Professional Documents
Culture Documents
The Impact of Private School Competition On Public School Outcomes in Wisconsin
The Impact of Private School Competition On Public School Outcomes in Wisconsin
edu April13,2009
2 Abstract:Ianalyzetheeffectofmorecompetitiveprivateschoolsinacounty
schoolsonpublicschooloutcomesinWisconsin.Iusetheshareofstudentsina countyattendingprivateschoolasaproxyforprivateschoolcompetitiveness.This isbecausefixedcostsmakeupalargepartofschoolexpenses,andsoalarger numberofstudentsattendingaschooldecreasetheperstudentfixedcosts (Bowles).However,areaswithworsepublicschoolsmayhavehighersharesof studentsattendingprivateschools.Toremedythis,Iusethedensityofreligious adherentsasaninstrument.Thiscanbedonebecauseahighpercentageofprivate schoolsarerunbyreligiousorders,andtheyreceivealargeamountoftheirrevenue throughtheaffiliatedchurch(Kealey).Areaswithhigheradherentdensitiesareable todecreasetheirtuition,whichmakesthemamorecompetitivealternativeto publicschooling. Theresultsdonotshowprivateschoolcompetitivenesstohaveastatistically
II.BackgroundInformation
Americaspublicschoolsarenotoriouslylackluster.Inthe50biggestcitiesin
theUnitedStates,only53%ofstudentsgraduatehighschool(CitiesinCrisis).Many proposalshavebeenfloatedtoimprovepublicschooloutcomes,butincreasing competitionisgenerallyviewedwithsuspicion.Thiscouldpartlybeexplainedby thelackofstudiesaboutcompetitionsimpactonpublicschooloutcomes.Inthis paper,IuseanapproachproposedinCarolineHoxbys1994paperDoPrivate SchoolsProvideCompetitionforPublicSchools?toanalyzetheeffectofprivateschool competitiononpublicschooloutcomesinWisconsin.First,thegeneralmodelfor schoolingdecisionsmustbeoutlined. Afamilychooseswheretoliveprimarilybasedonemployment opportunities.Aworkerfirstfindsawillingemployer,andtheworkershousing choiceisthenconstrainedtoareasonablecommutingdistancefromtheirjob.The workercanchosefromavarietyofcitiesbasedonaTieboutsortingmodel.This modelsaysthateachcityofferstheworkerauniquesetoftaxratesandgovernment benefits,andthebenefitsofferedshouldbeequilibratedbythecityspropertytax ratesandpropertyvalues.Theworkerwillchoosethepackagethatmaximizesthe familysutility.Sincetherearemanycitieswithintheworkerscommuting constraint,theareaofanalysisisquiteimportant. Theagenthasavarietyofhousingchoicessowecannotsimplyanalyzeon
theschooldistrictlevel.Withincommutingdistanceofanemployer,therearea widevarietyofschooldistrictswithdifferenttaxratesandschoolquality.Adistrict
sharetheexpensesofstartingaschoolinadditiontopayingtheirexistingtaxes.The privateschoolsmustthenprovideagreatvaluetothesefamilies.Therearetwo mainwaysprivateschoolscanprovidethis.First,theprivateschoolsmayhave differentproductionfunctionsfromthepublicschools.Theschoolscouldhave greaterefficiencythanpublicschools,theymayprovideadifferenttypeofschooling likeahandsonapproach,ortheymayhavemorefreedominselectingand discipliningstudents.Anotherpossibilityistheprivateschoolshavethesame productionfunctionaspublicschools,buttheyprovideadistinctcurriculum.For example,theparentsmaydesiretheirchildrentogetamorereligiousbased educationoramoreartsbasededucation.Onceaschooliscreated,parentswith thatparticulardemandwouldchosetoliveclosetoandattendthatprivateschool (inmostcases,inthesamecounty).Sincefixedcostsmakeupalargepartofschool expenses,areaswithgreaterprivateschoolenrollmentshouldhavecomparatively lowertuitionwithallotherthingsequal(Bowles).Thisprovidesabasisforouridea
thatprivateschoolcompetitivenessmayvaryincertainareas.Nowwemust addresshowprivateschoolcompetitivenesscouldaffectpublicschools.
Whiletherearemanywayspublicschoolsmaybeaffectedbyprivateschool
competitiveness,Iwillmentionjustthreemainroutes.Thefirstisthatschooling outcomesandsatisfactionwithpublicschoolsarenotoriouslyhardtomeasure. Whenanareahasfewpublicschoolsorthosethatexisthaveexorbitanttuition, familiescaneitheraccepttheschoolsperformanceormovetoadifferentarea. Whenanareahasanumberofprivateschoolswithmorereasonabletuition,more familiesareonthemarginbetweenpublicschoolingandprivateschooling.Then,if anareahasnotablybadpublicschooloutcomes,morefamilieswillbeabletosend theirkidstoprivateschools.Thenumberofstudentsattendingprivateschoolscan thenprovideanimportantsignaltothepublicschoolsabouttheirown competitiveness.Generally,citieswillnotcompensateforthisbyleavingschool qualitypooranddecreasingtaxes.Anotherpossiblerouteisadirectlyfinancialone. Thecompetitivenessofacityspublicschoolsshouldbecapitalizedinto
affectcompetitivenessiftheschooladministratorsandteacherscaremoreabout theoverallschoolbudget(andthereforetheirsalaries)thanperpupilbudgets.
Athirdpossiblerouteisthroughthesortingeffect.Privateschoolsmaytake disruptivestudentsoutofthepublicschools,andtherebyincreasethepublicschool outcomes.Theprivateschoolsmayalsotakethebeststudentsoutofthepublic schools,andtherebyreducetheperceivedpublicschooloutcomes.Thereissome evidencethatmorecompetitiveprivateschoolsdodrawgoodstudentsfrompublic schoolsandthereforedecreasepublicschooloutcomes(McEwan).Thiscouldlead toanunderestimationoftheimpactofprivateschoolcompetitivenessonpublic schooloutcomes,butIdonotdirectlyaddressthisduetodataconstraints.Wehave thenexplainedhowprivateschoolcompetitivenessmayimpactpublicschool outcomes,buttherearedistinctstatisticalproblemsinanalyzingthiseffect. Inanareawithpoorpublicschooling,moreparentswillchoosetosendtheir
privateschoolstudentsintheUnitedStatesattendreligiouslyaffiliatedschools,and byfarthelargestsupplierofreligiousprivateeducationistheCatholicChurch.
Religiousprivateschoolsareuniqueinthattheyreceiveagreatamountofsupport
fromtheiraffiliatedchurch.Theabilityofalocalchurchtofunditsprivateschoolis directlylinkedtothedensityofadherentsinthegivenarea.Thedensityof adherentscanreducetheprivateschoolcostsandmaketheschoolsmore competitiveinanumberofways.First,churchesusetheirdonationrevenuesto fundtheirprivateschools.Anareawithmoreadherentswillpresumablyhave greaterdonations,andtheprivateschoolmayreceiveagreaterpercentageofits revenuefromthischaritablesource.Theschoolscanthendecreasetuition,since mostreligiousprivateschoolssettuitiontojustcovertheiroperatingcosts.Second, areaswithgreateradherentdensitycandecreasetheprivateschoolsfixedcostsby sharingmoreequipment,buildings,andpersonnel.Third,parentsofthesame denominationaremorelikelytosendtheirchildrentoasimilarlyaffiliatedprivate school.Anareawithgreateradherentdensitywillhavemoreparentsonthemargin betweenpublicandprivateschooling.Lastly,areaswithgreateradherentdensity canhaveagreaterschooldensity(Kealey).Sincetransportationcostsarealarge factorindecidingwhichschooltoattend,thiscertainlyincreasesprivateschool competitiveness. NationwideandWisconsinspecificprivateschooldataisunavailable publicly,buttheNCESreportsthatinthe20032004schoolyearreligiously affiliatedschoolshadsignificantlycheapertuitionthannonsectarianschools.Non sectarianschoolshadanaveragetuitionof$13,419whilereligious(NonCatholic) schoolshadanaveragetuitionof$5,839.Catholicschoolshadanevenless
ThejsubscriptindicatesthecountyinWisconsin.Theisubscriptindicates theyearofthedata.Yijisthemeasureofpublicschoolingoutcomesforaparticular year.Iusetheresultsofseveraldifferenttestsadministeredacrossthestateof WisconsintoassessstudentachievementincludingMath,Reading,andScience exams.Cijisthepercentageofstudentsinthecountythatareenrolledinprivate school.ThedifferentXsarevariousdescriptorsofthecounties.Theseincludethe percentagesofvariousracialgroups,thepercentageofresidentslivinginpovertyas definedbythestateofWisconsin,thepercentageofresidentsmakingover$75,000, andeducationindicators.Acomprehensivelistcanbefoundinattheendofthe paper.jisthecountyerrorterm.SinceCjcanclearlyberelatedtoYj,Iusethe densityofreligiousadherentsasaninstrument.Thesecondequationis: (2)Cij=Z1j+j
HeretheZjrepresentstheCatholicdensityintheparticularcountyof Wisconsin.jisthecountyspecificerrorterm.Theuseofreligiousadherentdensity
Instructiondatasite.AsrequiredbyWisconsinlawandtheNoChildLeftBehindact, Wisconsinmustreporttheresultsofavarietyofachievementmeasures.Iusedata fromthe20022003,20032004,and20042005schoolyears.FortheMathand Scienceexams,Imeasuredthepercentageofstudentswhoscoredproficientor advanced.FortheReadingexam,Iusedonlythepercentageofstudentswho scoredadvancedbecausetheproficiencyratesweremuchhigherforReadingthan ScienceandMath.Sinceeachschoolwasreportedindividually,Iaveragedthescores ofalltheschoolsinadistrict.Smallschoolswereautomaticallyexcludedbecause Wisconsindoesnotreportdatawithsmallsamplesasindividualresultsmaybe inferred.Irepeatedthisprocedurefromthedistricttocountyleveltogetafinal countyaverage.Ialsosummedthepublicschoolenrollmentfromthesereportsin conjunctionwiththeWisconsinNonPublicenrollmentdata. WhileIcouldnotobtainspecificprivateschooldata,Iwasabletofind
enrollmentdataforprivateschoolsbydistrict.Isummedthisanduseditin conjunctionwiththepublicschoolenrollmentdatatogetavalueforthepercentage
10
ofstudentsinagivencountywhoareenrolledinprivateschools.Iusedatafromthe 20022003,20032004,and20042005schoolyears.
duetothelackoftestscores.Thisisbecauseofthesmallschoolsizesinthese
11
impactonpublicschooloutcomes.Theonlyvariablethathadstatisticallysignificant resultsformorethanonetestscorewasthepercentageofpeopleinthecountywith ahighschooldiploma.Thiscouldbeanindicatoroftheoveralleducationallevelof thecitizensinthecounty,andalsoanindirectmeasureofthequalityofschoolsin thecountyiflargeportionsofhighschoolgraduatesremaininthecounty.Allthree testsshowedthata1%increaseinthepercentageofhighschoolgraduatesina countyincreasesthepercentageofstudentsscoringproficient(MathandScience) oradvanced(Reading)byroughly.5%.Severaloftheothervariableshad economicallysignificantresults.TheMathandScienceresultsweresimilar,andsoI willdiscussthemtogether. TheMathandSciencetestsbothhadmuchhighercoefficientsfortheimpact
privateschoolcompetition.Ihavefoundresearchsupportingbothsides,andthe
12
VI.TablesofResults SummaryofData(CountyLevel) Variable %Proficient orBetterin Science %Advanced inReading %Proficient orBetterin Math Mean SD Min Max
13
0.34% 24.16%
Percentage of students enrolled in private school Percentage of people living in poverty, as defined by Wisconsin Percentage of people making over $75,000 per year Percentage of people who have at least 12 year of education Percentage of people who identify as Caucasian Percentage of people who are foreign born Percentage of people who identify as Hispanic or Black
%StudentsProficientorBetterinMath Variable %privateschool %poor %rich %HSdiploma %white %foreign %blackandhispanic %StudentsProficientorBetterinScience Variable %privateschool %poor %rich %HSdiploma %white %foreign %blackandhispanic Coefficient SE ZStat Coefficient SE ZStat
14
0.531 0.360 1.470 0.022 0.659 0.030 0.309 0.213 1.450 0.603 0.294 2.050 0.278 0.230 1.210 1.100 0.785 1.400 0.000 0.238 0.000
0.492 0.336 1.470 0.250 0.615 0.410 0.180 0.199 0.900 0.495 0.275 1.800 0.042 0.215 0.190 0.884 0.734 1.200 0.259 0.222 1.160
%StudentsProficientorBetterinReading Variable %privateschool %poor %rich %HSdiploma %white %foreign %blackandhispanic Coefficient SE ZStat
15
0.130 0.296 0.440 0.289 0.537 0.540 0.011 0.172 0.060 0.509 0.239 2.140 0.447 0.188 2.380 0.080 0.635 0.130 0.137 0.193 0.710
WorksCited Bowles,TylerJ.,andRyanBosworth."ScaleEconomiesinPublicEducation:
16
EvidencefromSchoolLevelData."JournalofEducationFinance28(2002): 285300. CitiesinCrisis2009:ClosingtheGraduationGap.Rep.America?sPromiseAlliance, 2009. Hoxby,CarolineM."DoPrivateSchoolsProvideCompetitionforPublicSchools?" NBERWorkingPaperSeries4978(1994). Kealey,RobertJ.BalanceSheetforCatholicElementarySchools:1995Incomeand Expenses.Rep.NationalCatholicEducationalAssociation,1996. McEwan,PatrickJ."ThePotentialImpactofVouchers."JournalofEducation79 (2004):5780. NationalCenterforEducationStatistics.DigestofEducationStatistics.2007. OwusuEdusei,Kwame."SchoolQualityandPropertyValuesInGreenville,South Carolina."JournalofAgriculturalandAppliedEconomics(2007).