FinanceandEconomicsDiscussionSeries

FederalReserveBoard,Washington,D.C.

ISSN1936-2854(Print)

ISSN2767-3898(Online)

NavigatingHigherEducationInsurance:AnExperimentalStudyonDemandandAdverseSelection

SidhyaBalakrishnan,EricBettinger,MichaelS.Kofoed,DubravkaRitter,DouglasA.Webber,EgeAksu,andJonathanS.Hartley

2024-024

Pleasecitethispaperas:

Balakrishnan,Sidhya,EricBettinger,MichaelS.Kofoed,DubravkaRitter,DouglasA.Web-ber,EgeAksu,andJonathanS.Hartley(2024).“NavigatingHigherEducationInsurance:AnExperimentalStudyonDemandandAdverseSelection,”FinanceandEconomicsDis-cussionSeries2024-024.Washington:BoardofGovernorsoftheFederalReserveSystem,

/10.17016/FEDS.2024.024

.

NOTE:StafworkingpapersintheFinanceandEconomicsDiscussionSeries(FEDS)arepreliminarymaterialscirculatedtostimulatediscussionandcriticalcomment.TheanalysisandconclusionssetfortharethoseoftheauthorsanddonotindicateconcurrencebyothermembersoftheresearchstafortheBoardofGovernors.ReferencesinpublicationstotheFinanceandEconomicsDiscussionSeries(otherthanacknowledgement)shouldbeclearedwiththeauthor(s)toprotectthetentativecharacterofthesepapers.

NavigatingHigherEducationInsurance:AnExperimental

StudyonDemandandAdverseSelection*

SidhyaBalakrishnan?

,EricBettinger?

,MichaelS.

Kofoed§

,DubravkaRitter?

DouglasA.Webber|,|EgeAksu*,

*a

ndJonathanS.Hartley??

February21,2024

Abstract

Weconductasurvey-basedexperimentwith2,776studentsatanon-profituniversitytoanalyzein-comeinsurancedemandineducationfinancing.Weofferedstudentsahypotheticalchoice:eitherafederalloanwithincome-drivenrepaymentoranincome-shareagreement(ISA),withrandomizedfram-ingofdownsideprotections.EmphasizingincomeinsuranceincreasedISAuptakeby43%.Weobservethatstudentsareresponsivetochangesincontracttermsandpossiblestudentloancancellation,whichisevidenceofpreferenceadjustmentoradverseselection.Ourresultsindicatethatframingspecifictermscanincreasedemandforhighereducationinsurancetopotentiallyaddressriskforstudentswithvaryingoutcomes.

*TheauthorsappreciatethehelpfulcommentsandfeedbackofNeilBhutta,JuliaCheney,BarryCynamon,AndrewHertzberg,JeromeHodges,CarolineHoxby,RajeevDarolia,RobertHunt,JeffLarrimore,JosephMarchand,LoisMiller,KevinMumford,MarshallSteinbaum,andJoshuaPrice.WearealsogratefulforparticipantsatseminarandconferencepresentationsincludingtheAmericanEconomicAssociation,AppalachianStateUniversity,AssociationforEducationFinanceandPolicy,AssociationforPolicyAnalysisandManagement,BrighamYoungUniversity,NationalBureauofEconomicResearch(EconomicsofEducation),ProvidenceCollege,SouthernEconomicAssociation,UnitedStatesAirForceAcademy,andUniversityofTennessee,Knoxville.

?SidhyaBalakrishnanisthedirectorofresearchattheJainFamilyInstitute,email:sidhya.balakrishnan@

?EricBettingeristheConleyDeAngelisFamilyProfessorofEducationatStanfordUniversityandaresearchassociateatNBER,email:ebettinger@.

§MichaelS.KofoedisanassistantprofessorofeconomicsattheUniversityofTennessee,KnoxvilleandResearchFellowatIZA,andcorrespondingauthor:mkofoed1@

?DubravkaRitterisasenioradvisorandresearchfellowattheConsumerFinanceInstitute,FederalReserveBankofPhiladelphia,email:dubravka.ritter@.ThisPhiladelphiaFedworkingpaperrepresentspreliminaryresearchthatisbeingcirculatedfordiscussionpurposes.TheviewsexpressedinthesepapersaresolelythoseoftheauthorsanddonotnecessarilyreflecttheviewsoftheFederalReserveBankofPhiladelphiaortheFederalReserveSystem.Anyerrorsoromissionsaretheresponsibilityoftheauthors.

NotelssiBoardofGovernorsoftheFederalReserve,email:douglas.a.webber@.

TheanalysisandconclusionsinthispaperarethoseoftheauthorandshouldnotbeinterpretedasreflectingtheviewsoftheBoardofGovernorsortheFederalReserveSystem.

**EgeAksuisaPhDcandidateatCUNYGraduateCenterandfellowattheJainFamilyInstitute,email:ege.aksu@

??JonathanS.HartleyisaPhDcandidateatStanfordUniversity,email:hartleyj@

2

1Introduction

Insuranceproductsareimportanttoolsemployedbyindividualstohedgerisksintheirfinanciallives.Insur-ancemarketsallowindividualstopoolriskagainstunexpected,negativeoutcomesandarewelldevelopedinmanycontexts,likehealthcareorrealestate.

1

However,risk-hedgingopportunitiesarenotreadilyavailableinpost-secondaryeducation,eventhoughcollegeisanincreasinglyuncertaininvestment(Webber,

2022)

madeonlyonceinalifetime.Whilereturnstocollegearepositiveonaverage(LovenheimandSmith,

2022

),theirdistributionismorenuanced(Webber,

2016

,BroadyandHershbein,

2020)

.Financialout-comesforstudents,forexample,varyacrossinstitutiontypes(e.g.selectivevs.non-selective;four-yearvs.two-year),fieldsofstudy(e.g.educationvs.engineeringvs.economics),andmacroeconomicconditionsupongraduation(Rothstein,

2023

).Perhapsmoreimportantly,returnsvarywithineachofthesesegmentsgivenunobservablestudentskill–whichispotentiallydifficultforthestudentand/ortheinstitutiontoiden-tify–anduncertainlabormarketconditions.Particularlyforyoungerstudentsandthoseenteringlongerdegreeprograms,thereisuncertaintybothintheexpectedlevelofincomeandinitsvariability.

Students,educationalinstitutions,andgovernmententitiesunderstandandbeartheseriskstodifferentdegrees.Forexample,workbyStange(

2012

)showsthatmanystudentstreatattendingtheirfirstyearofcollegeakintopurchasinganoptionscontract—completinganinitialyearsoastodevelopabettersenseoftheirlikelyreturns,afterwhichtheydecidewhethertoexercisetheoptionforasecondyear.Policymakersandadvocatesoftenworktotransfertheriskinessofthereturntotaxpayers,e.g.viathefreecollegemovement,financialaidpolicy(bothgrantsandloans),ortheCovid-19studentloanrepaymentpausefor

loansguaranteedbythefederalgovernment.

Interestingly,individualeducationalinsurancepolicieswherestudentspayapremiumtoprotectthem-selvesfromincomeriskareeithernotwelldevelopedorarenon-existent.

2

Onereasonforthelowpreva-lenceofeducationalinsuranceinpost-secondarymarketsmaybelowdemand.Thereisevidencethatstu-dentscanbeover-optimisticaboutfutureearnings(e.g.Bakeretal.(

2018

)),failingtoadequatelyconsiderinsuranceriskatthetimeofenrollmentandfinancingbecauseofdifficultyinpredictingfutureincomesac-cordingtomajor(Arcidiaconoetal.,

2012

;Bakeretal.,

2018

;Conlon,

2021

).Onthesupplyside,anotherexplanationcouldbethepresenceofadverseselection(Einavetal.(

2023

))andmoralhazard(ZweifelandManning,

2000

)ininsurancemarkets.

1Closesttooursetting,individualspurchaseinsurancetomitigatefinanciallosses(e.g.Arrow,1963),bufferagainstincomeshocks(e.g.ChettyandSzeidl

(2007)),andforavarietyofotherreasons.GuisoandPaiella

(2008)documenttheincreasingpropensity

ofhouseholdstohedgeagainstlaborincomeriskinparticular,indicatingarisingawarenessofemployment/incomeuncertainties.Inthecaseoffinancialmarkets,diversification,includingtheuseofderivativeinstrumentslikefuturesandoptions,remainsaprimarystrategyforriskmanagementinthefaceofuncertaineconomicoutcomes(e.g.Bodie

(1994);GoyalandWelch

(2007))

.

2Throughoutthispaper,wewillassumethattheprimaryformofinsuranceinpost-secondaryeducationisagainsttheriskofloworuncertainincome,andwillrefertothisas"lowincomeinsurance"or"educationalinsurance."

3

Ourpapermakesauniquecontributiontoourunderstandingofstudentdemandforeducationalinsur-anceandthepotentialrelevanceofadverseselectionintheviabilityoflow-incomeinsuranceineducationmarkets.Studentswhoareunsureabouttheirprospectsmaydemandinsurancetoprotectthemselvesagainstdownsidelabormarketrisks.Adverseselectionisrelatedbutdifferent:theinsureduseinformationthattheycaneasilyconcealfromtheinsurancecompanytotakeadvantageoftheinsurance’sdownsideprotections.Ourmajorcontributionisaframingexperimentusedtoteaseoutinsurancedemand;oursurveyallowsustoseeconcealedinformationunavailabletothehypotheticalinsurertotestforadverseselection.Usingtheexperimentcoupledwiththesurvey,wecantesthowframingaffectsastudent’sdemandforeducationinsuranceandseeifstudentsuseconcealedinformationtotakeuptheprotectionsatdifferentrates.

Wepartneredwithalarge,non-profituniversity(hereafter,TheUniversity)thattypicallyservesnon-traditionallyagedstudentswhoareoftenworkingadults.Weconductedarandomizedsurveylabexperi-

mentwith2,776studentstounderstandtheirpreferencesoverdifferenteducationalfinancingchoices.

Inthesurvey,studentswereaskedtochoosebetweenahypotheticalfederalstudentloanwiththeoptionofanincome-drivenrepayment(IDR)planandahypotheticalincome-shareagreement(ISA).Bothoptionsprovidedinformationonmonthlyloanpaymentsthatarewaivedforverylowincomesandotherwisearecappedtoafixedshareofanindividual’sincome,withmarkedlydifferentimplementationandpathsforsatisfyingtheloanobligation.

Intheexperiment,studentswererandomizedintotwoequalgroupsand,similartoAbrahametal.

(2020),thepresentedhypotheticaloptionsdifferedintermsoflevelofdetailprovidedforeachofthe

choices.ThefirstgroupwasshowndescriptionsofthestudentloanwithoptionofIDRandtheISAwitharisk-neutralframingthatexplainedthedifferencesinmonthlypayments,generalstructureoftheloan,thebaselinepaymentterms,andthesourceoffunding.ThetermsofthestudentloanwithIDRandtheISAweresettobeactuariallyequivalent.Weexposedthesecondgroup–ourtreatmentgroup–tothesamedescriptionsofthestudentloanwithoptionofIDRandtheISA,butwithanadditionalemphasisontheinsurancefeatures(natureoftheincomecontingencyandmaximumrepaymentterm)ofthetwofinancingoptions.

TherearemanydifferencesbetweenfederalstudentloanswithIDRandISAs,andmanyreasonswhydifferentborrowersmightpreferoneovertheother.Withfederalstudentloans,borrowerswhodowellinthelabormarketwillpaylessintotalbypayingfixedmonthlypaymentsfortheminimumnumberofyears(120payments,or10yearswithnogapsinpayment).Sincethereisnoprepaymentpenaltyforfederalstudentloans,theycanalsobepaidofffasterthanscheduledandmaybeparticularlyattractivetostudentswhoexpectconsistentlyhighearningsaftercollege.Toaccesstheincomecontingency,borrowersmustfollowaseriesofadministrativehurdlesinordertoqualifyforreducedmonthlypaymentscappedata

4

certainpercentageoftheirincome,payingnothingiftheirincomefallsbelowasetthreshold,butpotentiallyextendingtheirtermupto20yearscomparedwiththestandardrepaymentplan

.3

WithanISA,borrowers’monthlypaymentsaresetasapre-agreedshareofincomebydesign,andtherepaymenttermistypicallyextendedtoalesserdegreethanIDRduetomonthsofnon-payment,makinganISAapotentiallyattractivepropositionforborrowerswithpersistentlyloworvariableearnings.Ontheotherhand,thereisnowayto“refinance”outofanISAandborrowerswhoendupearninghighincomeswillpayuptoamultipleoftheoriginalloanamount,describedinourexperiment.Finally,borrowersmayhavepreferencesoverborrowingfromthegovernmentversusaprivatelender.

Wefindthatstudentshaveasignificantpreferenceforthebuilt-inincomeinsuranceintheISAandthatourinsuranceframinggreatlyincreasesthedemandforthehypotheticalISA–byabout10percentagepoints,or43percent.Importantly,thereislimitedheterogeneityintreatmenttobefoundalongdemo-graphic,academic,orfinanciallinesforstudentsinoursample.

4

Theinsuranceframinghas,byfar,thelargesteffectontake-up.Ourresultssuggestthatstudentsarenotnecessarilythinkingaboutincomeriskoraboutthepotentialbenefitsofeducationalinsurancewhentheychoosehowtofinancetheirstudies,butthateducationalandloanproviderscanhelpmakethepotentialneedforeducationalinsurancesalientforborrowersbythoroughlyexplainingthecostsandbenefitsofsuchinsurance.

Oursurveyandfollow-upquestionsalsoallowustocharacterizehowadverseselectionmayentertheeducationalinsurancemarket(HerbstandHendren,

2021)

.5

ThesurveyallowsustosolicitinformationfromthestudentthatisunavailabletotheISAoriginator.Studentsmaybeconfidentabouttheirincomepotentialbutcaneasilyconcealthisinformationfromfinancingproviderswhodonothavetheabilitytopricediscriminate(i.e.mustchargethesameinterestrateorincomesharetoall).Insuchanenvironment,studentsexpectinglowincomeswillsortintotheISAwhilestudentsexpectinghighincomeswilloptforatraditionalloan.EducationalinsurancethatlooksmorelikeanISAwillnotbeasustainablepolicychoiceifstudentswhoachievesignificantreturnstocollegesystematicallychoosestudentloans.Overall,thereislessevidencesuggestiveofadverseselectionacrossavarietyofvariablesthanwesupposedexante.Employmentuncertainty,forexample,doesnotappeartoinfluencetake-up.However,wedofindstrong

suggestiveevidenceofadverseselectionbasedonlikelihoodoffutureincomesbeinglow.

Tofurthertestforadverseselection,weaskseveralfollow-upquestionstoinvestigatehowstudents

3RecentpolicychangesaroundtheSavingonaValuableEducation(SAVE)IDRplansimplifysomeoftheseprocessesforfederalstudentloans,buttheincomeprotectionisstillfarfrombuiltin.AlthoughsomecurrentlyavailableIDRplansofferedbytheDepartmentofEducationextendtherepaymenttermtoupto25years,wewantedtokeepthecomparisonsimpleforborrowersandchosethe(modal)maximumtermof20years.

4Thepreregisteredbaselinevariablesforheterogeneoustreatmenteffectsincludedrace/ethnicity(Black,Hispanic,white),gender(indicatorforfemalerespondent/recipient),householdsize,age(mediansplit),maritalstatus,riskaversion.

5Sincewecannotfollowstudentsafterthesurvey,wecannotshedlightonpotentialmoralhazardfromtheavailabilityofinsuranceforstudentsinthetreatmentgroup.

5

mightchangetheiranswersiftheoffertermsweremodestlydifferent.AftertheirinitialchoicebetweenastudentloanandanISA,weofferedstudentswhooriginallyselectedastudentloanwithIDRanactuarially-equivalentalternativeISAwithalowerincomeshareandalongerterm.Ifastudentmaintainedtheiroriginalchoiceofstudentloaninthesecondroundaswell,wethenofferedthemanotheralternativeISAinthethirdround–thistimewithahigherincomeshareandashortertermthantheoriginalISA.TostudentswhooriginallyselectedthehypotheticalISA,weseparatelyofferedbothalternativeISAsatthesametime.WefindthatbothoriginalISAchoosersandoriginalstudentloanchooserswereequallylikelytoswitchtothelongertermISA,with18%ofrespondentsselectingthelongertermISAovertheiroriginalchoice.Interestingly,respondentswhooriginallychosetheISAwereconsiderablymorelikely(61%)tochoosetheshortertermISAcomparedwiththerespondentswhochosethestudentloanwithIDRinboththefirstand

secondrounds(17%).

Wefurtherfindthattreatedstudentswhooriginallychosethestudentloanwere5.1percentagepoints(17%)morelikelytochoosetheISAoptionwithalowershareandlongerterm.ThetreatmenteffectforthoseofferedthealternativeISAwithshortertermandhighershareisnotstatisticallysignificant.ForstudentswhochosetheoriginalISAinourbaseexperiment,treatedstudentswere8.6percentagepoints(15%)morelikelytochoosethealternativeISAoptionwithahighershareandshorterterm.WefindnosignificanttreatmenteffectsonswitchingtowardthelowershareandlongertermISAforstudentswhooriginallychosetheISAoverthestudentloanwithIDR.Overall,ourresultssuggestthattheinsuranceframinghelpedreinforcestudentpreferencesovershortermaximumrepaymentterm(12yearsforISAv.20yearsforstudentloan),andthatstudentsselectedalternativeISAcontractsinourfollow-upquestionsinawaythatreflectedtheirstatedpreferencebetweentheinitialchoice.

Separately,weaskedallstudentswhethertheywouldselectastudentloaniftherewerea20%chancethatthe$10,000loantheyborrowedwouldbeforgiven.WefindthatstudentswhoswitchedfromaloantoanISAinthesecondroundwere6.4percentagepointsmorelikelytoswitchbacktoaloanwhenof-feredthechanceofstudentloanforgiveness.Conversely,whenstudentswhopickedtheoriginalISAwereofferedaloanwithachanceoffuturedebtforgiveness,theprospectoffuturebalancereductiondecreasedthewillingness/likelihoodofswitchingbacktothefederalstudentloan.ThestudentswhoswitchedtoanISAinthesecondroundaremarginallyattachedtotheISAandmaybeeasilyinducedtoswitchbetweenthetwofinancingchoicesgivenrelativelysmallchangestoterms,whiletheoriginalISAchoosersappeartobemoresetintheirISApreference.Takentogetherwithourresultsonstudentpreferencesoveralterna-tiveISAvariationscomparedwithastudentloanwithIDR,ourstudycontributestotheunderstandingofoptimaldesignofloanproductswithincomeinsurancefeatureswithregardtobothupsideprotections(likemaximumtermortotalpaymentamount)anddownsideprotections(likeincomeshare).

6

Thoughfocusedoneducationalinsurancemarkets,ourpapercontributestootherlinesofresearch,

includingtheliteratureonfinancialaid,educationfinance,andstudentdebt.Asthecostofhighereducationhasrisenandthepurchasingpowerofpublicsubsidieshavefallenalongwithpublicfinancialsupporttouniversities(Webber,

2017

),familieshavehadtoincurdebtorforegoconsumptiontoaffordpost-secondaryeducation.Recentresearchhasemphasizedtheburdenthatstudentloansplaceonstudents(Chakrabartietal.,

2020

)-includingontheirotherconsumerspending(e.g.MezzaandSommer,

2015

),"lifemilestones"(Mezzaetal.,

2020),andeducationaloutcomes(e.g.Blacketal.,

Forthcoming

,DenningandJones,

2021)

.Becausethemonthlypaymentisproportionaltoincome,educationinsurancesuchasIDRandISAscanhedgeagainsttheadverseeffectsofstudentloans.Assuch,take-upofsuchproducts,particularlyamongpopulationswherestudentloanshavehadadverseeffects(e.g.studentsattechnicalorpublicregionalcollegeswithhighervarianceincollegeoutcomes),isimportant.

Manyfamiliesdonotapplyforaid–meaningtheydonotcompletetherequiredfinancialaidforms–becauseofalackofinformationoruncertaintyofeligibility(e.g.Kofoed(

2017

),Bettingeretal.(

2012

)).Evenforfamiliesthatapplyforaid,theprotectionofsomeassetswithinthePelleligibilityformulaleadstolessfinancialaideligibilityforstudentsfromdisadvantagedfamilies(LevineandRitter,

2023

)resultinginhigherstudentloanburdensanddecreasedaccesstoselectiveinstitutions.Formsofbuilt-ineducationalinsurancearepotentiallymoreattractivetothesestudents,astheyautomaticallyreallocatesomeoftheriskfromthestudenttotheproviderandmayreduceuncertaintyaroundfinancialaideligibility.

Additionally,ourresearchcontributestothebehavioralliteratureonstudenttake-upoffinancialaidprogramsundervariedframing.

6

Abrahametal.(

2020

)andMarxandTurner(

2019

)demonstratethatframingmattersforgovernment-sponsoredIDRplansandtraditionalstudentloantake-up,respectively.Inthisliterature,researchersmanipulatethestudents’norms,theterms,risk,andthecostscommunicatedtostudentswithrespecttospecificfinancialinstruments.Ourpapercontributestothisdiscussionbyshowingthatstudentspreferincomecontingentfinancingwhenweemphasizethebuilt-ineducationalinsuranceoftheISA,withtreatmenteffectsfortheinsuranceframingcomparableinmagnitudetoAbrahametal.(

2020)

.Thelessonsfromourstudyareapplicabletothedesignofanyincome-contingenteducationfinancingproductandareparticularlysalienttoongoingpolicydiscussionsaroundtheDepartmentofEducation’s

income-drivenrepaymentplansforfederalstudentloans.

Ourpaperisorganizedasfollows.Section

2

reviewsstudents’collegefinancingandtheprospectforeducationinsuranceinfinancingeducation.Section

3

detailstheexperimentaldesignofourresearchquestions.Section

4

laysoutourempiricalstrategyandelaboratesonthedatacollection.Section

5

provides

6Coxetal.

(2020)examinewhystudentsdon’tchooseIDRwhentheyareworriedaboutfutureincomeexpectations.They

conductalaboratoryexperimentwheretheyprovideinformationaboutIDRanddefaultstudentsintotheplan.Theyfindthatextrainformationandcorrectdefaultingdoesincreaseenrollment.

7

empiricalresults.Section

6

offersconcludingremarksandpolicyconsiderations.

2CollegeFinancingandEducationalInsurance

Thissectiondescribesborrowerlabormarketexpectationsandavailable(studentloanwithIDR)andrel-

ativelynovel(ISAs)highereducationfinancingoptionswithincomeinsurancethatmotivatedourexper-iment.InSection

2.1

,wedescribeborrowers’income/employmenttrajectoriesandpotentialrisksanddisruptionstofutureincomeandemployment.WethenproceedtoexplainthemechanicsofanIDRoptionforatraditionalstudentloaninSection

2.2

andthetypicalISAinSection

2.3.

InSection

2.4

,wediscusshowthefeaturesofthetwofinancingoptionsmightinfluencechoicesbetweenthemandwhichtypesofborrowersmightrespondtowhichincentives.Finally,Section

2.5

discussesthemotivationbehindourexperimentandthecomparisonweofferedtostudentsinourstudy.

2.1Students’EducationalRisks,EarningsTrajectories,andRepaymentShocks

Untilrecently,studentsandparentstypicallyrepaidgovernmentstudentloansinfixedmonthlypaymentsoveragivenrepaymentperiod,akintoatraditionalmortgageloan(Karamchevaetal.,

2020

).Thispaymentremainedconstantregardlessofage,income,employmentstatus,orfamilysituation.Mostnon-governmentlendershaveofferedprivateloanswithastandard,mortgage-stylepaymentschedule,thoughselectlendersarebeginningtoofferorcontemplatealternativeoptions.Some95%ofoutstandingstudentdebtisguaran-teedbythefederalgovernment,sorepaymentplansdesignedandofferedbytheDepartmentofEducation

dominatethesetofchoicesavailabletostudents.

Itisimportanttoconsiderrisktothereturnstocollegeattendance(includingborrowingforthatatten-dance)whenthinkingaboutfutureincomeandemploymentprospectsforstudents(Webber,

2016

;Balakr-ishnanandCynamon,

2018

;HendricksandLeukhina,

2018

;Akers,

2021

).Perhapsthelargestriskfactortotherepaymentofeducationaldebtthatstudentsfaceinvolvestheriskofnon-completion.Historically,6-yearcompletionrateshavehoveredaround60%,andareevenlowerfornontraditionalcollegestudentsandnon-selectiveinstitutions(Bowenetal.,

2009

).Financialcircumstances,lackofacademicprepared-ness,andahostofacademicallyorientatedchallengesmayputstudents’financialinvestmentincollegeatrisk.Additionally,thereare“l(fā)ife”risksthatstudentsface,includingemergenciesarisingfromphysicaltoemotionalhealthtofamilycircumstances.Adultlearners,inparticular,reportthatchild-careemergencies,children’shealth,andeventransportationemergenciescanderailtheireducationalcareers(Markle,

2015)

.Evenforcompleters,riskstoincomeandemploymentaremany.Theaveragefinancialreturnforthe

mediangraduateofa4-yearcollegeoruniversityislarge,andhandilyoutweighstheimplicitandexplicit

8

costsofattendingcollege,whichiswhyenrollmentinapostsecondaryprogramofstudymakessensefor

moststudentsex-ante.Butex-post,returnsareheterogeneousacrossmanydimensionsincludingmajor,institutiontype,andinstitutionprestige.Sincestudentpopulationsparticularlyatriskoflowornegativereturnstocollegeenrollmenttendtoskewtowardvulnerablegroups,addressingtheriskinessofcollegeattendancewithproduct/programdesignandeffectivepublicpolicyisimperative.

Borrowersalsofaceavarietyofincomeshocksduringrepayment,suchthatrepaymentburdenscanvary

widelyforindividualswithvariableoruncertainincomeand/oremployment(ChapmanandLounkaew,

2015

;ChapmanandDearden,

2017

).Borrowersmayfacetemporaryrepaymentchallenges(e.g.duetoperiodsofunemploymentorunderemployment)orchronicrepaymentstrugglesduetolowincomes(e.g.becauseofdegreenon-completion,ordegree/majorwithpoorfinancialreturnoninvestment).

Fixedpaymentsovera10-yearperiodforborrowerswhorecentlycompletedordroppedoutofapro-gramofstudymaynotbeoptimalgiventypicalearningstrajectories,either.Formostborrowers,studentloandebtserviceratios(i.e.,scheduledpaymentsasashareofaborrower’sincome)aretypicallygreaterearlyintherepaymentterm,whenaborrower’sincomeislower.Thisisparticularlytrueforstudentloanborrowerswithlittleworkexperienceuponenteringrepayment,forborrowerswhotypicallybeginrepay-mentinlowerpaidearly-careerpositionsbutultimatelyearnsubstantialamounts(e.g.,medicaldoctors),andforborrowerswithdegreesinmajorsthattraditionallyhavesteepearningstrajectories(e.g.,biology).Consideringallofthesefactors,standard