school-age children in rental units in new jersey: results

School-Age Children in Rental Units in New Jersey:Results from a Survey of Developers and Property Managers

Morris A. Davis, Ph.D.

David Frame, Ph.D.

Ronald S. Ladell

Debra Tantleff

July 2018

WHITE PAPER SERIES

©2018 Rutgers, the State University of New Jersey. May not be reproduced without express consent.

TableofContents

ExecutiveSummary........................................................................1

SurveyDesign.................................................................................5

Analysis..........................................................................................8

Conclusions..................................................................................15

AppendixA...................................................................................16

AppendixB...................................................................................19

Executive Summary

Inthispaperweprovidenewestimatesofthenumberofschool-agechildrenassociatedwithnewdevelopmentsofmarket-rateandaffordablerentalunitsinNewJersey.Givenourknowledgeofandrelationshipsacrosstheindustry,wedesignedandconductedalarge-scalesurveycompletedbydevelopersandpropertymanagersofmulti-familyrentalbuildings.Amongotherquestions,weaskedsurveyrespondentstoprovidecountsofmarket-rateandaffordableunits,childrenaged5-17(“school-agechildren”),andaveragehouseholdincome.

Ourdataandanalysisshowthataone-sizefitsallapproachisinappropriateforestimatingtheexpectednumberofschool-agechildrenarisingfromanewdevelopment.Instead,weshowthefollowingvariablesareessentialtoaccuratelypredictthenumberofschool-agechildrenarisingfromnewdevelopment:(i)thedistributionofthenumberofbedrooms,separatelyforaffordableandmarket-rateunits,(ii)theproducttypeofthedevelopment–High-rise,Mid-riseorLow-rise1–and(iii)theexpectedhouseholdincomeofmarket-rateresidents.Withthisinformation,theexpectednumberofschool-agechildrencanbedeterminedusingtheinformationinTable1below:2

Table1:School-AgeChildrenper100Units

MarketRateUnits–AverageHHIncomeAffordable <$50K* $50K-$100K* >$100K*

AllUnits High-riseorMid-rise Low-rise

High-riseorMid-rise Low-rise High-rise Mid-rise Low-rise

(1) (2) (3) (4) (5) (6) (7) (8)studioor1br 10.3 2.6 11.4 1.6 7.6 0.4 1.3 1.9

2br 72.1 43.6 126.4 13.4 56.7 2.2 8.9 28.23brandlarger 108.9 100.0 137.9 17.6 63.0 4.3 23.9 61.8

* Valuesrefertotheaveragehouseholdincomeofresidentsofmarket-rateunitsinthebuilding.

Eachelementofthetablerepresentsexpectedschool-agechildrenper100unitsforthespecificcharacteristicslisted.3TherowsofTable1refertonumberofbedrooms.Thefirstcolumnreferstoschool-agechildrenforaffordableunitsonly.Columns(2)through(8)areestimatesformarket-rateresidents.Columns(2)and(3)arefordevelopmentswithaverageincomeofresidentsoflessthan$50thousandperyear,columns(4)and(5)arefordevelopmentswith

1Low-riseisdefinedasaTownhomeorabuildingwith1-3floors,Mid-riseas4-9floorsandHigh-riseas10+floors.2Theresultsinthisdataarecomputedusingunitsbuiltinanyyear.Forthelowest-incomecategory,columns2and3,samplesizesaretoosmalltocomputethistableusingonlydataonunitsbuiltafter2000.Fortheotherincomecategories,columns4-8,resultsaresimilarbutnotidenticalwhenwerestrictthesampletounitsbuiltafter2000.3Ouranalysiscoversallschool-agechildrenattendingeitherpublicorprivateschools.Datafromthe2015(5-Year)AmericanCommunitySurveyascollectedbytheUnitedStatesCensusBureausuggests12percentofallschool-agechildrenattendprivateschoolsinNewJersey.

1|RUTGERSCENTERFORREALESTATEWHITEPAPERSERIES

SCHOOL-AGECHILDRENINRENTALUNITSINNEWJERSEY|JULY2018


SCHOOL-AGE CHILDREN IN RENTAL UNITS IN NEW JERSEY | JULY 2018

averageincomeofresidentsbetween$50and$100thousandperyearandcolumns(6)through(8) arefordevelopmentswithanaverageincomeofresidentsofover$100thousandperyear. Thispartitionofincomesplitsoursampleroughlyintothirds,althoughwedemonstratelater theaverageincomeoftheresidentsofmorethan90%oftheunitsbuiltsince2000isatleast$100,000peryear.Duetosample-sizelimitations,wecombineHigh-riseandMid-rise developmentswhentheaverageincomeofthemarket-rateresidentsislessthan$100,000.

Therearefourfairlynon-controversialresultsthatareimmediatelyapparentinTable1.First,foranyincomelevelandbuildingproducttype,thenumberofschool-agechildrenincreaseswiththenumberofbedrooms.Second,foranygivennumberofbedroomsandproducttype,thenumberofschool-agechildrendecreasesasincomeincreases.AscanbeseeninFigures1and2below,higher-incomehouseholdshavelowerbirthratesandaremorelikelytoown(ratherthanrent)theirhomes.Third,holdingincomeandnumberofbedroomsfixed,school-agechildrenincreasesastheproducttypebecomeslessdense,i.e.therearemorechildreninLow-risethaninHigh-risedevelopments.Finally,theresultsforaffordableunitsandformarket-unitsonbuildingswithaverageincomelessthan$50,000peryeararequitesimilar,addingcredibilitytotheaccuracyofthesurveyasresidentslivinginaffordableunitsare,bydefinition,lowincomehouseholds.

Figure1



WeillustratehowtousetheinformationinTable1withtwoexamples.Theseexampleshighlightthepossibilityoflargedifferencesintheestimatednumberofschool-agechildrenassociatedwithdifferenttypesofdevelopmentsandresidents.

Example1–Highincome,High-riseA200unitHigh-riseprojectwithaverageincomeofresidentsofmarket-rateunitsof$125,000:

• 30affordableunits(15%):6onebedroom,17twobedroom,7threebedroom• 170marketrateunits:85onebedroom,76twobedroom,9threebedroom

Theexpectednumberofschool-childrenis23,calculatedasfollows:

• Affordableunits(useestimatesincolumn1):Expect20.5childreno 6onebedroomx(10.3/100)=0.6o 17twobedroomx(72.1/100)=12.3o 7threebedroomx(108.9/100)=7.6

• Marketunits(useestimatesincolumn6):expect2.4childreno 85studioandonebedroomx(0.4/100)=0.3o 76twobedroomx(2.2/100)=1.7o 9threebedroomx(4.3/100)=0.4

Giventhebuildingtype,theincomeofthemarket-rateresidents,andthedistributionofbedrooms,theaffordableunitsinthisbuildingareassociatedwith68.3childrenper100units,computedas100*(20.5/30),andthemarket-rateunitsareassociatedwith1.4childrenper100units,computedas100*(2.4/170).

Example2–Middle-income,Low-riseA200unitLow-riseprojectwithaverageincomeofresidentsofmarket-rateunitsof$75,000.Assumethenumberofunitsandthedistributionofbedroomsisthesameaswithexample1

• 30affordableunits(15%):6onebedroom,17twobedroom,7threebedroom• 170marketrateunits:85onebedroom,76twobedroom,9threebedroom

Theexpectednumberofschool-childrenis75,calculatedasfollows:

• Affordableunits(useestimatesincolumn1):Expect20.5children,sameasExample1o 6onebedroomx(10.3/100)=0.6o 17twobedroomx(72.1/100)=12.3o 7threebedroomx(108.9/100)=7.6

• Marketunits(useestimatesincolumn5):expect55.2childreno 85studioandonebedroomx(7.6/100)=6.5o 76twobedroomx(56.7/100)=43.0o 9threebedroomx(63.0/100)=5.7



Comparingexamples1and2highlightstheimportanceofcontrollingforproducttype,numberofbedrooms,andincomeofmarket-rateresidentswhendeterminingtheexpectednumberofschool-agechildren.Themarket-rateunitsinexample2areassociatedwith32.5childrenper100units(computedas100*55.2/170),a23timesincreaseintheincidenceofschool-agechildrenresidinginmarket-rateunitsrelativetotheestimateprovidedinexample1.

Figure2



SurveyDesign,DataCollectionandDataCleaning

AppendixAshowstheexact13-questionsurveythatwedistributed.Thesurveywasdesignedtobeshortandtothepointtoencourageasmanyrespondentsaspossible.Thefirst5questionscoveraddress,buildingcharacteristics,yearbuiltandcondo/rentalstatus;questions6and7coverrents,unitsandoccupancyformarket-rateunits;question8coversaffordableunitsandoccupancy;question9coversbuildingamenities;questions10and11coverschoolagechildrenandquestions12and13coverincomeofresidents.Questions11and13weredesignedtobe“backup”questionsintheeventtherespondentscouldnotanswer10and12.Wedidnotusequestions11and13asinalmosteverycaserespondentsthatansweredquestions11and13alsoanswered10and12.

ThefirstsurveysweredistributedinOctober,2016anddatacollectioncontinueduntilaboutJanuary,2018.4Mostrespondentscompletedtheformbyhandandsenttousascanoftheresults;afewrespondentssubmittedExcelspreadsheetscontainingtheinformation.Manyofthedeveloperswecontactedwereeagertoparticipatebutsome(wedonothaverecordsoftheexactnumber)didnotrespondtoourrequestsforinformationastheyeitherdidnotwanttosharetheinformationwithusforcompetitivereasons,ortheydidnotcollecttheinformation.Wepromisedalldevelopersandsurveyrespondentsthatthedatacollectedwouldnotbesharedorposted.

Therawsurveyresponsesneeded”cleaning”beforewecouldbeginanalysis.Insomeinstancesregardingthenumberofoccupiedandvacantunits,questions7and8,wereceivedresponsesof“norecord”or“unknown,”orinsomeresponses,theindividualrowswerecrossedout.Inthesecases,andincaseswhentherowswereleftblank,weassumeavalueof0units.Forquestion10,theschool-agechildrenquestion,manyrowswereleftblankorcrossedout.Ifthesurveyrespondententeredatleastonenumberforschool-agechildreninquestion10,including0,weassumealltheothernon-numericentriesforthisquestionincludingblankresponsesare0andkeepthesurveyinourworkingsample.Ifthesurveyrespondentfailedtoenteratleastonenumericvalueforquestion10,weexcludethesurveyfromtheworkingsample.5Inthecaseofquestion12(householdincome),weassumethesedataaremissingwhentheresponseisblankoriftherespondentwrote“unknown”andultimatelydonotusethesesurveyswhencomputingschool-agechildrenformarket-rateunits.

AsummaryofoursampleexclusionsandcoverageisgiveninTable2a.Wecollected265surveysintotal.Twosurveyswerediscardedbecausetheywerefromage-restricteddevelopments(55+or62+).Onesurveywasdiscardedbecauseitwascompletedin2018andmanyoftheunitswerevacant.Foursurveyswerediscardedbecausetheywerelistedascondo

4WereceivedafewadditionalsurveysinApril,2018.5Werepeatthisstepseparatelyformarket-rateandaffordableunitsinthesamesurvey.Inotherwords,thesamesurveymighthaveuseableinformationformarket-rateunitsandnoneforaffordable,andvice-versa.



orcondoandrentaldevelopments.Sevenmoresurveyswerediscardedastheylistnovacantoroccupiedunitsatall.Theremaining251surveyscovermorethan40,000market-rateunitsandmorethan4,000affordableunits.Werefineoursamplebyeliminatingsurveyswithoutinformationaboutchildren6(row2)orincomeforhouseholdsresidinginmarket-rateunits(row3),yielding32,200marketrateunitsfrom143surveysand1,968affordableunitsfrom42surveys.Rows4and5listoursamplesizesofunitsbuiltbeforeorafter2000,respectively.

Table2A:SurveyResponsesandSampleSizes

Market Affordable TotalSample Units Surveys Units Surveys Surveys

Totalunits (1) 42,425 222 4,111 79 251*Unitsw/infoonchildren (2) 33,140 157 1,968 42 168(2) +infoonincome** (3) 32,200 143 1,968 42 157(3) forunitsbuilt<2000 (4) 21,138 92 677 13 94(3) forunitsbuilt>=2000 (5) 11,062 51 1,291 29 63

* Somesurveyscoverbothmarket-rateandaffordableunits.**Filterforavailabilityofincomedataonlyappliedtomarket-rateunits.

Table2B:StatisticsonUnitsperSurvey

Market AffordableProductType #Surveys Avg.Units Std.Dev. Range #Surveys Avg.Units Std.Dev. Range

(1) (2) (3) (4) (5) (6) (7) (8)High-rise (1) 11 721 1,238 140-4433 1 140 X XMid-rise (2) 30 203 119 6-504 12 36 20 8-63Low-rise (3) 102 178 210 2-1492 29 48 40 3-173

Table2BshowsstatisticsconcerningthenumberofunitseachsurveyrepresentsseparatelyforHigh-rise(column1),Mid-rise(column2)andLow-riseunits(column3).Thefirstfourcolumnsshowthestatisticsformarket-rateunits:Thenumberofsurveys,theaveragenumberofunitspersurvey,thestandarddeviationofthenumberofunitspersurvey(ameasureofspread),andthelowestandhighestnumberpersurvey,respectively.Columns5-8showthesamestatisticsfortheaffordableunitsinthesample.Generallyspeaking,asdensitydeclineseachsurveycoversfewerunits,butwehavemoresurveys.Ourmeasuresofspread,columns3-4and7-8showthatoursurveyscoverawidevarietyofapartmentandcommunitysizes.

6Thesearethesurveysforwhichquestion10wasleftentirelyblank,i.e.therespondentfailedtoenteratleastonenumericvalue(including0)forquestion10.



WecompareourestimatesoftheuniverseofrentalunitsthatwederivefromdatafromtheU.S.CensusBureauinAppendixB.AppendixTableB1showsoursampleforunitsbuiltinanyyear(row3ofTable2)andTableB2restrictstheanalysistounitsbuiltafter2000(row5oftable2).Overall,oursampleincludes3.5%ofallrentalunitsinNewJersey(TableB1).Wehavemorerepresentationofunitsbuiltafter2000,asouroverallcoverageratefortheseunitsis10.9%relativetothetotalbuiltasestimatedusingdatafromtheCensus(TableB2).TableB2showsthatoursampleincludesdataformorethan15percentoftherentalunitsbuiltsince2000inBergen,Hudson,Mercer,MonmouthandMorriscountiesandmorethan9percentoftheunitsbuiltinHunterdon,Passaic,SomersetandWarrencounties.Webelieveoursamplesizesaresufficientlylargeandrepresentativetoprovideinformativeandrobustresults.



AnalysisTable3belowreportsthevaluesofschoolagechildrenper100unitsfortheentiresample(rows1and2),unitsbuiltbefore2000(3and4)andunitsbuiltonorafter2000(5and6).Theoddrowsreportvaluesformarket-rateunitsandtheevenrowsreportvaluesforaffordableunits.Thefirstcolumnreportschildrenper100unitsandthelasttwocolumnsreporttotalchildrenandtotalunits.Childrenper100unitsiscomputedsimplyas100timeschildrendividedbyunits.Afewobservationsfromthistableareworthmentioning.Firstabouttwothirdsofaffordableunitsinoursamplewerebuiltafter2000(826outof1,238)whereasonlyonethirdofmarket-rateunitswerebuiltafter2000(11,062outof32,200).Second,thenumberofschool-agechildrenper100affordableunitsissignificantlyhigherthanthenumberofschool-agechildrenper100marketunits,regardlessofthetimeperiod.Third,thenumberofchildrenper100marketrateunitsforunitsbuiltafter2000,9.8,ismuchlowerthanforunitsbuiltbefore2000,25.9.

Table3:School-AgeChildrenper100Units

Children Total per100units Children Units

BuiltAnyYear (1) Market 20.4 6,561 32,200 (2) Affordable 62.9 1,238 1,968

BuiltBefore2000 (3) Market 25.9 5,477 21,138 (4) Affordable 60.9 412 677

BuiltAfter2000 (5) Market 9.8 1,084 11,062 (6) Affordable 64.0 826 1,291

Wearguethatolderandnewermarket-rateunitsdifferbyproducttype,thecompositionofthenumberofbedrooms,andtheincomeofmarket-rateresidentsandthesedifferencesaccountforthevariationinthenumberofschool-agechildrenper100unitsshowninTable3.ThefirstthreerowsofTable4showthedistributionofmarket-rateunitsbytypeofunit:studioandonebedroom(row1),twobedroom(row2)orthreebedroomsormore(row3).Theserowsshowthattheolderunitstendtohavemorestudioandonebedroomsandfewertwobedrooms,andallelseequalthisshouldimplythattheolderunitsshouldhavefewerchildren.Butallelseisnotequal:Thedistributionofincomeandthetypesofbuildingsareverydifferentandthesevariablesmatter.Rows4-6showthepercentagesofmarket-rateunitsoccupiedbyhouseholdsearninglessthan$50thousandperyear(row4),between$50and$100thousandperyear(row5),andmorethan$100thousandperyear(row6).Residentsofmarket-rateunitsbuiltafter2000havemuchhigherincomesthanresidentsofunitsbuiltbefore2000andasTable1demonstrates,foranygivennumberofbedrooms,higher-incomeresidentslivinginmarket-raterentalstendtohavefewerchildrenthanlower-incomeresidents(reflectingthe



relationshipbetweenbirthratesandhouseholdincomeshownearlierinFigure1).Finally,rows7-9showthatmorethan70percentoftheunitsbuiltbefore2000inoursampleareLow-riseunits,but70percentoftheunitsbuiltafter2000areeithermid-orhi-riseunits.Table1showsthatresidentsofLow-riseunitshavemorechildren.

Table4:DistributionofMarket-RateUnitsbyYearBuilt

BuiltBefore2000 BuiltAfter2000 #Units Percent #Units Percent

studio+1br (1) 13,016 61.6 5,570 50.42br (2) 7,080 33.5 4,956 44.8

3brandlarger (3) 1,042 4.9 536 4.8<$50K (4) 5,339 25.3 121 1.1

$50K-$100K (5) 9,946 47.1 861 7.8>$100K (6) 5,853 27.7 10,080 91.1High-rise (7) 5,373 25.4 2,561 23.2Mid-rise (8) 896 4.2 5,184 46.9Low-rise (9) 14,869 70.3 3,317 30.0

Table1reportsresultsusingdatafromourentiresample,whichimplicitlyassumestherearenointrinsicdifferencesinchildrenperunitbetweenbuildingsbuiltearlierthanoraftertheyear2000oncewecontrolforproducttype,numberofbedroomsandincomesofmarket-rateresidents.Table5showstheresultsfromaformalstatisticaltestofthisassumptionforaffordableunitsbuiltbeforeandafter2000.Row1showstheresultsforstudioandonebedroomapartments,row2fortwobedroomapartmentsandrow3forthreebedroomandlargerapartments.Column1showsestimatesfromthefullsample,column2showsresultsfortheolderunits,column3showsresultsforthenewerunitsandthelastcolumnshowstheresultofthestatisticaltestoftheequalityoftheestimatesincolumns2and3.A“No”meansthat,afterrunningaproperstatisticaltest,wecannotrejectthattheestimatesarethesame;a“Yes”meansthatwerejectthattheestimatesarethesame.7ThelastcolumnofTable5showsthatwecannotrejectthattheestimatesofchildrenper100unitsareequalforanybedroomtype.Eventhoughtheexactbefore-andafter-2000estimatesofschool-agechildrendiffer,thereisenoughuncertaintyaroundeachestimatethatonceweformallyaccountforthisuncertaintyastatisticaltestcannotrejecttheequalityoftheestimates.8

7The“nullhypothesis”isthattherearenodifferencesinschool-agechildrenbetweendifferenttypesofunitsandwetestifthathypothesiscanberejected.Exceptwhennoted,ourrejectioncriteriaisa5%probabilitylevel.Atthatlevel,5%ofthetimewewillrejectthattheestimatesarethesamewheninfacttheyareidentical.8Wehavealsoconfirmedthatforaffordableunitswecannotrejecttheequalityoftheestimatesofschool-agechildrenper100unitsacrossproducttypesateithera5%or10%probabilitylevel(notshown).



Table5:School-AgeChildrenper100Units,AffordableUnitsbyYearBuilt

FullSample BuiltBefore2000 BuiltAfter2000 Rejectat5%?* (1) (2) (3) (4)

studio+1br (1) 10.3 13.6 6.9 No2br (2) 72.1 96.0 61.5 No**

3brandlarger (3) 108.9 106.8 109.5 No*Thismarkswhetherwecanrejectthenullhypothesisoftheequalityofthe“Before2000”and“After2000”estimatesata5%probabilitylevelusinganF-testthatexplicitlyaccountsforsamplinguncertainty.**Wecanrejectata10%probabilitylevelbutnotat5%.

Table6:School-AgeChildrenper100Units,Market-RateUnitsbyBuildingProductType

Income Bedrooms Full

Sample**High-rise+Mid-rise

High-rise Mid-rise Low-rise Rejectat5%?*

(1) (2) (3) (4) (5) (6) studio+1br (1) 11.0 2.6 11.4 Yes

<$50K 2br (2) 120.0 43.6 126.4 Yes 3brandlarger (3) 135.7 100.0*** 137.9 X studio+1br (4) 6.8 1.6 7.6 Yes

$50K-$100K 2br (5) 49.3 13.4 56.7 Yes 3brandlarger (6) 55.1 17.6 63.0 Yes studio+1br (7) 1.0 0.4 1.3 1.9 Yes

>$100K 2br (8) 12.2 2.2 8.9 28.2 Yes 3brandlarger (9) 26.5 4.3 23.9 61.8 Yes

*Thismarkswhetherwecanrejectthenullhypothesisoftheequalityofthe“High-rise”,“Mid-rise”and“Low-rise”estimatesata5%probabilitylevelusinganF-testthatexplicitlyaccountsforsamplinguncertainty.**Notethatweexclude205unitswithoutareportedproducttypeand53unitswithareportforproducttypeas“Townhome,Mid-rise”).***Thisestimateisfromexactlyonesurvey;weignoretheoutcomeofthestatisticaltestforthisrow.

Table6providessomeperspectiveforwhyweemphasizeproducttypeasanimportantdeterminantofschool-agechildren.Thistableshowstestsoftheequalityofschool-agecoefficientsformarket-rateunitsforvariousproducttypesaftercontrollingforincomeandnumberofbedrooms.Resultsforincomelessthan$50thousandareshowninrows1-3;rows4-6showresultsforincomesfrom$50to$100thousand;androws7-9showresultsforincomesgreaterthan100thousand.Column1showsestimatesofschool-agechildrenwhenallproducttypesarecombinedandcolumns2-5showresultswhensplitbyproducttype.Whenwestudyincomeslessthan$100thousand,duetosmallsamplesizeswecombineHigh-riseand



Mid-riseunits(column2)andcomparethattoLow-riseunits(column5);whenwestudyincomesof$100thousandormore,wehaveenoughdatatoseparatelycompareHigh-rise,Mid-riseandLow-riseunits(columns3-5).Column6showstheresultsofaformalstatisticaltestfortheequalityofthereportedestimatesincolumns2-5.Wheneverwerejecttheequalityoftheestimatesoftherateofschool-agechildren,wemarktheresultsusingredtext.Column6showsthatinalmostallcaseswecanrejecttheassumptionthatthenumberofschool-agechildrenisindependentofproducttypeaftercontrollingforincomeandnumberofbedrooms.Simplyput,therearemorechildreninlow-risebuildingsthaninmid-orhigh-risebuildings,evenaftercontrollingforincomeandnumberofbedrooms,andthisexplainswhyTable1includesproducttypeasafactor.

Unfortunately,itisverydifficultto“drilldown”furthertoseeifotherfactorsmatteroncewecontrolforproducttype,incomeandnumberofbedroomsaswelacksufficientobservationstorunrobuststatisticaltests.Thatsaid,weinvestigate,asbestaspossible,whether(a)livingwithinone-halfmileofatransitstopor(b)locatinginareaswithgoodschoolsaffectsestimatesofschool-agechildrenformarket-rateunits.OurresultsfortransitareinTables7and8andourresultsforschoolsareinTables8and9.

InTable7,wecontrolforincomeandnumberofbedrooms,temporarilyignoringproducttype,toseeifthereisanyobviousrelationshipbetweendistancetoatransithubandnumberofchildren.Column1showsestimatesofschool-agechildrenfortheentiresample.Columns2and3showestimatesofschool-agechildrenper100market-rateunitswhenthesampleissplitintounitsoutsideone-halfmileofatransithub(column2)andwithinone-halfmile(column3).Column4showstheresultofastatisticalprocedurethatteststheequalityoftheestimatesincolumns2and3.Thecellshighlightedinredshowthebedroom-incomecombinationswherewecannotruleoutthatproximitytoaTODiscorrelatedwithdifferencesinthenumberofchildren.Row2indicatesthatlower-incomerenterslivingproximatetoaTODtendtohavemoreschool-agechildrenandrows8-9showthathigher-incomerenterslivingclosetoaTODtendtohavelessschool-agechildren.

InTable8,wetrytoinvestigatefurtherbycontrollingforproducttype,butstudyingonly2bedroommarket-rateunitsduetosmallsamplesizesforone-andthree-bedroomunits.Wecombinemid-andhi-riseunitsforallincomecategoriestoincreasesamplesizes.Rows1-2reportresultsforincomeslessthan$50thousand,rows3-4areforincomesbetween$50and$100thousandandrows5-6areforincomesmorethan$100thousand.Eachrowrepresentstheresultsofadifferentproducttypefor2-bedroomunitsholdingincomefixed.TheresultsofTable8supportthoseofTable7.Focusingonrow2ofTable8:column3showsthatlower-incomehouseholdsrenting2-bedroomunitsinLow-risebuildingstendtohavemorechildrenwhenlocatedwithin0.5milesofatransitstop(147.4perhundredunits)thanwhenlocatedfartheraway(40.6childrenperhundredunits,shownincolumn1).Row5showsthattheexactoppositeistrueforhigh-incomerenterslivinginmid-orhi-risebuildings;theserenterstendto



havefewerchildrenwhenlocatednearatransithub.Thislatterresultisnotsignificantatthe5%probabilitylevel,butissignificantatthe10%levelwhichiswhyweusebluetextcolor.

Table7:School-AgeChildrenper100Units,Market-RateUnitsbyTOD

Income Bedrooms FullSample OutsideTOD InsideTOD Rejectat5%*? (1) (2) (3) (4) studio+1br (1) 11.0 5.8 13.2 No

<$50K 2br (2) 120.0 41.6 138.8 Yes 3brandlarger (3) 135.7 85.9 300.0** X studio+1br (4) 6.8 7.0 6.8 No

$50K-$100K 2br (5) 49.3 60.7 37.8 No 3brandlarger (6) 55.1 56.9 51.5 No studio+1br (7) 1.0 0.9 1.0 No

>$100K 2br (8) 12.2 24.4 7.1 Yes 3brandlarger (9) 26.5 56.1 16.3 Yes

*Thismarkswhetherwecanrejectthenullhypothesisoftheequalityofthe“OutsideTOD”and“InsideTOD”estimatesata5%probabilitylevelusinganF-testthatexplicitlyaccountsforsamplinguncertainty.**Thisestimateisfromexactlyonesurvey;weignoretheoutcomeofthestatisticaltestforthisrow.

Table8:School-AgeChildrenper100Units,2-BedroomMarket-RateUnitsbyTOD

Income ProductType OutsideTOD InsideTOD Rejectat5%?* (1) (2) (3)

<$50K Hi-andMid-Rise (1) 56.5** 41.3 X Low-rise (2) 40.6 147.4 Yes

$50K-$100K Hi-andMid-Rise (3) 17.1** 13.2 X Low-rise (4) 61.6 50.0 No

>$100K Hi-andMid-Rise (5) 16.0 3.7 No*** Low-rise (6) 28.8 27.2 No

*Thismarkswhetherwecanrejectthenullhypothesisoftheequalityofthe“OutsideTOD”and“InsideTOD”estimatesata5%probabilitylevelusinganF-testthatexplicitlyaccountsforsamplinguncertainty.**Thisestimateisfromexactlyonesurvey;weignoretheoutcomeofthestatisticaltestforthisrow.***Wecanrejectata10%probabilitylevelbutnotat5%.



Tables9and10havethesamelayoutasTables7and8,butshowresultsbasedonameasureofthequalityofthelocalschoolsystem.Foreachbuildinginoursample,welookuptheschoolscores(from1-10,with10thebest)forthepublicelementary,middleandhighschoolasassignedbygreatschools.org.9Weaveragethe3scoresandsplitoursamplebasedonwhethertheaverageislessthanorgreaterthan7.Asbefore,whenwecanrejecttheequalityoftheestimatesofschool-agechildrenper100unitsata5%probabilitylevel,wehighlighttheestimatesusingredtext;ifwecanonlyrejectequalityata10%probability,weusebluetext.Wefindtheresultsforschoolsareessentiallytheoppositeofthoseoftransit.Lower-incomerenterslivinginanareawithgoodschoolshavefewerchildren(rows2and5,Table9)andhigh-incomerenterslivinginanareawithgoodschoolshavemore(rows8and9,Table9).

SimilartoTable8,Table10expandsontheresultsofTable9byfocusingonlyon2bedroomunits,whichenablesustocontrolforbothincomeandproducttype.Rows2and4ofTable10confirmthathouseholdsearninglessthan$100thousandandlivinginLow-riseapartmentstendtohavefewerchildreniftheyliveinanareaswithgoodpublicschools.Rows5and6ofTable10suggeststhatoncewecontrolforproducttype,thepresenceofgoodpublicschoolsdoesnotimpacttheincidenceofschool-agechildrenforhigh-incomerenters.10

Summingup,webelievethatschool-qualityandproximitytoaTODmightbothbeimportantindeterminingschool-agechildren,andtheeffectsofeachmightvarybyincome,buthesitatetoalterthebaselineestimateswereportinTable1duetomixedresultsandsmallsamplesizes.

9Therankingsareavailableathttps://www.greatschools.org/new-jersey/.Thiswebsitedoesnotassigneveryaddresstoalocalpublicelementaryormiddleschool.Inthesecases,weusejudgmentastotherelevantpublicschool.10Onewaytoreconciletheresultsofrows8and9oftable9(showingthatpublicschoolsimpactschool-agechildrenforhigh-incomerenters)withtheresultsofrows5and6ofTable10(showingthatgoodpublicschoolsdonotimpactschool-agechildren)involvesastoryofsorting:high-incomerentersinareaswithgoodpublicschoolstendtoresideinLow-risebuildingsandhigh-incomerentersinareaswithworsepublicschoolstendtoresideinMid-andHigh-risebuildings.



Table9:School-AgeChildrenper100Units,Market-RateUnitsbyAvg.SchoolQuality

Income Bedrooms FullSample School<7 School>=7 Rejectat5%?* (1) (2) (3) (4) studio+1br (1) 11.0 11.2 X X

<$50K 2br (2) 120.0 120.7 44.4 Yes 3brandlarger (3) 135.7 135.7 X X studio+1br (4) 6.8 7.3 4.0 No

$50K-$100K 2br (5) 49.3 52.4 25.8 No** 3brandlarger (6) 55.1 53.9 400.0*** X studio+1br (7) 1.0 1.0 1.4 No

>$100K 2br (8) 12.2 9.5 23.9 No** 3brandlarger (9) 26.5 17.6 65.0 Yes

*Thismarkswhetherwecanrejectthenullhypothesisoftheequalityofthe“School<7”and“School>=7”estimatesata5%probabilitylevelusinganF-testthatexplicitlyaccountsforsamplinguncertainty.**Wewouldrejectata10%probabilitylevel.***Thisestimateisfromexactlyonesurvey;weignoretheoutcomeofthestatisticaltestforthisrow.

Table10:School-AgeChildrenper100Units,2-BedroomMarket-RateUnitsbyAvg.SchoolQuality

Income ProductType School<7 School>=7 Rejectat5%?* (1) (2) (3)

<$50K Hi-andMid-Rise (1) 43.6 X X Low-rise (2) 127.2 44.4 Yes

$50K-$100K Hi-andMid-Rise (3) 12.2 22.5*** X Low-rise (4) 60.8 26.5 No**

>$100K Hi-andMid-Rise (5) 5.7 2.6 No Low-rise (6) 25.9 30.8 No

*Thismarkswhetherwecanrejectthenullhypothesisoftheequalityofthe“School<7”and“School>=7”estimatesata5%probabilitylevelusinganF-testthatexplicitlyaccountsforsamplinguncertainty.**Wewouldrejectata10%probabilitylevel.***Thisestimateisfromexactlyonesurvey;weignoretheoutcomeofthestatisticaltestforthisrow.



ConclusionsBasedontheresultsofalargesurveyofbuilders,developersandpropertymanagersandusingnon-controversialmethodsweconstructa“matrix”inTable1toenabledevelopersandotherinterestedpartiestoestimatetheexpectednumberofschool-agechildrengeneratedbynewrentaldevelopment.Wecomputeschool-agechildrenforaffordableunitsbynumberofbedrooms;andformarket-rateunitswecomputeschool-agechildrenbynumberofbedrooms,producttype(Low-rise,Mid-riseandHigh-rise),andbyaveragehouseholdincomeoftheresidents.Weshowthatschool-agechildrenincreaseswiththenumberofbedrooms;decreaseswithhouseholdincome;anddecreaseswiththedensityoftheproducttype.Wedemonstrateusingtwoexamplesthata“onesizefitsall”approachfordeterminingthenumberofschool-agechildrenisinappropriate,asthevariationinducedbyallthesefactorscanbelarge.Ineachexample,weconsideranew200-unitdevelopmentwiththesamedistributionofbedroomtypesandwitha15%affordableset-aside.ThefirstdevelopmentisaHigh-riseprojectcateringtohigh-incomehouseholds,leadingto23school-agechildren.TheseconddevelopmentisaLow-riseprojectcateringtomid-incomehouseholds,leadingto75school-agechildren–anincreaseofnearlyafactorof3.Wealsostudytheimpactoflocationnearatransithubonschool-agechildren,aswellastheimpactoflocationassociatedwithahighly-regardedpublicschoolsystem.Ourresultsabouttransitandschoolsaresuggestive,butnotconclusiveandfurtherstudyandresearchonthesetopicsisneeded.



AppendixAQuestions on Bui lding Characterist ics

1. Name of Community and Street and City Address

____________________________________________________________ ____________________________________________________________

2. Development type (check any that apply): o TOD (less than 0.5 mile away from transit stop) o Mixed Use (at least 2 uses)

3. Product type: o Townhome (A multilevel unit with interior stairs, but not a loft) o Lowrise 1-3 stories o Midrise 4-9 stories o Highrise 10 or more stories

4. When was the Year Built/Completed _____________________

5. Is the community a condominium or a rental? _____________________

6. Market rate rents: Base rent range (net of any other fees or concessions, if applicable); Skip if this is a condominium community Monthly Base Rent Range Studio _________________ 1BD _________________ 2BD _________________ 3BD _________________ 4BD _________________



7. Market rate units and populat ion: Total units and population (according to lease documents) Total Vacant Units Total Occupied Units Total Current Residents Studio _______________ _______________ _______________ 1BD _______________ _______________ _______________ 2BD _______________ _______________ _______________ 3BD _______________ _______________ _______________ 4BD _______________ _______________ _______________ Total Units and Residents (sum rows) _______________ _______________ _______________

8. Affordable units and populat ion : total units and population (according to lease documents) Affordable units are deed restricted as to the amount someone can pay for the unit (condominium) or amount of rent that can be charged (rental) Total Vacant Units Total Occupied Units Total Current Residents Studio _______________ _______________ _______________ 1BD _______________ _______________ _______________ 2BD _______________ _______________ _______________ 3BD _______________ _______________ _______________ 4BD _______________ _______________ _______________ Total Units and Residents (sum rows) _______________ _______________ _______________

9. Does your community offer the following amenities Yes No

o Pool ____ ______ o Fitness Center ____ ______ o Kids play area ____ ______ o On-site leasing/management ____ ______ o Shuttle to transit ____ ______



Questions on School Age Chi ldren Answer question 10 if known, otherwise answer question 11

10. Total school age chi ldren (ages 5-17) by unit type. Market Units Affordable Units

o Studio __________ _______________________ o 1BD __________ _______________________ o 2BD __________ _______________________ o 3BD __________ _______________________ o 4BD __________ _______________________

11. Answer the following about school age chi ldren (ages 5-17) The sum of these 4 rows should be 100% for Market Units and 100% for Affordable Units

Market Units Affordable Units o Percent of units with no school age children ___________ ___________ o Percent of units with 1 school age child ___________ ___________ o Percent of units with 2 school age children ___________ ___________ o Percent of units with 3+ school age children ___________ ___________

Questions on Demographics – Market Rate Units Only

12. What is average household income per unit? _______________

13. Can you estimate the percent of households earning the following (totals should sum to 100) The sum of these 6 rows should be 100%

o Less than $50K per year ______________________ o Between $50K - $100K per year ______________________ o Between $100K - $150K per year ______________________ o Between $150K - $200K per year ______________________ o Between $200K - $250K per year ______________________ o More than $250K per year ______________________



AppendixB

TableB1:SurveyCoverage,UnitsBuiltinAnyYear

Market Affordable Total Census* %ofCensusAtlantic 838 0 838 5,042 16.6%Bergen 3,055 92 3,147 103,937 3.0%

Burlington 2,296 162 2,458 30,811 8.0%Camden 2,933 0 2,933 52,543 5.6%CapeMay 0 0 0 1,324 0.0%Cumberland 0 0 0 2,175 0.0%

Essex 1,759 173 1,932 145,284 1.3%Gloucester 0 0 0 13,205 0.0%Hudson 9,709 176 9,885 164,992 6.0%

Hunterdon 145 12 157 5,048 3.1%Mercer 1,733 439 2,172 40,777 5.3%

Middlesex 2,307 81 2,388 90,822 2.6%Monmouth 1,589 110 1,699 45,893 3.7%Morris 2,902 37 2,939 37,753 7.8%Ocean 346 174 520 26,736 1.9%Passaic 626 193 819 68,463 1.2%Salem 0 0 0 43,689 0.0%

Somerset 450 98 548 23,749 2.3%Sussex 0 0 0 5,726 0.0%Union 1,372 149 1,521 68,840 2.1%Warren 140 72 212 8,863 2.4%TOTAL 32,200 1,968 34,168 985,672 3.5%

*Censusdataderivedfromthe2015AmericanCommunitySurvey.Unitsrefertonon-group-quarterrentalsintown-homesormulti-familybuildings.



TableB2:SurveyCoverage,UnitsBuiltSince2000

Market Affordable Total Census %ofCensusAtlantic 0 0 0 653 0.0%Bergen 1,785 92 1,877 10,101 18.6%

Burlington 0 100 100 5,099 2.0%Camden 0 0 0 5,103 0.0%CapeMay 0 0 0 99 0.0%Cumberland 0 0 0 106 0.0%

Essex 625 64 689 15,381 4.5%Gloucester 0 0 0 1,961 0.0%Hudson 4,772 176 4,948 23,030 21.5%

Hunterdon 81 12 93 704 13.2%Mercer 770 227 997 4,350 22.9%

Middlesex 47 61 108 11,087 1.0%Monmouth 803 110 913 5,771 15.8%Morris 1,191 34 1,225 4,271 28.7%Ocean 0 47 47 6,058 0.8%Passaic 165 189 354 3,772 9.4%Salem 0 0 0 3,922 0.0%

Somerset 294 98 392 3,370 11.6%Sussex 0 0 0 570 0.0%Union 529 9 538 7,472 7.2%Warren 0 72 72 768 9.4%TOTAL 11,062 1,291 12,353 113,648 10.9%

*Censusdataderivedfromthe2015AmericanCommunitySurvey.Unitsrefertonon-group-quarterrentalsintown-homesormulti-familybuildings.

RutgersRE.com

Rutgers Center for Real Estate’s innovative model of real-world education, industry-focused research and the cross-pollinating exchange of ideas has created a dynamic hub for students, academics and professional trailblazers alike at Rutgers, the leading national research university and the state of New Jersey’s preeminent public institution of higher education.

For more information, please contact:

Lorri LaytonDirector of Marketing & Outreach [email protected]

https://www.rutgersrealestate.com

school-age children in rental units in new jersey: results

Documents