You are on page 1of 36

Perceivedfinancialknowledgeandactualfinancialbehavior:

Aninternationalassessment
*DraftasofSeptember2012.

ColleenTokarAsaad
DoctoralCandidate

KentStateUniversity
CollegeofBusinessAdministration
DepartmentofFinance
Kent,OH44242

office:(330)6721211
cell:(440)5546001
fax:(330)6729806
email:ctokar@kent.edu

Abstract

Thispaperseekstobetterunderstandhowbothactualfinancialknowledgeand
perceivedfinancialknowledgeaffectfinancialdecisions.Usingtwointernational
data sets, results indicate that both types of financial knowledge influence
financialbehaviors,whereperceivedknowledgeappearstobejustasimportant
asactualknowledge.Additionally,overconfidentindividuals,orthosewithhigh
perceived(orselfassessed)knowledgebutlowactualknowledge,haveahigher
propensitytoengageinriskyfinancialbehaviors.Together,resultssuggestthat
financial literacy initiatives should focus not only on factual knowledge, but on
motivating and inspiring individuals to take action. With enough confidence,
individualsaremorelikelytomakegoodfinancialchoices.Butwithtoomuch
confidence and too little actual knowledge, individuals are apt to make risky
financialchoices.Thestrugglethenistohelpindividualsfindahealthydoseof
confidence.

Key words: financial literacy; perceived knowledge; overconfidence; risk; decisionmaking;


behavioralfinance;international;

Perceivedfinancialknowledgeandactualfinancialbehavior
1.Introduction
Financial literacy is a measure of the degree to which one understands key
financialconceptsandpossessestheabilityandconfidencetomanagepersonal
financesthroughappropriateshorttermdecisionmakingandsound,longrange
financial planning, while mindful of life events and changing economic
conditions.(Remund,2010,p.284)

Financial decisionmaking is an essential component of daytoday life, from minor decisions


suchasdecidingwhetherornottopurchasealattetomajordecisionssuchastakingonahome
mortgage.Severaldefinitionsoffinancialliteracyhighlightthefactthatinordertomakesound
financialdecisions,individualsmustnotonlypossessthenecessaryknowledge,butmustalso
have the ability to apply their knowledge to develop a course of action. Huston (2010)
emphasizesthatfinancialliteracyconsistsofbothknowledgeandapplication:Likegeneralor
health literacy, financial literacy could be conceptualized as having two dimensions
understanding (personal finance knowledge) and use (personal finance application) (p. 306).
KozupandHogarth(2008)discusstheroleoffinancialeducationinhowindividualscanattain
financial security, arguing that core financial literacy involves numeracy, critical thinking, and
economicawareness:
Consumers need to (1) know what pieces of information they need; (2) process
thosepieceswithfactorsrelatingtotheirsituation,tastes,andpreferences;and
(3)usetheoutputtomakedecisionsaboutwhatfinancialproductstopurchase.
(p.131)

Becauseofthepervasivenessandimportanceoffinancialdecisions,financialliteracyinitiatives
havereceivedattentionoverthelastdecade,tryingtohelpindividualsacquiretheknowledge
needed to make sound, informed consumer choices. However, the evidence regarding the
lasting effectiveness of the initiatives is mixed (Choi, Laibson, and Madrian 2005; Hilgert and
Hogarth, 2003; Lusardi, 2004; Mandell, 2009), begging the question: If financial knowledge is
notenough,whatotherfactorsinfluencethefinancialdecisionmakingprocess?
Thispaperseekstoexplorehowperceivedfinancialknowledge,inadditiontoactualfinancial
knowledge, influences financial decisions and behaviors. While actual financial knowledge

reflectscorrectanswerstomultiplechoicequestions,perceivedfinancialknowledgereflectsa
selfassessedleveloffinancialknowledge,whichmayormaynotcoincidewithactualfinancial
knowledge. This paper shows that both actual and perceived financial knowledge influence
manytypesoffinancialbehaviors,andsurprisingly,theaffectofperceivedfinancialknowledge
on financial behaviors is just as important, if not more important, than the affect of actual
financial knowledge. In instances where perceived knowledge exceeds actual knowledge (i.e.,
overconfident),anindividualhasagreaterlikelihoodofengaginginriskyfinancialbehavior.
Whilepsychologystudieshavedemonstratedthatperceivedknowledgeisrelatedtocognitive
functioning,fewstudieshaveshownhowperceivedknowledgeinfluencesfinancialbehaviors.
Thisstudydemonstrateshowperceivedfinancialknowledgeinfluencesamultitudeoffinancial
behaviors, ranging from saving and investment decisions to consumer credit and insurance
decisions.Asanindividualsconfidenceinhisorherknowledgeincreases,itappearsthatthe
individual is more likely to take action. Furthermore, by examining the interaction of and
relationshipbetweenactualfinancialknowledgeandperceivedfinancialknowledge,thisstudy
expands on the literature related to overconfidence, or the tendency to overestimate ones
accuracyandtounderestimaterisk.Individualswithlowlevelsofactualfinancialknowledge
but high levels of perceived financial knowledge are more likely to engage in risky financial
behaviors,suchastakingoutatitleloanorusingarenttoownfacility.
The structure of the paper is organized as follows. Section 2 provides an overview of the
literature, touching ontraditionalfinancialandeconomic theory, financialliteracy,behavioral
finance,andperceivedknowledge.Section3providesthehypotheses,aswellasanoverviewof
thedata,measures,andmethodology.Section4presentstheresultsandSection5concludes.
2.LiteratureReview
2.1FinancialLiteracy
Mandell (2008) describes financial literacy as the ability of consumers to make financial
decisions in their own best short and longterm interests (p. 257). At its most basic level,
financial literacy relates to a persons competency for managing money and is typically
measuredattheindividuallevelandthenaggregatedbygroups(Remund,2010,p.279).Due

to the changing economic environment, financial literacy initiatives have received much
attention. The OECD (2005) emphasizes the importance of financial education in light of
financialmarketdevelopmentsanddemographic,economicandpolicychanges(p.27).

Becauseofthepervasivenessoffinancialdecisionmaking,itisimportantthatindividualshave
the knowledge and empowerment needed to make sound, financial decisions. However,
knowledgemaynotbeenough.Insteadoffollowinghisrationalallocationstrategytomaximize
returns,NobelPrizewinnerHarryMarkowitzinvestedhispersonalfundstominimizehisfuture
regret.Markowitzisquotedassaying(fromZweig,2007,p.4):

Ishouldhavecomputedthehistoricalcovariancesoftheassetclassesanddrawn
anefficientfrontier.Instead,Ivisualizedmygriefifthestockmarketwentwayup
andIwasntinitorifitwentwaydownandIwascompletelyinit.Myintention
was to minimize my future regret. So I split my contributions 50/50 between
bondsandequity.

Ifthemostknowledgeablearemakingmistakes,whatotherfactorsareinfluencingbehavior?

2.2Behavioralfinanceandotherpsychologicalfactors
Neoclassical economic theory assumes that individuals are rational, calculating agents.
However,researchshowsthatindividualsarenotpurelyrationalbeingsandmakesuboptimal
choices.Forexample,anindividualspreferencesarenotstable,butrather,dependonthetime
frameunderconsiderationandtheframeofreferenceinvolved(ReadandvanLeeuwen,1998;
Berns et al, 2007; Laibson, 2007; Loewenstein, 1996; Loewenstein et al, 2003; Tversky and
Kahneman,1981).Moreover,additionalinformationdoesnotalwaysassistthedecisionmaking
processsincetoomuchinformationmaybeoverwhelminganddemotivating(Malhotraand
Bazerman, 2005; Cain, Loewenstein and Moore, 2005; Iyengar and Lepper, 2000; Schwartz,
2004; Mottola and Utkus, 2003). Individuals also may incorporate irrelevant information into
thedecisionmakingprocess,suchasrelyingonpriceanchorsortheirownpersonalexperience,
and are likely to be overconfident in their abilities (Choi et al, 2009; Grinblatt and Keloharju,
2009; Glaser and Weber, 2007; Barber and Odean, 2002; Allen and Evans, 2005). As the
psychology literature illuminates, individuals are not fully rational and for this reason it is
critical to understand how psychological factors influence or bias an individuals decision

making. This paper seeks to understand how perceived knowledge influences financial
behavior,withspecificattentiongiventoriskybehaviors.

2.3Actualversusperceivedknowledge
Researchers often emphasize what people actually know at a given time, yet understanding
perceived knowledge is also important. Park, Gardner, and Thukral (1988) emphasize that
perceived knowledge is related to cognitive functioning, including recognition (Schachter,
1983),identification(Nelson,Gerler,andNarens,1984),andproblemsolving(Metcalfe,1986).
Economicsresearchshowsthatbothanindividualsactualfinancialknowledgeandperceived
financialknowledgeinfluencecreditcardbehaviors(AllgoodandWalstad,2011).
Oftenthereisadiscrepancybetweenanindividualsactualknowledgeandanindividualsself
assessment,orperceivedknowledge.Itisinterestingtolookattheinteractionsanddifferences
between these two measures of knowledge, specifically in situations where perceived
knowledge exceeds actual knowledge. The differences between actual and perceived
knowledge are likely to be largest at an inbetween stage, when an individual is not nave
about a particular subject matter nor is the individual an expert (Park, Gardner, and Thukral,
1988).Understandinggapsinknowledgeisvaluablesincemotivationtolearnisafunctionof
bothactualknowledgeandperceivedknowledge(Bransford,1979;Glenbergetal,1982;Park,
Gardner,andThukral,1988).
Overconfidence refers to an individuals propensity to overestimate the accuracy of their
estimates, meaning that there is a positive difference between assessed confidence and
observed achievement (Campbell, Goodie, and Foster, 2004, p. 299). Put another way,
individuals who are overconfident have narrow confidence intervals and therefore tend to
overestimateprecisionandunderestimaterisk(GoelandThakor,2008).Thosewhotakemore
risk are not necessarily riskseeking, but are less aware of the risk involved due to cognitive
biases such as the illusion of control or relying on small sample sizes (Simon, Houghton, and
Aquino,2000).GoelandThakor(2008)theorizethatoverconfidentindividualstendtobecome
CEOsbecausetheyunknowinglychooseprojectswithhigherrisks.OverconfidentCEOsareapt

to make valuedestroying acquisitions, not because they are empire building but rather
becausetheybelievetheyareactingintheshareholdersbestinterests(MalmendierandTate,
2008).CEOoverconfidenceeffectsfirmvaluenonmonotonically,meaningthatsomedegreeof
overconfidence is good but too much is harmful (Goel and Takor, 2008). When it comes to
financial decisionmaking, overconfidence is associated with excessive trading (Grinblatt and
Keloharju, 2009; Glaser and Weber, 2007), trading that is hazardous to your wealth due to
the increased fees (Barber and Odean, 2000). Likely, these findings will hold for individual
financialdecisionsconsideredinthisanalysis:thatsomeoverconfidenceisgood,buttoomuch
overconfidenceisharmful.
3.Methods
3.1Hypothesesandmethodology
Researchers in psychology and behavioral finance have demonstrated that human decision
makingisnotpurelyrational.Thispaperseekstoexpandonthesefinding,attemptingtoshow
howactualfinancialknowledgeandperceivedfinancialknowledgeinfluencefinancialdecisions
andfinancialbehaviors.Specifically,thispaperseekstoassesstwohypotheses:
Hypothesis1:Bothactualandperceivedfinancialknowledgepredictfinancialbehavior.
Hypothesis2:Overconfidentindividuals(highperceivedknowledge,lowactual
knowledge)aremorelikelytoengageinriskyfinancialbehaviors.
In considering the two hypotheses, two main methodologies are employed. Crosstabulations
explore and highlight behavioral differences between knowledge groups. Logistic regression
attempts to predict financial behaviors, whereby the four types of financial knowledge
(describedinSection3.3.3)arethevariablesofinterest.
3.2Data
Dataisobtainedfromtwosources:theNationalFinancialCapabilityStudy(FINRA2009)andthe
German SAVE Study (SAVE 2009). The National Financial Capability Study, with support from
theU.S.DepartmentoftheTreasuryandthePresidentsAdvisoryCouncilonFinancialLiteracy,
aimstomeasureAmericansmoneyskills.Thestatebystatesurveycollecteddatafrom28,146

respondentsviaanonlinesurvey.TheSAVE(2009)survey,conductedbyTheMunichCenterfor
theEconomicsofAging,collectsdataonGermanhouseholdsavingsbehavior,surveying2,222
households on topics ranging from financial assets and consumption behavior to social
environments and psychological factors. Table 1 provides descriptive statistics for the two
samples.TheGermansampleisolderthantheAmericansample,withagreaterpercentageof
thesamplemarriedwithchildren.
3.3Measuringfinancialliteracy
3.3.1Actualfinancialknowledge
Three questions, regarding knowledge about risk, interest and inflation, are used to measure
basicfinancialknowledge(correctanswersareitalicized):

Do you think that the following statement is true or false? Buying a single company
stockusuallyprovidesasaferreturnthanastockmarketmutualfund.False

Supposeyouhad$100inasavingsaccountandtheinterestratewas2%peryear.After
5years,howmuchdoyouthinkyouwouldhaveintheaccountifyouleftthemoneyto
grow:morethan$102,exactly$102,lessthan$102?
Imagine that the interest rate on your savings account was 1% per year and inflation
was2%peryear.After1year,wouldyoubeabletobuymorethan,exactlythesameas,
orlessthantodaywiththemoneyinthisaccount?

ThethreefinancialknowledgequestionsdetailedabovearecommontoboththeFINRA(2009)
andSAVE(2009)surveys.PanelAofTable2documentsthatintheUnitedStatesjustoverone
halfofrespondentsunderstandtheconceptofdiversification,abouttwothirdsofrespondents
understand the concept of inflation and about threefourths of respondents understand the
conceptofcompoundinterest.ComparedtotheUnitedStates,Germansappeartohavehigher
levelsoffinancialknowledgeonaverage,whereapproximately65%ofrespondentsunderstand
theconceptofdiversificationandover80%ofrespondentsunderstandtheconceptsofinterest
andinflation(seeTable2PanelA).Consistentwithpreviousfindings(seeforexampleLusardi,
2008andMandell,2006)theresultsinbothsamplespointtodifferencesinknowledgeacross
gender whereby males tend to have higher levels of financial literacy than females.
Additionally,olderandmarriedindividualstendtohavehigherlevelsofknowledge.Sincethe

Germansampleconsistsofagreaterpercentageofolderandmarriedindividuals,itispossible
that the higher financial literacy level of this sample is attributed to these characteristics.
However, Panel A of Table 2 shows that the Germans have higher levels of knowledge no
matterhowthesampleissliced.

Panel B in Table 2 shows that in the United States approximately 39% of the respondents
answered all three financial literacy questions correctly with males faring best. Fortynine
percentofmalesansweredallthreequestionscorrectlycomparedto29%offemales.PanelB
also shows that 12% of the respondents did not answer any of the three financial literacy
questions correctly and 20% of the respondents answered only one of the three financial
literacy questions correctly. In Germany, about 57% of respondents answered all three
questionscorrectlywheremaleshavehigherlevelsoffinancialknowledgethanfemales.More
than 80% of the German respondents answered at least two financial literacy questions
correctly.
3.3.2Perceivedfinancialknowledge
IntheFINRA(2009)surveyintheUS,participantsratedtheirownfinancialknowledgeona7
pointLikertscalewherebya1reflectslowselfassessedlevelsoffinancialknowledgeanda
7reflectshighselfassessedlevelsoffinancialknowledge.Thequestion,aspresentedinthe
survey, reads, On a sale from 1 to 7, where 1 means very low and 7 means very high, how
wouldyouassessyouroverallfinancialknowledge?Table3showsthat,onaverage,individuals
rated their financial knowledge as 4.95 on a 7point scale. Only 11.73% of individuals self
assessed their knowledge level as below average, while 18.98% of individuals rated their
knowledgeasaverageand69.3%ratedtheirknowledgeasaboveaverage.
Similarly,theGermanSAVE(2009)surveyasksparticipantstojudgetheirpersonalknowledge
aboutseveraltypesofknowledge,includingfinancialknowledge.1Answersrangefrom1(very
low)to7(veryhigh).Onaverage,individualsratedtheirfinancialknowledgeasa4.62ona7

The survey is only available in the German language. Google translate was used to manually translate the
completesurveyfromGermantoEnglish.

point scale, slightly lower than the American sample. Additionally, Table 3 illustrates that
19.09% of individuals selfassessed their financial knowledge as below average, while 23.18%
and57.75%ofindividualsratedtheirknowledgeasaverageandaboveaverage,respectively.It
is interesting to note that while Germans appear to have higher average levels of financial
knowledge, as measured by the three financial literacy questions, Americans appear to have
higherlevelsofperceivedknowledge,pointingtooverconfidence.
3.3.3Actualandperceivedfinancialknowledge
Allgood and Walstad (2011) develop a measure that accounts for both an individuals actual
financialknowledgeandperceivedfinancialknowledge,arguingthatthecombinationofactual
financialknowledgeasmeasuredbycorrectresponsestotestquestionsandperceivedfinancial
knowledgeasmeasuredbyrespondentsselfassessmentsprovidesmorerobustandnuanced
insights about how financial knowledge affects financial behavior (p. 3). Following the
methodologyofAllgoodandWalstad(2011),acompositemeasureiscreated.First,highand
lowgroupsareestablishedforbothactualknowledgeandperceivedknowledge,wherethose
individualswithaboveaverageknowledgearecategorizedashighandthoseindividualswith
belowaverageknowledgearecategorizedaslow.2Thenfouradditionalvariablesarecreated
to represent the four types of combined knowledge: 1) high actual financial knowledge and
high perceived financial knowledge, 2) high actual financial knowledge and low perceived
financialknowledge,3)lowactualfinancialknowledgeandhighperceivedfinancialknowledge,
and4)lowactualfinancialknowledgeandlowperceivedfinancialknowledge.
Table3showsthepercentageofeachsamplethatfallswithineachofthesefourcategories.In
the US, 41% of individuals have low actual financial knowledge and low perceived financial
knowledge while only 18% of individuals have high actual financial knowledge and high

In measuring actual financial knowledge, Allgood and Walstad (2011) use two additional financial literacy
questions, for a total of five questions. The two additional questions are (correct answers are italicized): 1) If
interestratesrise,whatwilltypicallyhappentobondprices?(a)theywillrise;(b)theywillfall;(c)theywillremain
the same; (d) there is no relationship between bond prices and the interest rate, and 2) A 15year mortgage
typicallyrequireshighermonthlypaymentsthana30yearmortgage,butthetotalinterestpaidoverthelifeofthe
loanwillbeless.(a)true;(b)false.Thispaperfollowstheirmethodology,usingallfivequestionsasameasureof
actualfinancialknowledgefortheFINRA(2009)dataandusingthethreeoriginalquestionsasameasureofactual
financialknowledgefortheSAVE(2009)data.

perceived financial knowledge. In Germany, however, a greater proportion of the sample


population has high actual financial knowledge and high perceived financial knowledge (36%)
thanhaslowactualfinancialknowledgeandlowperceivedfinancialknowledge(22%).Incases
whereactualandperceivedfinancialknowledgearenotaccuratelycalibrated,individualstend
to overestimate their financial knowledge: in both samples, there are more instances where
individualswithlowactualfinancialknowledgeperceivetheirknowledgeashighthanthereare
instances where individuals with high actual financial knowledge perceive their knowledge as
low.Thissuggestsoverconfidence,althoughtheevidenceisstrongerintheUSsample.
Crosstabulationsofactualfinancialknowledgewithperceivedfinancialknowledgeshowthat
the two types of knowledge are related. Pearsons ChiSquare (2) Test (crosstabulation)
examines the relationship between two categorical variables by comparing observed
frequencies to chance frequencies. If the two categorical variables are independent, the
observedandexpectedfrequenciesshouldbesimilar;however,ifthetwocategoricalvariables
arerelated,atleastsomeoftheobservedfrequenciesshouldbedifferentfromtheexpected
frequencies.Table4illustratesthatresultsaresignificantfortheUS,2(1)=599.17,p<.001,
and for Germany, 2 (1) = 42.144, p < .001. Thus, there is an association between perceived
financial knowledge and actual financial knowledge (i.e., the variables are related). For
example,thosewithlowactualfinancialknowledgearemorelikelythanchancetoselfassess
their financial knowledge as low and are less likely than chance to selfassess their financial
knowledgeashigh.

Table5suggeststhatgeneralintelligence,asproxiedbythethreelogicquestions,isassociated
withfinancialliteracy.TheGermanSAVE(2009)posesthreebrainteasers,orlogicquestions,as
follows(correctanswersareitalicized):

A bat and a ball together cost 110 cents. The bat costs 100 cents more than the ball.
Howmuchdoestheballcost?5cents
Waterliliesgrowonapond.Everydaythewaterliliesdoubletheareathattheycover.It
takes48daysforthepondtobecompletelycoveredwithwaterlilies.Howlongdoesit
takeforhalfofthepondtobecoveredwithwaterlilies?47days
10

Five machines need five minutes to manufacture five products. How long do 100
machinesneedtomanufacture100products?5minutes

For each of the three logic questions, those individuals who correctly answered the question
have higher average financial literacy levels than those who answered incorrectly. Perhaps
then, individuals rely on their level of general intelligence, which they likely assess over the
years, to gage their knowledge in certain areas (e.g., financial knowledge). However, Table 3
pointstoaninterestingpattern:whenactualandperceivedfinancialknowledgedonotalign,
individuals tend to overestimate their abilities. What affect does this have on their financial
decisionsandfinancialbehaviors?

3.4Measuringfinancialbehaviors
The FINRA (2009) surveys financial topics including, but not limited to: banking and
investments, retirement and pension savings, home ownership and mortgages, credit cards,
consumer loans, and insurance coverage. Appendix 1 details the survey topics and questions
usedinthisanalysis,wherebyeachofquestionsareposedtoallowforayesornoresponse.It
is therefore possible, in most cases, to discern whether an individual is engaging in a good
financial behavior such as making regular contributions for retirement or a bad financial
behaviorsuchasonlymakingtheminimumpaymentonacreditcard.TheSAVE(2009)survey
asks individuals who they talk to about financial matters. Multiple answers are possible,
including:relatives,friends,colleagues,neighbors,andbanks.
4.Results
Table6presentsthepercentageofthesamplethatengagesineachtypeoffinancialbehavior,
showingthebehaviorofthetotalsample(column2)aswellasthebehaviorofeachofthefour
financialknowledgegroups(columns36).Thebehaviorsofthefourknowledgegroupscanbe
compared with the average behavior as well as the behavior of each of the other knowledge
groups.

11

4.1 High perceived, high actual. Not surprisingly, individuals with both high actual financial
knowledge and high perceived financial knowledge engage in good financial behaviors and
havethebestfinancialbehaviorscomparedtootherknowledgegroups.Forexample,Table6
showsthat36.88%ofallindividualshaveanemergencyfundand39.02%ofallindividualshave
calculatedhowmuchtheyneedtosaveforretirement.However,ofthoseindividualswithboth
high actual and high perceived financial knowledge, 60.46% have an emergency fund and
65.62% have calculated how much they need to save for retirement, reflecting better than
averagefinancialbehaviors.Thehighperceivedandhighactualknowledgegroupisalsolessapt
toengageinbadfinancialbehaviorssuchasobtainingapaydayloan(4.96%)orusingapawn
shop(5.33%)comparedtothebehaviorsofallgroups(9.44%and12.15%,respectively).
4.2 Low perceived, low actual. Individuals with low perceived financial knowledge and low
actualfinancialknowledgearelesslikelytoengageingoodfinancialbehaviorsandmorelikely
toengageinbadfinancialbehaviors(seeTable6).Forinstance,comparedtothebehaviorsof
allindividuals,asmallerpercentageofthelowlowknowledgegrouphaspaidcreditcardbillin
full (31.94% v. 41.83%), established an emergency fund (22.96% v. 36.88%), or compared
mortgage lenders (55.66% v. 63.64%). However, a greater percentage of the lowlow
knowledgegrouphascarriedabalanceontheircreditcard(64.56%v.57.52%)orusedapawn
shop (16.06% v. 12.15%). Compared to the highhigh knowledge group, the lowlow group is
lesslikelytodiscussfinancialmatterswithbanks(26.20%v.35.15%).
4.3 High perceived, low actual. Interestingly, holding actual financial knowledge constant,
perceived financial knowledge appears to influence financial behaviors. Allgood and Walstad
(2011) noted that, holding actual knowledge constant at the low level, individuals with high
perceivedknowledgearemorelikelytoengageingoodcreditcardbehaviorsandlesslikelyto
engageinbadcreditcardbehaviorsthanthoseindividualswithlowperceivedknowledge.The
creditcardbehaviorssectionofTable6replicatesAllgoodandWalstadsfindings.Table6also
showsthatAllgoodandWalstadsfindingsholdacrosscertaintypesofotherfinancialbehaviors
aswell.Holdingactualknowledgeconstantatthelowlevel,anindividualwithhighperceived
knowledgeismorelikelythananindividualwithlowperceivedknowledgetoengageingood

12

behaviors, such as establishing an emergency fund or comparing mortgage and auto lenders,
andislesslikelytoengageinbadbehaviors,suchascarryingabalanceonacreditcard.
4.3.1 Overconfidence. However, Table 6 also shows that overconfident individuals have an
increased propensity to engage in risky financial behaviors. For each of the six behaviors
classifiedasrisky,agreaterpercentageofindividualswithbothlowactualandlowperceived
levels of financial knowledge engage in such behaviors than those individuals with both high
actualandhighperceivedlevelsoffinancialknowledge,e.g.,4.96%ofthehighhighknowledge
grouptookoutapaydayloanwhile11.84%ofthelowlowknowledgegrouptookoutapayday
loan.Interestingly,whenholdingactualknowledgeconstantatlowlevels,moreindividualswith
highperceivedknowledgeengageinriskybehaviorsthanthoseindividualswithlowperceived
knowledge.Forexample,8.52%ofthelowperceivedandlowactualknowledgegroupuseda
renttoownstorewhile9.47%ofthehighperceivedandlowactualknowledgeusedarentto
own store. Or, 6.95% of the lowlow knowledge group has taken out an auto title loan while
8.24%ofthehighperceivedandlowactualknowledgegrouphastakenoutanautotitleloan.In
fouroutofthesixriskybehaviorsconsidered,agreaterpercentageofoverconfidentindividuals
engaged in the risky behaviors than those with low perceived and low actual knowledge.
Overconfident in their abilities, the high perceived and low actual knowledge group
underestimatesrisk.
4.4 Low perceived, high actual. Holding actual financial knowledge constant at a high level,
individualswithlowperceivedknowledgearelesslikelytoengageingoodfinancialbehaviors
thanthoseindividualswithhighperceivedfinancialknowledge.Forinstance,comparedtothe
behaviors of the highhigh knowledge group, a smaller percentage of the low perceived
knowledgeandhighactualknowledgegrouphaspaidcreditcardbillinfull(40.21%v.55.72%),
establishedanemergencyfund(37.83%v.60.46%)orcomparedmortgagelenders(65.05%v.
71.37%);however,agreaterpercentageofthelowperceivedandhighactualknowledgegroup
has carried a balance on their credit card (60.60% v. 45.87%) or used a pawn shop (8.65% v.
5.33%).Suchdifferencesinbehavior,whileholdingactualknowledgeconstantatahighlevel,
highlights the important role perceived knowledge plays in the decisionmaking processes. In

13

manyinstances,individualswithlowactualknowledgebuthighperceivedknowledgeengagein
betterfinancialbehaviorthanthoseindividualswithhighactualknowledgebutlowperceived
knowledge: paid credit card bill in full (47.06% v. 40.21%), established an emergency fund
(44.90% v. 37.83%) or compared mortgage lenders (66.08% v. 65.05%). These differences
suggestthattheroleofperceivedknowledgemaybeasimportantastheroleofactualfinancial
knowledge.
4.5Perceivedfinancialknowledge.AsTable6illustrates,perceivedknowledgeappearstobe
an important factor in financial decisionmaking. As an individuals confidence in his or her
knowledge increases, it appears that the individual is more likely to take action. Social
psychology research shows that positiveillusions are overall beneficial, specifically for
motivation and achievement. Beliefs about ones self are motivated toward accuracy,
consistency, and enhancement, with selfenhancement being the strongest goal (Sedikides,
1993).Peoplehaveunrealisticallyhighselfevaluations,perceptionsofcontroloverevents,and
optimism for future events (Taylor and Brown, 1988). Therefore, as suggested earlier, having
positiveillusionsorbeingoverconfidentcanbegood,uptoapoint.
Table 6 also suggests that perceived financial knowledge may be related to with whom an
individualhasfinancialdiscussions.Comparedtothelowlowknowledgegroup,thosewithhigh
perceived knowledge and low actual knowledge are more likely to consult banks (28.54% v.
26.20%)andlesslikelytoconsultfamily(23.54%v.39.77%)andfriends(16.25%and17.26%).
Similarly, compared to the low perceived and high actual knowledge group, the highhigh
knowledgegroupismorelikelytoconsultbanks(42.96%v.37.77%)andlesslikelytoconsult
family(28.14%v.35.15%)andfriends(22.67%v.29.48%).
4.6Logisticregressions.WhileTable6illuminatessomeinterestingpatterns,theanalysisdoes
notcontrolforotherfactorsthatmayinfluencebehavior.Forexample,healthinsurancemay
beareasonableinvestmentforanyone,butlifeinsuranceneedsmaydifferdependingonthe
ageofanindividualandtheirlevelofobligations,e.g.,asinglefemaleinherearlytwentiesmay
not need life insurance whereas it may be a wise investment for a male who is the sole
breadwinner for a family of five. The logistic regressions in Tables 7 through 12 control for

14

these other factors that may influence financial decisions. The dependent variable in each
regressionisabinaryvariableandthemodelsattempttopredictwhetherornotanindividual
engagesinaspecifictypeoffinancialbehavior.
Inadditiontothefourtypesoffinancialknowledgegroups,severalotherdemographicfactors
are considered (all of which are dummy variables), including: gender (female=1); age (1824,
2434, 3544, 4554, 5564, 65+); race (nonwhite=1); education (<high school, high school
graduate only, some college, college graduate only, postgraduate); marital status (single,
married, divorced or separated, widowed or widower); dependent children (no children or
dependentchildren,onechild,twochildren,threeormorechildren);andannualincome(less
than $15K; $1525; $2535; $3550; $5075; $75100; $100150; $150K or more). The use of
thesedemographiccontrolsisestablishedintheliterature(AllgoodandWalstad,2011;Lusardi,
Mitchell, and Curto, 2010; Lusardi and Mitchell, 2011; van Rooij, Lusardi, and Alessie, 2011;
Fornero and Monticone, 2011; BucherKoenen and Lusardi, 2011). The omitted variables are:
the low perceived knowledge and low actual knowledge group, 1824 age group, college
graduate,married,nodependentchildren,andannualincomeofatleast$50,000butlessthan
$75,000.Allvariablesareinterpretedinreferencetothesegroups.
AllgoodandWalstad(2011)summarizethreereasonswhyreversecausalityisnotanissue,i.e.,
discuss why knowledge influences behaviors, not the other way around. First, other studies
using different methodologies have not found reverse causality to be an issue (Courchane,
Gailey,andZorn,20088;vanRooij,Lusardi,andAlessie,2011).Second,reversecausalitywould
suggestthatthoseindividualsthathaveexperiencedpoorfinancialoutcomeswouldthenself
assess their knowledge as low. However, Allgood and Walstad (2011) do not find consistent
statisticalrelationshipsacrossmultiplefinancialoutcomes,indicatingthatfinancialoutcomes
would not be a clear predictor of financial knowledge (p. 15). Third, while some financial
behaviors can broadly be classified as good or bad, it is not possible to make normative
judgmentsaboutfinancialbehaviorswithoutconsideringindividualcircumstances.
Inlogisticregressions,thecoefficientsaredifficulttointerpret.Forthisreason,theoddsratio,
amoreusefulmeasureofeffectsize,isreported.Themeasureistheratioofthelikelihoodof

15

aneventoccurringinonegrouptothelikelihoodofthesameeventoccurringinanothergroup.
Forexample,ifanoddsratioof1.345isreportedforthefemaledummyvariable,thismeans
that females are 1.345 times more likely to engage in that specific financial behavior than
males.Ifacoefficientisnotsignificantatthe5%level,theoddsratioisreportedasNS(not
significant). For example, if the coefficient for the dummy variable for female is not
statisticallysignificantinaregression,thentheoddsratioisreportedasNS,meaningthatthere
nodiscernabledifferenceinthefinancialbehaviorsofmalesandfemales.
4.6.1Creditcardbehaviors.3TheresultsinTable7confirmthefindingsofAllgoodandWalstad
(2011)whosummarizethatasidefromthemarginaleffectofincomedrop(andperhapsage),
the variables for perceived and actual knowledge have the largest marginal effect on the
probability that a person engages in each behavior (p. 15). Regression 6, which predicts
whetheranindividualusedacreditcardtoobtainacashadvance,indicatesthatanindividual
with high perceived knowledge and low actual knowledge is 1.264 times more likely to get a
cashadvancethananindividualwithlowperceivedknowledgeandlowactualknowledge(the
omitted variable). Because the actual financial knowledge of the two individuals is the same,
the results point to overconfidence: it is the perceived knowledge that is driving the risky
behavior.
4.6.2 Retirement & savings behaviors. Perceived and actual financial knowledge are both
important predictors of retirement and savings behaviors. Moreover, in each of the seven
regressions in Table 8, perceived knowledge appears to have a greater influence on behavior
thanactualknowledge.Sincetheomittedvariableisthelowlowknowledgegroup,interpreting
the odds ratio of the high perceived and low actual knowledge group reflects a difference in
perceived knowledge and interpreting the odds ratio of the perceived low and high actual
knowledgegroupreflectsadifferenceinactualknowledge.Forexample,anindividualwithhigh
perceived knowledge and low actual knowledge is 2.453 times more likely than an individual
withlowperceivedknowledgeandlowactualknowledgetoestablishanemergencyfund,while

Allgood and Walstad (2011) use probit analysis to examine the same behaviors. See their paper for a more
detaileddescriptionandinterpretationoftheseresults.

16

an individual with low perceived knowledge and high actual knowledge is 1.318 times more
likely than an individual with low perceived and low actual knowledge to establish an
emergency fund (see Table 8). These findings, which are similar across all seven regressions,
showthatfinancialknowledgeisnotenough.Anindividuallikelyneedstheconfidence(i.e.,a
good selfperception) to apply whatever knowledge he or she possesses to the financial
decisions.ThesefindingshighlightHustons(2010)twoprongeddefinitionoffinancialliteracy
as understanding (i.e., actual knowledge) and use (i.e., perceived knowledge), and support
Remunds (2010) definition of financial literacy as a measure of the degree to which one
understands key financial concepts and possesses the ability and confidence to manage
personalfinances(p.284).
Table8illustratesthatinadditiontoperceivedandactualfinancialknowledge,ageandincome
appear to be important predictors of retirement and savings behaviors. Not surprisingly,
comparedtoanindividualbetweentheagesof18and24,anindividualwhois65yearsorolder
is 2.273 more likely to have established an emergency fund, 3.059 times more likely to have
calculated how much he or she needs to save for retirement, and 2.602 times more likely to
have investments (not including a retirement account). Compared to a household with an
annual income between $50,000 and $75,000, a household with an income greater than
$150,000is4.385timesmorelikelytohaveinvestments(notincludingaretirementaccount)
and 3.746 times more likely to have saved for a childs college education. Compared to a
householdwithanannualincomebetween$35,000and$50,000,ahouseholdwithanincome
between$50,000and$75,000is1.429timesmorelikelytohaveinvestmentsand1.502times
morelikelytohavesavedforachildscollegeeducation.4
4.6.3Credit&loanbehaviors.Table9depictsthesamepatternsasTable8:1)perceivedand
actual knowledge are both significant predictors of financial behavior and 2) perceived
knowledgehasalargeinfluenceonfinancialbehavior,evenlargerthantheinfluenceofactual
knowledge. While the two types of financial knowledge are significant factors in each
regression,someotherinterestingrelationshipsemerge:thosewithhigherannualincomesare

UsingtheoddsratiosfromTable7,calculatedas:1.429=(1/.700);and1.502=(1/.666)

17

morelikelytocomparemortgages;thosewithlowerannualincomesarelesslikelytocalculate
their mortgage as a percentage of income; having children and/or experiencing a drop in
income increases the likelihood of making a mortgage payment late; individuals aged 2534
withchildrenaremostapttoobtainanautoloan;financialknowledgeandhigherincomesare
associatedwithcomparingautoloansandobtainingcreditreportsandscores.
4.6.4Riskybehaviors.Becauseofanunderestimationofrisk,overconfidentindividuals,i.e.,the
high perceived and low actual knowledge group, have an increased propensity to engage in
riskybehaviors.Aftercontrollingformanyotherfactors,Table10showsthatindividualswith
highperceivedknowledgeandlowactualknowledgearemorelikely(inmostcases)thanother
knowledgegroupstotakethesefinancialrisks,echoingthefindingsofTable6.Keepingactual
knowledgeconstantatalowlevel,anindividualwithhighperceivedknowledgeis1.341times
morelikelytodeclarebankruptcy,1.151timesmorelikelytotakeoutatitleloan,1.309times
more likely to receive a tax advance, and 1.316 times more likely to use a renttoown store
thananindividualwithlowperceivedknowledge.Lowerlevelsofeducationandhavingchildren
alsoappeartobehighlyassociatedwiththehighriskloans.However,duetothelowincidences
oftheseriskybehaviors(seeTable6),models1through6inTable10doapoorjobofcorrectly
classifyingindividualsasengaginginsuchbehaviors.TheSensitivityrowatthebottomofthe
table shows that the models correctly classify only up to 1.4% of individuals engaging in the
risky behaviors. The models do a good job of classifying those who do not behave in such
manners (see the Specificity row). Taken together, these models correctly classify between
88% and 98% of individuals (see the % correctly classified row), yet this classification is no
better (or in one case worse) than classifying all individuals as not engaging in the risky
behaviors(seethe%originalrow).
4.6.5 Insurance behaviors. Table 11 once more shows the importance of both perceived and
actualfinancialknowledgeonfinancialbehavior.Forexample,anindividualwithhighperceived
knowledge and low actual knowledge is 1.351 times more likely than an individual with low
perceived knowledge and low actual knowledge to have homeowners or renters insurance,
while an individual with low perceived knowledge and high actual knowledge is 1.381 times

18

more likely than an individual with low perceived and low actual knowledge to have
homeowners or renters insurance (see Table 8). Higher ages and incomes also appear to be
associatedwithinsurancebehaviors.Infact,individuals65yearsorolderare8.985timesmore
likely than an 1824 year old to have health insurance. Unlike the Risky Behavior models in
Table 10, the factors in Table 11 improve the models classification. In models 2 and 3, the
classification percentage increases over 10%, with over 50% accuracy for predicting if an
individualhashealthinsuranceorlifeinsuranceandpredictingthatanindividualdoesnothave
healthinsuranceoflifeinsurance.
4.6.6Financialdiscussions.Table6suggeststhathigherperceivedfinancialknowledgemaybe
associated with consulting banks about financial information rather than family and friends.
However,thelogisticregressionresultsinTable12donotindicatethattherearedifferencesin
financialdiscussionbehaviorbetweenthehighperceivedknowledgeandlowactualknowledge
group with the low perceived knowledge and low actual knowledge group, suggesting that
perceivedknowledgeisnotamajorfactorinfluencingthesebehaviors.Comparedtothelow
lowknowledgegroup,individualswithlowperceivedandhighactualknowledgearemorelikely
to consult relatives (1.396), friends (1.822), colleagues (2.560), and banks (1.523). Perhaps
thosewithhighperceivedknowledgeareconfidentintheirknowledgeanddonotfeeltheneed
toconsultothers,andtheindividualswithlowperceivedknowledgebuthighactualknowledge
consult others because they are unsure of their knowledge and are not confident in their
abilities.
5.Conclusion
Becauseoftheimportanceoffinancialdecisionmakinganditsprevalenceineverydaylife,itis
critical to understand the relationships between financial literacy, behavioral biases and
financial behavior. This analysis emphasizes the importance of two types of financial
knowledge, perceived financial knowledge and actual financial knowledge. Not surprisingly,
individuals with both high actual and perceived knowledge are more likely to make good
financialdecisionsthanindividualswithbothlowactualandperceivedknowledge.Somewhat
surprising,however,ishowinfluentialperceivedknowledgeisonfinancialbehavior.Inmany

19

cases, perceived knowledge has a greater affect on behavior than actual knowledge.
Additionally, when perceived knowledge is high and actual knowledge is low, individuals are
morelikelytotakefinancialrisks.

Understanding that perceived knowledge is just as important as actual knowledge is of


paramount value to educators and policy makers. Individuals benefit not only from actual
financial knowledge, but from perceived financial knowledge as well. If individuals are more
confident in their knowledge and abilities, they are more apt to apply their knowledge and
abilitiestofollowthroughonafinancialdecision.Knowingthatindividualsdeviatefromrational
decisionmaking in certain systematic ways, economic policies have been developed that
attempttoexploitthebehavioralbiasestoinducepositiveoutcomes.SaveMoreTomorrow
(Bernartzi and Thaler, 2007) is a policy designed to increase participation in employment
sponsored retirement plans by creating the program to play off of cognitive and behavioral
biases.Similarly,ThalerandSustein(2008),undertheassumptionthathumansarenotalways
rational, propose policies of libertarian paternalism, arguing that default options assist
complexdecisionmaking,withoutlimitingchoice.Perhapsfinancialliteracyinitiativescanfind
ways to address selfperceptions and selfconfidence. This study shows that with a little
additional motivation and inspiration, individuals are more likely to make good financial
choices. But with too much confidence and too little actual knowledge, individuals are apt to
make risky financial choices. The struggle then is to help individuals find a healthy dose of
confidence.

20

References

Allen, W. D., & Evans, D. A. (2005). Bidding and overconfidence in experimental financial
markets.JournalofBehavioralFinance,6,108120.
Allgood,S.,&Walstad,W.B.(2011).Theeffectsofperceivedandactualfinancialknowledgeon
credit card behavior. Networks Financial Institute at Indiana State University. Working
PaperJuly2011.
Barber, B. M., & Odean, T. (2000). Trading is hazardous to your wealth: The common stock
investmentperformanceofindividualinvestors.JournalofFinance,55,773806.
Barber,B.M.,&Odean,T.(2002).Onlineinvestors:Dotheslowdiefirst?ReviewofFinancial
Studies,15,455487.
Bernartzi,S.,&Thaler,R.(2007).Heuristicsandbiasesinretirementsavingsbehavior.Journal
ofEconomicPerspectives,21,81104.
Berns, G. S., Laibson, D., & Loewenstein, G. (2007). Intertemporal choice toward an
integrativeframework.TrendsinCognitiveSciences,11,482488.
Bransford,J.D.(1979).Humancognition:Learning,understandingandremembering.Belmont,
CA:Wadsworth.
BucherKoenen,T.,&Lusardi,A.(2011).FinancialliteracyandretirementplanninginGermany.
JournalofPensionEconomicsandFinance,10,565584.
Cain,D.M.,Loewenstein,G.,&Moore,D.A.(2005).Thedirtoncomingclean:Perverseeffects
ofconflictsofinterest.JournalofLegalStudies,34,125.
Campbell,W.K.,Goodie,A.S.,&Foster,J.D.(2004).Narcissism,confidence,andriskattitude.
JournalofBehavioralDecisionMaking,17,297311.
Chabris,C.,&Simons,D.(2010).Theinvisiblegorilla.CrownPublishers,NY:NY.
Choi, J. J., Laibson, D., & Madrian, B. C. (2005). Are empowerment and education enough?
Underdiversificationin401(k)plans.BrookingsPapersonEconomicActivity,2005,151
198.
Choi,J.J.,Laibson,D.,Madrian,B.C.,&Metrick,A.(2009).Reinforcementlearningandsavings
behavior.JournalofFinance,64,25152534.
Courchane,M.,Gailey,A.,&Zorn,P.(2008).Consumercreditliteracy:Whatpriceperception?
JournalofEconomicsandBusiness,60,125138.
Fornero, E., & Monticone, C. (2011). Financial literacy and pension plan participation in Italy.
JournalofPensionEconomicsandFinance,10,547564.
Glaser, M., & Weber, M. (2007). Overconfidence and trading volume. The Geneva Risk and
InsuranceReview,32,136.
Glenberg,A.M.,Wilkinson,A.C.,&Epstein,W.(1982).Theillusionofknowing:Failureinthe
selfassessmentofcomprehension.Memory&Cognition,10,597602.
Goel,A.M.,&Thakor,A.V.(2008).Overconfidence,CEOselection,andcorporategovernance.
JournalofFinance,63,27372784.
Grinblatt,M.,&Keloharju,M.(2009).Sensationseeking,overconfidence,andtradingactivity.
JournalofFinance,64,549578.
Hilgert, M A., Hogarth, J. M., & Beverly, S. (2003). Household financial management: The
connectionbetweenknowledgeandbehavior.FederalReserveBulletin,89,309322.

21

Huston,S.J.(2010).Measuringfinancialliteracy.JournalofConsumerAffairs,44,296316.
Iyengar, S. S., & Lepper, M. R. (2000). Choice is demotivating: Can one desire too much of a
goodthing?JournalofPersonalityandSocialPsychology,79,9951006.
Kozup, J., & Hogarth, J. M. (2008). Financial literacy, public policy, and consumers self
protectionsmorequestions,feweranswers.JournalofConsumerAffairs,42,127136.
Laibson, D. (2007). The Psychology of Saving and Investment: Sticky Biases and the Curse of
Education. The London School of Economics and Political Science Public Lectures and
Events.21November2007.
Loewenstein,G.(1996).Outofcontrol:Visceralinfluencesonbehavior.Organizationalbehavior
andhumandecisionprocesses,65,272292.
Loewenstein,G.,ODonoghue,T.,&Rabin,M.(2003).Projectionbiasinpredictingfutureutility.
QuarterlyJournalofEconomics,118,12091248.
Lusardi,A.(2004).Savingandtheeffectivenessoffinancialeducation.InMitchell,O,&Utkus,S.
(Ed.) Pension Design and Structure: New Lessons from Behavioral Finance (157184).
OxfordUniversityPress.
Lusardi,A.(2008).Financialliteracy:Anessentialtoolforinformedconsumerchoice?Available
athttp://www.dartmouth.edu/~alusardi/pub.html.
Lusardi,A.,&Mitchell,O.(2011).Financialliteracyaroundtheworld:Anoverview.Journalof
PensionEconomicsandFinance,10,497508.
Lusardi, A., Mitchell, O. & Curto, V. (2010). Financial Literacy Among the Young. Journal of
ConsumerAffairs,44,358380.
Malhotra, D., & Bazerman, M. (2005). Economics wins, psychology loses, and society pays.
Negotiation,OrganizationsandMarketsResearchPapers.HarvardNOMResearchPaper
No.0507.SSRN.
Malmendier, U., & Tate, G. (2008). Who makes acquisitions? CEO overconfidence and the
marketsreaction.JournalofFinancialEconomics,89,2043.
Mandell,L.(2008).Financialeducationinhighschool.InLusardi,A.(Ed.)Overcomingthesaving
slump(pp.257279).TheUniversityofChicagoPress,Chicago.
Mandell,L.(2009).Theimpactoffinancialliteracyeducationonsubsequentfinancialbehavior.
JournalofFinancialCounselingandPlanning,20,1524.
Metcalfe,J.(1986).Feelingofknowinginmemoryandproblemsolving.JournalofExperimental
Psychology,12,288294.
Mottola, G. R., & Utkus, S. P. (2003). Can there be too much choice in a retirement savings
plan?VanguardCenterforRetirementResearch:TheVanguardGroupInc.Malvern,PA.
Nelson, T. O., Gerler, D., & Narens, L. (1984). Accuracy of feelingofknowing judgments for
predictingperceptualidentificationandrelearning.JournalofExperimentalPsychology:
General,113,282300.
OECD(OrganisationforEconomicCooperationandDevelopment)(2005).Improvingfinancial
literacy.Analysisofissuesandpolicies.OECDPublishing,Paris,France.
Park,C.W.,Gardner,M.P.,&Thukral,V.K.(1988).Selfperceivedknowledge:Someeffectson
informationprocessingforachoicetask.AmericanJournalofPsychology,101,401424.
Read, D., & van Leeuwen, B. (1998). Predicting hunger: The effects of appetite and delay on
choice.OrganizationalBehaviorandHumanDecisionProcesses,76,189205.

22

Remund, D. L. (2010). Financial literacy explicated: The case for a clearer definition in an
increasinglycomplexeconomy.JournalofConsumerAffairs,44,276295.
Schachter, D. L. (1983). Feeling of knowing in episodic memory. Journal of Experimental
Psychology:Learning,Memory,andCognition,9,3954.
Schwartz,B.(2004).Theparadoxofchoice:whymoreisless.NewYork,NY:HarperCollins.
Sedikides, C. (1993). Assessment, enhancement, and verification determinants of the self
evaluationprocess.JournalofPersonalityandSocialPsychology,65,317338.
Simon,M.,Houghton,S.M.,&Aquino,K.(2000).Cognitivebiases,riskperception,andventure
formation:Howindividualsdecidetostartcompanies.JournalofBusinessVenturing,15,
113134.
Taylor,S.E.,&Brown,J.(1988).Illusionandwellbeing:Asocialpsychologicalperspectiveon
mentalhealth.PsychologicalBulletin,103,193210.
Thaler, R. & Sustein, C. (2008). Nudge: Improving decisions about health, wealth and
happiness.YaleUniversityPress.
Tversky, A., & Kahneman, D. (1981). The framing effect of decisions and the psychology of
choice.Science,211,453458.
vanRooij,M.,Lusardi,A.,&Alessie,R.(2011).Financialliteracyandstockmarketparticipation.
JournalofFinancialEconomics,101,449472.
Zweig,J.(2007).YourMoneyandYourBrain.Howthenewscienceofneuroeconomicscanhelp
makeyourich.NewYork,NY.

23

Table1CharacteristicsoftheSurveySamples(inpercentages)

UnitedStates
FINRA(2009)
Sex/Gender

Male
48.67
Female
51.33
Age*

1824
13.52
2534(Under35)
17.08
3544(3650)
18.23
4554
19.61
5564(5165)
16.31
65+
15.20
MaritalStatus

Married
53.37
Single
28.24
SeparatedorDivorced
13.97
Widowed
4.41
Children

1
15.89
2
13.31
3+
9.20
Nochildren
31.17

Germany
SAVE(2009)

47.98
52.02

11.61
28.35

30.11
29.93

60.47
15.75
14.22
9.54

21.24
33.39
25.39
19.98

*The age groupings differ between the two samples. The SAVE (2009) survey groupings are
reported in italics. The 65+ age group reflects both the FINRA (2009) and SAVE (2009) surveys.

24

Table2FinancialLiteracySummaryStatisticsbyTopic(PanelA)andbyNumberofQuestionsAnsweredCorrectly(PanelB)

UnitedStates(FINRA,2009)*
%respondentsknowledgeable
Gender/Sex
Male
Female
Age
1824
2534
3544
4554
5564
65+
Maritalstatus
Married
Single
SeparatedorDivorced
Widowed

Germany(SAVE,2009)
%respondentsknowledgeable
Gender/Sex
Male
Female
Age
Under35
3650
5165
65+
Maritalstatus
Married
Single
SeparatedorDivorced
Widowed

Diversification

53.4

62.2
45.0

39.8
47.2
54.3
55.1
60.1
62.0

58.8
45.8
49.4
50.0

64.9

71.8
58.6

70.5
71.9
67.4
53.5

68.8
66.3
61.7
42.5

PanelA
Interest

77.7

83.3
72.4

72.6
76.8
77.6
79.7
79.3
79.0

80.2
74.9
75.2
73.7

84.8

87.2
82.6

88.0
86.7
85.5
81.2

87.6
84.9
81.3
72.6

Inflation

64.5

72.6
45.0

43.7
52.2
64.0
72.0
76.4
75.0

69.2
54.2
65.5
70.5

80.1

85.5
75.2

74.0
81.0
81.5
80.3

82.6
78.6
77.2
71.2

PanelB
Numberoffinancialliteracyquestionsansweredcorrectly
0
1
2
3

0
11.7
20.0
29.4
38.9

7.7
15.5
27.8
49.0
15.5
24.2
30.9
29.4

17.3
31.2
29.8
21.8
14.1
27.1
27.2
31.6
11.1
20.1
30.5
38.3
10.0
15.9
31.5
42.6
8.4
15.1
28.8
47.7
10.3
12.4
28.1
49.1

9.6
17.3
28.3
44.7
14.5
26.0
29.5
30.0
13.0
19.3
32.3
35.4
14.3
15.3
32.2
38.1

8.4
10.1
24.8
56.8

6.6
7.0
21.8
64.6
10.1
12.9
27.5
49.5

8.9
10.4
19.8
60.8
7.5
9.2
19.7
63.7
7.0
9.1
26.3
57.5
10.5
11.7
29.9
47.8

6.4
9.2
23.5
60.9
10.0
9.7
20.9
59.4
10.4
10.4
27.5
51.6
15.6
16.0
34.9
33.5

*The FINRA statistics are weighted to match US Census distributions based on age, gender, race, ethnicity, education and census region. The SAVE statistics are unweighted and based solely on the
surveyssamplepopulation.

25

Table3TypesofFinancialKnowledge:ActualandPerceived

Actualaverage*
Lowactual
Highactual

Perceived
1verylow
2
3
4
5
6
7veryhigh

Perceivedaverage**
Lowperceived
Highperceived

PerceivedActual
LowLow
LowHigh
HighLow
HighHigh

UnitedStates

2.99(of5)
57.81
42.19

2.35
2.50
6.88
18.98
35.17
23.92
10.21

4.95
65.89
34.13

41.00
15.98
24.87
18.15

28,146
16,271
11,875

643
684
1,884
5,202
9,640
6,556
2,799

27,548
18,053
9,355

11,238
4,379
6,815
4,976

Germany

2.30(of3)
43.23
56.75

2.57
4.46
12.06
23.18
32.18
18.23
7.34

4.62
42.26
57.74

21.65
20.61
21.60
36.14

2,222
961
1,261

57
99
268
515
715
405
163

2,222
939
1,283

481
458
480
803

*See Table 2 for more details about actual knowledge. For the US (FINRA 2009) sample, an individual is classified as having high
actualfinancialknowledgeiftheyanswered4or5of5questionscorrectlyandaslowactualfinancialknowledgeiftheyanswered3
questionsorlesscorrectly.FortheGerman(SAVE2009)sample,anindividualiscategorizedashighactualfinancialknowledgeifhe
orsheansweredallthreefinancialliteracyquestionscorrectlyandaslowactualfinancialknowledgeotherwise.
**FortheUS(FINRA2009)sample,anindividualisclassifiedashavinghighperceivedfinancialknowledgeiftheyselfassessedtheir
financialknowledgeasa6or7ona7pointscaleandclassifiedashavinglowperceivedfinancialknowledgeiftheyselfassessed
theirfinancialknowledgeas5orlowerona7pointscale.FortheGerman(SAVE2009)sample,anindividualisclassifiedashaving
highperceivedfinancialknowledgeiftheyselfassessedtheirfinancialknowledgeasa5,6,or7ona7pointscaleandclassifiedas
havinglowperceivedfinancialknowledgeiftheyselfassessedtheirfinancialknowledgeas4orlowerona7pointscale.

26

Table 4 Crosstabulations of perceived financial knowledge and actual financial


knowledge

PerceivedActual
LowLow
LowHigh
HighLow HighHigh

UnitedStates

ExpectedCount
10287
7767
5330
4024
ActualCount
11238
6815
4379
4976
ChiSquare
599.17

p
<.001

Germany

ExpectedCount
406
533
555
728
ActualCount
481
458
480
803
ChiSquare
42.144

p
<.001

Pearsons ChiSquare ( ) Test examines the relationship between two categorical variables by comparing
observed frequencies (Actual Count) to chance frequencies (Expected Count). If the two categorical variables
are independent, the actual and expected frequencies should be similar; however, if the two categorical
variablesarerelated,atleastsomeoftheactualfrequenciesshouldbedifferentfromtheexpectedfrequencies.
The ChiSquare statistic and its significance report whether there is a statistically significant association
betweenthetwovariables.

27

Table5OnewayAnalysisofLiteracybyThreeBrainTeasersfortheSAVE(2009)Survey

Logicquestion*
Bat&ball
Incorrect
Correct
Lily&pond
Incorrect
Correct
Machine
Incorrect
Correct

1749
473

1099
1123

1185
1037

Mean

2.20526
2.64271

2.01365
2.57703

2.10549
2.51880

Std.Dev.

0.992913
0.713991

1.02731
0.78991

1.00749
0.84452

Lower95%

2.1587
2.5782

1.9528
2.5308

2.0481
2.4673

Upper95%

2.2518
2.7072

2.0745
2.6233

2.1629
2.5703

ThisOneWayAnalysisconsiderstherelationshipbetweenfinancialliteracyandgeneralintelligence.Themeanscorerepresents
theaveragefinancialliteracyscore(outof3).Threelogic/brainteaserquestionswereposedandforeachofthethreelogictopics,
theaveragefinancialliteracyscoresarereportedfortwogroups:thoseindividualsthatcorrectlyansweredthebrainteaserand
thoseindividualswhodidnotcorrectlyanswerthebrainteaser.
*Thelogicquestionsareasfollows:1)Bat&ball:Abatandaballtogethercost110cents.Thebatcosts100centsmorethanthe
ball.Howmuchdoestheballcost?5cents2)Lilyandpond:Waterliliesgrowonapond.Everydaythewaterliliesdoublethearea
thattheycover.Ittakes48daysforthepondtobecompletelycoveredwithwaterlilies.Howlongdoesittakeforhalfofthepond
to be covered with water lilies? 47 days 3) Machine: 5 machines need 5 minutes to manufacture 5 products. How long do 100
machinesneedtomanufacture100products?5minutes

28

Table6Typesoffinancialknowledge(perceivedactual)bytypeoffinancialbehavior(percentage)

Creditcardbehaviors*
Paidinfull
Carrybalance
Minimumpayment
Latefee
Exceedlimit
Cashadvance

Retirement&savingsbehaviors
Emergencyfund
Collegesavings
Educationaccount
Calculate
Otherinvestments
Regularcontribution
Lifecycle

Credit&loanbehaviors
Comparedmortgagelenders
Calculatemortgageas%income
Latemortgagepayment
Haveautoloan
Compareautoloans
Creditreport
Creditscore

Riskybehaviors
Declaredbankruptcy
Autotitleloan
Paydayloan
Taxadvanceonrefund
Usedpawnshop
Usedrenttoown

Insurancebehaviors
Healthinsurance
Homeownersorrenters
Lifeinsurance
Autoinsurance

Financialdiscussions
Relatives
Friends
Colleagues
Neighbors
Banks
Nofinancialtalks

Low
Low

31.94
64.56
51.51
34.30
21.03
13.96

22.96
22.44
20.31
24.31
23.06
72.19
38.59

55.66
80.92
41.71
33.49
38.54
34.82
33.79

2.65
6.95
11.84
7.74
16.06
8.52

73.47
54.86
50.82
80.70

30.77
17.26
2.50
0.62
26.20
43.04

High
Low

47.06
53.65
40.37
24.12
16.07
16.67

44.90
38.47
33.69
42.93
36.67
77.61
51.10

66.08
86.68
16.48
33.19
51.40
47.83
45.44

3.39
8.24
10.94
8.57
14.80
9.47

78.33
65.96
62.72
85.09

23.54
16.25
2.71
2.92
28.54
46.67

Low
High

40.21
60.60
37.25
26.21
14.21
11.61

37.83
33.48
35.70
45.44
41.60
76.21
31.36

65.05
82.28
26.83
36.17
46.06
45.87
44.59

2.27
5.82
8.04
3.64
8.65
4.27

84.11
74.69
66.02
91.66

35.15
29.48
8.30
1.97
37.77
31.66

High
High

Total

55.72
45.87
25.68
15.42
8.75
10.93

60.46
48.70
44.97
65.62
61.39
79.20
24.25

71.37
81.12
14.98
33.92
53.07
55.94
52.58

1.67
4.04
4.96
2.66
5.33
2.66

89.65
84.48
72.12
94.95

28.14
22.67
8.47
1.62
42.96
33.87

41.83
57.52
40.22
26.36
15.72
13.13

36.88
32.02
33.14
39.02
37.00
76.10
33.06

63.64
82.04
21.38
33.86
45.23
43.15
41.44

2.47
6.33
9.44
5.93
12.15
6.59

79.63
66.65
60.09
86.36

29.16
21.51
5.90
1.76
35.15
38.16

20,034
20,056
20,111
20,087
20,072
20,127

26,953
10,468
3,177
21,896
25,912
11,078
5,586

4,672
4,731
10,469
27,827
9,733
27,616
27,658

27,956
27,660
27,870
27,800
27,869
27,884

27,806
27,533
27,340
27,920

2,222
2,222
2,222
2,222
2,222
2,222

UsingFINRA(2009)data,exceptforthefinancialdiscussiontestsinthelastsixrowswhichcomesfromtheSAVE(2009)data.
*AsreportedbyAllgoodandWalstad(2011).

29

Table7PredictingCreditCardBehaviors*

PerceivedHigh,ActualHigh
PerceivedHigh,ActualLow
PerceivedLow,ActualHigh
Female
Age2534
Age3544
Age4554
Age5564
Age65+
Nonwhite
<Highschool
Highschool
Somecollege
Postgraduate
Single
Divorced/Separated
Widowed
Onechild
Twochildren
Threeormorechildren
Income<$15K
$15K<Income<$25
$25K<Income<$35
$35K<Income<$50K
$75K<Income<100K
$100K<Income<150K
Income>$150K
Incomedrop
Constant

N
ChiSquare
df
Sig
%original
%correctlyclassified
Specificity
Sensitivity

1
Paidinfull
1.945
1.814
1.209
.861
.558
.497
.522
.618
NS
NS
.673
.760
.696
1.269
NS
.765
NS
.680
.665
.593
NS
.866
.855
.871
1.152
1.380
2.345
.727
1.657

20,204
2098.89
28
<.001
58.3
65.3
80.4
44.1

2
Carry
balance
.590
.668
NS
1.114
1.695
1.759
1.712
1.461
.840
NS
NS
1.169
1.343
.812
NS
1.345
NS
1.332
1.340
1.479
.774
NS
1.136
NS
.894
.787
.506
1.410
.753

20,221
1525.53
28
<.001
57.8
63.6
41.6
79.7

3
Minimum
payment
.511
.690
.715
NS
1.297
NS
NS
.649
.395
1.232
1.603
1.331
1.272
.823
NS
NS
NS
1.594
1.713
1.760
1.304
1.365
1.286
1.181
.796
.654
.418
1.974
.431

20,279
3049.12
28
<.001
59.9
67.3
79.7
48.8

4
Latefee
.508
.639
.844
1.237
NS
NS
NS
NS
.550
1.472
1.357
1.026
1.120
.969
1.145
1.129
.862
1.529
1.582
1.961
1.218
1.206
1.264
1.212
.938
.858
.663
2.118
.112

20,245
2043.31
28
<.001
73.7
74.4
95.2
16.2

5
Exceed
limit
.563
.760
.795
NS
NS
NS
.829
.664
.398
1.433
1.807
1.257
1.299
NS
1.199
1.248
NS
1.426
1.491
1.921
1.388
1.292
1.318
1.395
NS
.795
.621
2.087
.077

20,242
1548.79
28
<.001
84.3
84.2
99.6
1.5

6
Cash
advance
NS
1.264
NS
.721
1.433
1.449
1.498
1.406
NS
1.537
1.530
1.330
1.161
NS
1.539
1.363
1.384
1.239
1.399
1.416
1.397
1.234
1.240
1.200
NS
NS
.843
1.674
.049

20,279
706.12
28
<.001
86.9
86.9
100.0
0.0

Theoddsratio(notthebetacoefficient)isreported,aratioofthelikelihoodofaneventoccurringinonegrouptothelikelihoodofthesame
eventoccurringinanothergroup.Ifthebetacoefficientisnotsignificant,NSisreported.TheChiSquarestatisticanditssignificancereport
how well the model fits the data. % original refers to how well a model would do by simply predicting that all individuals engage in a
particularbehaviorordonotengageinaparticularbehaviorbasedonthesampleaverage.%correctlyclassifiedreferstotheresultsofthe
modelusingthepredictors.Specificityreferstothe%ofnegativecasescorrectlyidentifiedwhilesensitivityreferstothe%ofpositivecases
correctlyidentified.
*AllgoodandWalstad(2011)showed,usingaprobitmodel,thatactualandperceivedfinancialbehaviorspredictcreditcardbehaviors.

30

Table8PredictingRetirementandSavingsBehaviors

PerceivedHigh,ActualHigh
PerceivedHigh,ActualLow
PerceivedLow,ActualHigh
Female
Age2534
Age3544
Age4554
Age5564
Age65+
Nonwhite
<Highschool
Highschool
Somecollege
Postgraduate
Single
Divorced/Separated
Widowed
Onechild
Twochildren
Threeormorechildren
Income<$15K
$15K<Income<$25
$25K<Income<$35
$35K<Income<$50K
$75K<Income<100K
$100K<Income<150K
Income>$150K
Incomedrop
Constant

N
ChiSquare
df
Sig
%original
%correctlyclassified
Specificity
Sensitivity

1
Emergency
2.611
2.453
1.318
.882
.831
.768
.863
1.225
2.273
NS
.594
.694
.714
NS
NS
.691
NS
.738
.693
.567
.376
.404
.586
.705
1.349
1.845
3.460
.622
NS

26,339
5569.93
28
<.001
62.8
72.3
85.3
50.4

2
3
College
529or
savings
otherplan
1.603
1.661
1.955
1.745
NS
1.314
.773
.723
.611
NS
.445
NS
.379
NS
.226
NS
.169
NS
NS
NS
.369
.225
.494
.485
.668
.533
NS
1.399
NS
NS
.835
.589
1.467
NS
NA
NA
NS
NS
.825
NS
.460
NS
.437
.521
.604
.545
.666
.612
1.734
NS
2.278
1.995
3.746
1.742
.718
1.211
1.856
.578

10,574
3,453
1788.70
455.02
27
27
<.001
<.001
67.7
66.5
73.1
71.0
88.9
87.1
40.1
39.1

4
Calculate
3.311
2.116
1.742
.926
1.387
1.482
2.046
2.425
3.059
NS
.438
.678
.873
1.161
NS
.874
NS
NS
NS
NS
.411
.599
.631
.730
1.299
1.779
2.286
1.385
.344

21,669
3899.23
28
<.001
60.3
69.6
81.8
51.0

5
Invest
ments
2.444
1.635
1.406
.876
NS
1.258
1.396
1.760
2.602
.928
.444
.570
.720
NS
1.196
.799
1.172
NS
NS
.809
.239
.326
.496
.700
1.562
2.280
4.385
.885
.665

25,232
5791.61
28
<.001
62.5
72.6
84.7
52.5

6
Regular
contrbns
NS
1.254
NS
NS
NS
.743
.616
.437
.185
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
.177
.275
.518
.674
1.261
1.830
1.933
.595
7.309

10,959
1055.19
28
<.001
76.1
77.6
17.6
96.4

7
Lifecycle
.645
1.595
NS
1.143
NS
NS
.666
.575
.333
1.471
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
NS
.898
1.292
.376

5,542
371.70
28
<.001
66.8
68.4
92.5
20.1

Theoddsratio(notthebetacoefficient)isreported,aratioofthelikelihoodofaneventoccurringinonegrouptothelikelihoodofthesameeventoccurring
inanothergroup.Ifthebetacoefficientisnotsignificant,NSisreported.TheChiSquarestatisticanditssignificancereporthowwellthemodelfitsthedata.
% original refers to how well a model would do by simply predicting that all individuals engage in a particular behavior or do not engage in a particular
behaviorbasedonthesampleaverage.%correctlyclassifiedreferstotheresultsofthemodelusingthepredictors.Specificityreferstothe%ofnegative
casescorrectlyidentifiedwhilesensitivityreferstothe%ofpositivecasescorrectlyidentified.
*For regressions 2 and 3, which are specific to college savings, all respondents have children. For this reason, the one child is the omitted variable.

31

Table9PredictingCreditandLoanBehaviors

PerceivedHigh,ActualHigh
PerceivedHigh,ActualLow
PerceivedLow,ActualHigh
Female
Age2534
Age3544
Age4554
Age5564
Age65+
Nonwhite
<Highschool
Highschool
Somecollege
Postgraduate
Single
Divorced/Separated
Widowed
Onechild
Twochildren
Threeormorechildren
Income<$15K
$15K<Income<$25
$25K<Income<$35
$35K<Income<$50K
$75K<Income<100K
$100K<Income<150K
Income>$150K
Incomedrop
Constant

N
ChiSquare
df
Sig
%original
%correctlyclassified
Specificity
Sensitivity

1
Compare
mortgages
1.523
1.476
1.213
.813
NS
.658
.529
.551
.529
1.191
.456
.767
NS
NS
NS
NS
NS
NS
.846
.797
NS
.644
NS
NS
1.272
1.617
1.627
1.171
2.514

4,607
206.17
28
<.001
63.8
65.1
17.9
91.8

2
3
Calculate Latemort.
%income payment
NS
.607
1.560
.860
NS
.769
NS
1.048
NS
.934
NS
.942
NS
1.004
NS
.924
NS
.465
NS
1.650
NS
1.687
NS
1.343
NS
1.303
NS
.926
NS
1.314
NS
1.160
NS
.862
NS
1.629
NS
1.799
NS
2.118
.530
1.474
NS
1.676
NS
1.575
NS
1.429
NS
.823
NS
.705
NS
.548
NS
2.623
3.612
.064

4,667
11,307
58.205
1297.72
28
28
.001
<.001
82.4
78.7
82.4
79.7
0.0
96.8
100.0
16.0

4
Autoloan
.747
.893
.876
NS
1.499
1.195
NS
NS
.675
.848
.742
1.114
1.089
.806
.502
.674
.510
1.331
1.328
1.342
.171
.440
.612
.764
NS
NS
.797
1.101
1.051

26,941
3265.40
28
<.001
65.6
68.9
87.5
33.4

5
Compare
auto
1.631
1.687
1.271
NS
NS
NS
NS
NS
NS
NS
NS
.847
NS
NS
.849
.794
NS
NS
1.148
NS
.719
NS
NS
NS
1.172
NS
NS
NS
.718

9,560
219.20
28
<.001
54.7
57.6
74.6
37.1

6
Credit
report
1.932
1.643
1.348
1.083
1.428
1.213
NS
NS
.978
1.117
.595
.759
NS
NS
.898
NS
NS
NS
NS
1.152
.470
.722
NS
NS
1.162
1.177
1.265
NS
.499

26,796
1622.47
28
<.001
56.3
61.0
74.2
44.0

7
Credit
score
1.716
1.544
1.311
NS
1.181
NS
.891
.893
.792
1.072
.654
.787
NS
NS
.847
NS
NS
1.088
1.088
1.188
.455
.678
.875
.889
1.236
1.242
1.275
1.045
.681

26,833
1510.01
28
<.001
57.9
61.2
78.5
37.3

Theoddsratio(notthebetacoefficient)isreported,aratioofthelikelihoodofaneventoccurringinonegrouptothelikelihoodofthesameeventoccurringin
anothergroup.Ifthebetacoefficientisnotsignificant,NSisreported.TheChiSquarestatisticanditssignificancereporthowwellthemodelfitsthedata.%
originalreferstohowwellamodelwoulddobysimplypredictingthatallindividualsengageinaparticularbehaviorordonotengageinaparticularbehavior
basedonthesampleaverage.%correctlyclassifiedreferstotheresultsofthemodelusingthepredictors.Specificityreferstothe%ofnegativecasescorrectly
identifiedwhilesensitivityreferstothe%ofpositivecasescorrectlyidentified.

32

Table10PredictingRiskyBehaviors

PerceivedHigh,ActualHigh
PerceivedHigh,ActualLow
PerceivedLow,ActualHigh
Female
Age2534
Age3544
Age4554
Age5564
Age65+
Nonwhite
<Highschool
Highschool
Somecollege
Postgraduate
Single
Divorced/Separated
Widowed
Onechild
Twochildren
Threeormorechildren
Income<$15K
$15K<Income<$25
$25K<Income<$35
$35K<Income<$50K
$75K<Income<100K
$100K<Income<150K
Income>$150K
Incomedrop
Constant

N
ChiSquare
df
Sig
%original
%correctlyclassified
Specificity
Sensitivity

1
Bankruptcy
NS
1.341
NS
NS
2.052
2.454
2.174
1.773
NS
NS
1.913
NS
NS
NS
.634
NS
NS
NS
1.445
1.353
.632
NS
NS
NS
.560
.322
.232
1.836
.008

27,027
373.15
28
<.001
97.5
97.5
100.0
0.0

2
TitleLoan
.723
1.251
NS
.733
1.211
NS
.740
.641
.562
NS
NS
1.288
NS
NS
.768
NS
NS
1.470
1.409
1.738
.755
1.293
1.261
NS
NS
NS
NS
1.467
.073

26,784
526.97
28
<.001
93.6
93.6
100.0
0.0

3
Payday
loan
.696
NS
.874
.913
1.638
1.494
1.356
NS
.365
1.614
1.510
1.354
1.457
NS
1.155
1.568
2.007
1.609
1.929
2.219
NS
1.710
1.479
1.402
.680
.488
.326
1.572
.021

26,981
1761.19
28
<.001
90.5
90.5
100.0
0.2

4
Tax
Advance
.704
1.309
.675
.855
1.712
1.380
.721
.361
.249
1.215
2.238
1.901
1.634
NS
NS
1.273
1.494
2.373
2.870
4.049
1.209
1.918
1.850
1.595
NS
.638
.695
1.805
.013

26,923
1936.17
28
<.001
94.1
94.1
100.0
0.1

5
Pawn
shop
.680
NS
.788
.827
NS
.816
.582
.366
.140
1.249
2.210
1.656
1.438
.777
1.235
1.461
1.506
1.433
1.587
1.868
1.977
2.096
1.715
1.396
.618
.518
.427
1.820
.064

26,985
2818.78
28
<.001
88.0
87.9
99.8
1.4

6
Rentto
own
.662
1.316
.765
.876
NS
NS
.699
.403
.204
1.171
2.718
1.715
1.400
NS
.824
1.198
NS
1.671
2.025
2.894
1.655
2.228
1.950
1.389
.611
.547
.450
1.734
.027

26,995
1713.47
28
<.001
93.5
93.5
100.0
0.0

Theoddsratio(notthebetacoefficient)isreported,aratioofthelikelihoodofaneventoccurringinonegrouptothelikelihoodofthesame
eventoccurringinanothergroup.Ifthebetacoefficientisnotsignificant,NSisreported.TheChiSquarestatisticanditssignificancereport
how well the model fits the data. % original refers to how well a model would do by simply predicting that all individuals engage in a
particularbehaviorordonotengageinaparticularbehaviorbasedonthesampleaverage.%correctlyclassifiedreferstotheresultsofthe
modelusingthepredictors.Specificityreferstothe%ofnegativecasescorrectlyidentifiedwhilesensitivityreferstothe%ofpositivecases
correctlyidentified.

33

Table11PredictingInsuranceBehaviors

PerceivedHigh,ActualHigh
PerceivedHigh,ActualLow
PerceivedLow,ActualHigh
Female
Age2534
Age3544
Age4554
Age5564
Age65+
Nonwhite
<Highschool
Highschool
Somecollege
Postgraduate
Single
Divorced/Separated
Widowed
Onechild
Twochildren
Threeormorechildren
Income<$15K
$15K<Income<$25
$25K<Income<$35
$35K<Income<$50K
$75K<Income<100K
$100K<Income<150K
Income>$150K
Incomedrop
Constant

N
ChiSquare
df
Sig
%original
%correctlyclassified
Specificity
Sensitivity

1
Health
insurance
1.322
NS
1.197
1.485
.786
NS
1.135
1.394
8.985
NS
.491
.595
.736
NS
.777
.761
NS
NS
NS
NS
.265
.238
.348
.520
1.565
2.281
2.588
.439
8.071

26,881
5311.51
28
<.001
80.1
80.8
22.9
95.3

2
Home
insurance
1.769
1.351
1.381
1.203
1.393
2.133
3.008
4.511
7.410
.532
.515
.746
.808
NS
.471
.436
.671
NS
NS
NS
.127
.206
.358
.590
1.328
2.217
2.749
.915
4.500

26,643
9955.41
28
<.001
67.3
79.4
59.5
89.0

3
Life
insurance
1.147
1.384
1.162
NS
1.213
1.549
2.264
2.650
2.655
NS
.574
.836
.874
NS
.627
.728
NS
1.126
1.211
NS
.216
.332
.495
.704
1.400
1.752
1.563
.671
1.798

26,468
5448.02
28
<.001
60.6
71.2
52.2
83.6

4
Auto
insurance
1.820
1.124
1.588
1.254
NS
NS
NS
1.445
1.901
.530
.453
.741
NS
.727
.280
.331
.322
NS
NS
.758
.184
.300
.499
.789
1.296
1.774
3.072
NS
45.533

26,977
4752.08
28
<.001
86.9
87.3
18.3
97.7

Theoddsratio(notthebetacoefficient)isreported,aratioofthelikelihoodofaneventoccurringinone
group to the likelihood of the same event occurring in another group. If the beta coefficient is not
significant, NS is reported. The ChiSquare statistic and its significance report how well the model fits the
data.%originalreferstohowwellamodelwoulddobysimplypredictingthatallindividualsengageina
particular behavior or do not engage in a particular behavior based on the sample average. % correctly
classifiedreferstotheresultsofthemodelusingthepredictors.Specificityreferstothe%ofnegativecases
correctlyidentifiedwhilesensitivityreferstothe%ofpositivecasescorrectlyidentified.

34

Table12PredictingFinancialDiscussionBehaviors**

PerceivedHigh,ActualHigh
PerceivedHigh,ActualLow
PerceivedLow,ActualHigh
Female
Age3650
Age5165
Age65+
Divorce/Separated
Single
Widowed
Onechild
Twochildren
Threeormorechildren
Less1000
1015
1520
3040
Morethan4000
Constant

N
ChiSquare
df
Sig
%original
%correctlyclassified
Specificity
Sensitivity

1
Relatives
NS
NS
1.396
1.540
.487
.280
.290
1.682
1.810
2.348
NS
NS
NS
1.596
NS
NS
NS
NS
.491

2,222
214.16
18
<.001
70.8
72.4
94.2
19.4

2
Friends
1.566
NS
1.822
1.259
NS
.546
.314
2.480
2.726
NS
NS
1.505
NS
NS
NS
NS
NS
NS
.207

2,222
228.08
18
<.001
78.5
78.5
97.2
10.3

3
Colleagues*
3.152
1.080
2.560
.695
1.010
.398
.029
1.211
1.620
.882

2,222
140.36
18
<.001
94.1
94.1
100.0
0.0

4
Banks
1.668
NS
1.523
NS
NS
NS
NS
.677
NS
NS
.745
NS
NS
.349
.580
NS
NS
NS
.549

2,222
151.79
18
<.001
64.9
91.3
17.9
65.6

5
Notalks
.595
NS
.592
.719
1.751
2.130
1.955
.747
.579
.705
1.393
NS
NS
NS
NS
NS
NS
NS
.579

2,222
113.57
18
<.001
61.8
62.1
87.3
21.2

Theoddsratio(notthebetacoefficient)isreported,aratioofthelikelihoodofaneventoccurringinonegrouptothelikelihood
ofthesameeventoccurringinanothergroup.Ifthebetacoefficientisnotsignificant,NSisreported.TheChiSquarestatistic
anditssignificancereporthowwellthemodelfitsthedata.%originalreferstohowwellamodelwoulddobysimplypredicting
thatallindividualsengageinaparticularbehaviorordonotengageinaparticularbehaviorbasedonthesampleaverage.%
correctlyclassifiedreferstotheresultsofthemodelusingthepredictors.Specificityreferstothe%ofnegativecasescorrectly
identifiedwhilesensitivityreferstothe%ofpositivecasescorrectlyidentified.

Theomittedgroupsare:perceivedlow,actuallow;ageunder35years;married;nochildren;andmonthlynetincomeofat
least2000butlessthan3000.
*Only 131 individuals have consulted colleagues about financial matters. Because of this limited behavior, not all of the
predictorvariablescanbeusedandthereforechildrenandincomearenotincludedinthisregression.
**Only39individualstalktoneighborsaboutfinancialmatters.Duetothesmallnumberofpeoplewhoengageinthisbehavior,
itisnotpracticaltorunalogisticregression.

35

Appendix1FINRA(2009)Survey:SelectedTopicsandQuestions

Creditcardbehaviors
Paidinfull
Carrybalance
Minimumpayment
Latefee
Exceedlimit
Cashadvance

Savings&retirementbehaviors
Emergencyfund
Collegesavings
Educationaccount
Calculate
Otherinvestments
Regularcontribution
Lifecycle

Loanbehaviors
Comparedmortgagelenders
Calculatemortgageas%income
Latemortgagepayment
Haveautoloan
Compareautoloans
Creditreport
Creditscore

Riskybehaviors
Declaredbankruptcy
Autotitleloan
Paydayloan
Taxadvanceonrefund
Usedpawnshop
Usedrenttoown

Insurancebehaviors
Healthinsurance
Homeownersorrenters
Lifeinsurance
Autoinsurance

YesorNo

Ialwayspaidmycreditcardsinfull
Insomemonths,Icarriedoverabalanceandwaschargedinterest
Insomemonths,Ipaidtheminimumpaymentonly
Insomemonths,Iwaschargedalatefeeforlatepayment
Insomemonths,Iwaschargedanoverthelimitfeeforexceedingmycreditline
Insomemonths,Iusedthecardsforacashadvance

Haveyousetasideemergencyorrainydayfundsthatwouldcoveryourexpenses
incaseofsickness,jobloss,economicdownturn,orotheremergencies
Areyousettingasideanymoneyforyourchildrenscollegeeducation
Areyourusinga529PlanorCoverdellEducationalSavingsAccounttosavefor
college
Haveyouevertriedtofigureouthowmuchyouneedtosaveforretirement
Notincludingretirementaccounts,doesyourhousehold/doyouhaveany
investmentsinstocks,bonds,mutualfunds,orothersecurities
Doyouregularlycontributetoaretirementaccountlikea401(k)orIRA
Areyourretirementassetsprimarilyinvestedinalifecycleortargetdatefund

Whenyouweregettingyourmortgage,didyoucompareoffersfromdifferent
lendersormortgagebrokers
Whenyouweregettingyourmortgage,didyouconsiderhowmuchthemonthly
paymentswouldbeasapercentageofyourincome
Howmanytimeshaveyoubeenlatewithyourmortgagepaymentsinthelast2
years?(recodedas0=never;1=onceormore)
Doyou/Doesyourhouseholdcurrentlyhaveanautoloan?WeareNOTreferring
toanautolease
Thinkingaboutyourmostrecentautoloan,didyoucompareoffersfrom
differentlenders
Inthepast12months,haveyouobtainedacopyofyourcreditreport
Inthepast12months,haveyoucheckedyourcreditscore

Haveyoudeclaredbankruptcyinthelasttwoyears
Haveyoutakenoutanautotitleloan
Haveyoutakenoutashorttermpaydayloan
Haveyougottenanadvanceonyourtaxrefund
Haveyouusedapawnshop
Haveyouusedarenttoownstore

Areyoucoveredbyhealthinsurance
Doyouhavehomeownersorrentersinsurance
Doyouhavealifeinsurancepolicy
Doyouhaveautoinsurance

36

You might also like