Research Funding of Australian Universities: Are There ... · that is driven by its reliance on...
Transcript of Research Funding of Australian Universities: Are There ... · that is driven by its reliance on...
ResearchFundingofAustralianUniversities:
AreThereIncreasingConcentration?
ThiMaiAnhNguyena,AliciaRambaldi
a,KamKiTang
a
a.UniversityofQueensland,Australia
Abstract
Australia’shighereducationsectorisfacingawatershedmomentofitsresearchfundingregime.
TheFederalGovernmenthasproposedtochangefromthelong-standingfundingmodelthat
heavilybasedonpublicationoutput,toonebasedonpublicationplusindustryengagement.Inthis
paper,wetakestockofhowresearchfundingisraisedandallocatedwithinthesectoroverthepast
twodecades.Itisfoundthattheshareoftotalresearchfundingbyuniversitygroupshavebarely
changed.ButthedisciplineofBiologicalandClinicalScienceshasincreasinglydominated
competitivefundingschemes.
Keywords:Researchfunding,ARC,NHMRC
1 Introduction
Australia’shighereducationsectorisfacingunprecedentedchallengesfrombothdomesticand
internationalfronts.Inthepast20years,thesectorhasexperiencedcutsinfederalfundingona
perstudentbasis(Marginson2001),atrendthatputspressureonAustralianuniversitiesto
increasinglyrelyonstudentfees.Australianuniversitiesareincreasinglylookingtooverseas
educationmarketsasasourceoffunding.However,competitionforinternationalstudentshas
intensifiedrapidlywiththeentryofnon-EnglishspeakingcountriesfromEurope(e.g.France)and
Asia(e.g.China)inadditionaltoEnglishspeakingcountriesliketheUSandSingapore.Whataddsto
thisalreadycomplexenvironmentisthespectacularriseofglobaluniversityrankingsinthepast
decade.Theserankingsaredeemeda‘consumerguide’byinternationalstudentsorevenpublic
organizations(Hazelkorn2007;HEFCE2008),andtheytypicallyputoverwhelmingweightson
research,especiallyTheAcademicRankingofWorldUniversities(commonlyunknownasShanghai
JiaoTongUniversityranking).Consequently,theyhaveputhugepressureonuniversitiesworldwide
tolifttheirgameonresearch.
ThetotalresearchfundingallocatedtoAustralianuniversitieswasaboutone-thirdofapercentage
pointofGDPin2013(seeTable2).AstheFederalGovernmentfacesthechallengesofbudget
repair,thefundingpieisunlikelytogetbigger–ifnotsmaller–intheforeseeablefuture.
Therefore,howthepieisdistributedbecomesevenmorecriticalforthefutureresearch
performanceofindividualuniversities.Inrelationtothat,certainsegmentofthetertiarysectorhas
arguedthatduetotherelativelysmallsizeofAustraliainglobalresearch,thewaytoensure
Australiacanpunchaboveitsweightistochannelscarcefundingtothenation’s‘bestperformers’
sothattheycanachievethecriticalmasstobegloballycompetitive(Dodd,2015).Notsurprisingly,
sandstoneuniversitiesaremoresupportiveofthisviewwhilesmallerandregionaluniversities
2
disagree.However,littleevidencehasbeenproducedtovalidateorinvalidatethecriticalmass
argumentintheAustraliancontext.1
Furthermore,theFederalGovernmentrecentlyannouncedchangestothefundingmodelof
Australianuniversitieswhichwillseelessemphasisonpublicationandanincreasedweightof
industrialengagement:‘thenewarrangementsaredesignedtodrivegreaterresearch-industry
engagementbysubstantiallyboostingincentivesforcollaborationwithbusinessandorganisations
whichusetheoutcomesofresearch’(DepartmentofEducationandTraining,2016).Againstthis
background,andconsideringthetwodecadesfrom1992to2013,theobjectiveofthispaperisto
investigate:
(a) howmuchresearchfundingthepublicandprivatesectorshavecontributedtothe
Australianhighereducationsectorandhowdiditchangeovertime;
(b) howfundinghasbeendistributedamongstuniversitiesandwhetherthereisincreasingor
decreasingfundingconcentrationovertime;and
(c) howthecompetitivefundingfromthetwomajorgrantbodies,theAustralianResearch
Council(ARC)andNationalHealthandMedicalResearchCouncil(NHMRC),hasbeen
distributedamongstdisciplinesandwhetherthereisincreasingordecreasingfunding
concentrationovertime.
Tothebestofourknowledge,thesequestionshavenotbeenansweredbeforeusingthedatain
recentdecades.Akeycontributionofthispaperisthecompilationofapaneldataofresearch
fundingovertwodecadesforallAustralianuniversities.Wecollectedandprocesseddatafromthe
HigherEducationResearchDataCollection(HERDC),ARCandNHMRCdatabases.Acontributionof
thepaperistoreconciletheFOR2008andRFCD1998codesforresearchdisciplines(seeappendix
fordetails),withwhichwecalculatetheamountofARCorNHMRCresearchfundingattributedto
eachdisciplinewithineachuniversityeachyear.MostanalysesontheallocationofARCand
NHMRCfundingoverdisciplinesamalgamategrantsbycommencementyear.Weinsteadlookat
annualallocationsandupdateexpectedallocationswithactualoneswheneverpossible.Thedata
collectionworkischallengingbecauseofthemultiplicityofdatasourcesandchangingdata
collectionmethodsbygovernmentagenciesovertime.Wewentthroughthefundedprojectsone
byonetocorrectanydouble-ormis-reporting.
StudiesonresearchfundingallocationamongstAustralianuniversitiesarefewandfarinbetween.
AbbottandDoucouliagos(2003)studytechnicalandscaleefficiencyofAustralianuniversitiesusing
DataEnvelopmentAnalysis(DEA).Theirstudyusesthenumberofequivalentfull-timestudents
(EFTS),thenumberofpostgradstudentsandundergraduatedegrees,andResearchQuantum
Allocationasmeasuresofoutput,whileinputmeasuresincludethenumberofstaff(academicand
non-academic),expenditureonallinputsotherthanlabourinputs,andthevalueofnon-current
assetsasaproxyforcapitalstock.TheyfindthatAustralia’suniversitiesperformfairlyevenly
regardingefficiency.However,thestudysuffersfromthesameproblemfacedbyotherstudiesin
theliterature(includingthecurrentpaper),whichisitsinabilitytocaptureoutputquality.Our
1ArecentstudybyWilliams(2016)assessesAustralia’sinvestmentinR&Dactivities,researchoutputandconnectivity
withindustryfromaninternationalperspective.Itconcludesthatthecountryisbiasedtowardsbasicresearch,afact
thatisdrivenbyitsrelianceoninternationalstudents’tuition,whichinturnimpliesitsrelianceoninternational
rankings.Thestudysuggestsadiversesystem,inwhichagroupofinstitutionspursuesexcellenceofinternational
standardsinbasicresearch,andothersfocusmoreonappliedstudiesinvolvinglocalissues.
3
studydoesnotassessuniversityproductivity.Instead,wefocusondistributionalmovementsof
totalresearchfundingovertime.
Anotherstudy,Villeetal.(2006),focusesonthedistributionalmovementsofmajorinputand
outputmeasurescompiledbyHERDCusingcalculatedGinicoefficients.However,theirstudycovers
theperiodfrom1992to2003,excludingblockgrantsanddetailedbreakdownsofCommonwealth
competitiveincomeintoARC,NHMRC,andotherschemesaswellastheirdisciplinecomponents.
TheirstudypointstoagradualconvergencepatterninoutputbyGo8andsmalleruniversities,but
continuingunevendistributioninfundingbetweenthetwogroupsfrom1992to2003.Ouranalysis
from2001to2013revealssimilarpatterns.
Therestofthepaperisorganizedasfollows.Section2describesthedatausedintheanalysis.
Section3presentsadescriptivesummaryofthedistributionoffundingbysources.Section4
examinestheallocationoffundingacrossuniversitygroups,withineachofthesegroupsandfinally
acrossalluniversities.Thepaperconcludeswithabriefdiscussionontheimplicationsofourstudy
andasummaryofitskeyfindings.
2 Data
Sinceallthefundingdatausedinthispaperareaboutresearchingfunding,inwhatfollowsweuse
theterm‘funding’forbrevity.Wecompilethedatafromfourmainsources:theHigherEducation
ResearchDataCollection(HERDC)fromtheDepartmentofEducationandTraining(DET),the
NationalCompetitiveGrantsProgramDataset(NCGP),theResearchManagementDatabaseofthe
NationalHealthandMedicalResearchCouncil(NHMRC),andtheDET’sdocumentonResearch
BlockGrant(RBG)allocations.Detailsofthesedatasetsareprovidedintheappendix.
Figure1providesaschematicpresentationofthecompositionofresearchfundingforAustralian
universities.Thetotalresearchfundingmakesupofthreemaincomponents:CompetitiveIncome,
RBG,andOtherIncome.TwomainsubcomponentsofCompetitiveIncomeareAustraliaResearch
Council(ARC)grantsandNHMRCgrants.OurrecordofARCgrantsonlyincludesthosecommencing
noearlierthan2002,andNHMRCgrantsincludethosecommencingnoearlierthan1990.This
meansthatanalysesofannualfundingflowsinlevelswilldownwardlybiasfundingamountsin
earlyyearsoftheperiod,andinadditioninflationaryeffectswouldneedtobeaccountedfor.To
reducetheformerbiasandcastsasidethelatterissueweusesharesoffunding.
Thedatasetcovers39universitiesoutof41universitieslistedasHEPsundertheHigherEducation
SupportAct2003.2FiguresforUniversityofNotreDameAustraliaareonlyavailableforyearssince
2008.Thishasnegligibleeffectontheanalysesoffundingallocationbyuniversitygroupbecause
thisuniversity’sshareintotalresearchfundingisverysmall.Tomaketheanalyticalresultseasierto
comprehend,wegroupsimilaruniversitiestogetherasaunit.Sinceuniversitiesareheterogeneous
inmanyaspects,thereisnosinglewaytogroupthem.Therefore,weadoptthegroupingchosenby
universitiesthem.AnumberofAustralianuniversitieswithsimilarityinsize,historyandorientation
haveformedtheirowngroupsandusethemasplatformtopursueissuesoftheirinterests.They
includetheAustralianTechnologyNetwork(ATN),theGroupofEight(Go8),andtheInnovation
2TheBachelorInstituteofIndigenousTertiaryEducationandMCDUniversityofDivinityareexcludedbecauseof
inconsistentrecordbyHERDC.AustralianMaritimeCollege,whichwasmergedtotheUniversityofTasmaniain2008is
alsoexcluded.
4
ResearchUniversities(IRU).3TheircompositionisshowninTable1.Wegrouptherestasthe
‘Other’.Go8universitiesareknownasthe‘sandstone’universities,astheytendtobetheoldest
andlargest.Theyarealsomoreresearchorientedingeneraland,asaresult,tendtodominate
researchfundingallocation.
3 DistributionofFundingbySource
Inthissection,welookatthedistributionoffundingbysource;inthenextsection,welookatthe
distributionbyuniversity.
Figure1showstheamountandshareoffundingreceivedbyAustralianuniversitiesbysourcesasin
2013.Itcanbeseenfromthemiddlelayerofthechartthat,CompetitiveIncome,RGB,andOther
Sourcesareofroughlycomparablemagnitudes.Asshowninthebottomlayerofthechartthat,
ARCandNMHRCgrantstogethermakeupthebulkofthecompetitivegrants:46%and37%
respectively.WithintheOtherSourcescategory,OtherPublicIncomeandIndustryandOther
Income(e.g.contracts,donations,bequests)makeup44%and48%respectively.
Figure2showstheshareoffundingreceivedbydifferentuniversitygroupsin2013.ATN’sfunding
sourcesaremostdiverse,whileGo8’sareleastdiverse.Groupswithlessdiversefundingsources
arepotentiallymorevulnerabletochangesinindividualgrantcategories.HalfofGo8’sresearch
fundingisfromARC,NHMRCorsimilarcompetitivesources,whileonlyathirdofATN’sfundingis
fromthesesources.Moreover,althoughCRCremainsasmallportionoftotalfunding,itssharein
ATN’stotalfundingismuchhighercomparedtothoseoftheothergroups,probablyreflectingthe
former’sfocusonappliedresearch.
Figure3showstheallocationofeachfundingbyuniversitygroupin2013.AlthoughGo8dominates
everyfundingcategoryatthebottomlevelofthechart,itsshareisparticularlylargefortheARC
andNHMRCones.ATNperformsbestintheCRCcategory,andIRUdoeswellintheOtherACGR
Schemescategory;butthesetwocategoriesarerelativelysmallinabsolutedollarterms.
4 AllocationofFundingbyUniversity
Thelastsectionpresentsthefundingallocationofasingleyear;inthissection,welookathowthe
allocationchangesovertime.Thefocusisonwhetherthereisanysignofresourceconcentration
acrossaswellaswithinuniversitygroups.
Table2expandsthedataofFigure1backto1992,andexpressesitasashareofnominalGDP.For
somecategories,dataarenotavailableinearlieryears.Therefore,weonlydothesumifdataforall
componentsofthedataareavailabletoavoidinconsistentandpotentiallymisleadingresults.This
isthereasonTotalFundingfiguresarenotreportedtill2001,forinstance.
Overtheyears,fundinggrewinsomecategoriesanddeclinedinothers.From2004to2013,
CompetitiveIncomeandOtherSourceincreasedby32%and24%respectively,whileRBG
decreasedby13%(allintermsofshareofGDP,notabsolutedollars).Themaindriveroffunding
3Thegroupmembershipinformationisdrawnfromeachofthefirstthreegroups,withtheremainingonebeing
groupedtogetheras“Others”:https://go8.edu.au/page/member-information,https://www.atn.edu.au/about-us/our-
universities/,andhttp://www.iru.edu.au/about/history/.
5
growthfromOtherSourcesisthegrowthinOtherPublicIncome(63%).Incontrast,CRCfunding
decreasedby47.24%.WithinCompetitiveIncome,ARCfundinggrewmuchlessrapidlycompared
toNHMRCfunding,withincreasesof23%and47%respectivelyovertheten-yearperiod.
4.1 AllocationofFundingacrossUniversityGroups
Figure4showstheshareofthetotalfundingbyuniversitygroupsfrom2001to2013.Duringthis
period,theshareofGo8increasedslightlyby2.5percentagepointsfrom64.6%to67.1%,andthat
ofATNincreasedby0.9percentagepointfrom8.5%to9.4%.Onthecontrary,theshareofIRU
declinedby1.9percentagepointsfrom12.5%to10.6%,andthatofOthersalsodeclinedby1.5
percentagepointsfrom14.4%to12.9%.
Abreakdownofthefundingbysub-categoryshowsthattheincreaseinthetotalfundingshareby
Go8duringthisperiodcamelargelyfromitsrisingsharesofCompetitiveIncome(from70.8%to
73.0%)andRGB(from61.8%to64.5%);itsshareofOtherSourcesincreasedonlymarginallyin
comparison(from64.0%to64.1%).ForATN,itsexpansionspreadquiteequallyacrossthethree
sources:from4.8%to6.6%forCompetitiveIncome,from8.9%to9.8%forRGB,andfrom10.5%to
11.5%forOtherSources.
Amongstalltypesofresearchincome,thecompetitionforARCandNHMRCfundingtendtoattract
mostattentionwithinthesection.Thisisbecausetheymakeupthebulkofcompetitivefunding,
whichfurtherfeedsintothecalculationofRBGallocation,andsohavingtheeffectofdouble-
dipping.Furthermore,becauseoftheirassessmentmethodsheavilyrelyonpeerreview,ARCand
NHMRCareperceivedtobeamuchstrongerindicatorofresearchexcellenceintheconventional
sensethanothertypesoffunding.Therefore,welookatthesharesofthesetwosourcesoffunding
inmoredetails.
Figure5showstheshareofARCfundingbyuniversitygroupover2002-2013.TheshareofGo8in
ARCgrantshasasignificantgainof7.7percentagepoints–from65.2%to72.9%.Asaresult,all
otherthreegroupssufferedalossoffundingshare.However,thedropofIRUwasparticularbigof
4.4percentagepoints–from12.8%to8.4%.ThedeclineinfundingsharesbyATNandOtherswere
1percentagepoint(from7.7%to6.7%)and2.2percentagepoints(from14.3%to12.1%)
respectively.
DataforNHMRCgrantsisavailableforamuchlongerperiod,from1992to2013,andtheresults
areshowninFigure6.Interesting,unlikeARCfunding,theshareofGo8inNHMRCfundingwas
essentiallystaticataround83.5%.ThisfindingisevenmorestrikinggivenANUwasnotallowedto
participateinNHMRC(orARC)grantcompetitionpriortoyear2000.Overthesetwodecades,the
biggestgainwasexperiencedbyATN:anoverthreepercentagepointsincreaseinfundingshare
from0.9%to4.2%.Others’sharewasmorethandoublefrom2.3%to5.1%.Onthecontrary,IRU
sufferedabigloss;itssharewasalmosthalvedfrom13.3%to7.1%.
4.2 AllocationofFundingwithinUniversityGroups
Inthissub-section,weexaminetheallocationoffundingwithineachuniversitygroup.Tothatend,
wecalculatethecoefficientofvariationoffundingwithineachgroup.Coefficientofvariation(CV)is
equaltothestandarddeviationdividedbythemean,andacommonlyusedmeasureofdispersion.
AlargerCVmeansthatfundingisdistributedlessequallywithinthegroup.Becausethemeasureis
normalizedbythemean,wecancomparethevaluesacrossgroups.
6
Figure7showstheCVofTotalFundingbyuniversitygroupover2001-2013.TotalFundingwithin
ATN,Go8andIRUismuchlessdispersecomparedtothatwithintheOther,whichhadconsistent
CVfiguresofaround0.9-1.0throughouttheperiod.ThisreflectsthefactthattheOtherisless
homogeneousthanthethreemajorgroups.Amongthethreemajorgroups,thereappearstobea
slightpatternofconvergenceinfundingdispersion.WhiletheCVofATNincreasedfrom0.13in
2001to0.27in2013,thatofGo8decreasedfrom0.33to0.27,andthatofIRUdecreasedfrom0.40
to0.25overthesameperiod.ComparedtothefiguresforARCandNHMRCfundinginthefollowing
discussion,thesefiguresandvariationsarehowevernegligible.
Again,welookatARCandNHMRCfundingmoreclosely.Figure8showstheCVofARCfundingby
universitygroupfrom2002to2013.ByandlargetherearelittlechangesofCVovertime.If
anything,theCVofGo8reducedfrom0.34to0.28overtheperiod.Likewise,Figure9showstheCV
forNHMRCfundingbyuniversitygroupover2002-2013.ComparingtoARCfunding,therearelot
moredynamicsintheNHMRCfundingallocationwithineachgroup.TheCVofOthersreducesfrom
1.93to1.39overtheperiod,andthatofIRUfrom1.40to0.88,meaninganincreaseinequalityof
fundingdistributionwithineachofthesetwogroups.Thedistributionoffundingbecomesless
equalwithinATN,withitsCVincreasedfrom0.85to1.02.Thesestrongwithingroupinequality
dynamicscouldbeduetothefactthatGo8universitieshadalionshareofNHMRCfunding(see
Figure4),andsoanoccasionalsuccessbynon-Go8universitiesintheNHMRCschemecouldtrigger
inabigchangeindistributionwithinGo8.WedonotobservethesamesituationinARCevenwhen
Go8alsodominatesthere(buttoalessextent).ThisisprobablybecauseARCcoversamuch
broaderspectrumofdisciplinesthanNHMRCandsoitiseasierforauniversitytosmoothoutthe
fundingincomesacrossallitsdisciplinesunderARC.
Last,Figure10showstheCVoffundingacrossalluniversities,notjustwithinuniversitygroups.The
variationofTotalFundingacrossalluniversitiesincreasedoverthesampleperiod,astherising
variationinCompetitiveIncomeandRGBoverwhelmedthedeclinevariationinOtherSources.
5 AllocationofARCandNHMRCFundingbyDiscipline
Inthissection,weshiftthefocusfromuniversitygroupstodisciplines,whichhasneverbeen
investigatedbefore.TheanalysisislimitedtoARCandNHMRConlyduetotheavailabilityofdata
Figure11showsthatbetween2002and2013thereappearedtobeashiftinARCfundingfrom
naturalsciences,includingPhysical,ChemicalandEarthScience(PCE),Engineeringand
EnvironmentalScience(EE)andBiologicalandBiotechnologicalSciences(BB),towardsother
disciplines.TheshareofPCEexhibitsmostvariation,decliningsignificantlyfrom29%to19%over
thefirsteightyearsbeforetherecentrecoveryto22%.Furthermore,EEandBBeachhaslostabout
threepercentagepointsintheshareofARCfunding.Incontrast,asteadyincreaseofsimilar
magnitudeinsharesofARCfundingwasenjoyedbyHumanitiesandCreativeArts(HCA),Social,
BehaviouralandEconomicsSciences(SBE)andBiomedicalandClinicalHealthSciences(BCH).
However,thebulkoffundingforBCHandPublicandAlliedHealthSciences(PAHS)comesfrom
NHMRCinsteadofARC.Inordertogainamorecomprehensivepictureweneedtoconsiderboth
fundingsources.
7
Figure12showsthevariationineachdiscipline’sshareinthecombinedpieofARCandNHMRC
funding.William(2016)findsthatAustralia’shighereducationsectormaintainsrelativelyweak
connectivitywiththeindustry,incomparisontocountrieswithgoodresearchpoliciesorhighlevel
oftechnologicaladvances.Thestudypointsoutthatengineeringandtechnologytakeupa
relativelysmallshareofresearchexpenditure5byAustralianhighereducationinstitutions.Thisisin
starkcontrastwithexpenditureonmedicalresearch,whereAustraliaonlycomessecondafter
Sweden.ThesituationmanifestsitselfinFigure12.Sharesoffundingonmedicalresearchare
predominantandrisingfast.TheshareoffundingbyBCHincreasedsignificantlyfrom26%to30%,
leavingagapof15percentagepointsby2013tothenextpredominantdiscipline-BB.Asimilarrise
canbeseeninPAHS,with8%and11%in2002and2013respectively.Sharesoffundingon
engineeringandtechnologydecreasedsteadilyovertheperiod.However,futurechangestothe
fundingschemewithemphasisesoninnovationandapplicationofresearchmaybreakthispattern.
6 Conclusion
Inthispaper,wehavecompiledthedataofresearchfundingforAustralianuniversitiesbetween
2001and2013.Althoughtheunderlyingdataarepubliclyavailable,differentsourcesarenot
necessaryconsistentand,insomecases,evendatafromthesamesourceareinconsistentover
time.Therefore,thecompilationofthisdatasetisinitselfacontributiontothedisciplineasthe
datasetshouldproveusefulforfutureresearchinthisarea.
Inthispaper,wehaveprovidedacomprehensivethoughbasicanalysisofthedata.Wehave
examinedthedistributionoffundingbysources,universitygroupsanddisciplinesovertime.One
strikingresultisthat,over13years,thesharesoftotalresearchfundingbythefouruniversity
groupshavebarelychanged,thoughthechangesofindividualfundingcategoriessuchas
competitivegrantandindustryfundingareslightlybigger.Changesinfundingdistributionwithin
eachofthefouruniversitygroupsovertimearealsolimited,withtheexceptionofNHMRCfunding.
Ouranalysisrevealedthat,atleastforARCandNHMRCfunding,thedynamicsliedwithdisciplines,
notuniversities.BiologicalandClinicalScienceshasincreasinglydominatedthetwocompetitive
grantschemes,absorbingupto30%ofthefundingin2013,whilethesetwoschemestogether
accountedformorethanaquarter(26.4%)ofthetotalresearchfundingtouniversitiesintheyear.
Amongstthefouruniversitygroups,Go8relyparticularlyheavilyoncompetitivegrants.Thisisnot
surprisinggiventhesetypesofgrantsrequireinvestigatorswithtrackrecordsbasedonquality
basedtraditionalresearchoutputsuchaspublicationsratherthan,say,socialimpacts.Assuch,Go8
universitiescouldbemorevulnerablethantheircounterpartstochangesincompetitivegrant
allocationcriteriaawayfrompublicationsinhighrankedjournals.
7 References
5TheremarkisbasedondatafromtheUNESCOScience,TechnologyandInnovationdatabaseat
<http://data.uis.unesco.org>.Thedatarefertoyear2011-2013.
8
Abbott,M.&Doucouliagos,C.(2003).TheefficiencyofAustralianuniversities:adataenvelopment
analysis.EconomicsofEducationReview,22(1):89-97.
DepartmentofEducationandTraining(2016).Newarrangementforresearchblockgrantsin2017.
PressRelease,21October.URL:https://education.gov.au/news/new-arrangements-research-block-
grants-2017.
Dodd,T.(2015).FundthebestPhDsandforgettherest,sayeliteunis.AustralianFinancialReview,
5September.URL:http://www.afr.com/news/policy/education/fund-phds-at-the-best-unis-and-
forget-the-rest-say-elite-universities-20150901-gjcozi
Gowthorp,L.(2016).WinningEdgefailstodeliver,sowhatnowforAustralia’sOlympichopes?The
Conversation.URL:https://theconversation.com/winning-edge-fails-to-deliver-so-what-now-for-
australias-olympic-hopes-64051.
Hazelkorn,E.(2008).Learningtolivewithleaguetablesandranking:Theexperienceofinstitutional
leaders.HigherEducationPolicy,21(2):193–215.
HigherEducationFundingCouncilforEngland(HEFCE)(2008).Countingwhatismeasuredor
measuringwhatcounts?Leaguetablesandtheirimpactonhighereducationinstitutionsin
England.London:HEFCE.
Marginson,S.(2001).TrendsinthefundingofAustralianhighereducation.TheAustralianEconomic
Review,34(2):205-215.
Norton,A.(2016).MappingAustralianhighereducation2016.GrattanInstitute.Melbourne.URL:
http://grattan.edu.au/wp-content/uploads/2016/08/875-Mapping-Australian-Higher-Education-
2016.pdf.
Ville,S.Valadkhani,A.&O’Brien,M.(2006).Thedistributionofresearchperformanceacross
Australianuniversities,1992-2003,anditsimplicationsforbuildingdiversity.AustralianEconomic
Papers,45(4):343-361.
Williams,R.(2016).EvaluatingthecontributionofhighereducationtoAustralia’sresearch
performance.TheAustralianEconomicReview,49(2):174-83.
9
Appendix
TheHERDCdataset–providesdataonCompetitiveIncome,OtherPublicIncome,Industryand
OtherIncome,CooperativeResearchCentre(CRC)Income.CompetitiveIncomeconsistsofincome
fromresearchschemesandprogramsregisteredontheannualAustralianCompetitiveGrants
Register(ACGR).OtherPublicIncomeincludesincomefromtheAustralianGovernmentnoteligible
forinclusioninCompetitiveIncome;incomefromStateorTerritoryGovernment;incomefromlocal
government;incomefromgovernmentbusinessenterprisesthatarepubliclyownedorfunded,
eitherwhollyorpartly;CRCsIncomewherethereportinghighereducationprovider(HEP)isnot
specifiedaseither‘TheResearcher’or‘Participant’intheCommonwealthAgreement.Industryand
OtherResearchIncomeincludescontracts,grants,donations,bequestsandfoundationsfrom
Australianindustryornon-governmentagencies;HDRfeesfornon-Commonwealthsupported
domesticstudents;competitive,peer-reviewedresearchgrantincomefromnon-Australian
agencies;otherincomefromnon-Australianagencies;HDRfeesfornon-Commonwealthsupported
internationalstudents.CRCIncomeincludesincomefromaCRCwherethereportingHEPisacore
participant.ItisimportanttonotethatCRCIncomeisreportedbythelatestfinancialyearwhile
incomefromothersourcesisreportedbythelatestcalendaryear.10Weusethecompiledtime
series(from1992to2013)maintainedbyUniversitiesAustralia.11
TogetherwithRBG,thesefoursourcesofincomemakeupTotalResearchFundingasusedinthe
analysisofreturnstoscaleinsection6ofthepaper.Thetwo‘return’variables,weighted
publicationsandHDRcompletions,arederivedfromtheHERDCdataset.Asweonlyusethe2001-
2013timeseriesofTotalResearchFundingforconsistency,weonlyneeddataonweighted
publicationsfrom2001onwards.Duringthisperiod,weightedpublicationsconsistentlycoverfour
maincategories:books,bookchapters,conferencepapers,andjournalarticles(withaweightof
fivegiventobooks).OurHDRcompletionvariableiscalculatedasaweightedsumofthenumberof
mastersandthenumberofPhDs,givingweightsof0.5and1respectivelytothetwocomponents.
TheRBGdataset–isdownloadedfromtheDepartmentofEducationandTraining(DET)website.
RBGfiguresareonlyavailablefrom200112onwards.Thefiguresareinitialallocations,and
componentprogramsmaychangefromyeartoyear.Duetothenarrowperiodofavailabilityof
RBGdata,weonlyuse2001-2013timeseriesofTotalResearchFunding.Furthermore,the
structuralbreakin2000,inwhichANUstartedtocompeteforresearchincome,andthe
inconsistentrecordofCRCIncomebeforeandafter2000deemnarrowingthetimeseriesto2001-
2013necessary.
TheNCGPdataset13–providesdataonresearchgrantsbytheAustralianResearchCouncil(ARC)from2002onwards,withspecifiedannualallocationsofeachandeverygrant.OurmainARC
figuresarethetotalARCfundingannuallyallocatedtoeachuniversity.Insection5,welookintothe
allocationofARCfundingacrossdisciplines.Tothisend,weusethedisciplineclustersspecifiedin
10HERDCspecificationmaychangeslightlyfromyeartoyear.Thedescriptiongivenrelieson2015specification.
11Thedatasetcanbefoundathttps://www.universitiesaustralia.edu.au/australias-universities/key-facts-and-
data/Research-Intensity---Output/Research-Intensity---Output#.Ve67nkvlfwJ.12ThecurrentsuiteofRBGwasestablishedin2001toreplacethepreviousresearchquantumandresearchrelated
programs,comprehensiverecordsofwhicharenotheldbytheDET.13Theoriginaldatasetsaredownloadablefromhttp://www.arc.gov.au/grants-dataset.
10
the2010ExcellenceResearchAustralia(ERA)EvaluationemployedbytheFederalGovernment.
‘Disciplines’aredeterminedbythefour-digitFieldsofResearch(FOR)codeasidentifiedinthe
AustralianandNewZealandStandardResearchClassification(ANZSRC2008).
Animportantcontributionofourpaperisinreconcilingthe2008FORsystemwithitspredecessor,
the1998ResearchFields,CoursesandDisciplines(RFCD)classificationsystem.Forthispurpose,we
utilisetheone-to-multiplemapbetweenRFCDcodesandFORcodesthatcanbefoundinthe
ANZSRCcorrespondencetables.14Togetherwithinformationontheproportionofeachproject’s
researchcontentattributabletoeachofthecomponentRFCD/FORcode,themaphelps
consistentlyidentifytheproportionoffundingtakenupbyeachFORcode.Wesubsequently
calculatetheamalgamatedARCfundwithineachdisciplineclustertoeachuniversityeachyear.
TheNHMRCdataset15–providesdataonresearchgrantsbytheNationalHealthandResearchCouncil.ThemethodofcompilationissimilartothatoftheNCGPdataset.Theonlydifferenceis
thatdataonthepercentagecontributedbyeachRFCD/FORcodetoeachprojectisnotavailablein
theNHMRCdataset.Wethushavetoattributetheentirefundingofaprojecttoitsmaindiscipline,
acompromisethatmayslightlybiasthealreadypredominantclusters–BiomedicalandClinical
HealthSciences(BCH)andPublicAlliedHealthSciences(PAHS).However,sincewearemost
concernedwithvariationsovertimeoftheallocation,thisbiasshouldbesmall.Finally,grants
allocatedtoprojectswithoutspecifiedresearchdisciplineisclassifiedas‘unspecified’,andthosefor
infrastructureareclassifiedas‘infrastructure’.
14Forexample,ifanRFCDcodecorrespondstotwoFORcodes,theARCfundallocatedtothatparticularRFCDcodeis
brokenintotwoequalamountsassignedtoeachoftheseFORcodes.15Theoriginaldatasetsseparatedbycommencementyear(1990-1999,2000-2009,2010-2019)aredownloadablefrom
http://www.nhmrc.gov.au/grants-funding/research-funding-statistics-and-data.
11
Figure1:Researchfundingbysource,2013
*Figuresfor‘OtherACGRSchemes’iscomputedastheresidualoftheCompetitiveIncomesubtractingARCandNHMRCgrants.
TotalFunding($5291m)
CompetitiveIncome($1680m;31.75%of
total)
ARC($770m;14.55%of
total)
NHMRC($625m;11.87%of
total)
OtherACGRSchemes*
($285m;5.39%oftotal)
RBG($1670m;31.56%of
total)
OtherIncome($1941m;36.68%of
total)
OtherPublicSources($845m;15.97%of
total)
IndustryandOtherSources
($922m;17.43%oftotal)
CRCs($104m;1.97%of
total)
12
Figure2:Fundingsourcebyuniversitygroups
Figure3:Fundingallocationbyuniversitygroups
33.94%
28.75%
29.84%
7.475%
competitive publicindustry CRC
ATN
50.62%
22.16%
25.37%1.851%
competitive publicindustry CRC
Go8
41.54%
30.84%
24.77%2.849%
competitive publicindustry CRC
IRU
44.06%
23.03%
27.22%
5.689%
competitive publicindustry CRC
Other
Reference year: 2013
13
9.374%
67.11%
10.62%
12.9%
ATN Go8IRU Other
Total Funding
6.677%
72.88%
8.366%
12.08%
ATN Go8IRU Other
ARC4.243%
83.61%
7.073%5.074%
ATN Go8IRU Other
NHMRC
11.09%
63.66%
13.61%
11.64%
ATN Go8IRU Other
Public Funding
10.55%
66.82%
10.02%
12.62%
ATN Go8IRU Other
Industry funding
23.37%
43.13%
10.19%
23.32%
ATN Go8IRU Other
CRC
Reference year: 2013
14
Figure4
Figures5
15
16
Figure6
Figure7
17
18
Figure8
Figure9
19
20
Figure10
Figure11
21
22
Figure12
23
Table1:UniversityGroupsATN QueenslandUniversityofTechnology,UniversityofTechnologySydney,RMITUniversity,Universityof
SouthAustralia,CurtinUniversity.Go8 UniversityofWesternAustralia,MonashUniversity,AustralianNationalUniversity,Universityof
Adelaide,UniversityofQueensland,UniversityofMelbourne,UniversityofNewSouthWales,UniversityofSydney.
IRUa CharlesDarwinUniversity,FlindersUniversity,GriffithUniversity,JamesCookUniversity,LaTrobeUniversity,MurdochUniversity,UniversityofNewcastle.
Other AustralianCatholicUniversity,BondUniversity,CentralQueenslandUniversity,CharlesSturtUniversity,DeakinUniversity,EdithCowanUniversity,MacquarieUniversity,SouthernCrossUniversity,SwinburneUniversityofTechnology,FederationUniversityAustralia,UniversityofCanberra,UniversityofNewEngland,UniversityofNotreDameAustralia,UniversityofSouthernQueensland,UniversityofTasmania,UniversityofWesternSydney,UniversityofWollongong,UniversityoftheSunshineCoast,VictoriaUniversity.
a:TheUniversityofNewcastleleftIRUinDecember2014.Inourstudy,groupmemberstatusistakenasinyear2013.
24
Table2:Researchfunding(asapercentageofGDP)bysources,1992-2013
Year TotalFunding(A=B+C+D)
CompetitiveIncome
(B=B1+B2+B3)
RBG(C) OtherSources(D=D1+D2+D3)
ARC(B1) NHMRC(B2) OtherACGRSchemes(B3)
OtherPublicIncome(D1)
IndustryandOtherIncome
(D2)
CRC(D3)
1992 0.0631 0.0162 0.0246
1993 0.0624 0.0162 0.0258
1994 0.0636 0.0142 0.0182 0.0369
1995 0.061 0.0153 0.0208 0.0386
1996 0.0649 0.0163 0.0199 0.0387
1997 0.0658 0.0171 0.0217 0.0436
1998 0.0649 0.0181 0.0211 0.0429
1999 0.0673 0.0195 0.0217 0.0432
2000 0.0692 0.0845 0.0174 0.0245 0.0481 0.0119
2001 0.2891 0.0672 0.1295 0.0924 0.0209 0.0275 0.054 0.011
2002 0.2916 0.0672 0.1288 0.0956 0.0245 0.0291 0.0545 0.012
2003 0.3055 0.0784 0.1268 0.1003 0.0268 0.0322 0.0546 0.0135
2004 0.3026 0.0823 0.123 0.0974 0.0403 0.0275 0.0145 0.0335 0.0512 0.0127
2005 0.3079 0.088 0.1181 0.1017 0.0467 0.0295 0.0118 0.0373 0.051 0.0135
2006 0.3252 0.0921 0.1128 0.1203 0.0497 0.0303 0.0120 0.0474 0.0603 0.0126
2007 0.3259 0.093 0.106 0.1269 0.0453 0.0331 0.0146 0.0566 0.0592 0.0111
2008 0.3264 0.0931 0.0992 0.134 0.0421 0.036 0.015 0.0616 0.0624 0.01
2009 0.3216 0.0972 0.101 0.1234 0.0447 0.0405 0.012 0.0608 0.0529 0.0098
2010 0.3521 0.118 0.1047 0.1294 0.0451 0.0392 0.0337 0.0621 0.0585 0.0088
2011 0.3274 0.1006 0.1041 0.1227 0.0465 0.0393 0.0149 0.0583 0.057 0.0074
2012 0.3356 0.1067 0.1085 0.1204 0.0496 0.0392 0.0179 0.0575 0.055 0.0078
2013 0.3366 0.1085 0.1074 0.1206 0.0496 0.0403 0.0186 0.0545 0.0594 0.0067
Averageannualchangefromthefirstdatayearto2013 1.28% 2.61% -1.55% 2.77% 2.33% 5.64% 2.81% 5.95% 4.29% -4.32%Changefrom2004to2013 11.24% 31.83% -12.68% 23.82% 23.08% 46.55% 28.28% 62.69% 16.02% -47.24%
Note:DataforB,C,D1,D2andD3isobtainedfromHERDCdatasetasdescribedinsection2.DataforB1andB2iscompiledseparatelyfromtheARCandNHMRCdatasetsrespectively.DataonnominalGDPiscollectedfromtheAustralianBureauofStatistics,5206.0Australian
25
NationalAccounts.ThedatainabsolutetermsforARCandNHMRCmaybedownwardlybiasedinthebeginningyearsofthecorrespondingdatasets,duetothefactthatgrantsarerecordedbycommencementyear.However,sincethemajorityofARCgrantsspanaperiodofthreeyearsorless,weexcludeonlydatapointsinthefirsttwoyearsofthecorrespondingdataset.NHMRCdatasetstartsfrom1990,howeverwealsoexcludethedatapointsin1992and1993ofthisdataset,duetothelongerspanofNHMRCgrantscomparedtothoseARCones.Foranalysesonthesharesoffunding,wehoweverkeeptheentiredatasets,assumingthatthissortofbiasaffectsalluniversitiesanddisciplinesequally.