Financial (il)literacy vs. individual’s behavior: evidence

on credit card repayment patterns

Gustavo Barbozaa,b, Paola Bonginic, Monica Rossolinic,*

aJack and Vada Reynolds Endowed Chair Professor of International Business, Loyola University of

New Orleans, 6363 St Charles Avenue, New Orleans, LA 70118, USA
bDepartment of Business and Law, University of Milano-Bicocca, Via Bicocca degli Arcimboldi,

8 20126, Milan, Italy
cDepartment of Business and Law, University of Milano-Bicocca, Via Bicocca degli Arcimboldi,

8 20126, Milan, Italy

Abstract

We explore the role that financial (il)literacy and personal traits have on financial behavior. Using a sample

of 156 college students from the United States, we provide unique empirical evidence by specifically differen-

tiating between individuals with higher levels of financial literacy versus individuals declaring not knowing the

answers to financial literacy questions and those answering incorrectly. Thus, we assess the implications of

revealed lack of financial knowledge on financial behavior regarding credit card use in comparison with two

other cohorts; cohort one answering correctly, and cohort two failing to answer correctly. A novelty of our

study is that we contrast these results to the behavioral factors of over spending and surprised levels of spend-

ing—proxies for personality traits—when using credit card. Our exploratory empirical findings indicate that

among personal-traits considered in this study overspending results in lack of payment in full in credit card

debt, and more importantly these effects dominate any gains derived from financial literacy. To this extent fi-

nancial literacy appears to only play a marginal role avoiding month-to-month credit card debt. Furthermore,

financial knowledge derived from parents has a strong positive effect on individuals’ financial behavior espe-

cially for students characterized by a relevant financial illiteracy. The implications of this research support the

argumentation that early financial literacy may have the strongest effect in shaping individuals inherent behav-

ior patterns; that is, early exposure to financial education is strictly preferred and should be promoted at early

stages of the educational system. © 2021 Academy of Financial Services. All rights reserved.

JEL classification: I22; G41

Keywords: Financial literacy; Credit card debt; College students; Financial behavior

*Corresponding author. Tel.: +39 02 6448 3014; fax: +39 02 6448 3072

E-mail address: monica.rossolini@unimib.it

1057-0810/21/$ – see front matter © 2021 Academy of Financial Services. All rights reserved.

Financial Services Review 29 (2021) 247–276



1. Introduction

Issues relating to financial knowledge and individual behavioral patterns are at the core of

every individual’s economic decision (Thaler & Shefrin, 1981). It is well documented that

financially illiterate individuals incur in costly or even improper financial decisions with sig-

nificant negative spillover effects on other aspects of their life (see Lusardi & Mitchell,

2007; and Van Rooij, Lusardi, & Alessie, 2011, 2012, among others). For instance, previous

studies demonstrate (see Lusardi & Mitchell 2014, for an excellent review of the literature)

that when young people possessing low levels of financial literacy and combined with a

widespread use and easy access to credit cards (especially among college students), results

in making the wrong financial decisions; most notably carrying over large balances in credit

card debt. As a matter of fact, credit card debt coupled with financial illiteracy is associated

with unhealthy behaviors (Adams & Moore 2007; Berg et al., 2010; Lyons & Rogers, 2004),

lower academic performance (Pinto et al., 2001), and lower financial well-being (Grable &

Joo 2006; Norvilitis et al., 2006; Nelson et al., 2008). The negative effects of poor financial

decisions are well stated; however, what causes and fuels this behavior is a much lesser

study field.

In this context, the role of personality traits, individual preferences with hyperbolic

discounting, and attitudes in influencing financial decision-making have been less inves-

tigated to this date in the literature. This is more interesting as these issues are increas-

ingly attracting the attention of psychologists and, and more recently yet to a lesser

extent economists (see Brown & Taylor, 2014 and Barboza, 2018, for instance).

Similarly, issues relating to one self’s perception (i.e., confidence about one’s knowledge

or ability to conduct financial decision in particular) has been explored in the realm of be-

havioral economics and finance (Thaler, 1980; Thaler & Shefrin, 1981). However, de-

spite these advances, it is only until recently that the combined elements have become a

subject of analysis in conjunction. More specifically the actual financial knowledge

showing that overconfident individuals, or those with high self-assessed knowledge but

low actual knowledge, have a higher propensity to engage in risky (costly) financial

behaviors (Tokar Asaad, 2015; Brown & Taylor, 2014; Chu et al., 2017; Kramer, 2016;

Porto & Xiao, 2016; Xia et al., 2014).

On the basis of these considerations and given the importance that financial literacy plays

in optimal decision-making processes, our paper aims at studying the relationship between

individual behavior and financial (il)literacy. More specifically, we aim at providing robust

empirical evidence useful to fill the gap where little attention has been given to study the

impact of both financial literacy and of personality traits, preferences and attitudes on overall

financial behavior measured as repayment patterns in credit card debt.

Specifically, we hypothesize that regardless to one’s financial knowledge level, specific

personality traits, such as present-bias preferences or impulse behavior, may negatively

impact the financial decision-making processes and overpower the (potential) positive

effects from higher level of financial literacy. Secondly, we hypothesize that self-awareness

of not possessing a strong financial knowledge (admitted by choosing the Do Not Know

option in a financial literacy questionnaire) may lead to more conservative financial

248 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



decisions. In turn these decisions could prevent individuals from assuming too much finan-

cial risk, resulting in a “correct” or less costly financial decision.

This paper studies credit cards use among college students as we investigate the financial

behavior and decisions that college students make in terms of credit card repayment patterns.

That is, whether students select to pay the credit card balance in full every billing cycle, or

decide to carry a month-to-month balance, either by paying less than the full balance, by

only the minimum payment (anchoring) or by falling behind on their payments. We use a

sample of 156 college students from the United States and consequently estimate a series of

ordered Probit models. Our results provide unique empirical evidence by creating a distinc-

tive differentiation between individuals with higher levels of financial literacy versus indi-

viduals declaring not knowing the answers to financial literacy questions and those

answering incorrectly. Due to the relevance that perception about oneself has on actual

behavior, we bring forth the hypothesis that incorrectly answering financial questions versus

answering “I do not know” may have potentially large implications relating to the way indi-

viduals react in the face of making—important—financial decisions. Besides, we account

for a series of personality traits (namely overspending, anxiety, and mental accounting

issues) to test for their effects on financial behavior (credit card repayment). An important

finding of our study is that these personality traits may have stronger (negative) effects not

easily overcome by higher levels of financial literacy.

As a spoiler alert, our exploratory results find robust evidence indicating that individuals’

personal traits are the main driver for individuals to accumulate and carry over a month-to-

month balance in their credit card(s). Specifically, attitude towards overspending is the main

factor negatively affecting the capability of individuals to repay in full credit card debt every

billing cycle. This result is in line and provides statistical support to the importance of men-

tal accounting issues faced by individuals with present bias preferences. We also find, con-

trary to previous literature (Norvilitis et al., 2006; Robb, 2011; Shim et al., 2009, 2010), that

a higher level of financial literacy is not a fundamental and it only controls on the margin for

personal traits and attitudes (overspending, lack of self-control, and issues related to poor

mental accounting). Furthermore, financial knowledge derived from parents has a strong

positive effect on individuals’ financial behavior especially for people characterized by high

levels of financial illiteracy. Our findings reinforce the evidence of those studies that specifi-

cally investigated the relationship between the role of parents and the (mis)use of credit

cards by college students (Hancock et al., 2013; Xiao et al., 2011). Our findings may serve

policy makers when designing specific policies aimed at avoiding, or reducing, debt traps

and socio-economic vulnerability of the borrower.

Our paper contributes to the extant literature in at least three strands. At this point we do not

make any claim to make a theoretical contribution to the field. Our contributions are empirical

and they are as follows. First, we provide a unique and innovative empirical break down

between financial literacy and financial illiteracy. In particular we separate wrong answering to

a standard financial literacy questionnaire (five questions in total; see Section 2 for more details

on the questionnaire) between incorrect answering and Do Not Know responses. The former is

classified as an attempt to respond and measures an overconfidence about one’s actual knowl-

edge which at the end is revealed to be poor (Overconfidence). Whereas the latter represents a

direct answer to not knowing and therefore to not possess overconfidence on his or her

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 249



knowledge level. It is relevant to point out that while at present previous research has studied

the role of financial literacy (Correct/Incorrect) in decision making, little has been dedicated to

study the relation to how much people actually Do Not Know or believe they do not know.

Therefore, our central contributing focuses on accounting the magnitude and explicit separation

between an inaccurate perception of knowledge level leading to incorrectly answering, versus

individuals actually not knowing and clearly revealing this lack of knowledge.1 To the best of

our knowledge, we are the first to explore this issue.2

Secondly, we study the effect of consumption and spending patterns, as they relate to over-

spending (present bias) and surprised factors (inadequate mental accounting) as they lead to

credit card debt accumulation. The issues at stake here are similar to those first introduced by

Thaler (1980) and Thaler and Shefrin (1981) on the different effects that individual mental

accounting patterns may have between a planner (sophisticated self) and doer (naı̈ve self). More

specifically, the decisions individuals have to make include but are not limited to borrowing

money, using credit cards, and adjusting consumption to income flows and avoiding falling prey

of present-bias behavior and consequently accumulate a month-to-month balance on their credit

card. In this context we then hypothesize that those with higher levels of self-awareness have

the knowledge to answer correctly (basic financial literacy questions) or recognize their limita-

tions and face the reality of not knowing the answer, and openly recognizing that by answer Do

Not Know. Consequently, here our interest is to uncover the differences between those that are

aware of their limitations versus those that are or possess excessive self-confidence, while being

wrong. It is in this difference that we expect to reach relevant findings as to why individuals

with certain personality traits are more likely to make incorrect and costly financial decisions.

We believe this is the first paper that directly addresses this very important and timely issue.

Finally, our third contribution is to study the role of financial literacy in ameliorating neg-

ative effects of costly personality traits on credit card repayment patterns. The extant litera-

ture assumes that higher levels of financial literacy are a predominantly determinant of

superior financial performance. However, while significant contributions have been made in

advancing the role of financial literacy, previous works in this field have paid little attention

to the role that personality traits (aptitude and attitude) play in financial decision and sec-

ondly what the role of financial (il)literacy is in shaping these inherent behavioral patterns.3

We organize the rest of the paper as follows. The next section reviews the most relevant

literature on financial literacy and personal traits and sets forth our testing hypotheses. The

third section describes the research design and the estimation model. The fourth section ana-

lyzes estimation results, which are then discussed in the fifth section, while the last section

presents some general conclusions and policy recommendations.

2. Literature review

Recently the topic of financial literacy has received increased and extensive attention in

the literature. The literature argues in favor of the relevance of adequate and timely financial

literacy as paramount on individual decision-making process and their outcomes. In fact, the

evidence points out in support of the argument that low levels of financial literacy are not

only linked with high levels of personal and household debt (Lusardi & Tufano, 2009;

250 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



Moore, 2003; Stango & Zinman, 2009), inadequate retirement planning (Hilgert, Hogarth, &

Beverly 2003; Lusardi & Mitchell, 2007), or inadequate stock market participation (van

Rooij, Lusardi, & Alessie 2011), but also to poor health (Joo & Garman, 1998) or adverse

health choices (Peters et al., 2007) and in general poorer overall life outcomes. In this

respect, widespread financial illiteracy among young people is of particular concern for two

main reasons. First, as they enter adulthood, a number of important financial decisions are to

be undertaken (such as financing college studies; moving away from home; purchasing their

first car; using credit cards; saving for retirement; etc.), for which they might not be

adequately prepared. Misguided financial decisions in the early stage of their lives could

have potentially disastrous consequences (huge debt, a poor credit rating, and inadequate

retirement plans) for the remaining of their whole life (Schagen & Lines, 1996; Lusardi,

Mitchell, & Curto 2010). Second, a lack of financial literacy seems to impact students’ uni-

versity performance as noted by Kezar and Yang (2010) whom suggest that a student’s aca-

demic achievement is negatively affected by financial distress, which, in turn, is a more

likely outcome in presence of low levels of financial literacy. In particular the literature

documents that inadequate financial skills (especially in the area of cash management) result

in higher level of stress and even anxiety with significant (negative) impact on academic per-

formance (Kapoor et al., 2006; Razafimahasolo et al., 2016; Xiao et al., 2011). Low financial

capabilities generate financial stress and anxiety, which in turn create negative spillover

effects into other life dimensions.

In more recent research, particularly in the last decade, the majority of theoretical and em-

pirical literature on financial literacy has addressed and investigated many different topics.

These topics cover from the influencing factors that drive financial literacy to the methodo-

logical approaches to best treat survey questions that try to effectively measure a latent vari-

able such as financial literacy.

A general agreement has been reached such as on the necessary prerequisites to gauge fi-

nancial literacy (including the types of knowledge that best motivate and facilitate financial

action). It has now become common knowledge that financial literacy among both adults

and the young is low; and influenced by socio-demographic factors, such as gender, educa-

tion, income, employment status, and age. It is also recognized that informal sources of edu-

cation such as family background and interaction with peers are of particular importance in

this regard (see Lusardi & Mitchell, 2014, for a review of theory and empirical evidence).

More specifically, individuals’ financial literacy seems to be significantly and positively

associated with parental educational attainment and with the presence of forms of financial

socialization within the family and the group of peers (for instance children observing their

parents’ saving behavior or receiving a more formal financial education from them). In gen-

eral, the literature also agrees that young adults receive financial literacy through two main

sources, parents and the educational system (Lusardi et al., 2009). In particular, Lusardi et al

(2009) note that young adults with college-educated parents tend to have a better under-

standing of financial concepts. Mandell (1997) provides evidence on the role that proper

types of financial education play as a significant factor in achieving financial literacy, while

others indicate that much of the financial education is being conducted through business and

community organizations, and not through educational institutions (Fox et al., 2005).

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 251



However, despite that the role of diversity (e.g., gender, ethnicity, age, and education) has

been well documented, including considering different settings and time spans, our under-

standing of the mechanisms and channels of how these forms of diversity continue to result

in significant gaps in financial literacy and consequently results in pervasive effects, remains

a conundrum. The quest to find solutions to these differences is far from complete. This chal-

lenge is more complicated given the plethora of methodologies used to assess financial liter-

acy. To date, the issue of assessing financial literacy has concentrated on the conceptual

definition of this latent variable. Houston (2010) and Remund (2010) provide a thorough lit-

erature review that helps frame the issue of the conceptual definition of what financial liter-

acy is or should be. Indeed, financial literacy has been variably defined as specifically

referring to a form of knowledge (e.g., Hilgert, Hogarth, & Beverly, 2003), the ability to

apply that knowledge (e.g., Mandell, 2008), or good financial behavior (e.g., Moore, 2003).

The constructs used to measure financial literacy vary quite substantially according to the

different conceptual definitions adopted. In fact, the construct either covers a wide variety of

financial topics, including debt, insurance, spending, investments and retirement savings,

budgeting, and inflation, or focuses on a few financial issues. Accordingly, the number of

questions used to assess financial knowledge levels varies widely, ranging from three to 45

total items.

Across studies, both multiple-choice questionnaires and self-report questions have been

employed to measure financial literacy; where the former are knowledge based and the lat-

ter assess perceived knowledge. More recently, surveys have been designed to gauge both

objective knowledge and perceived knowledge. In general, considerable progress has been

achieved in the design of surveys aimed at identifying individual levels of financial liter-

acy through the effort made by the OECD and its International Network on Financial

Education (INFE). Jointly they develop and promote a common questionnaire based on

the experience of a large number of previous rigorous national and international surveys.

The OECD/INFE (2012) report describes the questionnaire’s underlying methodology and

Kempson (2009) provides further details on this subject.4 Due to its importance, research

on financial literacy has in fact inspired numerous public initiatives at both the national

and international level. Several countries now have financial literacy initiatives and strat-

egies in place to increase the levels of financial understanding and knowledge among all

citizens.

In contrast, it is not until very recently, that the process of data analysis (i.e., of analyzing

the information obtained through questionnaires) has taken a central role in exploring the

difference between financial literacy and illiteracy. That is to say, the emphasis has been on

what is known but not in what it is not known. In other words, the research focus has been

placed on people thinking that they know enough to be correct, but not realizing that they do

not know enough to be correct. Due to the large arrays of data sources, it has become neces-

sary to use both bivariate (ordered data from less to more) and multivariate techniques to

quantify financial (il)literacy. In general, responses to the stated questions are simply

summed to generate an index (score) of financial literacy, which typically ranges between

zero and the maximum number of correct answers.5

A common practice to most studies is to cluster the “Do Not Know” responding with

“incorrect” answers, in opposition to “correct” answers, notwithstanding the fact that it is

252 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



also widely recognized that these two types of responses (DNK and Incorrect) might refer to

two distinct kind of respondents, with diverse (financial) educational needs. Manton et al.

(2006) were the first to signal that college women tend to select “don’t know” response

more frequently than men, especially on more numerically oriented subjects. In the same

vein, Lusardi and Mitchell (2014) pointed out that “one twist on the differences by sex, (. . .)
is that while women are less likely to answer financial literacy questions correctly than men,

they are also far more likely to say they ‘do not know’ an answer to a question, a result that

is strikingly consistent across countries.” However, so far, the common choice throughout

the empirical literature is to include them in the same cluster of wrong answers for methodo-

logical issues. Exceptions to this practice are two recent papers by Chen and Garand (2018)

and Kim and Mountain (2019). Kim and Mountain specifically address the econometric

issue of misleading results obtained from ignoring DNK responses and suggest the use of bi-

nomial-latent regression models to prevents distortions from DK/RF responses. Chen and

Garand (2018) deepen the well-known issue of gender gap in financial literacy, by giving

specific attention to DNK answers. In particular, after having ascertained that women may

exhibit lower levels of financial knowledge because they lose the opportunity to hazard a

guess and arrive at a correct answer based either on partial knowledge or on random chance,

they consider the possibility that there are psychological processes at work involving risk ac-

ceptance and confidence in financial knowledge that prompt women to give DK responses at

a rate higher than men. As a result, they suggest that future research should consider the rela-

tive roles of DK and incorrect responses in measuring financial knowledge.

In this study, we follow such a suggestion and aim to contribute to the extant literature by

treating respondents, who admit not knowing, differently from those respondents who im-

plicitly consider themselves as knowledgeable but in fact possess a “wrong” knowledge. As

a matter of fact, when self-assessed questionnaires are included in multiple choice question-

naire (testing objective knowledge), the evidence indicates that most people are unaware of

their own shortcomings, as there is often a substantial mismatch between people’s self-

assessed knowledge versus their actual knowledge (Lusardi & Mitchell, 2014). This incor-

rect assessment is also directly related to the research on behavior as presented by Thaler

and Shefrin (1981). In particular, we take particular interest in separately scoring “Correct,”

“Incorrect,” and “Do Not Know” answers and conjecture that admitting of not possessing a

proper financial knowledge may lead to more conservative financial decisions, that could

prevent taking too much financial risk. In addition, we approximate the measures for present

bias and mental accounting biases as noted next.

We also take a cue from a growing body of literature acknowledging that the drivers of fi-

nancial choices are not constraint just by knowledge and the acquisition of basic informa-

tion. In this new stream of research cognitive biases, individual psychological traits and

aptitudes, motivations, and timing of the choice to be undertaken are all examples of the be-

havioral and psychological constraints that interact with economic decisions in general and

financial choices in particular. One specific cognitive bias that has attracted much attention

in empirical studies connecting financial decisions, behavioral biases, and financial knowl-

edge is overconfidence on one’s actual ability, performance, level of control, or chance of

success (Moore & Healy, 2008). Financial confidence, in particular, reflects a self-assessed

level of financial knowledge, which may or may not coincide with measured financial

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 253



knowledge. Financial literacy overconfidence has been linked to various risky behaviors

such as higher stock market participation (Xia et al., 2014), less use of financial advice

(Kramer, 2016; Porto & Xiao, 2016) and a preference for direct stock investment rather than

in less risky/more diversified mutual funds (Chu et al., 2017) or greater likelihood of engag-

ing in risky (costly) financial behaviors, such as taking out a title-loan, or a short-term pay-

day loan (Tokar Asaad, 2015).

From this perspective, we take particular interest in the literature on behavioral economics

which assumes, and dictates consequently, that individuals may acquire debt, adjust repay-

ment capabilities, and adhere to differentiated patterns of repayment away from the neo-

classical rational expectations teaching. The general consensus is that rationally behaving

individuals have perfect foresight, are rational and apply consistent discounting rules on con-

sumption. On the other hand, the literature on behavioral biases proposes the existence of

several factors governing individuals’ behavior. Particularly, issues relating to hyperbolic

discount functions in consumption – present bias –, naı̈ve behavior, lack of self-control and

impatience (Akerlof, 1991; Kuchler 2013; Laibson, 1997; O’Donoghue & Rabin, 1999;

Thaler & Shefrin 1981; Thaler 2018, among others) may lead to patterns of consumption,

which fueled by easy access to credit cards may result into too much debt accumulation and

patterns of procrastination on repayment (Barboza, 2018). Although aware that the psycho-

logical literature on personality traits is not confined to and is richer than that just mentioned

on behavioral biases, in this study we are interested in highlighting and measuring those

individual features that characterize one’s pattern of consumption and savings and manifest:

(1) in a tendency to overvalue immediate rewards (I buy what I like now, disregarding the

issue of affordability of the purchase), while putting less worth in long-term consequences (I

will think tomorrow how to find the money to afford the purchase); (2) in a process, known

as mental accounting, whereby individual expenses will not be considered in conjunction

with the present value of one’s total wealth.6 Instead it is considered in the context of the

current budgetary period and the category of expenses, leading to constraints/relaxations of

purchases irrespective to the whole—same—fungible resource that is income plus (eventu-

ally) personal net worth ( Cheema & Soman, 2006; Zhang & Sussman, 2018) and a greater

willingness to pay for goods when using credit cards than cash (Prelec & Simester, 2001).

At the same time the procrastination of debt payment can cause pain and anxiety concerns

which feedback into the repayment behavior individuals display next period. We follow a

recent strand of literature (Andrews & Wilding, 2004; Fiksenbaum et al, 2017; Marjanovic

et al., 2013) who define an emotional state, such as financial threat, referring to self-reported

fearful-anxious uncertainty regarding one’s current and future financial situation. In this

study we take particular interest in the effect that higher financial anxiety may have on credit

card repayment behavior.

Thus, we propose to study the combined and interaction effects of financial (il)literacy

and personality traits, as they relate to credit card repayment patterns. With these consid-

erations in mind, then the backbone of our analysis is driven by the following set of

hypotheses.

Hypothesis 1: Higher levels of financial literacy—measured as higher number of correct answers—
lead to better repayment rates in credit card debt.

254 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



Hypothesis 2: The higher the transgenerational transmission of financial education from parents to
children results in improved repayment in credit cards.

Hypothesis 3: Individuals with self-assessed lower level of financial literacy (DK answers) display
diverse behaviors (credit card repayment patterns) as opposed to individuals who do
not possess basic financial literacy (wrong answers).

Hypothesis 4: Individuals displaying present bias, improper mental accounting and or impulse
behavior and financially derived anxiety are less likely to display proper financial
behavior (i.e., pay credit cards in full) and consequently carry month-to-month
balance.

Hypothesis 5: Positive effects from higher levels of financial literacy could be overpowered by
personal traits. This is to say, that financial literacy may or may not be enough to
counter inherent negative traits that individuals already possess.

3. Method

3.1. Data and variables

Data for this research comes from a survey administered to three samples of business col-

lege students in the United States, attending a Midwest Higher Education University and a

Mid-Atlantic University. The survey was paper based and administered in person to a total

of 1,149 students. Data were collected in 2015 and the total complete sample size useful for

our analysis is of 156 respondents.

The sample is composed as follow: 45.50% are female students and 54,50% are male stu-

dents; minority represents 21.2% whereas White race students are 78.8% of the sample. In

terms of academic status, freshman students are 32.05% of the sample, Sophomore students

8.97%, Junior students are 20.51%, Senior students are 36.54%, and Graduate students only

1.93%.

The main interest of this study is to explore the effects that financial (il)literacy and indi-

vidual personal traits related to purchasing behavior have on their financial behavior as man-

ifested in credit card repayment patterns. Thus, we measure financial behavior as an

individual’s credit card repayment pattern and ask questions regarding credit card repayment

behavior, such as: Pay if full every month; Pay some in full and then only the minimum

required; Pay more than minimum required but not in full; Pay Minimum required; or Pay

less than Minimum required. We create the categorical variable used to estimate the ordered

Probit model and corresponding probabilities of occurrence.

The survey also includes a combination of questions related to personal traits and finan-

cial (il)literacy elements. Regarding financial literacy we post five financial literacy ques-

tions to students (see Fig. 1 for details) that tests the basic financial concepts traditionally

tested in the financial literacy literature since Lusardi and Mitchell (2007) proposed them:

inflation, time value of money, diversification, and interest rate compounding.

The survey also includes a set of questions designed to capture specific personality traits.

These variables serve as proxies to assess individuals’ time preferences (present bias issues),

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 255



F
ig
.
1
.
F
in
an
ci
al

li
te
ra
cy

q
u
es
ti
o
n
s.

256 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



issues relating to mental accounting, and potential issues relating to procrastination and lack

of commitment. Based on the review of the literature we argue that these topics may mani-

fest in turn in behavioral delays on credit card repayment, as purchase may be in excess to

monthly repayment capabilities. Therefore, to further understand individuals’ decision-mak-

ing process, subjects were asked if they were surprised at the end of the billing cycle with

the balance the credit cards has reached, and secondly if they have engaged in purchases

knowing that they did not have money to pay it in full when the balance was due. We define

these two variables as Surprised and Overspending, respectively. In both cases, our tentative
hypothesis is to expect that the higher the level of Surprised and the higher the amount of

Overspending lead to a worsen in credit card repayment patterns in the next billing cycle.

To identify one’s personal attitude toward debt repayment we include the variable

Anxiety where the variable takes the values according to the following scale: 1= not anxious

at all and 5 = extremely anxious. In our survey, students self-report the level of their finan-

cial anxiety answering to a specific question of the survey. Anxiety is in relation to how the

person feels when he or she has to pay the credit card bill at the end of the billing cycle.

Finally, in addition to financial literacy questions our survey also considers a financial

education question. In particular we analyze the role that parents may play in the buildup of

the financial literacy levels of their children. The survey asks whether their parents were the

primary source of financial education and we define a dummy variable FePar when parents

are reported as the main source of financial education. In line with the transgenerational

effect found elsewhere (Barboza, Smith, & Pesek 2016), we argue that if parents are the

main source of financial education, parents’ financial experiences serve as a mechanism to

develop knowledge spillover effects and possibly avoiding painful self-experiences that

could result in lower levels of financial-based anxiety and translate onto better credit card

repayment patterns.

Based on the number of correct answers to the financial literacy questions, we first pro-

ceed to construct the cumulative number of Total Correct Answers. The number of correct

answers is thus the actual level of financial literacy (that we called Financial literacy rate),

as it is presented in most of the extant literature. We take particular interest in the objective

separation between Correct, Incorrect, and Do Not Know answers. Therefore, we construct

the variable total number of incorrect answers given by each individual, and call it Financial

Illiteracy Level A. Finally, we construct the total number of answers under the Do Not

Know category. As noted earlier, we want to emphasize that answering incorrectly is differ-

ent than answering Do Not Know. Correspondingly, answering DNK is labeled as Financial

Illiteracy Level B. We argue that those answering incorrectly (Incorrect) have attempted to

answer assuming that they have an adequate level of knowledge; however, clearly failing to

achieve a correct answer. On the other hand, those answering I Do Not Know (DNK) openly

acknowledge that they are not prepared to even attempt answering recognizing a higher level

of financial illiteracy with no fear to state it as such. The fundamental difference is that the

DNK group may be less ready to engage in financial decision-making processes, as their

self-awareness indicates a clear lack of preparation. Conversely, those responding incor-

rectly (Level A) implicitly assume knowing the answer but failing. The potential implica-

tions of the separation of answering not correctly into these two distinctive groups could be

relevant in understanding the implications of different levels of financial illiteracy. We also

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 257



provide a decomposition of answers (under the three categories) by question as these relate

to different financial concepts.

3.2. Descriptive statistics

Table 1 presents a summary of results from the questionnaire. These results indicate that a

large majority of individuals answer correctly questions on Compound interest, Inflation and

Purchasing Power. A large number of students were not capable of answering correctly the

question regarding Inheritance. In terms of the inflation question, a relative similar proportion of

individuals answer DNK or incorrectly. For the risk question a significant number of students

responded DNK. In general, the data appears to indicate that DNK answering is a prevalent

issue. In addition, a large number of student answer incorrectly. The combination of both levels

of financial illiteracy confirms the prevalence of wide spread lack of financial knowledge.

When we look at the cumulative answering (see Table 2, below), we observe that 5% of

the population are not capable of answering any question correctly. By the same token, only

12% of the individuals are capable of answering all questions correctly. In addition, a large

proportion of individuals answer DNK to one and two questions. When combined with those

also answering incorrectly one to two questions, we are able to observe a large proportion of

the population struggling with about 50% of the questions asked. In general, this evidence

indicates that financial illiteracy is highly present among the population under study. This

statistic is in line with previous literature and with most recent valuations made across the

U.S. population (see Lusardi & Mitchell 2014). Table 3 disassembles financial literacy rate

by demographics. The interpretation of the results thus indicates that females are more likely

to answer correctly since people able to provide five correct answers are mainly females

(62.5%); this is a result that contradicts the empirical evidence found in the literature for the

United States (Borden et al. 2008; Chen & Volpe, 2002; Danes & Tahira, 1987; Lusardi,

Table 2 Financial (il)literacy indicators (total answers)

No. of answers Correct Do Not Know Incorrect

0 4.9% 52.9% 22.1%
1 10.0% 31.4% 41.9%
2 18.0% 9.2% 25.9%
3 27.4% 3.1% 7.6%
4 27.7% 2.2% 1.5%
5 11.9% 1.0% 1.0%

Table 1 Financial (il)literacy indicators per question

Correct Do Not Know Incorrect

Compound interest 70.3% 6.0% 23.7%
Inflation 66.8% 16.1% 17.1%
Inheritance 33.5% 3.1% 63.3%
Purchasing power 77.2% 8.2% 14.6%
Risk 51.0% 40.1% 8.9%

258 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



Mitchell, & Curto, 2010; Markovich & DeVaney, 1997) where it is argued that females pos-

sess lower levels of financial literacy. However, outside the United States, this is not a stand-

ard evidence (Bongini et al, 2016; Koshal et al., 2008; Wagland & Taylor, 2009) when

specifically studying Business Students.

In addition, older students are more likely to answer correctly, as well as upper classmen

or classwomen. Among different academic status graduate students provide at least three

correct answers and the 33% of students answers correctly to all the financial literacy ques-

tions. This interpretation of the results speaks in favor of the educational process. As

expected, one can argue that attending college should increase knowledge in field specific

subjects, financial literacy being one of them. Finally, the evidence does indicate that minor-

ities are at a disadvantage and more likely to answer incorrectly or do not know. This last pi-

ece is in agreement with the extant literature.

Table 4 below presents some basic descriptive statistics on the variables used in the em-

pirical estimation section. A full description of each variable and its corresponding coding

could also be found there. With these considerations in mind, we then proceed to outline the

model specification and the corresponding model estimations expectations and restrictions.

3.3. The model

As we research the effects of personal traits and financial literacy on financial behavior,

the basic model description has the following general specification:

y�i ¼ x
0
ib þ « i (1)

where « i are assumed independent and identically distributed random variables as usual, x
0
i

is the matrix of explanatory variables (financial (il)literacy and personality traits), b is the

vector of coefficients to be estimated, and y�i is unobserved yet described by the answers to

our survey questions relating to credit card repayment habits.

In fact, we code students’ responses on credit card repayment capability using a discrete

categorical scale as follows: 1 = pay if full every month, 2 = pay some in full and other only

Table 3 Financial literacy rate by demographics

Financial literacy rate Gender mean AS mean Age Race

0 0.333 1.939 20.152 0.697
0.2 0.313 2.119 20.373 0.731
0.4 0.430 2.421 20.554 0.777
0.6 0.492 2.415 21.196 0.837
0.8 0.478 2.595 21.978 0.892
1 0.625 2.838 23.213 0.938

Source: Demographic variables are defined as follows. Gender G is a dummy variable taking value of 1 if

female and 0 otherwise. Academic Status-AS is the 1 = Freshman, 2 = Sophomore, 3 = Junior, 4 = Senior, and

5 = Grad Student. Race is 1 if White and 0 if minority. Age is expressed in years at the time of the survey.

FLITRATE is the percentage of correct answers provided, based on a five question questionnaire as presented

in Figure 1.

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 259



T
ab
le

4
D
es
cr
ip
ti
v
e
st
at
is
ti
cs

o
f
v
ar
ia
b
le
s
b
y
ca
te
g
o
ry
,
d
es
cr
ip
ti
o
n
,
an
d
co
d
in
g

C
at
eg
o
ry

D
es
cr
ip
ti
o
n

C
o
d
e

M
ed
ia
n

M
ea
n

S
D

M
ax

M
in

O
b
s

D
em

o
g
ra
p
h
ic
s

A
ca
d
em

ic
st
at
u
s
(F

=
1
,
S
=
2
,
J
=
3
,
S
r
=
4
,
G
ra
d
=
5
)

A
S

3
2
.6
7
3

1
.3
1
1

5
1

1
5
6

G
en
d
er

(f
em

al
e
=
1
)

G
0

0
.4
5
5

0
.5
1
2

2
0

1
5
6

A
g
e

A
G
E

2
0

2
0
.6
0
3

2
.8
5
3

4
2

1
8

1
5
6

R
ac
e
(m

in
o
ri
ty

=
0
)

R
A
C
E

1
0
.7
8
8

0
.4
1
0

1
0

1
5
6

F
in
an
ci
al
ed
u
ca
ti
o
n
an
d
li
te
ra
cy

P
ar
en
ts
ar
e
m
ai
n
so
u
rc
e
o
f
fi
n
an
ci
al
ed
u
ca
ti
o
n

F
E
P
A
R

1
0
.6
2
8

0
.4
8
5

1
0

1
5
6

N
u
m
b
er
s
o
f
co
rr
ec
t
an
sw

er
s

T
o
tC
o
r

3
.0
0

2
.8
7
8

1
.2
6
2

5
0

1
5
6

N
u
m
b
er
s
o
f
in
co
rr
ec
t
an
sw

er
s

T
o
t_
In
c

1
.0
0

1
.3
9
1

1
.0
3
2

5
0

1
5
6

N
u
m
b
er

o
f
“D

o
N
o
t
K
n
o
w
”
an
sw

er
s

T
o
tD
K

1
.0
0

0
.7
3
1

0
.8
6
8

4
0

1
5
6

B
eh
av
io
ra
l

L
ev
el
o
f
an
x
ie
ty

o
n
re
p
ay
m
en
t
ca
p
ac
it
y

A
n
x

2
2
.0
1
9

1
.1
1
6

5
1

1
5
6

F
re
q
u
en
cy

o
f
p
ay
m
en
t
o
n
cr
ed
it
ca
rd

R
fr
eq

1
1
.6
2
2

1
.0
1
2

4
1

1
5
6

S
u
rp
ri
se
d
o
n
C
C
b
al
an
ce

le
v
el

S
u
rp

1
0
.8
7
8

0
.9
1
8

3
0

1
5
6

F
re
q
u
en
cy

o
f
o
v
er
sp
en
d
in
g

O
v
er
sp
d

0
0
.4
6
2

0
.8
2
2

3
0

1
5
6

N
u
m
b
er

o
f
cr
ed
it
ca
rd

N
C
C

1
1
.5
6
4

1
.0
6
1

6
1

1
5
6

S
o
u
rc
e:

D
em

o
g
ra
p
h
ic
v
ar
ia
b
le
s
ar
e
d
efi
n
ed

as
fo
ll
o
w
s.
G
en
d
er

G
is
a
d
u
m
m
y
v
ar
ia
b
le
ta
k
in
g
v
al
u
e
o
f
1
if
fe
m
al
e
an
d
0
o
th
er
w
is
e.
A
ca
d
em

ic
S
ta
tu
s-
A
S
is

th
e
1
=
F
re
sh
m
an
,
2
=
S
o
p
h
o
m
o
re
,
3
=
Ju
n
io
r,
4
=
S
en
io
r,
an
d
5
=
G
ra
d
S
tu
d
en
t.
R
ac
e
is
1
if
W
h
it
e
an
d
0
if
m
in
o
ri
ty
.
A
ge

is
ex
p
re
ss
ed

in
y
ea
rs
at

th
e
ti
m
e

o
f
th
e
su
rv
ey
.
F
in
an
ci
al

L
it
er
ac
y
,
E
d
u
ca
ti
o
n
,
an
d
re
la
te
d
v
ar
ia
b
le
s
in
cl
u
d
e
th
e
fo
ll
o
w
in
g
.
T
ot
C
O
R

is
th
e
n
u
m
b
er

o
f
fi
n
an
ci
al

li
te
ra
cy

q
u
es
ti
o
n
an
sw

er
s
co
r-

re
ct
ly
;
T
ot
D
K

is
th
e
n
u
m
b
er

o
f
an
sw

er
s
w
it
h
a
"D

o
N
o
t
K
n
o
w
"
an
sw

er
;
an
d
T
ot
_I
N
C

is
th
e
n
u
m
b
er

o
f
in
co
rr
ec
t
an
sw

er
s.
F
o
r
ea
ch

in
d
iv
id
u
al

th
e
su
m

o
f

T
O
T
C
O
R
+
T
O
T
D
K
+
T
O
T
_
IN

C
=
5
.
T
h
e
F
in
an
ci
al

L
it
er
ac
y
R
at
e-
F
L
itR

at
e
m
ea
su
re
s
th
e
p
er
ce
n
ta
g
e
o
f
co
rr
ec
t
an
sw

er
s
o
u
t
o
f
th
e
fi
v
e
li
te
ra
cy

k
n
o
w
le
d
g
e

q
u
es
ti
o
n
s.
A
n
x
ie
ty
-A
nx

is
th
e
le
v
el

o
f
an
x
ie
ty

re
p
o
rt
ed

o
n
a
sc
al
e
fr
o
m

1
=
n
o
an
x
ie
ty

to
5
=
ex
tr
em

el
y
an
x
io
u
s
in

re
la
ti
o
n
to

h
is
o
r
h
er

ca
p
ac
it
y
to

re
p
ay

y
o
u
r
cr
ed
it
ca
rd

m
o
n
th
ly

b
il
l.
F
E
P
ar

ta
k
es

a
v
al
u
e
o
f
1
if
p
ar
en
ts

ar
e
th
e
m
ai
n
so
u
rc
e
o
f
fi
n
an
ci
al

ed
u
ca
ti
o
n
,
0
o
th
er
w
is
e.

R
F
re
q
d
efi
n
es

th
e
re
p
ay
m
en
t

b
eh
av
io
r
o
n
cr
ed
it
ca
rd
s.
T
h
e
v
ar
ia
b
le

ta
k
es

th
e
v
al
u
es

o
f
1
=
P
ay

in
F
u
ll
;
2
=
P
ay

m
o
re

th
an

m
in
im

u
m
,
b
u
t
ca
rr
y
m
o
n
th
-t
o
-m

o
n
th

b
al
an
ce
;
3
=
P
ay

o
ff

so
m
e
cr
ed
it
ca
rd
s,
b
u
t
p
ay
s
m
in
im

u
m

o
n
th
e
re
st
;
4
=
P
ay

m
in
im

u
m

in
al
l;
5
=
P
ay

le
ss

th
an

m
in
im

u
m

in
al
l.
6
=
1
=
P
ar
en
ts
p
ay

in
fu
ll
w
il
l
b
e
th
e
as
su
m
p
-

ti
o
n
.
Su

rp
ri
se
d
d
efi
n
es

th
e
le
v
el
o
f
su
rp
ri
se

th
e
in
d
iv
id
u
al
re
p
o
rt
s
o
n
h
o
w
h
ig
h
o
f
a
b
al
an
ce

is
o
n
th
e
m
o
n
th
ly

st
at
em

en
t.
T
h
e
v
ar
ia
b
le
ta
k
es

th
e
v
al
u
es

o
f
0
=

n
ev
er
,
1
=
R
ar
el
y
,
2
=
S
o
m
et
im

es
,
3
=
F
re
q
u
en
tl
y
.
O
ve
rS
pd

d
efi
n
es

th
e
b
eh
av
io
r
o
f
th
e
in
d
iv
id
u
al

re
g
ar
d
in
g
th
e
in
d
iv
id
u
al

k
n
o
w
in
g
th
at

sh
e
o
r
h
e
u
se

th
e

cr
ed
it
ca
rd

k
n
o
w
in
g
th
at
sh
e
o
r
h
e
d
id

n
o
t
h
av
e
m
o
n
ey

to
p
ay

w
h
en

th
e
b
il
l
ca
m
e
d
u
e.
T
h
e
v
ar
ia
b
le
ta
k
es

th
e
v
al
u
es

o
f
0
=
n
ev
er
,
1
=
ra
re
ly
,
2
=
so
m
et
im

es
,

3
=
fr
eq
u
en
tl
y
.
F
in
al
ly
,
N
C
C
is
a
co
u
n
t
v
ar
ia
b
le
m
ea
su
ri
n
g
th
e
n
u
m
b
er

o
f
cr
ed
it
ca
rd
s
an

in
d
iv
id
u
al
p
o
ss
es
se
s.

260 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



minimum required, 3 = pay more than minimum required but not in full, 4 = pay Minimum

required, 5 = pay less than Minimum required. Therefore, our dependent variable (Rfreq)
takes discrete values along the scale 1 to 5. Notice that Rfreq of values ranging from 2 to 5

results in accumulation of month-to-month balance, with the risk of incurring high financial

costs related to interest rates and possibly other assessed fees.

In consideration of the ordinal nature of the dependent variable, the most appropriate model

to use is an ordered Probit model. The main difference among the models regards the financial

literacy variable. Due to our interest to test for both knowledge and lack of knowledge, we

decompose the responses given by individuals between correct, incorrect and do not know clas-

sifications, per our discussion in the previous section. In the first model we consider as

Financial literacy variable the number of correct answers. In the second model we consider the

number of questions where the students said that he does not know the answers, and in the last

model we consider the number of incorrect answers. Furthermore, because each of the financial

literacy questions measures knowledge of different financial issues (more or less directly

related to our dependent variable, i.e., credit card repayment), we also proceed to conduct esti-

mations by question.7 Thus, we propose to estimate the impact of each financial issue on the fi-

nancial behavior under investigation: our intuition is that some questions may be more relevant

(have a larger impact) and the separation of effects may yield relevant results, when under-

standing credit card repayment patterns.

In all the specifications, we consider all the personality trait variables and the presence of

financial education from parents; plus, as control variable we include the total number of

credit cards possessed by our sample students (NCC). This latter variable proves to be very

much correlated to the usual demographics variables (age, gender, race, and academic sta-

tus) used as controls; we choose to use NCC as our single control variable because the num-

ber of observations suggested to contain the number of covariates.8

We then proceed to incorporate interaction effects between the overall scores of financial

literacy, financial illiteracy Level A (Incorrect) and Level B (DNK), in relation to the finan-

cial education from parents and personality trait variables. In addition, and due to the possi-

bility of endogeneity in the data, we conduct Hausman testing as a robustness indicator. The

argument regarding endogeneity is justified as the possibility of a self-selection problem in

terms of the characteristics of those selecting to apply and obtain a credit card. However, the

counter argument indicates that in the U.S. market, young adults need to apply for credit

cards as a requirement to begin building their credit history and create a credit score. Due to

this apparent controversy, we apply the endogeneity testing.

4. Results and discussion

Estimations are presented in Table 5 where each of the three models is further decom-

posed to study the impact of each financial issue included in the financial literacy index on

credit card repayment behavior of individuals.

When looking at the coefficients (sign and magnitude) of the total number of questions

answered (Correctly, Incorrectly, and DNK, respectively) we observe that correct/DKN hold

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 261



both a negative sign, while only Correct being statistically significant at the 10% level of

confidence. The negative sign indicates that higher levels of literacy and/or declaring DKN

result in better repayment patterns. Higher financial literacy having a positive impact on

repayment is an expected outcome (confirming Hypothesis 1); however, the second compo-

nent (DNK) is clearly a puzzling result. Notice that higher levels of financial literacy are

only marginally statistically significant at the 10%. Regarding incorrect answering (financial

illiteracy) the result is statistically significant and with the expected positive sign. This result

supports the hypothesis that lack of financial literacy creates a negative effect on individuals’

financial performance, resulting in an accumulating month-to-month balance in their credit

card debt (Hancock et al., 2013; Robb, 2011; Xiao et al., 2011). Furthermore, results in

Table 5 Consumer behavior and financial decision making process with credit card repayment as dependent

variable

Model 1
Correct answers

Model 2
Incorrect answers

Model 3
Do Not Know answers

1.1 1.2 2.1 2.2 3.1 3.2

Financial educ parents �0.939 �0.973 �0.930 �0.969 �0.912 �0.888
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***

Anxiety 0.143 0.124 0.158 0.141 0.163 0.159
(0.18) (0.25) (0.13) (0.18) (0.12) (0.11)

Surprised 0.143 0.130 0.141 0.133 0.182 0.175
(0.28) (0.33) (0.25) (0.27) (0.16) (0.17)

OverSpending 0.533 0.533 0.506 0.497 0.509 0.511
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***

NCC 0.161 0.156 0.164 0.169 0.135 0.144
(0.08)* (0.10)* (0.04)** (0.04)** (0.13) (0.08)*

Financial literacy questions

Total answers �0.150 0.222 �0.022
(0.10)* (0.03)** (0.88)

Compound interest �0.017 0.020 �0.051
(0.95) (0.94) (0.91)

Inflation �0.385 0.501 �0.014
(0.12) (0.05)** (0.97)

Inheritance �0.107 0.163 �0.184
(0.68) (0.47) (0.41)

Purchasing power �0.046 0.118 �0.125
(0.87) (0.72) (0.82)

Risk �0.139 0.297 0.022
(0.55) (0.35) (0.92)

Pseudo R2 0.209 0.212 0.212 0.217 0.197 0.199
Obs 158 157 156 156 156 156
LR statistic 65.06 65.71 65.59 67.29 61.10 61.66
Probability (LR stat) (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
Akaike info criterion 1.674 1.726 1.681 1.722 1.710 1.758
Schwarz criterion 1.848 1.979 1.857 1.976 1.886 2.012
Hannan-Quinn criterion 1.744 1.829 1.753 1.825 1.782 1.861

Note: ***, **, * statistically significant at the 1%, 5%, and 10% level, respectively.

262 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



Models 2.1 and 3.1 clearly indicate that individuals responding incorrectly are significantly

more likely to incur in poor credit card repayment behavior. The estimated coefficient is

larger than the Correct counterpart and holds a positive sign and has a higher level of statisti-

cal significance. Perhaps the most revealing result of the estimations in Table 5 is the support

to our third hypothesis that Incorrect answering and DNK answering have different implica-

tions on Credit Card repayment patterns. That is, the common practice in previous research

to cluster any answering that is not correct as incorrect, seems to be not an appropriate way

to understand the effects of different levels (A and B) of financial illiteracy. In summary, our

findings confirm Hypothesis 1: higher levels of financial literacy (measured as higher num-

ber of correct answers) lead to better repayment rates in credit card debt but also Hypothesis

3 individuals with self-assessed lower level of financial literacy (DKN answers) display

diverse behaviors (credit card repayment patterns) as opposed to individuals who do not pos-

sess basic financial literacy (wrong answers).

Our findings also confirm Hypothesis 2 and to a lesser extent Hypothesis 4. First, we

observe that the variable FePar, being the main source of financial education, is highly sig-

nificant in all models and negatively associated with repayment habits. This result indicates

that individuals that receive their education from parents are more likely to have better

repayment behavior on her or his credit card.9 This result also confirms the importance of

parental involvement in kids overall financial education aspects. Our findings confirm

Hypothesis 2: the higher the transgenerational transmission of financial education from

parents to children results in improved repayment in credit cards.

With respect to personality traits also control for the presence of personality trait

(Surprised, Anxiety, and Overspending) our estimates highlight that individuals’ capability

to repay their credit cards is driven by the use factor and only marginally by the knowledge

component. In all the models, the role of Overspending impacts negatively the capability to

repay credit card debt. The other variables, Anxiety and Surprised, are not statistically sig-

nificant. As a consequence, our Hypothesis 4 is partially confirmed, at least with respect to

present-bias preferences. Recall that overspending is the response to individuals actually

buying when they knew a priori that he or she would not have enough money to repay the

credit card at the end of the billing cycle. We argue, thus, that the overspending coefficient

serves as a measure of present bias behavior as well as an indicator of lack of self-control.

This lack of self-control becomes a materialized purchase due to the availability of credit

cards. If the individual were not to have a credit card, he or she could not complete the pur-

chase. While this may seem obvious, it is relevant to point out that this behavior clearly

results in increased financial costs of the purchases, as individuals know ahead of time that

they will incur in a rolling debt. Notice also that the estimated coefficient value for

Overspending is consistent across all alternative models, even when controlling for financial

(il)literacy.

Finally, the NCC yields the expected positive sign indicating that individuals having

more credit cards are more likely to hold a month-to-month balance: in our estimations an

increase in the number of credit cards made the capacity of repayment worse. As hypothe-

sized earlier, a higher number of credit cards may be the result of maxing out of credit on

one card and consequently apply for more; or using new credit cards to transfer balances (at

promotional interest rates).

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 263



Due to the possible gains in understanding individuals’ behavior on credit card repayment

patterns we proceed with a further decomposition of the financial (il)literacy by question.

First, notice that under the Correct model (1.2), all five questions have the expected negative

sign (answering correctly results in better credit card repayment patterns) even though none

individually is statistically significant. In contrast, model (2.2), yields opposite signs in all

five questions in relation to the Correct model version. These differences in results are con-

sistent with expectations, and apparently strong enough to explain the overall positive sign

of the total number of incorrect answers. Lastly, the DNK model (3.2) provides results that

are more congruent with the correct estimations than the incorrect estimations; thus provid-

ing further evidence in favor of the hypothesis that clustering incorrect with do not know is

not an appropriate way to understand different levels of financial illiteracy.

However, given the results from the alternative estimations in Table 5, we confirm that

the role of financial literacy, or lack of, appears not to dominate behavioral/personal traits

(Hypothesis 5). This is to say, that financial decisions seem to be made primarily along pat-

tern of behavioral traits and financial education from parents; and in this context only mar-

ginally ameliorated by financial education. Assuming that these results are consistent and

robust (after accounting for possible sample issues) then, they indicate that while financial

literacy is an important or fundamental element for all individuals to learn, behavioral varia-

bles drive the accumulation of credit card debt and result in less than optimal repayment. In

this respect, we confirm the findings of Xiao et al. (2011) who, applying the Theory of

Planned behavior to investigate risky credit behavior among college students, found that be-

havioral intentions were the single most important factor in whether students make responsi-

ble credit decisions (risky borrowing behavior, risky paying behavior, and holding credit

card debt). Under these conditions, it then becomes relevant to explore the drivers of

overspending.

Given the robustness of the personal trait variables, in conjunction with the lack of strong

statistical significance of the financial literacy estimates, several possible scenarios come to

mind. For instance, we can argue as mentioned above that behavioral traits are just too

strong and clearly individuals have a hard time controlling them. Particularly, issues relating

to present-bias, preferences and gaps in mental accounting are strong and present. In addi-

tion, one can argue that this behavior may be prevented or ameliorated with early interven-

tion in the form of exposure to financial literacy. That is, there is the possibility that students

in our sample may be receiving financial education too late in life, and thus personality traits

are harder to counterbalance. This interpretation is compliant also with the relevance of fi-

nancial education transmitted by parent in ameliorating the credit card repayment.

4.1. Interaction effects

The results thus far have provided very useful information in the advancing the under-

standing of the relationship between financial (il)literacy and personality traits into credit

card repayment decisions. However, the relationships may also be shaped by the presence of

interaction effects deriving from Financial (il)literacy and the behavioral or personality vari-

ables. As noted, we uncover that personality traits are dominant in the decision individuals

264 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



make, particularly overspending (+) and Financial Education deriving from parents (�). In
this context, and given the strong interest in the effects that financial (il)literacy may have

on financial decision making, we proceed to compute and report the alternative models

including interaction effects between Financial (il)literacy and financial education from par-

ent and all four behavioral variables. The results are presented in Tables 6, 6a, and 6b, for

Correct, Incorrect (Level A), and DNK (Level B), respectively.

The first set of estimations in Table 6 corresponds to the Financial Literacy-Correct

answering sample. In these estimations, we observe that all previous results hold, overspend-

ing having a negative effect and worsening repayment, increased the number of credit cards

also worsening repayment. FEPAR has a significant effect improving credit card repayment

when the level of financial literacy is low whereas higher financial literacy resulting in better

repayment patterns and lower chance of accumulation month-to-month debt when students

did not receive financial education from their parents. In addition, the interaction effect of

Financial Literacy and FEPAR has a positive and statistically significant coefficient: when

students receive education from parents the positive effect of financial literacy disappears.

Notice also that the Akaike information criterion (AIC) for the interaction model is lower

than the corresponding model in Table 5 without the interaction. This is to say that the inter-

action model specification is preferred. In addition, none of the other interaction effects

prove to be statistically significant, and therefore the models are inferior.

For the second set of interaction models, reported in Table 6a, for the Level of Financial

Illiteracy Type A, we also observe results that are in accordance to the estimations reported

in Table 6. However, under the interaction effect Model 1, the level of FEPAR is now not

statistically significant, despite holding the expected sign. When the financial illiteracy is

low the role of financial education from parents is not significant in improving the credit

card repayment. The increase in the number of total incorrect answers results in credit card

repayment worsening. When we interact Financial illiteracy and education from parent the

coefficient is negative and statistically significant the 5% level of confidence. Our intuitive

interpretation of this effect is that for student characterized by high level of financial illiter-

acy the education received from parents is fundamental in improving the credit card repay-

ment. Notice that as it was the case for the Financial Literacy models, all other interaction

effects are not statistically significant and the AIC values are also higher. We argue that the

Model 1 is the preferred model.

For the third estimations, where Financial Illiteracy is measured through DNK answers

(Table 6b), we observe that the interaction effects model for FEPAR and Financial Illiteracy

Level B is significant and holds a negative sign. Our intuitive rationale for the coefficient in

this model is that students that mainly answered DNK are able to improve their credit card

repayment habits if they received proper financial education from parents. In addition, it is

relevant to point out that the Financial Illiteracy coefficient is not statistically significant (as

it was in Model 3.1–Table 5), but it now present a reversal of sign. In addition, the AIC for

Model 1–Table 6B is lower than the corresponding model in Table 5, without interaction

effects. Thus, we preferred the interaction model specification in this regard. As it was with

other models, interaction effects for all other variables are not statistically significant. Our

findings do not confirm completely Hypothesis 5. We do not detect that positive effects from

higher levels of financial literacy could be overpowered by personal traits. Furthermore,

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 265



T
ab
le

6
C
o
n
su
m
er

b
eh
av
io
r
w
it
h
cr
ed
it
ca
rd

re
p
ay
m
en
t
fr
eq
u
en
cy

as
d
ep
en
d
en
t
v
ar
ia
b
le

(C
o
rr
ec
t
an
sw

er
s)

C
o
rr
ec
t
an
sw

er
s
m
o
d
el
s
w
it
h
in
te
ra
ct
io
n
ef
fe
ct
s

1
2

3
4

5

F
in
an
ci
al
ed
u
c
p
ar
en
ts

�2
.1
4
5
(0
.0
0
1
)*
*
*

�0
.9
3
4
(0
.0
0
1
)*
*
*

�0
.9
3
6
(0
.0
0
1
)*
*
*

�0
.9
2
7
(0
.0
0
1
)*
*
*

�0
.9
3
9
(0
.0
0
1
)*
*
*

A
n
x
ie
ty

0
.1
2
5
(0
.2
4
)

0
.2
3
8
(0
.4
0
)

0
.1
4
0
(0
.1
8
)

0
.1
4
4
(0
.1
7
)

0
.1
4
4
(0
.1
7
)

S
u
rp
ri
se
d

0
.1
2
0
(0
.3
7
)

0
.1
3
5
(0
.3
1
)

0
.1
4
7
(0
.2
7
)

0
.3
9
5
(0
.1
3
)

0
.1
4
4
(0
.2
8
)

O
v
er
S
p
en
d
in
g

0
.5
7
7
(0
.0
0
1
)*
*
*

0
.5
3
3
(0
.0
0
1
)*
*
*

0
.4
3
9
(0
.1
0
)*

0
.5
2
8
(0
.0
0
1
)*
*
*

0
.5
3
0
(0
.0
0
1
)*
*
*

N
C
C

0
.1
7
1
(0
.0
6
)*

0
.1
6
2
(0
.0
8
)*

0
.1
5
5
(0
.0
9
)*

0
.1
6
7
(0
.0
7
)*

0
.0
8
1
(0
.7
5
)

F
in
an
ci
al
li
te
ra
cy

q
u
es
ti
o
n
s

T
o
ta
l
an
sw

er
s

�0
.4
1
0
(0
.0
0
1
)*
*
*

�0
.0
8
6
(0
.6
7
)

�0
.1
7
4
(0
.1
1
)

�0
.0
4
2
(0
.7
5
)

�0
.1
9
1
(0
.2
1
)

In
te
ra
ct
io
n
ef
fe
ct
s
o
f
fi
n
an
ci
al
li
te
ra
cy

w
it
h

P
ar
en
ts
fi
n
an
ci
al
ed
u
ca
ti
o
n

0
.4
2
8
(0
.0
2
)*
*

A
n
x
ie
ty

�0
.0
3
2
(0
.7
2
)

O
v
er
sp
en
d
in
g

0
.0
3
5
(0
.6
9
)

S
u
rp
ri
se
d

�0
.0
9
7
(0
.2
6
)

N
C
C

0
.0
2
6
(0
.7
3
)

O
b
s

1
5
8

1
5
8

1
5
8

1
5
8

1
5
8

P
se
u
d
o
R
2

0
.2
2
7

0
.2
0
9

0
.2
0
9

0
.2
1
3

0
.2
0
9

L
R
st
at
is
ti
c

7
0
.6
9
2

6
5
.1
9
2

6
5
.2
2
3

6
6
.3
4
6

6
5
.1
7
5

P
ro
b
ab
il
it
y
(L
R
st
at
)

(0
.0
0
1
)*
*
*

(0
.0
0
1
)*
*
*

(0
.0
0
1
)*
*
*

(0
.0
0
1
)*
*
*

(0
.0
0
1
)*
*
*

A
k
ai
k
e
in
fo

cr
it
er
io
n

1
.6
5
1

1
.6
8
5

1
.6
8
5

1
.6
7
8

1
.6
8
6

S
ch
w
ar
z
cr
it
er
io
n

1
.8
4
4

1
.8
7
9

1
.8
7
9

1
.8
7
2

1
.8
7
9

H
an
n
an
-Q

u
in
n
cr
it
er
io
n

1
.7
2
9

1
.7
6
4

1
.7
6
4

1
.7
5
7

1
.7
6
4

N
o
te
:
*
*
*
,
*
*
,
*
st
at
is
ti
ca
ll
y
si
g
n
ifi
ca
n
t
at
th
e
1
%
,
5
%

an
d
1
0
%

le
v
el
,
re
sp
ec
ti
v
el
y
.

266 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



T
ab
le

6
a

C
o
n
su
m
er

b
eh
av
io
r
w
it
h
cr
ed
it
ca
rd

re
p
ay
m
en
t
fr
eq
u
en
cy

as
d
ep
en
d
en
t
v
ar
ia
b
le

(I
n
co
rr
ec
t
an
sw

er
s)

F
in
an
ci
al
il
li
te
ra
cy

L
ev
el
A

In
co
rr
ec
t
an
sw

er
s
m
o
d
el
s
w
it
h
in
te
ra
ct
io
n
ef
fe
ct
s

1
2

3
4

5

F
in
an
ci
al
ed
u
c
p
ar
en
ts

�0
.2
8
3
(0
.4
5
)

�0
.9
4
0
(0
.0
0
1
)*
*
*

�0
.9
2
2
(0
.0
0
1
)*
*
*

�0
.9
4
2
(0
.0
0
1
)*
*
*

�0
.9
1
9
(0
.0
0
1
)*
*
*

A
n
x
ie
ty

0
.1
3
2
(0
.2
0
)

0
.2
2
2
(0
.1
3
)

0
.1
4
9
(0
.1
5
)

0
.1
6
3
(0
.1
1
)

0
.1
6
0
(0
.1
3
)

S
u
rp
ri
se
d

0
.1
4
2
(0
.2
5
)

0
.1
4
9
(0
.2
3
)

0
.1
4
4
(0
.2
4
)

�0
.0
5
1
(0
.7
9
)

0
.1
4
6
(0
.2
3
)

O
v
er
S
p
en
d
in
g

0
.5
3
7
(0
.0
0
1
)*
*
*

0
.5
0
6
(0
.0
0
1
)*
*
*

0
.6
1
2
(0
.0
0
1
)*
*
*

0
.4
9
8
(0
.0
0
1
)*
*
*

0
.4
8
8
(0
.0
0
1
)*
*
*

N
C
C

0
.1
8
5
(0
.0
2
)*
*

0
.1
6
4
(0
.0
4
)*
*

0
.1
5
4
(0
.0
5
)*
*

0
.1
6
7
(0
.0
4
)*
*

0
.2
7
1
(0
.0
1
)*
*
*

F
in
an
ci
al
li
te
ra
cy

q
u
es
ti
o
n
s

T
o
ta
l
an
sw

er
s

0
.5
3
0
(0
.0
0
1
)*
*
*

0
.3
4
8
(0
.2
1
)

0
.2
8
0
(0
.0
4
)*
*

0
.0
7
6
(0
.6
3
)

0
.3
6
8
(0
.0
3
)*
*

In
te
ra
ct
io
n
ef
fe
ct
s
o
f
fi
n
an
ci
al
li
te
ra
cy

w
it
h

P
ar
en
ts
fi
n
an
ci
al
ed
u
ca
ti
o
n

�0
.4
3
4
(0
.0
5
)*
*

A
n
x
ie
ty

�0
.0
5
4
(0
.6
0
)

O
v
er
sp
en
d
in
g

�0
.0
6
1
(0
.4
5
)

S
u
rp
ri
se
d

0
.1
2
3
(0
.2
2
)

N
C
C

�0
.0
8
0
(0
.1
6
)

O
b
s

1
5
6

1
5
6

1
5
6

1
5
6

1
5
6

P
se
u
d
o
R
2

0
.2
2
3

0
.2
1
3

0
.2
1
3

0
.2
1
6

0
.2
1
5

L
R
st
at
is
ti
c

6
9
.1
7
2

6
5
.8
7
0

6
6
.0
5
0

6
6
.9
8
1

6
6
.5
9
0

P
ro
b
ab
il
it
y
(L
R
st
at
)

(0
.0
0
1
)*
*
*

(0
.0
0
1
)*
*
*

(0
.0
0
1
)*
*
*

(0
.0
0
1
)*
*
*

(0
.0
0
1
)*
*
*

A
k
ai
k
e
in
fo

cr
it
er
io
n

1
.6
7
1

1
.6
9
2

1
.6
9
1

1
.6
8
5

1
.6
8
8

S
ch
w
ar
z
cr
it
er
io
n

1
.8
6
7

1
.8
8
8

1
.8
8
7

1
.8
8
1

1
.8
8
3

H
an
n
an
-Q

u
in
n
cr
it
er
io
n

1
.7
5
1

1
.7
7
2

1
.7
7
1

1
.7
6
5

1
.7
6
7

N
o
te
:
*
*
*
,
*
*
,
*
st
at
is
ti
ca
ll
y
si
g
n
ifi
ca
n
t
at
th
e
1
%
,
5
%
,
an
d
1
0
%

le
v
el
,
re
sp
ec
ti
v
el
y
.

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 267



T
ab
le

6
b

C
o
n
su
m
er

b
eh
av
io
r
w
it
h
cr
ed
it
ca
rd

re
p
ay
m
en
t
fr
eq
u
en
cy

as
d
ep
en
d
en
t
v
ar
ia
b
le

(D
o
N
o
t
K
n
o
w
an
sw

er
s)

F
in
an
ci
al
il
li
te
ra
cy

L
ev
el
B

D
o
N
o
t
K
n
o
w
an
sw

er
s
m
o
d
el
s
w
it
h
in
te
ra
ct
io
n
ef
fe
ct
s

1
2

3
4

5

F
in
an
ci
al
ed
u
c
p
ar
en
ts

�0
.5
6
0
(0
.0
5
)*
*

�0
.9
1
4
(0
.0
0
1
)*
*
*

�0
.9
1
6
(0
.0
0
1
)*
*
*

�0
.9
1
3
(0
.0
0
1
)*
*
*

�0
.9
0
8
(0
.0
0
1
)*
*
*

A
n
x
ie
ty

0
.1
8
3
(0
.0
9
)*

0
.1
3
0
(0
.3
7
)

0
.1
6
7
(0
.1
2
)

0
.1
6
3
(0
.1
2
)

0
.1
6
5
(0
.1
2
)

S
u
rp
ri
se
d

0
.1
4
7
(0
.2
7
)

0
.1
8
0
(0
.1
6
)

0
.1
8
5
(0
.1
5
)

0
.1
8
4
(0
.2
8
)

0
.1
8
4
(0
.1
5
)

O
v
er
S
p
en
d
in
g

0
.5
1
9
(0
.0
0
1
)*
*
*

0
.5
1
0
(0
.0
0
1
)*
*
*

0
.5
5
1
(0
.0
0
1
)*
*
*

0
.5
0
9
(0
.0
0
1
)*
*
*

0
.5
0
4
(0
.0
0
1
)*
*
*

N
C
C

0
.1
3
0
(0
.1
5
)

0
.1
3
6
(0
.1
3
)

0
.1
3
8
(0
.1
3
)

0
.1
3
4
(0
.1
4
)

0
.1
0
8
(0
.3
4
)

F
in
an
ci
al
li
te
ra
cy

q
u
es
ti
o
n
s

T
o
ta
l
an
sw

er
s

0
.2
7
5
(0
.2
0
)

�0
.1
0
6
(0
.7
1
)

0
.0
2
3
(0
.8
9
)

�0
.0
1
8
(0
.9
3
)

�0
.0
9
8
(0
.6
9
)

In
te
ra
ct
io
n
ef
fe
ct
s
o
f
fi
n
an
ci
al
li
te
ra
cy

w
it
h

P
ar
en
ts
fi
n
an
ci
al
ed
u
ca
ti
o
n

�0
.5
4
8
(0
.0
7
)*

A
n
x
ie
ty

0
.0
4
1
(0
.7
4
)

O
v
er
sp
en
d
in
g

�0
.0
7
3
(0
.6
1
)

S
u
rp
ri
se
d

�0
.0
0
3
(0
.9
8
)

N
C
C

0
.0
4
6
(0
.7
0
)

O
b
s

1
5
6

1
5
6

1
5
6

1
5
6

1
5
6

P
se
u
d
o
R
2

0
.2
0
9

0
.1
9
8

0
.1
9
8

0
.1
9
7

0
.1
9
8

L
R
st
at
is
ti
c

6
4
.6
7
4

6
1
.2
1
4

6
1
.3
5
9

6
1
.1
0
1

1
.7
2
2

P
ro
b
ab
il
it
y
(L
R
st
at
)

(0
.0
0
1
)*
*
*

(0
.0
0
1
)*
*
*

(0
.0
0
1
)*
*
*

(0
.0
0
1
)*
*
*

(0
.0
0
1
)*
*
*

A
k
ai
k
e
in
fo

cr
it
er
io
n

1
.7
0
0

1
.7
2
2

1
.7
2
1

1
.7
2
3

1
.7
2
2

S
ch
w
ar
z
cr
it
er
io
n

1
.8
9
6

1
.9
1
8

1
.9
1
7

1
.9
1
8

1
.9
1
7

H
an
n
an
-Q

u
in
n
cr
it
er
io
n

1
.7
7
9

1
.8
0
2

1
.8
0
1

1
.8
0
2

1
.8
0
1

N
o
te
:
*
*
*
,
*
*
,
*
st
at
is
ti
ca
ll
y
si
g
n
ifi
ca
n
t
at
th
e
1
%
,
5
%
,
an
d
1
0
%

le
v
el
,
re
sp
ec
ti
v
el
y
.

268 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



financial knowledge derived from parents has a strong positive effect on individuals’ finan-

cial behavior especially for people characterized by a relevant financial illiteracy.

Based on the findings from the interaction effects models, and the corresponding superior-

ity of Models 1, in Tables 6, 6a, and 6b, we decided to use these series of models to compute

the marginal effects and corresponding Probabilities. We present the overall probabilities for

each model in Table 7, followed by the computation of the marginal changes in the overall

probabilities when the independent variables change by one unit. It is also relevant to note

that the estimated probabilities are computed at the mean value of the independent variables.

For the especial case of the FEPAR variable (dummy 1,0) we compute the marginal effects by

taking the difference between the overall probabilities when FEPAR = 1, minus FEPAR = 0.

There are several interesting results that spring out the analysis of Table 7. For instance,

the first element to notice is that all probabilities are very similar across alternative models

of financial (il)literacy with the probability to paying in full being in the range of 67.27

(DNK answers – Level B) to 68.82% (Incorrect answers – Level A), with 67.55% (Correct

answers). And the rest of probabilities being almost identical for the other ranges across

models. In other words, when evaluated at the mean values, we observe very little discrepan-

cies across alternative models, when the only difference is based on the level of financial (il)

literacy. This result on itself is surprising, as one would expect that different levels of finan-

cial (il)literacy would yield much larger differences in repayment patterns.

Secondly, when we look at the changes in probabilities due to variations in RHS varia-

bles, we now observe potentially large differences across models. The first effect to study

here is that of FEPAR. Notice that the effect of this dummy variable has a significant mar-

ginal effect on the probability of repaying in full and has the largest effect of all variables. In

this sense, we also observe that those suffering the most from Level B financial illiteracy are

the group that would benefit the most with an increase learning deriving from their parents.

The counter result indicates that those suffering from Level A financial illiteracy would still

benefit from further interaction with their parents as the main source of financial literacy but

with the smallest effect among the three possible groups. In addition, the counter result of

variations in the FEPAR variable indicate that both Level B Financial Illiteracy and those

answering correctly would be the groups that would suffer the most in their repayment capa-

bilities should FEPAR were not to be the main source of financial literacy as reflected by the

negative sign of the probability changes for the remaining categories.

The fourth element that we pay attention in this analysis is the effect of changes in the

level of overspending. In this case, we observe that changes in overspending have the sec-

ond largest effect of all variables. More specifically, an increase of one unit in overspend-

ing (following the stated categories in the descriptive statistics) have a negative effect in

repayment capabilities of about 20% decrease in repayment in full; and incidentally an

increase in all other repayment categories with paying less than full balance every month

being the most affected. In this sense, a marginal change in overspending patterns, results

in a large increase in month-to-month debt accumulation and falling behind in repayment

patterns. Furthermore, this negative effect is larger than an increase in financial literacy as

we will discuss next.

Financial literacy has been championed as the main way to improve financial decision-

making process, and specifically research studying credit card debt emphasize on its

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 269



T
ab
le

7
P
ro
b
ab
il
it
ie
s
an
d
m
ar
g
in
al

ef
fe
ct
s,
fo
r
m
o
d
el

w
it
h
in
te
ra
ct
io
n
ef
fe
ct
s

P
ro
b
(y

=
1
jx)

/∂
P
1
/∂
x

P
ro
b
(y

=
2
jx)

/∂
P
2
/∂
x

P
ro
b
(y

=
3
jx)

/∂
P
3
/∂
x

P
ro
b
(y

=
4
jx)

/∂
P
4
/∂
x

F
in
an
ci
al
li
te
ra
cy

m
o
d
el

6
7
.5
5
%

1
9
.6
7
%

7
.6
9
%

5
.0
9
%

P
ar
en
ts
fi
n
an
ci
al
ed
u
c

0
.3
1
3

�0
.1
1
6

�0
.0
9
1

�0
.1
0
6

O
v
er

sp
en
d
in
g

�0
.2
0
7

0
.0
8
7

0
.0
6
0

0
.0
6
0

N
C
C

�0
.0
6
1

0
.0
2
6

0
.0
1
8

0
.0
1
8

C
o
rr
ec
t
an
sw

er
s

0
.1
4
7

�0
.0
6
2

�0
.0
4
3

�0
.0
4
3

In
te
ra
ct
io
n
ef
fe
ct

�0
.1
5
4

0
.0
6
4

0
.0
4
5

0
.0
4
5

L
ev
el
A
-
F
in
an
ci
al
il
li
te
ra
cy

6
8
.8
2
%

1
9
.1
4
%

7
.3
2
%

4
.7
2
%

P
ar
en
ts
fi
n
an
ci
al
ed
u
c

0
.2
9
5

�0
.0
5
5

�0
.0
8
5

�0
.0
9
5

O
v
er

sp
en
d
in
g

�0
.1
9
0

0
.0
8
2

0
.0
5
5

0
.0
5
3

N
C
C

�0
.0
6
5

0
.0
2
8

0
.0
1
9

0
.0
1
8

In
co
rr
ec
t
an
sw

er
s

�0
.1
8
7

0
.0
8
1

0
.0
5
4

0
.0
5
2

In
te
ra
ct
io
n
ef
fe
ct

0
.1
5
4

�0
.0
6
7

�0
.0
4
4

�0
.0
4
3

L
ev
el
B
-
F
in
an
ci
al
il
li
te
ra
cy

6
7
.2
7
%

1
9
.2
5
%

7
.8
4
%

5
.6
4
%

P
ar
en
ts
fi
n
an
ci
al
ed
u
c

0
.3
4
8

�0
.1
2
0

�0
.0
9
8

�0
.1
3
0

O
v
er

sp
en
d
in
g

�0
.1
8
7

0
.0
7
5

0
.0
5
4

0
.0
5
9

N
C
C

�0
.0
4
7

0
.0
1
9

0
.0
1
3

0
.0
1
5

D
o
N
o
t
K
n
o
w
an
sw

er
s

�0
.0
9
9

0
.0
4
0

0
.0
2
8

0
.0
3
1

In
te
ra
ct
io
n
ef
fe
ct

0
.1
9
8

�0
.0
7
9

�0
.0
5
7

�0
.0
6
2

S
o
u
rc
e:

M
ar
g
in
al

ef
fe
ct
s
o
n
ly

re
p
o
rt
ed

fo
r
M
o
d
el

1
T
ab
le

6
,
6
a,
an
d
6
b
.
y
=
1
re
fe
rs

to
p
ay

in
fu
ll
,
y
=
2
re
fe
rs

to
p
ay

le
ss

th
an

fu
ll
b
u
t
m
o
re

th
an

m
in
i-

m
u
m
,
y
=
3
re
fe
rs
to

p
ay

m
in
im

u
m
,
an
d
y
=
4
re
fe
rs
to

p
ay

le
ss

th
an

m
in
im

u
m
.

270 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



importance. While we do not dispute the inner importance of financial literacy in improving

decision making, our results indicate that financial literacy plays a secondary role, when con-

trolling for individuals’ personality traits and consequent behavior. In particular, marginal

effects analysis from Table 5, indicate that an increase in the number of correct answer in

the financial literacy questionnaire result in a positive improvement (pay in full) in repay-

ment patterns on the approximate amount of 14.7%. While a significant improvement, this

amount is not enough to counterbalance the negative effects of behavior. By the same token,

the negative effect of increased financial illiteracy is decomposed in a worsen repayment

capability (away from full repayment every month) of 18.7% for Level A and 9.9% for

Level B. It is relevant to point out that those overestimating their financial knowledge, yet

answering incorrectly, are at the highest risk of falling behind and carry a month-to-month

balance. In addition, it comes as a relative surprise that those suffering from Level B illiter-

acy are less likely to carry an increase negative probability of repayment in full when

answering more DNK to the financial literacy questions. In this context, it becomes evidence

that individuals with a lack of financial literacy Level A and personality traits dominated by

overspending are highly more likely to fall behind, as reflected by the estimated values of

the Prob y = 2, 3, and 4 categories.

A generalization of the results also indicate that the estimated probabilities and corre-

sponding marginal effects are lower for the Level B financial illiteracy individuals than the

other two groups. For Level B, it appears that they benefit the most from increased interac-

tion with their parents (FEPAR), and recur less to more credit cards and consequently have a

lower negative effect on repayment as the number of credit cards is increased. Thus, it

appears that not knowing about financial literacy (DNK) makes them act in a more cautious

way in relation to credit card use.

Furthermore, the robustness of the personality traits manifestation in the form of over-

spending are statistically consistent across all estimated models. In this context, it is more

relevant to note that the estimated marginal effects of changes in this behavioral variable are

also consistent even after controlling for the different levels of financial literacy and finan-

cial (il)literacy Levels A and B. In other words, our estimated probabilities and related mar-

ginal changes of increases in overspending seem to be independent of financial literacy. This

result, as far as we are aware of, is unique and not present in the extant literature. As such,

we argue that financial literacy has a limited impact on modifying personality traits and

related behavior. It appears, that early intervention and an increase in financial literacy at

early stages in life, primarily through parental education, may have the largest offsetting

effects to personality traits leading to poor financial decision making.

5. Conclusions

The empirical evidence in this paper demonstrates that financial behavior, measured in

terms of credit card repayment patterns, is affected more by personal traits than by financial

literacy. This is to say, that financial decisions are made mainly based on personal traits or

behavioral factors. Behavioral variables, such overspending, drive the accumulation of credit

card debt and result in less than optimal repayment. Financial knowledge derived from

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 271



parental interaction with children seems to be the form of financial knowledge most relevant

in positively influencing credit card repayment. Individuals with self-assessed lower level of

financial literacy (DNK answers) display diverse behaviors (credit card repayment patterns)

as opposed to individuals who do not possess basic financial literacy (wrong answers) even

after controlling for cognitive or personal trait factors.

In our context, it appears that financial literacy has a limited positive benefit in shaping fi-

nancial decisions. Previous evidence is mixed on this specific issue: although there is ample

evidence supporting that higher financial knowledge translate into less risky credit card use

(Norvilitis et al., 2006; Robb, 2011; Shim et al., 2009, 2010), other researches have reported

greater financial knowledge was associated with lower fear about using credit cards and

greater levels of debt (Borden et al. 2008; Lyons & Rogers, 2004; Robb & Sharpe, 2009).

One possible explanation for these mixed results could depend on how financial knowledge

is operationalized and measured. For instance, our study levers up the traditional financial

literacy questions developed by Lusardi & Mitchell (2014) while Robb (2011) uses a finan-

cial knowledge score specifically designed to uncover specific knowledge about credit card

use. Different results could be related to imprecise measurement of a latent variable such as

financial knowledge.

Results from the parsimonious model specification also indicate the presence of robust

behavior patterns along personality traits. Present-bias behavior, overconfidence and lack of

control seem to be the main drivers of credit card use, and consequently of credit card

repayment.

Now, with the existent data, we can measure the impact of behavioral differences while

controlling for financial (il)literacy, but we cannot measure the possible gains in behavioral

actions/responses due to increases in financial literacy. It is important to acknowledge this

caveat in our study to both avoid incorrect data interpretation, and to set the future research

agenda as we move forward. In essence, the long-term goal of the financial education or lit-

eracy movement is to provide mechanisms to positively affect economic behavior that cur-

rently result in costly and inefficient financial decisions. This is so, as financial behavior is

highly determined by preferences, and these preferences may or may not be a function of fi-

nancial literacy. If financial literacy does affect individuals’ preferences and decision-mak-

ing processes then one would expect that higher levels of financial literacy result in a series

of benefits such as: increased saving, wiser investment decisions, lower to no month-to-

month credit card debt, higher wealth accumulation, and higher retirement savings. Our find-

ings are useful for policy makers to implement policies able to avoid, or reduce, deb trap

and socio-economic vulnerability of the borrower. For instance, it appears that early inter-

vention in terms of financial education may provide the necessary means to control endoge-

nous costly personality traits. Once the personality trait has fully developed alternative

means of positively impacting financial decision making might need to be implemented. As

Xiao et al. (2011) highlight, financial education programs should target the multiple psycho-

logical processes that lead to changes in attitude and the enhancement of self-confidence,

that is, they should target and develop students’ positive financial intentions.

Finally, while being outside the scope of this study, further research on the subject of fi-

nancial anxiety and its relation to credit card behavior will need to be explored, particularly

the determination of the causality between credit card repayment habit and financial anxiety.

272 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



The tentative hypothesis states that higher financial anxiety should result in worse credit

card repayment; thus, future research could analyze in more depth the role of psychological

aspects and personality traits, in particular financial anxiety, in the process of financial deci-

sion-making, that is the possibility and cause-effect that higher credit card balance may fuel

anxiety and result in a freeze effect where individuals may not improve repayment.

Notes

1 In several studies, most individuals report having higher believed level of financial

literacy than actual correct scoring on financial literacy questionnaire/surveys. See

recent studies by GFLEC at George Washington University center for some exam-

ples (Almenberg et al., 2016).

2 In a recent papers Kim and Mountain (2019) suggest a specific statistical approach

(binomial-latent regression models) to specifically tackle the issue of group differen-

ces that are hidden in DK/RF responses.

3 Aptitude is different from attitude: while attitude is a way of looking at an issue or

an object, a mental position or way of thinking about an issue (in our case financial

matters), the concept of aptitude is akin to natural or acquired talent or ability, incli-

nation, predisposition. In this sense, aptitude for financial matters could be learn

through financial education as reflected by higher levels of financial literacy.

4 The questionnaire has been successfully used to capture the financial literacy of

diverse populations since it was first piloted in 2010 as part of the first OECD inter-

national financial literacy and financial inclusion measurement exercise. In 2018, an

updated version was released that takes into consideration the changes in the finan-

cial landscape and the evolving state of knowledge; therefore, including questions

related to digital financial services and crypto-assets, trust, integrity and financial

consumer protection.

5 More recent studies have applied factor analysis (van Rooij, Lusardi, & Alessie,

2011). It is widely acknowledged, however, that more work, developed through rig-

orous psychometric analysis is needed (Bongini et al., 2015, 2016; Knoll & Houts,

2012).

6 For a comprehensive survey on personality psychology and economics the interested

reader can refer to Almlund, Duckworth, Heckman, and Kautz (2011) and to Brown

and Taylor (2014) for a specific application that analyzes the relationship between

personality traits and financial decision-making focusing on unsecured debt and fi-

nancial assets.

7 While the topics of “time value of money,” “inflation,” and “interest rate compound-

ing” do represent basics knowledge to make informed choices when deciding to pay

in full or accumulating a month-to-month balance, the issue of risk diversification is

less strictly correlated.

8 Estimations for the NCC variables as the dependent variable, with demographic

characteristics as RHS variables, are available from the authors upon request. We

conducted the same estimations with demographics instead of NCC, and obtained

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 273



very similar results, without affecting the estimated coefficients of the other

variables.

9 Recall that credit card repayment behavior is measured in a reverse scale, where the

lowest value implies repayment in full, and higher values otherwise.

References

Adams, T., & Moore, M. (2007). High-risk health and credit behavior among 18-to 25-year-old college students.

Journal of American College Health, 56, 101–108.

Akerlof, G. (1991). Procrastination and obedience. The American Economic Review, 81(2), 1–19.

Andrews, B., & Wilding, J. M. (2004). The relation of depression and anxiety to life-stress and achievement in

students. British Journal of Psychology, 95, 509–521.

Barboza, G. (2018). I will pay tomorrow, or maybe the day after. Credit card repayment, present biased and pro-

crastination. Economic Notes, 47, 455–494.

Barboza, G., Smith, C., & Pesek, J. (2016). Impact of gender and academic performance differences on financial

literacy. Journal of Financial Education,Winter, 205–242.

Berg, C., Sanem, J., Lust, K., Ahluwalia, J., Kirch, M., & An, L. (2010). Health-related characteristics and incur-

ring credit card debt as problem behaviors among college students. The Internet Journal of Mental Health,

6, 1.

Bongini, P., Trivellato, P., & Zenga, M. (2015). Business students and financial literacy: When will the gender

gap fade away? Journal of Financial Management, Markets and Institutions, 3, 13–30.

Bongini, P., Trivellato, P., & Zenga, M. (2016). Financial literacy and undergraduates. Application of latent

regression Rasch model. Journal of Financial Management, Markets and Institutions, 41, 23–42.

Borden, L. M., Le, S., Serido, J., & Collins, D. (2008). Changing college students’ financial knowledge, attitudes,

and behavior through seminar participation. Journal of Family and Economic Issues, 29, 23–40.

Brown, S., & Taylor, K. (2014). Household finances and the ‘Big Five’ personality traits. Journal of Economic

Psychology, 45, 197–212.

Cheema, A., & Soman, D. (2006). Malleable mental accounting: The effect of flexibility on the justification of

attractive spending and consumption decisions. Journal of Consumer Psychology, 16, 33–44.

Chen, H., & Volpe, R. P. (2002). Gender differences in personal financial literacy among college students.

Financial Services Review, 11(3), 289–307.

Chen, Z., & Garand, J. C. (2018). On the gender gap in financial knowledge: Decomposing the effects of don’t

know and incorrect responses. Social Science Quarterly, 99(5), 1551–1571.

Chu, C., Buchman-Schmitt, J. M., Stanley, I. H., Hom, M. A., Tucker, R. P., Hagan, C. R., & Joiner Jr, T. E.

(2017). The interpersonal theory of suicide: A systematic review and meta-analysis of a decade of cross-

national research. Psychological Bulletin, 143(12), 1313.

Danes, S. M., & Tahira, K. H. (1987). Money management knowledge of college students. Journal of Student

Financial Aid, 17, 4–16.

Fiksenbaum, L., Marjanovic, Z., & Greenglass, E. (2017). Financial threat and individuals’ willingness to change

financial behavior. Review of Behavioral Finance, 9, 128–147.

Fox, M. D., Snyder, A. Z., Vincent, J. L., Corbetta, M., Van Essen, D. C., & Raichle, M. E. (2005). The human

brain is intrinsically organized into dynamic, anticorrelated functional networks. Proceedings of the National

Academy of Sciences of the United States of America, 102, 9673–9678.

Grable, J., & Joo, S. (2006). Student racial differences in credit card debt and financial behaviors stress. College

Student Journal, 40, 400–408. Retrieved from http://www.eric.ed.gov/ERICWebPortal/detail?accno=

EJ765337

Hancock, A. M., Jorgensen, B. L., & Swanson, M. S. (2013). College students and credit card use: The role of

parents, work experience, financial knowledge, and credit card attitudes. Journal of Family and Economic

Issues, 34, 369–381.

274 G. Barboza et al. / Financial Services Review 29 (2021) 247–276



Hilgert, M., Hogarth, J., & Beverly, S. (2003). Household financial management: The connection between knowl-

edge and behavior. Federal Reserve Bulletin, 89, 309–322.

Houston, S. J. (2010). Measuring financial literacy. The Journal of Consumer Affairs, 44, 296–316.

Joo, S., & Garman, T. E. (1998). The potential effects of workplace financial education based on the relationship

between personal financial wellness and worker job productivity. Personal Finances and Worker

Productivity, 2, 163–173.

Kapoor, A., Dunn, E., Kostaki, A., Andrews, M. H., & Matthews, S. G. (2006). Fetal programming of hypothal-

amo-pituitary-adrenal function: Prenatal stress and glucocorticoids. The Journal of Physiology, 572, 31–44.

Kempson, E. (2009). Framework for the Development of Financial Literacy Baseline Surveys. Retrieved from

Oecd-library.org

Kezar, A., & Yang, H. (2010). The importance of financial literacy. About Campus: Enriching the Student

Learning Experience, 14, 15–21.

Kim, N., & Mountain, T. P. (2019). Financial Knowledge and “Don’t Know” Response. Journal of Consumer

Affairs, 53(4), 1948–1969.

Knoll, M. A. Z., & Houts, C. R. (2012). The Financial Knowledge Scale: An application of item response theory

to the assessment of financial literacy. Journal of Consumer Affairs, 46, 381–410.

Koshal, R. K., Gupta, A. K., Goyal, A., & Choudhary, V. N. (2008). Assessing economic literacy of Indian MBA

students. American Journal of Business, 23, 43–52.

Kuchler, T. (2013). Sticking to Your Plan: Hyperbolic Discounting and Credit Card Debt Paydown. Discussion

Papers 12-025. Stanford Institute for Economic Policy Research.

Laibson, D. (1997). Golden eggs and hyperbolic discounting. The Quarterly Journal of Economics, 112, 443–477.

Lusardi, A., & Mitchell, O. S. (2007). Baby boomer retirement security: The role of planning, financial literacy,

and housing wealth. Journal of Monetary Economics, 54, 205–224.

Lusardi, A., & Mitchell, O. S. (2014). The economic importance of financial literacy: Theory and evidence.

Journal of Economic Perspectives, 52, 5–44.

Lusardi, A., Mitchell, O. S., & Curto, V. (2009). Financial Literacy Among the Young: Evidence and

Implications for Consumer Policy. CeRP Working Papers 91. Turin, Italy: Center for Research on Pensions

and Welfare Policies.

Lusardi, A., Mitchell, S., & Curto, V. (2010). Financial literacy among the young. Journal of Consumer Affairs,

44, 358–381.

Lusardi, A., & Tufano, P. (2009). Debt Literacy, Financial Experiences, and Overindebtedness. NBER Working

Papers 14808. National Bureau of Economic Research, Inc.

Lyons, J. S., & Rogers, L. (2004). The US child welfare system: A de facto public behavioral health care system.

Journal of the American Academy of Child & Adolescent Psychiatry, 43, 971–973.

Mandell, L. (1997). Personal Financial Survey of High School Seniors. Washington, DC: Jump Start Coalition

for Personal Financial Literacy.

Mandell, L. W. (2008). The Financial Literacy of Young American Adults: Results of the 2008 National

Jump$tart Coalition Survey of High School Seniors and College Students. Washington, DC: Jump Start

Coalition for Personal Financial Literacy.

Manton, E. J., English, D. E., Avard, S., & Walker, J. (2006). What college freshmen admit to not knowing about

personal finance. Journal of College. Teaching &. Learning, 3, 43–54.

Marjanovic, Z., Greenglass, E. R., Fiksenbaum, L., & Bell, C. M. (2013). Psychometric evaluation of the Financial

Threat Scale (FTS) in the context of the great recession. Journal of Economic Psychology, 36, 1–10.

Markovich, C. A., & DeVaney, S. (1997). College seniors’ personal finance knowledge and practices. Journal of

Family and Consumer Sciences, 89, 61–65.

Moore, D. (2003). Survey of Financial Literacy in Washington State: Knowledge, Behavior, Attitudes, and

Experiences. Olympia, WA: Washington State Department of Financial Institutions.

Moore, D. A., & Healy, P. J. (2008). The trouble with overconfidence. Psychological Review, 115(3), 502.

Nelson, M., Lust, K., Story, MF., & Ehlinger, E. (2008). Credit card debts, stress and key health risk behaviors

among college students. American Journal of Health Promotion, 22, 400–412.

G. Barboza et al. / Financial Services Review 29 (2021) 247–276 275



Norvilitis, J. M., Merwin, M. M., Osberg, T. M., Roehling, P. V., Young, P., & Kamas, M. M. (2006).

Personality factors, money attitudes, financial knowledge, and credit-card debt in college students 1. Journal

of Applied Social Psychology, 36, 1395–1413.

O’Donoghue, T., & Rabin, M. (1999). Doing it now or later. American Economic Review, 89, 103–124.

Peters, M., Seeds, K., Goldstein, A., & Coleman, N. (2007). Parental Involvement in Children’s Education 2007,

DCSF Research Report RR034. London: Department for Children, Schools and Families. Retrieved from

www.dcsf.gov.uk/research/data/uploadfiles/DCSF-RR034.pdf

Pinto, M., Parente, D., & Palmer, T. (2001). College student performance and credit card usage. Journal of

College Student Development, 42, 49–58. Retrieved from http://www.eric.ed.gov/ERICWebPortal/detail?

accno=EJ624498

Porto, N., & Xiao, J. J. (2016). Financial Literacy Overconfidence and Financial Advice Seeking. Journal of

Financial Service Professionals, 70(4).

Prelec, D., & Simester, D. (2001). Always leave home without it: A further investigation of the credit card effect

on willingness to pay.Marketing Letters, 12, 5–12.

Razafimahasolo, M., Borromeo, R., Sausa, L., Carpizo, R., & Sabado, J. (2016). Impact of financial literacy on

level of stress and academic achievement among college students. AUP Research Journal, 19(2), 21–35.

Remund, D. L. (2010). Financial literacy explication: The case for a clearer definition in an increasingly complex

economy. The Journal of Consumer Affairs, 44, 276–295.

Robb, C. A. (2011). Financial knowledge and credit card behavior of college students. Journal of Family and

Economic Issues, 32, 690–698.

Robb, C. A., & Sharpe, D. L. (2009). Effect of personal financial knowledge on college students’ credit card

behavior. Journal of Financial Counseling and Planning, 20(1).

Schagen, S., & Lines, A. (1996). Financial Literacy in Adult Life. London: National Foundation for Educational

Research in England and Wales. NatWest Group Charitable Trust.

Shim, S., Xiao, J. J., Barber, B. L., & Lyons, A. C. (2009). Pathways to life success: A conceptual model of finan-

cial well-being for young adults. Journal of Applied Developmental Psychology, 30, 708–723.

Shim, S., Barber, B. L., Card, N. A., Xiao, J. J., & Serido, J. (2010). Financial socialization of first-year college

students: The roles of parents, work, and education. Journal of Youth and JAdolescence, 39, 1457–1470.

Stango, V., & Zinman, J. (2009). What do consumers really pay on their checking and credit card accounts?

Explicit, implicit, and avoidable costs. American Economic Review, 99, 424–429.

Thaler, R. (1980). Toward a positive theory of consumer choice. Journal of Economic Behavior & Organization,

1, 39–60.

Thaler, R. (2018). From cashews to nudges: The evolution of behavioral economics. American Economic

Review, 108, 1265–1287.

Thaler, R., & Shefrin, H. M. (1981). An economic theory of self-control. Journal of Political Economy, 89, 392–406.

Tokar Asaad, C. (2015). Financial literacy and financial behavior: Assessing knowledge and confidence.

Financial Services Review, 24(2).

Van Rooij, M. C., Lusardi, A., & Alessie, R. J. (2012). Financial literacy, retirement planning and household

wealth. The Economic Journal, 122(560), 449–478.

Van Rooij, M., Lusardi, A., & Alessie, R. (2011). Financial literacy and stock market participation. Journal of

Financial Economics, 101, 449–472.

Wagland, S. P., & Taylor, S. T. (2009). When it comes to financial literacy, is gender really an issue? The

Australasian Accounting Business and Finance Journal, 3, 13–23.

Xia, T., Wang, Z., & Li, K. (2014). Financial literacy overconfidence and stock market participation. Social indi-

cators Research, 119(3), 1233–1245.

Xiao, J. J., Tang, C., Serido, J., & Shim, S. (2011). Antecedents and consequences of risky credit behavior among

college students: Application and extension of the theory of planned behavior. Journal of Public Policy &

Marketing, 30, 239–245.

Zhang, C. Y., & Sussman, A. B. (2018). Perspectives on mental accounting: An exploration of budgeting and

investing. Financial Planning Review, 1, e1011.

276 G. Barboza et al. / Financial Services Review 29 (2021) 247–276