Discussionof Panel1: NewEvidenceonCo-HoldingPuzzles Hwan-sikChoi - - PowerPoint PPT Presentation

Discussionof Panel1: NewEvidenceonCo-HoldingPuzzles

Hwan-sikChoi DepartmentofEconomics BinghamtonUniversity October,2020

1 / 13

TheOutline

• 1. Newevidencefromthe twopapers.
• 2. Strengthandweaknessofbankdata.
• 3. Whatmight betherootcauseofco-holding?
• 4. Sometechnicalcommentsontheirregressions.
• 5. Mywishlistandconjectureoftherootcause.
• 6. Policydiscussion.

2 / 13

TwoPapers: FindingsfromObjectiveBankData

TheCo-HoldingPuzzle: NewEvidencefrom Transaction-LevelData by John Gathergood and Arna Olafsson (2020)- “PAPER1”

• 1. Co-holdingisnotsevere(annualizedmediancost$12).
• 2. Hardtoexplainco-holdingwithliquidity need,creditlinerisk,

limitedattention,shopper-savermodel.

• 3. Suggest“mentalaccounting”orself-controlasafactor.

UntanglingtheCreditCardDebtPuzzle by Erkki Vihriälä (2020)- “PAPER 2”

• 1. Evidence ofuncooperative behaviorwithincouples.
• 2. Anchoringtominimumpayments(withnoliquidity problem)

Frequentminpayments ) moreco-holding

3 / 13

MainFeature: ObjectiveBankData

Paper1: Financialaggregatorapp. Dailyfrequency,transaction-levelinformationfromIcelandicbanks. Paper2: MonthlydatafromalargeFinnishbank. Includesinterestrates,housingvalue,etc. Advantage:

• Verydetailedbank-heldasset/debtinformation.
• Objectiveincomeandexpendituremeasures.

Disadvantage: Limited personaldata & non-bankfnancials

• Lackofabilitymeasuressuchaseducationlevel,fnancial

knowledge,cognitiveand noncognitiveabilities,and

• ccupation. Nopsycho-socialmeasures.

) Diÿculttoexamine(orcontrolfor)ability and psycho-socialfactors.

• Limitedinformationon brokerageaccounts (stock, funds).

4 / 13

WhatistheRootCauseofCo-holding?

Personalv.s. situationalfactors. Ability (personal) Situational Liquidity needs ! Self-control ! Mentalaccounting ! Creditlinerisk ? ! Shopper-saver ! ? Whatistherootcause? Rootcause= Policytarget.

• Lackofinformation.
• Lackofcomprehensionandeducation.
• Lackofincomeorliquidity.
• Heuristicsand boundedrationality?

5 / 13

SomeTechnicalCommentsforPaper1

• 1. Likelihood ofstartingco-holding:
• Signifcancelevelswith 3 millionobservationsmaynot bevery

informative.

• Seeminglysmalleconomicmagnitudeofthee˙ect: e.g. 0.23%more

likelytostartco-holdingwithtemptationbuying(FEmodel).

• Mostlyunexplainedvariation: R-squared=0.003.
• Theon-setofco-holdingisatransientvariablewhereas

income/expenditure isrelatively persistent. Thechange of income/expendituremay bemorerelevantfortheon-set.

• Simultaneity (endogeneity)ofco-holdingandincome(or

expenditure)? Whatifcognitiveability infuence bothco-holdingand income?

• 2. Sampleselectionbias? Householdswith“atleastonedeposit

accountandoneoverdraftline” ) Openingoverdraftlinesmay beanendogenouschoice(alluding futureco-holding).

6 / 13

SomeTechnicalCommentsforPaper1 (Cont.)

• 3. “Plausibly,individualsmight prefertoassigntheirgamblingand

temptationpurchasestoaseparate(mental)account,asidefrom theirregularexpenditure.”

• Cancheckwhethergambling/temptationbuying occurs in the
• verdraftaccounts?
• Ifso,gamblingis highly costly. Isthisconsistentwiththedual-self

(shopper-saverwithin an individual) model or is itfnancial ignorance?

• 4. Whatdeterminesthedurationofco-holdingonceco-holdingstarts

(individuallevelanalysis)?

7 / 13

SomeTechnicalCommentsforPaper2

• 1. Mortgageo˙ertake-upregressionforintra-householdinteraction:
• Lowdepositorhighdebtshare⇒ highmortgageo˙ertake-up.
• Est. sample=co-holdinghouseholdsonly(why?).
• Whatifthee˙ectsaresimilaramongnon-coholdingcouples?

I suspectthatnon-coholdersarealsouncooperative,butthee˙ect size may be larger forco-holdingcouples?

• 2. Householdswitchingfromasingletoacouple(orviceversa)

) Statisticallysignifcant(andeconomicallylarge)changein co-holding behavior? Canweusethepanelnatureofthedatatoanswerthisquestion?

8 / 13

SomeTechnicalCommentsforPaper2 (Cont.)

• 3. Ismortgageo˙ertake-upratesimilarfornon-coholders?
• Ifso,then“debtaversion”may drivetheirmortgagetake-up

behaviors.

• However,co-holdersseemtohavenodebtaversion.
• Isthereamorefundamentalpsychologicalfactorthatcanexplain

bothdebtaversionandco-holding?

• 4. Intra-householdinteractioncanimplyeither(1)cooperation(2)

noncooperation(shopper-saver model). Study therelationship of cooperation/noncooperationwithage-gap oreducation/incomegap withinthe household.

• 5. Anchoring- makingminpaymentand co-holdingaresimultaneously

(endogenously)determined. Whatistherootcauseofthe anchoring? Information(education)orability tocomprehendthe fnancialsituation?

9 / 13

CurrentChallengesandtheWishList

• 1. Whatiscausing thementalaccounting? Information,education,or

cognitiveability?

• 2. Useofbank dataisnotsuÿcienttoseparatethesefactors.
• 3. Ideally,wewanttocombinebankdata,surveydata,andgovernment

databases.

• 4. Realistically, might try toextract personal data in mortgage

applications- education,creditscore,occupation,income,etc.

10 / 13

MyConjecture

• 1. Peopleunderstandthatco-holdingiscostly,butjustalittle.
• 2. Fromtheco-holder’s perspective,thecostseemssmallenoughto

notcare.

• 3. Mentalaccounting,self-control,minimumcreditcardpayment.

“Slightly”sub-optimalchoiceistheoptimalchoice. Mentalcost >monetarycost.

• 4. Rootcause: Limitedcognitiveandnoncognitiveability.
• Lesscognitivee˙ortandheuristics(Simon, Tversky,andKahneman).
• Lazinessandcarelessness(non-Conscientiousness)

Missingdeadlines. Leavingdirty dishes & trash. Missingclasses. Notcheckingbilledamounts. Losingumbrellasmultiple Impulsiveshopping. timeswithoutregrets.

11 / 13

PolicyImplications

• 1. Can we improvementalaccounting by education?
• 2. Doesineÿcientmentalaccountingstemfrommisunderstanding,

lackoffnancialliteracyormorefundamentalcognitiveor noncognitiveability?

• 3. Lazinessandcarelessness(noncognitiveability)?

) Enforcebankstonotifyconsumers. Wouldtextmessageshelpnudgingtopay o˙overdraftbalances?

12 / 13

PolicyImplications(Cont.)

Libertarian Paternalism

• 4. Freeoverdraft protection(ODP)withlinkedaccounts?

Expensiveservicefee=$25-35intheU.S.

• If not protected, need to pay $35 or more(non-suÿcientfunds fee).
• Mentalaccountingmay bepsychologically benefcial.
• Nonetheless,co-holdersmay stilldemandinsuranceagainstoverdraft.
• DefaulttofreeODP programwithopt-outoption

⇒ Less proft forbanks, buthigherconsumer welfare.

• 5. Other policysuggestions:

– Fernandesetal. (2014): Just-in-timeeducation(earlyintervention hasnolastinge˙ect). – Thaler(NYT,2013): Rules of thumb +Easyanduser-friendly system(easytoviewtheoptionsandopt-out). ⇒ Lesse˙ectiveforco-holding & impulsivebuyers⇒ Defaultoption is better.

13 / 13