Factors Affecting Residential Property Values in a Small Historic Canadian University Town

MPRA

Munich Personal RePEc Archive

Factors Affecting Residential Property

Values in a Small Historic Canadian

University Town

Janmaat, John A

University of British Columbia Okanagan

10. January 2007

Online at

http://mpra.ub.uni-muenchen.de/6145/

Histori Canadian University Town

John Janmaat

∗

Eonomis, IK BarberShool of Arts and Sienes

University of BritishColumbiaOkanagan

Deember6, 2007

Abstrat

Thetownof Wolfville, NovaSotiaisasmallhistori ommunity,eonomiallydominated

byAadiaUniversity. It isloatedonthenorthslope ofaridge,aordingviewsoftheMinas

Basin, at theeastern endof the Bay of Fundy. The upperboundary ofthe townis a major

provinial highway. Aset ofsound level observationswas usedto generateaverage andpeak

sound levelprolesfor thetown. Average andpeak soundlevel,as wellas preseneofaview

wereinludedina hedoni regressionof propertyvalues. View and average sound levelwere

notstatistiallyrelated tohomeprie. However,peaksoundlevelispried,withaonedeibel

inreasereduingtheaveragehousepriebyabouttwoperent. Beyondonventionalvariables

suhas ageand living spae,the zoninglassiation of thepropertywas foundto behighly

signiant,with homeszonedfor singlefamilyresidentialonlyommandingthehighest prie.

GiventhehighpopulationofstudenttenantsinWolfville,tenantsunlikelytoliveinareaszoned

singlefamilyresidential,theseresults suggeststhatrentalexternalities -eitherdueto student

tenantsorlandlordpraties-arehavingastrongnegativeimpatonpropertyvalues.

JEL: R21,R31,R52

Keywords: hedonipriing;universitytown;rentalexternalities;noisepollution;zoning

∗

Thispaperbeganasalassresearhprojet. Iamgratefultothestudentsofthe2003Eonomis3713lassfor

theirpreliminarydataolletionandanalysis.DataolletionwasfundedinpartbyaHumanResouresDevelopment

ThetownofWolfville,NovaSotiaisasmallhistoriommunity,eonomiallydominatedbyAadia

University. TwoexternalitiesareanedotallyonsideredimportantinWolfville. Therstexternality

derivesfromtheloaltopography. Thetownisloatedonthenorthslopeofaridge,aordingviews

oftheMinasBasin,attheeasternendoftheBayofFundy. ThereareanumberofhomesinWolfville

withaveryattrativeview. Popularwisdomwithin thetownsuggeststhatpropertieswithaview

ommand a higher prie. The seond externality is tra noise. The southern boundary of the

townisamajorprovinialhighway,Highway101,whilethetown'smajortraartery,MainStreet,

liesnear it's northernmargin. Both of these roadwaysareimportantsoures ofnoisepollution in

Wolfville. Asforviewproperties,onventionalwisdomholdsthatpropertieslosertothesesoures

ofnoisesellforless.

Thisresearhprojetsoughttomeasuretheimpatonpropertypriesofthesetwoexternalities,

thepreseneofaviewandthelevelofnoisepollution,usingHedoniregression.Giventhegeography

of the town, it wasseen to provide anideal loation for suh an analysis. Within Wolfville there

is noaess to Highway 101,while MainStreet is easily aessiblefrom anywhere in town. Thus,

loation within the towndoesnotdetermine aessibilitybenets, benets that ommonlyosets

noisepollution damages. Further, theundulating nature of theloal geography, aonsequeneof

thetownbeingbisetedbythreereekgullies,resultsinviewpropertiesnotbeingsimplyoinident

with distane from Highway101. These fats should provide theanalysis with suientpowerto

isolatetheeets ofbothexternalitiesonpropertyvalues.

Apreliminaryanalysis ofnoisepollutioneetsin Wolfvillewasonduted asanenvironmental

eonomis lass projet in the winter of 2003. Sine these results suggestedsound level aeted

property values, a more omprehensive set of measurements were taken in the summer of 2003.

At 27sites sattered aroundWolfville, asequene of 22 one hoursound level measurementswere

reorded with a Larson-Davis

TM

712 sound meter during the summer of 2003. Using polynomial

interpolation,soundprolesweregeneratedforthetownusingtheL

eq

(equivalentsoundlevel)and

L

peak

(peaksoundlevel)observations. Theproleswereusedtopreditasoundlevelfortheloation

ofeahpropertytransation betweenJuly1998andJune2003. Usingthesepreditedsoundlevels,

togetherwithhomedetailstakenfromtheMLSlistinginformationandadditionalobservationsmade

atthestreetfrontofeahtradedproperty,anumberofhedoniregressionfuntionswereestimated.

Thenalfuntionexplainsabout90%ofthevariationinpropertyvalues.

Thepreseneof aview wasnot foundto besigniantin anyof the regressionsrun. TheL

eq

observationswerealsonotsigniant,whileL

peak

observationswere. Fortheaveragepriedhomein

Wolfville,aninreaseinthepeaksoundlevelofonedeibelreduesthepriebyabouttwoperent.

Homes most subjetto noise pollution from highway 101 are pried some ten perent below the

average. When zoning lassiation is inluded in the regression, L

peak

eases to be statistially

signiant. This suggeststhat zoning lassiationsegregates homesexperiening dierent sound

levels. Thehighestpriezoninglassiationissinglefamilyresidential,whilelassiationswhih

allow homes to have up to four apartments are the lowest pried. Sine the prie dierene for

zoninglassiationexeedsthesoundlevelpriedierenes,zoningsegregationapturesmorethan

just noiseleveldierenes. As auniversitytownwithalargepopulationofstudenttenants,zoning

lassiationsalsoservestoseparatestudentrentalhousingfromthehomesofnon-studentWolfville

residents.

Theeienyofthissegregationdependsontherelativeimpatoftherelevantexternalitieson

theoupants-whether in multiple unit orsingle family. Ifoupants ofmultiple unit

aommo-dationarelesswillingtopaytoavoidnoiseleveldamagesthantenantsofsinglefamilyhomes,then

this segregation may be eient. Likewise, if oupantsof multiple unit aommodationare less

sensitivetopoormaintenaneandneighbourhoodharaterististhansinglefamilyresidents,then

segregationmaybeeient. Inontrast,ifbeinglosetosinglefamilyhomeshasbeneialspillovers

formultiplefamilytenants,suhasbetterenforementoflandlordmaintenaneresponsibilities,then

segregationmaynotbeeient. Regardlessoftheeonomieieny,thepresentpatternofzoning

segregationleadstooupantsofmultiplefamilyhousingbeingsubjetedtohigherlevelsofsound,

Wolfville is a small town, loated approximately 100 kilometers west-northwest of Halifax, Nova

Sotia. Itsprinipleeonomi driverisAadia University,with tourismplayingan importantrole

during thesummermonths. Thetourist appeal of Wolfville ispartlydue to themany stylish and

histori homes liningits main streetsand its proximity to theMinasBasin, at the easternend of

theBayofFundy. Thetownitself liesonthenorthernslopeofalowridge,aordingmanyhomes

an attrative view of the MinasBasin. However, alongthe southern boundary of the town, near

therestoftheridge,runs amajorprovinialhighway,Highway101. Theloationofthehighway

makesit asigniant soureof noise pollution, with tra noise being audible northof the town

site,morethanonekilometerfrom thehighwayitself.

The prominent highway south of Wolfville runs to the provinial apital, Halifax. Given the

proximity of Halifax, theeonomi hub of the provinial eonomy, many loal residentsroutinely

travel to the ity. There is onsiderablepolitial pressureto have thehighwayexpanded from its

urrenttwolanestatetoafourlanedividedhighway. Suhanexpansionisexpetedtobebeneial

totheloalarea,intermsofeasingtraveltoHalifaxandattratingmoreresidents. Thisstudywas

motivated by theonern that argumentsabout the'twinning'projetwere not onsidering some

potentialadverseeets, inpartiularinreasednoisepollution.

Themethodology of this analysis is Hedonipriing, an empirial implementation of the

Lan-aster harateristis model of a good (Lanaster, 1966), rst popularized by Rosen (1974). A

residentialpropertyisseenasabundleofharateristis. Purhaserspayattentiontothese

hara-teristis-lotsize,housearea,typeofzoning,distanefromamenities,et. whenpurhasingahouse.

Theyalso pay attention to environmental fatorssuh aspollution levels. This paperinvestigates

theimpatoftwoenvironmentalfators,theambientnoiselevelandthepreseneofaview,onthe

prieresidentialpropertytradesforin thetownofWolfville.

Anedotally, the adverse eet on property values of negativeexternalities suh asnoise level

is well known. These anedotes are reeted in the literature. Nelson (1982) reviews a number

of studies onduted in the 1970s, a time when onern about the noisepollution eets of large

infrastrutureprojetswasmounting. Inreviewingthe hedonipriingmethodologyused in these

studies,itispointedoutthatthreekeyassumptionsunderlaythisapproah. First,itisassumedthat

there issuientturnoverin themarket sothat buyershavethe'freedomto move'in response to

diereneinsoundlevel. Seond,theremustbesuientvariationinsoundlevelarossthesample

ofhousesforprieimpatstobedetetable. Third,itmustbepossibletomeasuresoundlevelsatan

appropriateresolution to beableto empiriallyestimate therelationshipbetween property values

andsoundlevels. Thestudiesreviewedmanagedtheseissuestovaryingdegrees. Theyndthat,on

average,aonedeibel(dB)inreaseinsoundlevelsleadstoa0.40%delineintheprieofahouse.

AmorereentreviewondutedfortheEuropeanCommission(Navrud,2002)surveysstudiesusing

hedonipriing,ontingentvaluation, hoieexperiment,andonjointanalysismethods. Thenoise

disountrangesbetween0.08%and 2.30% of theproperty prieperdeibel. Sine propertyvalue

impatsarepresentvaluesoftheongoingnoiseost,itisarguedthatanannualormonthlyimpat

isamoreappropriatemeasure. Fortranoise,noiseostsfallbetween2and99eurosperdeibel

perhouseholdperyear. Translated intoCanadiandollarsandassumingadisountrateof5%,the

presentvaluenoiseostisbetween$62and$3,100.

Areentstudy (Wilhelmsson,2000) onsidersthe impatof tra noiseonthe valueofsingle

familyhomesinSweden. Theauthorsonsideranumberofritiismsofthehedonipriingmethod,

inludingthepreseneofasymmetriinformationwithrespet tonoiselevels. Ifbuyersare

inom-pletelyinformedaboutnoiselevels,thenonewould expethigherturnoverratesnearnoisesoures

thanfurther away. Theyndnostatistialevidene tosupportdieringturnoverrates,suggesting

that asymmetri information with respet to noiseis not an issue. Theynd a noisedisount of

0.6%perdeibel,fromalog-linearmodel. Otherimportantvariablesinludehousesizeandquality,

andahousingprieindex. AstudybyTheebe(2004)usesspatialautoorrelationtehniquestolook

forarelationshipbetweennoiselevelsandpropertyvaluesforalargesampleoftransationsinthe

Netherlands. Theimplied per deibel disountis around 0.4%. Someweak evidene is found for

Sirmans (1992) examinetwo suburbs of Baton Rouge, Louisiana, omparing low and high tra

neighborhoods, and also looking fora diret relationship between tra ounts and housepries,

where ounts are available. They nd that there is a large and statistially signiant negative

relationshipbetween propertypries andtra, homes onhigh tra streetssell at adisountof

about8.8%. However,sinetheyrelyon tra levelitself asthevariableof interest,it is unlear

if tra noise, aident risk, pollution, orsome other fator related to tra level is driving the

delineinpropertyvalues.

Theimpat of airport noise on property valueshas reeived onsiderable attention. A reent

studybyLipsomb(2003)onsiderstheimpatofairportnoise,usingsoundlevelontoursreported

by aloal airport, onpropertyvalues in asmall ity near Atlanta, Georgia. Inontrastto many

otherstudies,it isfoundthatnoisedoesnotsigniantlyaetpropertyvalues. Thisis attributed

totheuniquedemographiharateristisofthisommunity,wheremanyhouseholdshavemembers

employedin airtravelrelatedoupations. Distane fromtheairportthereforedominatesnoiseas

adeisionvariableformanypurhasers.

A meta-analysis of the relationship between airport noise and property values onduted by

Nelson (2004) nds an average impat on selling pries of 0.58% per deibel, with the Canadian

subsetof thesamplegeneratingnoisedisountsofbetween0.8%and 0.9%perdeibel. The

meta-regressionattemptstoidentifywhetherdierentmethodsofdealing(orfailingtodeal)withmobility

andemploymentbenetsofairportproximity. Nodierenewasfoundamongthestudies,suggesting

that either thepositive eets of airport proximityare minimal, orthat none of the studieshave

eetivelyaountedforit. Thesurveyedstudiesalsoseemtoshowapositiverelationshipbetween

averageproperty prieandnoisedisount,with studieswhere theaveragepropertyprieishigher

ndingalargerdisount. Insofarashousingandquietarebothnormalgoods,thisisnotsurprising.

Onemethod that ommunities usetodeal withexternalitiesisthroughzoning. Asrestritions

on land use, zoning odes an prevent ativities whih generate large negative externalities from

loating where those externalities will be felt, and thereby protet ertain land uses from these

externalities. Theloationofommerialativitiesnearbusyroadwaysbothfailitatesaesstothe

businesses,and separatesresidentialpropertyfromthe externalitiesassoiatedwiththese business

ativities. Likewise, zoning low inome housing where negative externalities are more prevalent

servesto separate higherinome residents from boththe externalitiesdiretly related to building

designanddensity,andanyadditionalexternalities(rime,et.) assoiatedwithlowinomehousing.

Further,itmayalsoreduestheostofbuildinglowinomehousingbyreduingtheostofaquiring

thelandiflowinomehousingisloatedinplaes whereotherexternalitiesarestronger.

Anearlyempirialstudy(Creineetal.,1967)onsideredtheimpatofanumberofneighborhood

externalitiesonpropertyvalueforareaswithdierentzoninglassiations. Forsinglefamilyhomes,

noonsistenteet of possibleuseexternalitieswasfound in theperunit areaprie. Maseret al.

(1977) examine the impat of both zoning and a number of externalities on property values in

MonroeCounty,NewYork. Zoningdesignationisnotfoundtoaet propertyvalues,whileseveral

externalities(positivenearwater,positivenearpark,negativenearairport)do. Theauthorsonlude

thatexternalitiesarebeingappropriatelypriedbythemarket,andzoningrestritionsaretherefore

notontributingtoanoutomeanydierentfromthemarketoutome. PogodzinskiandSass(1991)

arguethat zoning restritionslimitbuyerhoieandsupplieroerings,and therebyimpatonthe

priing equation parameters. They nd that interations betweenzoning restritions and spei

harateristis an be signiant, and that the eet of zoning restritionsestimated absent these

interations an be biased. Basedontheir analysis ofSanta Clara County, zoning restritionsare

foundtosigniantlyaet thepriingequation.

Stull(1975)examinedtheimpatofneighborhoodexternalitiesbyomparingommunitieswithin

theBostonMetropolitanArea. Aerialphotoswereusedtoharaterizelanduseineahommunity.

The tradingprie of singlefamilyhomes wasnegativelyaeted byinreasesin the proportion of

mostotherlandusetypes. Thiseetistakenassupportfortheontentionthatzoningrestritions

an protet the value of single family homes. Asabere and Human (1997) examine the impat

of hierarhialzoning on property pries in entral Philadelphia. Hierarhial zoning provides a

butnotommerialuses,andanareazonedommerialwilladmit allthreeuses. It isargued that

with a hierarhialsystem, residential property in an area zoned to allow 'lower'uses should see

a priedisount. Fousing on the prie of apartment buildings, adisount of over15% is found.

In ontrast, for Santa Clara County, California, Cervero and Dunan (2004)nd a positiveprie

premiumformixeduseneighborhoodsrelativetosinglefamilyneighborhoods. However,theyargue

thatthismaybesomewhatunique,asSantaClaraisarapidlygrowingareawitharelativeshortage

ofaordablehousing. Asaresult,ondominiums sellwell, andsinglefamilyhomesin areaszoned

formixeduseareasmaybeapturingdevelopmentpotentialin theirprie.

Oneaspetoftheexlusionaordedbyzoningwithin theUnitedStateshasbeenasaneetive

means to segregateraial groups. Theprie depressingeet of beingin araially heterogeneous

neighborhood is ommonly seen. Creine et al. (1967) inlude the proportion of non-whites in

their various regressions, and nd that the eet of greater heterogeneity is generally negative.

Maseret al.(1977)inlude perentNegro in their regressions,and ndthat the eet on pries

isnegativeandsigniant. TheresultsofCerveroandDunan (2004)indiate that inreasingthe

raial mix in a neighborhood depressespries. Along another segregationdimension, Wang et al.

(1991) examine how the proximity of rental properties, aets sale pries. Theynd that owner

oupiedhomessellformorethanrentedhomes,that proximitytorental homesreduesprie,and

that theamountofrentalhomesin aneighborhoodalsoredues prie. Theirresultsare onsistent

with twoeets, atendenyoflandlords to invest lessin maintenane thanowneroupants, and

adesirefor higherinomeownersto segregate themselvesfrom lowerinomerenters. In asimilar

vein,AsabereandHuman (1997)inludes unemployment,andndsthathomesinneighborhoods

withhigherunemploymentratessellforless.

As a study site for examining environmental externalities suh as sound and view, Wolfville

providesseveralattrativeharateristis. Asauniversitytownwithnomajorindustrialativities,

varietyoflanduseisrelativelylimited. WithrespettotheassumptionslistedbyNelson(1982),the

relativelyhighinomemeans that budgetonstraintsarelikelyto havealimitedimpatonhouse

hoie, while the sound data olleted shows both a relatively large range and spatial variation,

with interpolation tehniques developing 'reasonable' estimates for eah property. The impat of

noise level in Wolfville is also less likelyto be onfounded by aess issues, as aess to highway

101 is not available within the town, and no major loal noise generator (exepting students) is

an important employer. Inomplete information on the part of buyers - partiularly new faulty

movingtoWolfvillefromfaraway-maybeaproblem. However,highwaysaregenerallywellknown

asnoisesoures,sothisisunlikelytobealargeissue. Further,therelativelyhighinomemakesthe

transationsostsassoiatedwithreloatingwithinthetownlessofanissueinWolfville,ompared

toothertowns. Wolfville,therefore,appearstobeanidealloationtomeasuretheimpatonhouse

priesofnoisepollution.

2 Data

TheompositionofthetownofWolfvilleisonsiderablydierentfromtheprovinialaveragesalong

manydemographidimensions. Althoughlikelyimportant,theseare notexpliitlyinludedin the

analysis,asdemographidataonindividualbuyersandsellersisnotavailable. However,itappears

toplayanimportantpartinexplainingsomeoftheresults. Somekeyfeatures,inludinginomeand

earnings,householdownership,eduation,andommutingmode,arehighlightedintable1. Among

thosewhoholddownafulltimejob,averageearningsare15%abovetheprovinialaverage.However,

themedianinomeis11%belowtheprovinialmedian. Studentearnings,whiharegenerallyquite

low, likelyexplains muhof this. Home ownership is well belowtheprovinialaverage,with 52%

ofdwellingsbeingrented. Again,thefat that Wolfvilleis auniversitytown, providing housingto

students, likelyaounts for muh of this. Anotheruniversity town eet is evidentin eduation

levels. The portion of the population with a university degree, diploma, orertiate is between

twoandthreetimestheprovinialaverage,dependingonagegroup. Likeeduation,theproportion

Wolfville NovaSotia

Total Perent Total Perent

Population 3,658 908,007 MedianInome 16,663 89 18,735 100 MedianAge 39.3 38.8 AverageEarnings 43,583 115 37,872 100 Privatehouseholds 1,615 100 360,020 100 Renteddwellings 840 52 103,305 29 Owneroupied 775 48 252,150 29

Perentofpopwithdegree,diploma,...

Aged20-34 38.7 22.8

Aged35-44 55.6 19.6

Aged45-64 59.0 18.1

Oupation-total 1,780 100 442,420 100

Soial siene,eduation,... 450 25 33,375 8

Art,ulture,rereation,andsport 165 9 11,125 3

Totaltripsto work 1,470 100 373,045 100

Tripsbyar, truk,orvan 1,045 71 280,365 85

Walkedorbiyled 365 25 33,130 9

mobilityis signiantlydierentin Wolfville, relativeto theprovinialaverage. Inpartiular,one

quarteroftheworkingpopulationommutesonfoot orbiyle.

During the summer of 2003, a student was hired to ollet sound measurements at various

loationsthroughoutthetown. UniversityemployeeswholivedinWolfvillewereaskedtovolunteer

theiryardsasasiteforanovernightmeasurement. Fromthevolunteeredproperties,asubsetwere

seletedtooerareasonablyomprehensiveoverage.Themeteringdevie,aLarson-Davis

TM Model

712soundmeter,waslokedtoanimmovableobjetin thebakyardof thevolunteeredproperty.

Thebakyardwas seletedbothforseurityofthereordingdevieandtobemorerepresentative

ofthat partofthe owner'syardwhere noiselevelswere mostlikelytobeaonern. A totalof27

sites weremonitoredin thisway,with twoextrapointsaddedto thedataset,dupliatingdatafor

the onehighwayobservation taken, and loated at two other points along thehighway. Figure 1

showstheloationofthesoundobservations,relativetothemajorroadsintheommunity. Ateah

site,thedataloggerreordedhourlymeasurementsforabout22hours. Fromthereordeddata,all

intervalsshorter than3600 seonds (onehour) were dropped,aswellasthe observations withthe

twohighest sound levels reorded. This wasto ontrol for ontattime withthe mahine,whih

ourred whenitwasset upand takendown, andto allowforshort durationextremeeventssuh

asheavydown-pours,lawnmowers,et. whihouldskewtheresults.

Asummary of the sound level data is reported in table 2. Soundlevelsare typially reported

in deibels(dB). Deibels area logarithmimeasurementsale, basedon thesquare of the sound

pressurelevel. Themeasurement is normallyaveragedoversometime interval. Forthis analysis,

measurementsarealulatedas anexponentialaverageoveraoneseondinterval,

L

p

(

t

) = 10 log

10

(1

/T

)

Z

t

t

s

p

(

ξ

)

2

e

−(t−ξ)/T

dξ/p

2

0

Thereferene level

p

0

forthe meter used is

20

µ

Pawith

t

s

=

t

−

T

and

T

= 1

seond. Thepeak

sound level reorded is the maximum L

p

measured overthe reording interval, whih was set to

onehour. Thisvalueis designatedL

peak

. A ommonlyused soundlevelmeasureis theequivalent

onstantsoundleveloverthereordinginterval. Thisisalulatedas

L

eq

= 10 log

10

"

Z

T

2

T

1

p

(

t

)

2

dt/p

2

0

(

T

2

−

T

1

)

#

(1)

−20

−10

0

10

20

30

−25

−20

−15

−10

−5

0

5

Easting

Northing

Monitoring Points and Transactions

Railroad Track

Main Street

Acadia University

Highway 101

10

19

24

27

Monitoring

Transactions

Figure1: MapofWolfvillewithsoundlevelmonitoringloationsandloationsofpropertiestraded.

Numbersidentifymonitoringsitesmentionedintable 2. Contoursmapaquadratiinterpolatedof

theL

peak

soundlevel.

for anintervalof length

T

2

−

T

1

. Sine sound levelsare measuredusing a logarithmisale, they

shouldproperlybemanipulatedgeometriallyratherthanarithmetially. Alternatively,thedeibel

measures an be onverted to a sound pressure level, and these values used for averagingand in

surfaeinterpolation. Thislatterapproahwasusedintheanalysisreportedinthispaper.

Togeneratesound levels forthesold properties, interpolationfrom theobservationstakenwas

neessary. Four dierentinterpolationmethods weretried. Forsets ofnearest neighbours,simple

average, inverse distane weighted average, and OLS forasting were used. Polynomial surfae

estimation wasalso applied to the entire set of sound observations. Based on explanatorypower

added to the hedoni regression model, and pereptions about the onsisteny of the graphially

representedprolewithloalpereptions,aquadratipolynomialsurfaewasused.

ThepointsofinterestweretheloationsofthepropertiesthathadbeensoldinWolfvillebetween

July1998andJune2003. Listingsdatawasolletedwiththehelpofaloalreal-estateagent. The

student assistant attempted to physially loateeah property, and if suessful assessed thesite

foranumberofqualitativevariablesnotinludedinthelistingdetail-preseneofagarage,paved

driveway,mature trees,aview ofthe MinasBasin,et. Thevariables measured,along with some

summarystatistis,arereportedintable4. Atotalof149propertytransationsarereordedinthe

datasetused. Duetomissingobservationsinkeyvariables,26ofthetransationsweredroppedfrom

thenal analysis. Betweentheyears1998and 2003,with noadjustmentfor ination,theaverage

prieforahomewas$136,770. WolfvilleisahistoriCanadiantown,whihisevidenedbythefat

thatamongthesoldhomes,theaverageagewas45.3years,withonehomeof176yearsoldtraded.

Wolfvilleisalsoauniversitytown,with theenrollmentatAadiauniversityrepresentingabout

half of the town's population during the university term. As suh, rental aommodation is an

important omponent of the loal real-estate market. A partiularly important form that rental

aommodationtakesinWolfville islargehousesonvertedinto multiple unitapartments. Within

thedata,theimpatofrentalaommodationisapparentasthepreseneofhomeswhih,forlisting

purposes, have up to 4 full bathrooms, 5 half bathrooms, and 7 bedrooms. The importane of

were reatedbyseleting twopointsalongthe highwayand assigning themthe sameobservations

asmade at the one sitethat was near the highway. The No. olumn reports identiers for map

sites(gure 1)

Averaging Site No. Mean St. Dev. Min. Max.

All Average - L

eq

47.6 6.16 35.3 67.9 Peak 82.8 9.40 61.2 110.2 Minimum 10 L

eq

41.8 2.39 38.0 46.1 Peak 80.1 8.62 65.5 101.7 Maximum Hwy L

eq

56.4 2.80 51.7 60.1 Peak 89.7 5.21 82.8 102.8 Day Average - L

eq

51.2 6.22 40.4 79.8 Peak 87.6 9.50 65.5 113.7 Minimum 27 L

eq

44.3 2.23 38.9 48.2 Peak 78.9 7.83 71.2 99.7 Maximum 24 L

eq

60.8 7.33 49.8 78.5 Peak 88.4 11.78 78.9 129.8 Night Average - L

eq

44.5 6.06 35.3 68.0 Peak 78.6 9.68 61.2 111.8 Minimum 19 L

eq

38.5 2.25 36.0 42.5 Peak 76.0 4.58 70.7 80.7 Maximum Hwy L

eq

54.2 2.36 51.7 58.7 Peak 87.9 3.54 82.8 95.6

legallyallowsomeformofrentalaommodation,and67werezoned insomeform ofmultipleunit

aommodation. The regressionresults presented below reet boththe importane that history

playsin theWolfville housingmarket,andtheimpatofthestudentrentalaommodation.

3 Results and Disussion

AsdisussedinCropperetal.(1988),itisunlearexatlywhatfuntionalformaHedoniregression

funtion should take. Several authors havetherefore used a Box-Cox transformation to evaluate

whetheralinear,logarithmi,orotherfuntionalformbesttsthedata. Figure2plotsthelikelihood

funtion for theBox-Cox transformof the selling prie asthe dependent variable and a Box-Cox

transformationofthesquarerootofthesellingprieasthedependentvariable. Independentvariables

were nottransformed. The95%ondeneintervalontainsneither

λ

= 1

(linear) nor

λ

= 0

(log-linear) forthe untransformedase. However,

λ

= 0

.

5

(square root)annot berejeted. Whenthe

dependent variable is transformed and the likelihood funtion is again alulated, the estimated

Box-Cox parameter is not signiantly dierent from one. A fully transformed model, with the

square rootof the ontinuousindependent variablesinluded ratherthan their levels, generateda

slightlysmallermaximumlikelihoodvalueforthe

λ

estimateonthetransformedmodelthanwhen

λ

wasestimatedforthemodelwithsquarerootappliedonlytothehouseprie. Therefore,thefully

transformedmodelisnotreported.

Ingeneral,theexplanatorypowerofallthreefuntionalformsishigh. Theregressiondiagnostis

arereportedintable5,fortworegressionsofeahfuntionalform. Whenzoningisnotinlude,the

R

2

valuesrangebetween0.842and0.885.With zoninglassiationsinluded,the

R

2

valuesrange

from0.892to0.912.Asahekforspeiationerrors,theDurbin-Watsonstatistiisreported. It's

valuesdonotsuggestaproblem. TheBreush-Pagentestforheterosedastiityissigniantforthe

log-linand squareroot-lin versions of themodel whenzoning is inluded, but insigniantforthe

others. Forompleteness,White's(1980)heterosedastiityorretedovarianeestimated

P

values

able 3: List of ratio sale and dumm y v ariables, together with desriptiv e statistis.

Variable Desription Mean Median Min Max SalePrie Prie atwhih homeatuallysold 136,770 123,500 28,500 399,000 Age Ageofhome 45.3 25 0 176 Floor Areaoflivingspae, in

m

2

148.0 127.7 53.1 447.6 LotSize Areaoflot whihhouseoupies,in

m

2

1,119.0 958.1 0.0 12,100.0 FullBath Numberofbathroomswithafullbath 1.67 2 1 4 HalfBath Numberofbathroomswithoutafullbath 0.36 0 0 5 CenterDist Straightlinedistanetotown enter,in

km

0.607 0.881 0.134 1.510 MainDist Shortestdistaneto MainStreet,in

km

0.317 0.375 0.978 0.024 AadiaDist Straightlinedistanetoenterofampus,in

km

0.688 0.853 0.211 1.906 Bedrooms Numberofbedrooms 3.34 3 1 7 DaysListed Numberofdayspropertyonmarket 124.2 128.2 0 596 Leq Measurementofaveragesoundlevel,

db

47.18 46.09 40.99 54.65 Peak Measurementofpeaksoundlevel,

db

87.25 87.56 79.62 90.55 WellDum Iswatersoureawell(well=1)? 0.02 Town 0 1 SemiDum Semi-detahedorsinglefamily(single=1)? 0.95 Single 0 1 Sto2Dum Oneortwostories(twostories=1)? 0.31 One 0 1 ViewDum ViewoftheMinasBasin(yes=1)? 0.21 None 0 1 VHwyDum Viewofthehighway(yes=1)? 0.05 None 0 1 HistDum Ispropertydesignatedhistori (no=1)? 0.02 Not 0 1 PaveDum Isdrivewaypaved(yes=1)? 0.76 Paved 0 1

able 4: List of ategorial v ariables. In the regression, dummies are inluded for ea h p ossible alue of ategorial v ariable.

Name Desription Categories

Eletri Oil Wood Other HeatFa HeatingSoure 54 75 17 3

R-1 R-1A R-2/4 R-8 RCDD ZoneFa ZoningClassiation 44 38 39 15 13

None Free Attahed GaraFa TypeofGarage 92 21 27

Single Condo TypeFa SingleFamilyorCondominium 129 20

None Young Mature TreeFa Trees 25 62 53

Freehold LeaseHold Other TitleFa TitletoProperty 127 2 20

1998 1999 2000 2001 2002 YearFa Year 12 32 30 35 40

Q1 Q2 Q3 Q4 QuarFa Quarter 32 54 36 27

0.0

0.5

1.0

1.5

2.0

−1590

−1570

−1550

−1530

λ

log−Likelihood

95%

Raw Prices

0.0

0.5

1.0

1.5

2.0

−732

−728

−724

λ

log−Likelihood

95%

Square Root Prices

Figure 2: Likelihoodasafuntionof

λ

foraBox-Coxtransformationofthemodel.

Table5: RegressionDiagnostis

Linear Logarithmi SquareRoot

noZ withZ noZ withZ noZ withZ

R

2

0.885 0.912 0.842 0.892 0.874 0.911

F

25.830 29.235 17.952 23.223 23.235 28.683

P

F

0.000 0.000 0.000 0.000 0.000 0.000

df

94.000 90.000 94.000 90.000 94.000 90.000 Durbin-Watson 1.923 2.166 1.808 2.196 1.871 2.208

P

DW

0.200 0.640 0.069 0.701 0.129 0.725 Breush-Pagen 25.098 39.757 28.054 53.495 23.469 56.834

P

BP

0.623 0.163 0.462 0.010 0.709 0.004 Moran'sI

−

0.004

−

0.019 0.002

−

0.016

−

0.001

−

0.018

P

I

0.535 0.069 0.101 0.175 0.235 0.088

The residuals were also tested for spatial orrelation by alulating Moran's I statisti (Moran,

1948;Anselin,1988;AnselinandBera,1998),aspatialanalogtotheDurbin-Watsonstatisti. The

reportedresultusesaweightingmatrixwithinverseneighbordistanesasweights,forallneighbors.

Squarerootandsquaredinversedistaneswerealsotried,aswellasrestritingthesetofneighbors

tothosewithin smallerradii. Fornoneofthesewassignianefoundat theveperentlevel.

Given the Box-Cox results, only estimates for the square root of selling prie regressions are

reported(table6). Anumberofdierentdependentvariableswereonsidered,andstepwise

regres-sionmethodswereexploredtoidentifyvariableswhihmadethelargestontributions. However,the

theoretialinterplaybetweensomeofthekeyvariables,partiularlysoundlevelandzoning

lassi-ation,meantthatexlusiverelianeonstepwiseresultsouldmaskimportantrelationships. Thus,

thenalmodelinludedallvariablesthattheoretialreetionsuggestareimportant,inpreferene

tothoseseletedbythestepwiseproedure.

Thevariablesinludedintheregressionsfallintothree generalategories: household

harater-istis,neighbourhood or amenityvalues, andnuisanevariables. Householdharateristisinlude

age,oorspae,lotsize,numberofbathroomswithafullbath,numberofbathroomswithoutafull

bath, numberof bedrooms, householdwatersupplied byawell, soureofheat (eletri,oil,wood,

orother),andifthepropertyhasbeendesignatedashistori. Age,watersoure,andhistori

desig-nationareexpetedto aetsellingprienegatively. Ageasolderhomesaremoreostly(heating,

et.) to oupy and maintain, water soure as operating osts of a well exeeds osts of supply

ase of heat soure, the omparison ase is eletri, whih during thestudy period wasthe most

ostly method ofheating ahome. In allases,quadrati termsare expeted to havethe opposite

signtotheirlinearomplement,reetingadiminishingmarginaleet.

Taxes and assessedvaluehavenotbeeninluded. As thisregressionfouses on onetown, the

taxrateisonstantthroughoutthetown. Wewouldthereforeexpetthetaxbilltoexplainmostof

thevariationinprie,to theextentthat thevariationisaptured byassessedvalue. Totheextent

that assessedvalueauratelytraksthetruevalueof homesin Wolfville, itis endogenous. Thus,

beyondlakoftaxratevariation,taxbillsthemselveswould alsobeendogenous.

Neighbourhood harateristis inlude distane to enter of town, distane to enter of Aadia

ampus,perpendiulardistanefromMainStreet,preseneofalearview,preseneofanobstruted

view,peaksoundlevel,astakenfromestimatedsoundprole,anddummyvariablesforzoning

las-siation. Thedistanevariablesareallexpetedtobenegative,astheseareimportantdestinations.

Preseneofaviewisexpetedtobepositive,withalearviewgeneratingalargerimpatthanan

obstrutedview. Peaksoundlevelisexpetedtobenegative,withitssquarepositive. Finally,from

anaiveperspetive,zoningodesareexpetedto bepositive,astheyprovidetheowneradditional

revenuegeneratingopportunities. However,thereviewedresearh suggeststhat zoning servesasa

segregationtoolandamethodofisolatingexternalties.Totheextentthatthiseetistakingplae,

zoningodedummiesmaybenegative.

Finally, dummy variables for year and quarter are inluded. These are onsidered nuisane

variables,astheirpreseneompliatestheregression,buttheirvaluesarenotthemain fous.

Most of the regression results are onsistent with expetations. In all ases where quadrati

termsare added,the expeteddiminishingeet is present. Among householdharateristis,age

and oorspae, together with thenumberof full bathbathrooms standoutpartiularly strongly.

Somewhatlessstrongin termsof

P

valuearethesize ofthelotandthenumberofhalfbathrooms.

Inpartiular, these variables loose signiane at

α

= 0

.

05

when theHCCM adjustmentis made.

Among variablessigniantat

α

= 0

.

10

, the'other' heat soure standsout. There areonly afew

observations in this ategory, with one being a geothermal heat exhange unit. This equipment

an substantially redue heating osts. The histori dummy is also signiant at

α

= 0

.

10

, and

this variable has a sign opposite to that expeted. Sine Wolfville is widely known as a histori

ommunity, perhapsthose hoosing to purhase property in Wolfville value this harateristi, in

spiteof therestritions imposed onmaintenane andrenovation. Of theremaining variables, well

hastheexpetedsignwhilebedroomsdoesnot. Althoughthe

P

valuesuggeststhatthisparameter

estimatehaslittleexplanatorypower,onepossibleexplanationfollowsfromthefatthatoorspae

hasbeenontrolledfor. Assuh,addingabedroomtoahomewithouthangingtheoorspaewill

reduethesizeofallotherroomsinthehouse.

Among the neighbourhood harateristis, to Aadia and to Main Street have the expeted

signs. Zoning lassiations are signiant and negative for three of the four ategory dummy

variables. The signs suggests that zoning is serving to protet the value of single family homes

from adverse impats more ommon where multiple family homes are permitted. The fat that

sound level beomes insigniantwhen zoning is inluded suggeststhat zoning is groupinghomes

intoategoriesexperieningwithsimilarnoiselevels. Whensoundlevelissigniant,theparameter

signsare oppositeto expetations. However,sinethe averagevalueof L

peak

is above80

dB

, the

marginalimpatat themeanisasexpeted. These marginalimpatsarereported below. Finally,

thedistanetotheenteroftownhasnoimpatonpropertyvalues,bothintermsofthemagnitude

oftheparameterestimateandin termsofitsstatistial signiane. Thetypeofviewalsofails to

besigniantat

α

= 0

.

05

, and itssign is theopposite ofexpetation. No learinterpretation for

thisresultisoered.

Forthe nuisanevariables, the year and quarter dummyvariables apture eets as expeted.

Overtime, theaverage prieat whih Wolfville homessellis inreasing. Also,relativeto therst

quarter (Januaryto Marh), home priesin theother quartersare higher. The ommon wisdom

holdsthatitisbest tosellinthespring. Fromtheresults,springpriesarehigherthanwinterand

summerpries. However,fallpriesarehighest. Again,asthese estimatesarefarfrom signiant,

are presented for regressions with and without zoning lassiations.

P

T r

signiane levels are

alulatedusing traditionalstandarderrors, while

P

H

are alulatedusingstandarderrors froma

heteroskedastiityorretedovarianematrix(HCCM).

WithZoning WithoutZoning

Fator

β

P

T r

P

H

β

P

T r

P

H

(Interept)

−

421.230 0.941 0.485

−

11843.983 0.039 0.062 Age

−

1.029 0.000 0.015

−

1.284 0.000 0.008 Age

2

0.008 0.001 0.041 0.008 0.001 0.055 Floor(

m

2

) 1.113 0.000 0.009 1.076 0.000 0.008 Floor

2

(

m

2

) -0.001 0.004 0.110 -0.001 0.008 0.103 Lot(

m

2

) 0.026 0.012 0.100 0.019 0.003 0.250 Lot

2

(

m

2

) -0.000 0.420 0.445 -0.000 0.124 0.459 FullBaths 31.317 0.000 0.000 37.140 0.000 0.000 HalfBaths 18.561 0.004 0.121 22.111 0.003 0.066 Bedrooms

−

2.811 0.490 0.311

−

2.107 0.647 0.361 Well

−

5.838 0.795 0.384

−

31.878 0.212 0.072 Heat: Oil 6.079 0.485 0.297 14.554 0.126 0.127 Heat: Other 43.609 0.072 0.325 36.692 0.187 0.330 Heat: Wood 3.661 0.757 0.405 16.186 0.211 0.147 Histori 46.487 0.070 0.296 51.780 0.079 0.255 to townenter(

km

) 0.002 0.628 0.352 0.003 0.456 0.293 toAadia(

km

)

−

1.328 0.015 0.022

−

1.249 0.038 0.049 toMainStreet(

km

)

−

3.421 0.005 0.007

−

2.082 0.065 0.051 Clearview

−

5.839 0.507 0.331

−

2.294 0.818 0.433 Obstruted view

−

13.465 0.077 0.073

−

10.528 0.227 0.154 Peak(

dB

) 8.551 0.949 0.487 279.518 0.037 0.059 Peak

2

(

dB

)

−

0.008 0.992 0.498

−

1.621 0.037 0.059 Zone: R-1A

−

26.034 0.007 0.023 Zone: R-2/4

−

58.668 0.000 0.000 Zone: R-8

−

14.244 0.566 0.310 Zone: RCDD

−

57.010 0.009 0.002 Year: 1999 14.105 0.264 0.167 19.056 0.193 0.154 Year: 2000 30.035 0.014 0.008 37.184 0.009 0.010 Year: 2001 50.444 0.000 0.000 59.134 0.000 0.001 Year: 2002 50.210 0.000 0.002 59.300 0.000 0.002 Quarter: Q2 7.415 0.416 0.245 6.042 0.564 0.318 Quarter: Q3 1.768 0.852 0.443 0.559 0.959 0.483 Quarter: Q4 10.100 0.376 0.278 2.108 0.870 0.453

househas theaverage valuesfor ratiosale variables. It issupplied with town water, has eletri

heat,doesnothaveaview,andwassoldintherstquarterof2000. Fortheregressionwithzoning,

itwasalsoasinglefamilyresidentialzoned home.

WithoutZoning With Zoning

Fator

∆

P rice

∆%

∆

P rice

∆%

Age

−

469.75

−

0.4

−

513.24

−

0.3 Floor(

m

2

) 483.30 0.0 517.16 0.0 Lot(

m

2

) 15.92 0.0 17.39 0.0 FullBaths 26,630.37 20.7 29,095.79 19.0 Half Baths 15,854.17 12.3 17,321.94 11.3 Bedrooms

−

1,510.87

−

1.2

−

1,650.74

−

1.1 Well

−

21,840.91

−

17.0

−

4,539.19

−

3.0 Heat: Oil 10,647.61 8.3 4,799.25 3.1 Heat: Wood 11,868.06 9.2 2,881.71 1.9 Heat: Other 27,655.59 21.5 36,065.30 23.5 Histori 39,808.61 31.0 38,579.43 25.1 totownenter (

km

) 1.95 0.0 2.13 0.0 toAadia(

km

)

−

895.50

−

0.7

−

978.40

−

0.6 toMainStreet(

km

)

−

1,493.05

−

1.2

−

1,631.27

−

1.1 Peak(

dB

)

−

2,618.05

−

2.0

−

2,860.43

−

1.9 Zone: R-1A

−

19,717.19

−

12.9 Zone: R-2/4

−

42,518.50

−

27.7 Zone: R-8

−

10,956.21

−

7.1 Zone: RCDD

−

41,411.97

−

27.0

Wolfville,forthesquarerootsaleprieregressions.Theaveragehouse,whihisalmost50yearsold,

suersapriedisountofabout$500foranadditionalyearofage. This disountis delining,and

beomespositiveataround80yearsofage. Anadditionalsquaremeterofoorspaeinreasesthe

priebyabout$500. Thisisapproximatelyhalf theareaostfornewonstrution. Anadditional

squaremeteroflotsizeaddslessthan$20totheprieofahome. Anadditionalfullbathroomadds

around20%totheprieoftheaveragehome,allotherthingsequal,andanadditionalhalfbathroom

adds about 12%to the prie. An additional bedroom redues the prieof an average home bya

littleoveroneperent. Theprieimpatsforwatersoureandheatsoureutuatesubstantiallyin

responsetowhetherornotzoningisinludedintheregression. Usingthemidpointoftheestimates,

thepresentvaluebenetof havingwood oroil heatis about $7,000. Ifthe relevantdisountrate

is 5%, then thehouseprie impatimplies that these heat souressaveabout $350peryear, and

iftherelevantdisountrateis10%,thentheysaveabout$700peryear. This isloosely onsistent

with anedotalevidene. Thenal homeharateristiishistori designation, whih inreasesthe

priebyalmost$40,000.

Amongneighbourhoodharateristis,thedistanetothetownenterhasasmallpositiveeet.

Theaveragehomebuyerpaysabouttwodollarstobeanextrakilometerawayfromthetownenter.

Inontrast,theaveragebuyerpaysalmost$1,000tobeakilometerlosertoAadiauniversity,and

around$1,500to bea kilometerloserto MainStreet. Onekilometeris approximatelythe width

ofthetown, andmovingonekilometerawayfromMainStreetrepresentsanelevationgainofmore

than 50 meters. Sine about one quarter of Wolfville residents biyle or walk to work, this hill

mayrepresentanimportantdeisioninhomeloationhoie. Aonedeibelinreaseinpeaksound

level dereases the prie of the average house by about $2,700, a little under two perent of the

prie. Thisis in therangereportedbyother studies. Inso far asquiet isanormalgood, andthe

averageinomeofWolfville homepurhasers ishigh, itseemsreasonablethat thepriedisountis

in theupperrangeof valuesreported in other studies. Finally, the impatof zoning lassiation

stands outpartiularly strong. Properties zoned R-1Aallow onerental suite, R-2/4 allowsup to

fourapartmentsinahouse,R-8allowsuptoeightapartments,andRCDDisageneraldevelopment

ategory, residential omprehensive developmentdistrit. The dierene between R-2/4and R-1,

more than $40,000, is greater than the prie dierene observed between the loudest and most

quite parts of Wolfville, about15%. In sofar aszoning is segregatingbased onexternalities, the

segregationisapturingmorethansoundleveleets.

Giventhat thesound leveldisountisnotadequateto explainthezoning ode priingimpat,

thisimpatlikelyreetsotherharateristisoftheWolfvillehousingmarket. Asdisussedabove,

oneof these is the importane of student rental aommodation. This rental market has reated

a pattern of zoning whih plaes a onentrationof multiple unit housing in the neighborhood of

the university ampus. In so far ashome buyersdo not desireliving with university students as

neighbors (externality eets suh asloud parties, fears aboutbehaviorshildrenmaybe exposed

to, et.), demand is likely lowerforhomes near theuniversitywhih arezoned for multiple units.

This fat may be ompounded by renovation osts. Many multiple unit houses are largersingle

familyhomeswhihhavebeenonvertedintosuites. Anyonepurhasingsuhapropertyforuseas

afamilyhomewouldfae signiantrenovationosts. Thesebuyerswouldthereforenotbewilling

topayashighaprieformanyoftheR-2/4orR-8zonedhomes,asforanR-1zonedhomewhih

requireslittleornomodiation. TheR-1Aeetissurprising,assuhahouseisunlikelytorequire

muhmodiation. However,sinetheownerofahouseanalwaysrentitto agroupofstudents,

proximity to the university may be a key variable as well in determining the presene of rental

housingrelatedexternalities.

Akeyquestioniswhether zoning in Wolfville iswelfareimproving. Ohls et al.(1974)desribe

two purposes for zoning restritions. Externality zoning is land use restritions to minimize the

impatofexternalities. SuhzoninganbeParetoimproving. Fisalzoningrestritionsaremanage

propertyusetoahieveasalobjetivesuhasminimizingtaxrates. Courant(1976)usesageneral

equilibriummodelofametropolitanarea,basedontheworkofOhlsetal.,toshowthatsalzoning

an only inrease property pries and thereby redue onsumer welfare. Whether or not zoning

Othermethodsofontrollingtheseexternalitiesinludenoiseandlitterregulationsandmaintenane

standards. Enforementoftenantbehaviorislikelydiultwithtransitorytenantssuhasstudents,

sothatusingsuhregulationsislikelytoinreaselandlordosts. Totheextentthatlandlordshave

disproportionatepolitialpower-notunlikelyin aommunitywithsuhahigh portion ofrenting

residents-zoningregulationswillbethepreferredinstrument.

Akeyquestionin analyzingtheeieny of zoning ishowtheexternalitiesaet the involved

parties. Ingeneral,theargumentisthatowner-oupiedpropertiesarenegativelyaetedbybeing

adjaent to renter oupied properties. Renters, or their landlords, are less likely to maintain

the rented property to thesamestandard asanowner-oupierwould. This generates anegative

externality to the owner-oupierneighbour. A question seldom disussed is whether the

owner-oupiergeneratesapositiveexternalityfortherenter. Twomehanismsmayexistforsuhaneet.

First,therentermayenjoyviewingthewellmaintainedhomesandyardsofnearbyowner-oupiers.

Seond, neighbouring owner-oupiers may demand a higher standard of their renter neighbours

and/ortheirlandlordsthanwouldbeexpetediftheneighbourisanotherrentalproperty. Ifthese

positiveexternalitiesexist, then theeientzoning pattern may involvemanysmall zones rather

thanasmallnumberoflargezoning ategories.

As pointed out by Pogodzinski and Sass (1991), it may be unreasonable to assume that shift

parametersaresuienttoapturetheimpatofzoningonthepriingequation. Regressionswere

thereforeruninteratingthezoninglassiationwithanumberofontinuousregressors-age,living

spae, lot size,distane to Aadia, et. Stepwiseregressionsretained anumberofthese interated

variables. However,theindividualparameterestimateswerefarfromsigniant. Thissuggeststhat

thepriingequationlikelydoesdierbetweenzoningtypes. However,multiollinearityand/orsmall

samplesizepreludeaurateestimationofthiseet. Further,sineboththesignsandmagnitudes

of the parameter estimates didnot hange substantially, results for theinteration termsare not

reported.

Severalvariables,suhastypeofownership(freeholdvsleasehold),styleofhouse(semi-detahed

ordetahed),typeofhouse(singlefamilyorondominium),et. wereinludedintheinitialmodels

asdummyvariables. Noneof thedummies generatedsigniantoeients,andallweredropped

throughthestepwiseproess. Itanbearguedthatdierentownershiptypes,housestyles,orhouse

typesmaygeneratedierentpriingfuntions. Thedatasetwasnotlargeenoughtoallowamodel

withthis diversityofeetsto beestimated. Tolimitpotentiallyonfoundingfators,nal results

wereestimatedwithoutinludingondominiumsoranypropertieswherethetitlewasnotfreehold.

WithrespettothepossibletwinningofHighway101,thisstudysuggeststhatpeaksoundevents,

suh aspassing trator-trailer units, are reeted in property pries. If twinning inreases tra

speed,then peak sound levels will alsoinrease. If300homes, about onequarterof thehomes in

Wolfville, experieneanaveragesound levelinreaseof onedeibel,thetotaldamageostisabout

$810,000. This amount needs to be ompared to the ost of measures to redue noise pollution

assoiatedwiththehighwayexpansion.

The results of this analysis suggest that the most important externalities aeting Wolfville

propertyvaluesrelatetostudenthousing. Whetherzoninglargetratsneartheuniversityfor

multi-familyresidentialisthemosteientmethodtomanagethisexternalityisnotlear. Thisapproah

hastheapparentadvantageofplaingtheburdenof theexternality onthosethat generateit, the

students. However,totheextentthattheexternalityisgeneratedbylandlordswhoareabletoinvest

relativelylittle in maintenane, this advantagemaybe illusory. Studentghettos permit landlords

to minimize maintenane asthetenantsare highly transitoryand unfamiliar with their rights. If

studenthousingwasinmixeduseneighborhoods,pressureonlandlordstomaintaintheirproperties

would likelybehigher. Ifthispressureissuientto raisethemaintenanestandardenough,then

thewelfare ofresident-ownersneednotbeadverselyaeted, while thewelfare of studenttenants

Theresultsoftheanalysisreportedinthispapersuggestthatmanyofthefatorsaetingproperty

values in Wolfville, NovaSotia, are the same as those found elsewhere. In partiular, property

valuesare inreasing in theareaof the house, the areaof the lot,and the numberof bathrooms.

Ofthe twoexternalitiesmeasured-soundlevelsandthepreseneofaview,onlypeaksound level

wasfoundtobesigniant. Attheaveragehouseprie,aonedeibelinreaseinpeaksoundlevels

redues the house prie by just under two perent. Two interestingresults stand out. First, the

impathouseagehasonprieisnotthatlarge,andreahesthemaximumdisountatabouteighty

years. Further,thereis apositivepremiumattahedto historiproperties. Purhasersin Wolfville

appearwillingtopayapremiumforolderhomes. Seond,thereisastrongnegativeeetofzoning

designationsthatallowrentalaommodation. SineWolfvilleisauniversitytown,thisislikelydue

toa'studentghetto'eet. Giventheuniquenatureofuniversitytowns-adisproportionatelylarge

numberofresidentswhoarebothhighlytransientandunfamiliarwithtenantrights-furtherwork

isneededtoestablishwhetherzoningthat aommodatesstudentghettosiswelfareimproving.

Referenes

Anselin,L.,1988.SpatialEonometris: MethodsandModels.KluwerAademiPublishers.

Anselin,L.,Bera,A.K.,1998.Spatialdependeneinlinearregressionmodelswithandintrodution

tospatialeonometris.In: Ullah,A.,Giles,D.(Eds.),HandbookofAppliedEonomiStatistis.

MarelDekker,Ch. 7,pp.237289.

Asabere, P. K., Human, F. E., 1997. Heirarhial zoning,inompatible usesand prie disounts.

RealEstateEonomis25(3),439451.

Cervero, R., Dunan, M., 2004.Neighbourhood ompositionand residentialland pries: Does

ex-lusionraisorlowervalues? UrbanStudies41(2),299315.

Courant, P.N., 1976. Onthe eet of salzoning onland and housing values.Journalof Urban

Eonomis3(1),8894.

Creine, J.P., Davis, O.A., Jakson,J.E., 1967.Urbanpropertymarkets: Someempirial results

andtheirimpliationsformuniipalzoning.JournalofLawandEonomis10,7999.

Cropper,M. L.,Dek,L.B., MConnel,K. E., 1988.Onthehoieof funtionalform forhedoni

priefuntions.TheReviewofEonomisandStatistis70(4),668675.

HughesJr.,W.T.,Sirmans,C.F.,1992.Traexternalitiesandsinglefamilyhousepries.Journal

ofRegionalSiene32(4),487500.

Lanaster,K.J.,1966.Anewapproahtoonsumertheory.TheJournalofPolitialEonomy74(2),

132157.

Lipsomb,C.,2003.Smallitiesmatter,too: Theimpatsofanairport andloalinfrastrutureon

housing priesin a small urban ity. Review of Urban & RegionalDevelopment Studies 15 (3),

255273.

Maser,S. M.,Rosett,R.N., Riker,W.H.,1977.Theeetsofzoningandexternalitiesontheprie

of land: An empirial analysis of monroeounty, new york. Journalof Law and Eonomis 20,

111132.

Moran, P., 1948.The interpretationof statistial maps.Journalof theRoyalStatistialSoietyB

10,24351.

Navrud, S.,2002. Thestate-of-the-artoneonomi valuation ofnoise.Teh. rep., European

Poliy16(2),117138.

Nelson, J. P., 2004. Meta-analysis of airport noise and hedoni property values: Problems and

prospets.JournalofTransportEonomis andPoliy 38(1),128.

Ohls, J. C., Weisberg, R. C., White, M. J., 1974. The eet of zoning on land value. Journal of

UrbanEonomis1(4),428444.

Pogodzinski,J., Sass, T. R., 1991. Zoning and hedoni housing priemodels. Journalof Housing

Eonomis1(3),271292.

Rosen,S.,1974. Hedonipriesand impliitmarkets: produt dierentiation in pureompetition.

JournalofPolitialEonomy82,3455.

StatsCan,2001.Statistisanadaommunityproles.http://www.statsan.a/.

Stull, W. J.,1975. Community environment, zoning,andthe market valueofsingle-family homes.

JournalofLawandEonomis 18(2), 535557.

Theebe,M.A.J.,2004.Planes,trains,andautomobiles: Theimpatoftranoiseonhousepries.

JournalofRealEstateandEonomis28(2/3), 209234.

Wang,K.,Grissom,T.V.,Webb,J.R.,Spellman, L.,1991.Theimpatofrentalpropertiesonthe

valueof single-familyresidenes.JournalofUrbanEonomis30(2),152166.

White, H., 1980.A heterskedastionsistent ovarianematrix estimatorand adiret test of

het-eroskedastiity.Eonometria 48,817838.

Wilhelmsson,M.,2000.Theimpatoftranoiseonthevaluesofsingle-familyhouses.Journalof

EnvironmentalPlanningandManagement43(6),799815.

A Alternative Surfae Generation Methods

With 27 noise level observation sites distributed unevenly around Wolfville, it was neessary to

projetfrom these loationsto theproperties that traded. Fourdierentmethods were used: (1)

simpleaverage,(2)distaneweightedaverage,(3)spatial OLSforeast,and(4)polynomialsurfae

generation.

Simple Average Thesimpleaveragewasalulatedas

L

i

= 10 log

10







1

P

j∈N

i

(n)

#

T

j

X

j

∈

N

i

(n)





X

k∈T

j

10

L

jk

/10

1/2











2

(2)

where

N

i

(

n

)

is a set indexing the

n

nearest neighbor measurement sites of sold property

i

,

T

j

is

a set indexing the observations made at site

j

,

L

jk

is the deibel sound levelmeasured at site

j

,

observation

k

,and

#

T

j

isthenumberof elementsinset

T

j

.

Distane Weighted Average Thedistaneweightedaveragewasalulatedas

L

i

= 10 log

10







X

j∈N

i

(n)

w

j





X

k∈T

j

10

L

jk

/10

1/2











2

(3) with

w

j

=

P

#

T

j

d

ij

k∈N

i

(n)

#

T

k

d

ik

(4)

P

wasformedwithallthesoundlevelobservationsforthe

N

i

(

n

)

nearestneighborobservationsites,

where

p

jk

= 10

L

jk

/10

1/2

. This vetorwas then regressedonan interept and vetors

X

and

Y

ontainingtheoordinatesoftheobservationsin

P

,as

P

=

β

0

+

β

X

X

+

β

Y

Y

+

U

(5)

where

U

isadisturbanevetor. Thesound levelatsoldproperty

i

wasthenforeastas

L

i

= 10 log

10

ˆ

β

0

+ ˆ

β

X

x

i

+ ˆ

β

Y

y

i

2

(6)

where

x

i

and

y

i

aretheoordinatesofsoldproperty

i

.

Polynomial Surfae Generation To generate this surfae, a polynomial regression was run

usingallofthesoundobservations. Theindividualobservationsweretransformedtosoundpressure

valuesasabove,andthenaregressionwasrunas

P

=

β

0

+

β

X

X

+

β

Y

Y

+

β

XX

X

2

+

β

XY

XY

+

β

Y Y

Y

2

+

. . .

+

U

(7)

forvariouspolynomialorders. Thedeibelsoundlevelat anysiteisthenforeastaordingto

L

i

= 10 log

10

ˆ

β

0

+ ˆ

β

X

x

i

+ ˆ

β

Y

y

i

+ ˆ

β

XX

x

2

i

+ ˆ

β

XY

x

i

y

i

+

. . .

(8)

where

x

i

and

y

i

aretheoordinatesofthesoldproperty

i

.

An example of the sound proles generated by an implementation of eah of the methods is

shown in gure3. Eah of themethods that usesnearest neighbors is implementedusing the six

nearestneighbors.Thepolynomialsurfaeisgeneratedusingaseondorder(quadrati)polynomial.

Thegreatestheterogeneityin soundlevelsoursfortheOLSprojetions. Theaveragingmethods,

simple and weighted, are less heterogeneous than the OLS approah, but not as smooth as the

polynomialsurfae.Giventhetopographyofthetown,knownloationsofsoundbarriers,alongthe

highway,andanedotalevidene aboutwhihparts oftownaremostquiet,thepolynomialsurfae

has thebest 't'. It is therefore used for thebalane of the analysesreported in the body of the

Easting

Northing

−25

−15

−5

0

5

a) Simple Average

Easting

Northing

−20

−10

0

10

20

30

−25

−15

−5

0

5

c) Least Squares

Easting

b) Weighted Average

Easting

−20

−10

0

10

20

30

d) Polynomial Surface

3: W olfville sound lev el proles generated using four dieren t in terp olation metho ds and eak sound lev el observ ations. Neigh b or based metho ds (a, b, and ) use 6 neigh b ors. P olynomial is seond order. Dark er olors orresp ond to a lo w er sound lev el. 20

MPRA http://mpra.ub.uni-muenchen.de/6145/