Appendix Table C1. The728estimates are very close to those in our baseline results, for both the magnitude and statistical significance.729Another possible concern about the baseline estimation result is that an individual's sentiment about a new730noise incident is dependent on previous exposure to noise. If the surrounding environment has been noisy for731years, residents may not notice additional noise incidents, while residents living in quiet housing units may be less732tolerant of a sudden increase in the noise level. To alleviate this concern, the average sentiment score of noise733complaints made in the same building with a 1‐year lag is included as an additional control. Results are reported in734Appendix Table C2. The estimates remain robust in magnitude and level of significance.735In addition, the occurrence time of noise incidents may impact the noise sentiment, because residents may736be disrupted more easily at night. Therefore, we use the report time of the noise complaints as a proxy for the occurrence737FAN ET AL.738|739585740741742===== PAGE 19 (score 43, numeric-paren 16, table-words 6, var-words 11) =====743744estimations, using ECP as IV. Results are presented in Panel B of Table 4. Columns (1) and (2) show estimates using745the dummy variable measuring closeness, while Columns (3) and (4) present estimates using distance variations.746The effects at buildings within 100 m of a bus stop show larger magnitude (0.511 for the binary closeness variable747and −0.288 for the continuous distance variable), with statistical significance at the 1% level. This indicates that the748noise effect from bus service is more severe around bus drop‐off points. A possible explanation is the influx of749visitors around bus stops. Also, it is possible that buses generate more noise when they stop at or leave from the750TABLE 4751Heterogeneous effects of bus route on noise complaints752Panel A: Noise sentiment across floor levels753(1)754(2)755(3)756(4)757(5)758(6)7591–4 Storey7605–8 Storey761Above 8762Storey7631–4 Storey7645–8 Storey765Above 8766Storey767Launch × Near7680.07387690.2141***770−0.0290771(0.0674)772(0.0799)773(0.0802)774Launch × Distance775−0.0552776−0.2428***7770.0282778(0.0527)779(0.0820)780(0.0608)781Block and time fixed782effect783Y784Y785Y786Y787Y788Y789First‐stage F‐stats79014.8379118.117927.8579332.7479420.6679511.65796Observations797221798144799162800221801144802162803R28040.4778050.6058060.5448070.4798080.5988090.564810Panel B: Noise sentiment with different distance to bus stops811(1)812(2)813(3)814(4)815Within 100 m816Out of 100 m817Within 100 m818Out of 100 m819Launch × Near8200.5114***8210.1315*822(0.1924)823(0.0739)824Launch × Distance825−0.2875***826−0.1330**827(0.0853)828(0.0607)829Block and time fixed effect830Y831Y832Y833Y834First‐stage F‐stats83517.2483621.3583738.5983826.27839Observations840179841348842179843348844R28450.4478460.2748470.5908480.314849Note: Columns (1)–(3) of Panel A report results for the binary closeness indicator. Columns (4)–(6) of Panel A report results850for the continuous distance variable. Columns (1) and (2) of Panel B report results for the binary closeness indicator.851Columns (3) and (4) of Panel B report results for the continuous distance variable. All estimations use ECP, the Euclidean852straight lines connecting bus stops, as the instrument. Unreported control variables include building age, floor and floor853squared, gender, morphology, RC center, and distance to train stations, LRT viaduct line, bus stops, expressways, major854roads, and shopping centers. Standard errors are clustered by building blocks. Robust standard errors in parentheses.855*p < 0.1.856**p < 0.05.857***p < 0.01.858588859|860FAN ET AL.861862863===== PAGE 20 (score 61, numeric-paren 42, table-words 3, var-words 8) =====864865TABLE 5866Net effects of the new bus service on housing price867(1)868(2)869log (price)870log (price)871Launch × Near8720.0129**873(0.0054)874Near875−0.0097**876(0.0038)877Launch × Distance878−0.0134***879(0.0050)880Distance8810.0104***882(0.0037)883Launch8840.01018850.0299***886(0.0077)887(0.0094)888Prime lending rate889−2.8766***890−2.8815***891(0.3558)892(0.3563)893Floor8940.0070***8950.0070***896(0.0003)897(0.0003)898Area8990.0030***9000.0030***901(0.0008)902(0.0008)903Building age904−0.0009905−0.0009906(0.0006)907(0.0006)908Distance to train station909−0.0035**910−0.0036**911(0.0015)912(0.0015)913Distance to bus stop914−0.0042915−0.0046916(0.0031)917(0.0032)918Distance to expressway9190.0025***9200.0024***921(0.0008)922(0.0008)923Distance to major road924−0.0014925−0.0010926(0.0016)927(0.0017)928Distance to shopping center9290.0037***9300.0037***931(0.0009)932(0.0009)933Block and time fixed effect934Y935Y936Observations93714509381450939R29400.9419410.941942Note: Column (1) reports results for the binary closeness indicator and column (2) reports results for the continuous943distance variable. Unreported control variables include building morphology and RC center. Standard errors clustered by944building blocks. Robust standard errors in parentheses.945*p < 0.1.946**p < 0.05947***p < 0.01.948FAN ET AL.949|950589951952953===== PAGE 21 (score 49, numeric-paren 25, table-words 4, var-words 13) =====954955bus stops. In addition, with frequent visual exposure to noise sources at bus stops, subjective feelings of annoyance956by surrounding residents may be intensified.9577958|959COST‐BENEFIT ANALYSIS: PUBLIC TRANSPORT AND HOUSING960PRICE961Public transport offers convenience to surrounding residents, and thus raises housing price in the neighborhood962(Chalak et al., 2016; Cohen & Paul, 2007; Holmgren, 2014; Monchambert & De Palma, 2014). Nonetheless, it also963generates noise pollution, which exerts a negative impact on the housing price (Diao et al., 2016). To implement a964cost‐benefit analysis, we start with an estimation on the overall impact of the new bus route on housing price.965Specifically, we follow the same specification in Equations (1) and (2), but replace the dependent variable of noise966sentiment with housing price (Diao et al., 2017):967β968β969β970θ971μ972φ973ω974ε975=976+977+978∗979+980+981+982+983+984′985′986Price987Near988Launch989Launch990Near991X992U993log (994)995,996ijt997i998t999t1000i1001i1002j1003t1004i1005ijt1006110072100831009(6)1010β1011β1012β1013θ1014μ1015φ1016ω1017ε1018=1019+1020+1021∗1022+1023+1024+1025+1026+1027′1028′1029Price1030Distance1031Launch1032Launch1033Distance1034X1035U1036log (1037)1038.1039ijt1040i1041t1042t1043i1044i1045j1046t1047i1048ijt1049110502105131052(7)1053TABLE 61054Effects of noise complaints on housing price1055(1)1056(2)1057log (price)1058log (price)1059Noise sentiment in previous 12 months1060−0.0300**1061−0.0262**1062(0.0149)1063(0.0132)1064Noise sentiment in previous 12 months × Low/high floors1065−0.00531066(0.0084)1067Distance to nearest bus stop connecting CBD1068−0.0027*1069−0.0027*1070(0.0016)1071(0.0016)1072Prime lending rate1073−3.5344***1074−3.5089***1075(0.4423)1076(0.4638)1077Floor10780.0064***10790.0065***1080(0.0005)1081(0.0006)1082Area10830.0075***10840.0075***1085(0.0002)1086(0.0002)1087Building age1088−0.0035***1089−0.0035***1090(0.0009)1091(0.0009)1092Block and time fixed effect1093Y1094Y1095Observations109689310978931098R210990.91911000.9191101Note: Unreported control variables include unit layout, building morphology, RC center, and distance to train stations, LRT1102viaduct line, bus stops, expressways, major roads, and shopping centers. Standard errors are clustered by building blocks.1103Robust standard errors in parentheses.1104*p < 0.1.1105**p < 0.05.1106***p < 0.01.11075901108|1109FAN ET AL.111011111112===== PAGE 22 (score 29, numeric-paren 0, table-words 9, var-words 11) =====11131114The definitions for the other variables are the same as in Equations (1) and (2). Table 5 displays the1115corresponding results using all of the resale transaction data in our study area from 2010 to 2018. Column1116(1) reveals that before the bus starts operation, the housing prices within 100 m of the route are 0.97% lower1117than those farther away (100–200 m). This is probably because the route is designed along the community1118road, generating traffic noise but offering limited convenience before the new bus service begins. After the1119introduction of bus service, the housing price within 100 m of the bus route increases by 1.3% compared1120to units within 100–200 m of the bus route. A similar conclusion can be drawn if we use a continuous variable1121of distance to estimate the benefit from the convenience of public transport (Column (2)). The prices1122for units closer to the new bus route by 100 m increase by 1.34%. The estimate is statistically significant at1123the 1% level.1124Nevertheless, this explicit benefit is likely to be offset by an implicit decrease in housing price due to the1125noise generated by the new bus route. We thus examine the negative effect of noise sentiment on housing1126price using the same set of resale transaction records. Column (1) in Table 6 presents the corresponding OLS1127regression results of Equation (3). We include all the transaction records in our study area which can be1128matched with the lagged noise sentiment. To estimate the exact economic costs of noise pollution on housing1129price, we also control for the distance to the nearest bus stop connecting the CBD before and after1130the launch of bus no. 972. It reveals that a 1‐scale‐point increase in noise sentiment over the past 12 months1131is correlated with a 3% decrease in housing price.9 This estimate is statistically significant at the11325% level.1133Finally, we analyze the cost and benefit for intensifying public transport on housing price. Since living closer to the1134new bus service route by 100 m has caused the noise sentiment score to increase by around 9.53% (Column (5)1135in Table 3), the negative effect of its noise on housing price can be interpolated as approximately 0.29% per1136100‐m distance to the bus route. Taking the explicit 1.34% increase in housing price per 100 m closer to the new bus1137route, we estimate that the implicit effect of traffic noise on housing value has offset around 17.79% of the economic1138benefit brought by the improvement of public transport.1139We also investigate the heterogeneous impact of noise from the new bus route on housing price across floor1140levels in two steps. First, we examine the heterogeneous impact of the new bus service on noise sentiment on1141low, medium, and high floors. We find that the noise nuisance from the new bus service is positive and statis-1142tically significant on medium floors only, as reported in Table 4. Second, we estimate the impact of sentiment1143from general noise on housing price across floor levels. Specifically, we modify Equation (3) by interacting1144−1145SIi t,114611147with a dummy variable denoting the units on the low and high floors. The estimation results are reported in1148Column (2), Table 6. We find that if the 1‐year lagged noise sentiment increases by 1 scale point, the housing1149price on medium floors decreases by 2.62%. This estimate is statistically significant at the 5% level. In addition,1150the overall impact of a 1‐scale‐point increase in sentiment from general noise on housing price is similar across1151housing units on low/high floors and on medium floors, which is reflected by the statistically insignificant1152coefficient for the interaction term between the lagged noise sentiment and the dummy variable for units on1153low/high floors.1154Combining the results from these two steps, we conclude that for units on medium floors, a decrease of1155distance to the bus route by 100 meters increases the noise sentiment by 24.3 percentage points (Column (5) in1156Panel A, Table 4), which translates to a 0.64% drop in housing price, or equivalently, a loss of around 2401 SGD1157(1766 USD) per housing unit. In comparison, for units on high/low floors, as the impact of the new bus service on1158noise sentiment is statistically insignificant (Columns (4) and (6) in Panel A, Table 4), the direct impact of noise from1159the new bus on housing price is likely small.11609The results remain robust if we use the noise sentiment over the past 6 months instead. The estimation results are reported in Appendix Table C8.1161FAN ET AL.1162|11635911164