Financial literacy: profiling a successful high school

outreach program

Greg Filbecka,*, Jason Pettnerb, Xin Zhaoc

aBlack School of Business, Penn State Behrend, Erie, PA 16563, USA
bMarket Analyst, C.S. McKee, 420 Fort Duquesne Blvd Suite 800, Pittsburgh, PA 15222, USA

cBlack School of Business, Penn State Behrend, Erie, PA 16563, USA

Abstract

The CFA Society Pittsburgh launched a high school financial literacy campaign resulting in sig-

nificant improvements in financial behavior, subjective and objective financial knowledge, and self-

esteem. Before the campaign, male students and students with higher grade point averages (GPAs)

show better objective knowledge. In addition, we find disconnect between actual and perceived fi-

nancial knowledge. Students exhibited gains in all aspects after completing the program. The subca-

tegories with the lowest pre-survey scores or female students show the greatest improvements in the

post-survey. Students with lower GPAs experienced greater improvement in financial behavior and

objective knowledge, while higher GPA students improved more in subjective knowledge. © 2020

Academy of Financial Services. All rights reserved.

JEL classifications: G53

Keywords: Financial literacy; Survey; Outreach; Education

1. Introduction

Based on the results of multiple academic studies, a significant lack of financial literacy

exists across nearly all demographics. While financial literacy statistics are important, the

implications of a lack of financial literacy and numeracy are far reaching due to their impact

on financial decisions. As a result, the potential implications of financial literacy and

*Corresponding author. Tel.: +1-814-898-6549.

E-mail address: mgf11@psu.edu

1057-0810/20/$ – see front matter © 2020 Academy of Financial Services. All rights reserved.

Financial Services Review 28 (2020) 315–340



numeracy will be explored in depth. In this paper, we demonstrate the effectiveness of intro-

ducing a program to improve financial literacy and numeracy for high school students.

According to Remund (2010, p. 278), “Financial literacy is a measure of the degree to

which one understands key financial concepts and possesses the ability and confidence to

manage personal finances through appropriate, short-term decision-making and sound, long-

range financial planning, while mindful of life events and changing economic conditions.”

To gauge financial literacy, studies use many different knowledge-based questions; however,

the overall concepts remain relatively consistent across financial literacy surveys. Three

common fundamental financial knowledge concepts include interest rates, inflation, and risk

diversification.

Numeracy is an important component of financial literacy that literature often relates to fi-

nancial behavior. Estrada-Mejia, de Vries, and Zeelenberg (2016, p. 53) define numeracy as

“the ability to understand and use numerical information.” Numeracy is closely related to

multiple aspects of financial decision making.

This study extends prior research on the effectiveness of financial literacy education by

providing direct evidence from a financial literacy campaign launched by the CFA Society

Pittsburgh. The study is based upon data collected from a financial literacy campaign of 53

high schools, across seven states, during the 2017-2018 academic year. The financial literacy

education campaign materials were created using the book The Missing Semester (Kabala &

Natali, 2013) as the main resource and curriculum. Before starting the course, students were

given a pre-survey to test their baseline in four major areas: subjective financial knowledge,

financial behavior, objective financial knowledge, and financial self-esteem. Following the

completion of the course, students were administered a post-survey to test changes in the

four major areas. To anonymously track the progress of students, they were assigned a

unique student ID code. The results display significant improvement in all areas of interest.

Each area of interest improves at a statistically significant level of at least 5% after the finan-

cial literacy campaign, indicating an effective effort at improving financial literacy.

2. Literature review

Since the effects of financial literacy are evident and the high percentage of financial illit-

eracy has been proven, steps must be taken to improve financial literacy. Green and Riddell

(2012) analyze data gathered from the Canadian component of the International Adult

Literacy and Life Skills Survey (IALSS), which sought to measure the skills of adults in

Canada. The survey asked skill-based questions that focused upon four cognitive skills:

prose literacy, document literacy, numeracy, and problem solving. Regression analysis

reveals an increase of 3.4% in literacy and numeracy scores for an additional year of school-

ing. They also find that completing an extra 4 years of education results in a 24% increase in

literacy, which increases a person’s literacy from the median to the 80th percentile.

Although understanding financial concepts is vital to financial behavior, possessing high

financial self-esteem and confidence is an essential key to successful financial decisions.

Using data collected on 12,686 individuals by the U.S. Bureau of Labor Statistics tracked

over a 30-year period, Tang and Baker (2016) create four main variables: financial behavior,

316 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340



self-esteem, objective, and subjective financial knowledge and covariates. Their results indi-

cate a direct and indirect relationship of self-esteem on multiple financial behaviors. As a

result, the effect of self-esteem is statistically significant; thus, self-esteem is a factor of fi-

nancial behavior. Their results indicate that to improve financial behavior, subjective knowl-

edge must be at least a portion of a financial education curriculum because objective

knowledge by itself does not have a complete positive impact on financial behavior.

An important facet of self-esteem is the extent to which a person’s confidence becomes
greater than their actual knowledge. Overconfident individuals have the tendency to overesti-
mate their own knowledge, leading to a higher risk of engaging in costly and risky financial
behaviors (Asaad, 2015). Asaad discovers confidence is an important fragment of financial lit-
eracy, but also finds that perceived knowledge without actual knowledge increases the risk of
suboptimal financial decisions. McCannon, Asaad, and Wilson (2016) conduct a study based
on subjects playing an experimental trust game (Berg, Dickhaut, & McCabe, 1995) and com-
pleting a risk assessment, background questionnaire, and financial literacy quiz. Results of the
risk assessment and financial literacy quiz were decomposed to generate overconfidence as a
variable. They find a statistically significant relationship between overconfidence and trusting
investments. Although the actual financial literacy in the United States has declined over the
past 5 years, Lin, Bumcrot, Ulicny, Lusardi, Mottola, Kieffer, and Walsh (2016) find the per-
centage of U.S. citizens who have a high-self assessment has actually increased. Thus, efforts
to improve financial behavior must address the going-concern of overconfidence.

Recent studies have identified the ability to increase financial literacy through education at

the university level. Gerrans and Heaney (2019) study the effects of financial literacy following

an undergraduate personal finance course at an Australian University. After completing the

course, students showed improvements in both objective and subjective financial literacy.

2.1. Gender gap in financial literacy

Several studies have found the existence of a financial literacy gender gap. In particular,

Cupák, Fessler, Schneebaum, and Silgoner (2018) find that women score lower than men on

financial literacy, with a more pronounced gap in developed countries. Additionally, Preston

and Wright (2019) examine the financial literacy gap in Australia. While the “human capital

variables” (age and education) were not significant in explaining the gap, “labor market vari-

ables” (including sector and occupation) were significant in explaining the gap. While these

findings identify an initial gender gap in financial literacy, Gerrans and Heaney (2019) find

female students benefit the greatest from financial literacy education.

3. Need for economic education

According to the 2015 National Report Card, financial literacy education in high school is

insufficient. On a scale from “A” to “F,” twenty-six states received grades of C, D, or F

(Pelletier, 2015). An “A” grade (five states) requires the schools to offer a one-semester per-

sonal finance course as a graduation requirement. A “B” grade (20 states) requires the

schools to include personal finance education within a required course, whether as a stand-

G. Filbeck et al. / Financial Services Review 28 (2020) 315–340 317



alone course or part of another course. A “C” grade (11 states) requires schools to offer per-

sonal finance topics in schools, but does not mandate students take the course. A “D” grade

(three states) means that the state has a “modest levels” of personal finance in its academic

standards. Finally, an “F” grade (12 states) goes to states with nearly no financial educational

requirements, meaning that a student can graduate high school without an introduction to

any financial literacy concepts. Because formal education has a statistically significant rela-

tionship with literacy and numeracy (Green & Riddell, 2012), the lack of high school finan-

cial education is a potential factor of low financial literacy scores.

Although the effects of financial literacy education are highly debated, research con-

ducted by Filbeck and Zhao (2018) illustrates that financial literacy initiatives have a posi-

tive impact on both financial knowledge and behavior. They find that teaching financial

concepts to high school students had a profound and statistically significant effect on both

subjective financial knowledge and financial behavior for the students involved in the sur-

vey. Based upon previous research conducted, perceived knowledge may differ from actual

knowledge (Asaad, 2015; Lin et al., 2016; McCannon et al., 2016). In addition, Tang and

Baker (2016) prove that self-esteem and subjective knowledge have a statistically significant

effect on financial behavior. As a result, further research could incorporate subjective educa-

tion as a piece of the financial initiative, as well as financial knowledge questions that ana-

lyze both perceived and actual financial knowledge.

This article extends the work of Filbeck and Zhao by including objective assessments

about acquisition of financial knowledge based on educational outreach. Our hypothesis is

as follows:

Hypothesis 1: The financial literacy outreach program, which targets subjective financial
knowledge, financial behavior, objective financial knowledge, and financial self-
esteem, will result in statistically significant improvement for participating
students.

In our article, we first analyze baseline pre-survey responses in financial behavior, subjec-

tive and objective financial knowledge, and self-esteem of high school students participating

in a financial literacy outreach program developed by the CFA Society Pittsburgh. We use

path analysis to assess the strength of direct and indirect links between these four aspects of

financial literacy and other control variables. Next, by comparing matched pre- and post-sur-

vey results, we analyze the impact of the financial literacy outreach program in financial

behavior, subjective and objective financial knowledge, and self-esteem. We first use a

paired t test to compare the matched pre- and post-survey results. Then, we use regression

analysis to analyze how the improvements in these four aspects of financial literacy in the

post-survey are impacted by control variables.

4. Data sample

The CFA Society Pittsburgh has been active in financial literacy outreach since 2010.

Over 50 individuals serve on the financial literacy committee, which directs curricular devel-

opment and training efforts. Each year, representatives from the Society participate in Act

318 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340



48 training sessions across the state of Pennsylvania as well as offering to provide an hour-

long presentation on core financial literacy concepts on request. The number of high schools

participating grew rapidly starting in 2015. In 2018, the State Treasurer of Pennsylvania

endorsed the program and encouraged all Pennsylvania high schools to participate. That

same year, representatives from the CFA Society Pittsburgh led a session for the National

Association of State Treasurers in Providence, Rhode Island, in an effort to further expand

outreach across more states. The CFA Society Pittsburgh provides participating schools with

instructional materials for a semester-long equivalent course based on The Missing Semester

(Kabala & Natali, 2012). Each school determines the best way to deliver their programs

(e.g., weekly for a semester, or daily lessons over fewer weeks). Schools are supplied with

PowerPoint resources to accompany the book, along with a web-based portal (available

through CFA Society Pittsburgh website) of best practices and exercises, submitted from

previous participating schools. In addition, members of the financial literacy committee

extend an invitation to present to the students in the classroom for a day. The financial liter-

acy member uses a PowerPoint presentation developed by the financial literacy committee,

which gives a broad overview of the main topics discussed within the curriculum, while

allowing students to ask questions relating to financial literacy subject material or real-world

applications. Committee members and other teachers act as a resource for the participating

schools throughout the semester, to provide any assistance or feedback about the

curriculum.

For the 2018-2019 school year, a total of 78 high schools, spread across seven states

(California, Michigan, New Jersey, Pennsylvania, West Virginia, Wisconsin, and Wyoming)

with 173 classes/teachers, were invited to participate in both a pre- and post-survey to exam-

ine the effectiveness of financial literacy education. Through our partnership with the CFA

Society Pittsburgh, ACT 48 training in Pennsylvania, multiple Intermediate Units, the

Pennsylvania State Treasurer, and other connections with individual teachers, we established

our subset of participating teachers. We offered each teacher financial literacy materials and

access to the CFA Society Pittsburgh portal for free, in exchange for their participation in

our survey. After gathering the list of enrolled teachers, we assigned each a unique class

code. Links for a pre- and post-surveys were provided to the participating teachers. Within

the introductory email, instructors were given directions to assign each student with a unique

ID number, allowing pre- and post-surveys to be matched for analysis. Participating schools

agreed to administer the presurvey before any instructional delivery. Post-surveys were to be

completed within a week after completion of the last instructional unit on financial literacy.

The distributed survey was designed as an extension of the work conducted by Filbeck

and Zhao (2018) with the addition of six objective financial knowledge questions. The full

pre-survey and post-survey can be found in Appendixes A and B, respectively. Of the origi-

nal population, 1,613 students from 31 participating schools and two states (Pennsylvania

and New Jersey) completed pre-surveys. A total of 1,050 post-surveys were completed by

students from 23 schools in Pennsylvania (91.2%) and New Jersey (8.8%).

Table 1 reports the descriptive statistics for the full sample and the test sample. The full

sample consists of 1,613 students completing the pre-survey, while the matched test sample

includes 829 students who submitted both a pre- and post-survey. The matched sample totals

show attrition in the survey process of approximately one-half, despite attempts to minimize

G. Filbeck et al. / Financial Services Review 28 (2020) 315–340 319



the loss through a series of six reminder emails to participating teachers during the period of

the program. Based on feedback from teachers, failure of students to complete post-surveys

were primarily associated with a failure of completion of the financial literary program or a

failure of teachers to oversee students in the completion of the post-survey. We also had 221

instances of students filling out post-surveys who previously did not complete a pre-survey.

Of the full sample, 1,425 (88%) students are in their junior or senior year; in the test sample,

718 (87%) students are in their junior or senior year. Female students account for approxi-

mately 47% in both the full sample and the test sample. Regarding their favorite subjects,

students within the full sample favored science (28%) and math (27%), a trend which

Table 1 Sample description

Grade

9th 10th 11th 12th Total

Panel A. Whole sample
Male
English 4 4 23 64 95
Math 18 13 59 171 261
Science 8 20 51 147 226
Social Studies 24 22 59 171 276

Female
English 12 10 47 155 224
Math 9 14 43 116 182
Science 6 11 34 170 221
Social Studies 9 4 22 93 128
Total 90 98 338 1,087 1,613

Panel B. Test sample
Male
English 3 1 12 31 47
Math 11 7 33 75 126
Science 4 14 31 69 118
Social Studies 12 14 34 84 144

Female
English 6 3 27 81 117
Math 6 8 27 61 102
Science 1 11 16 82 110
Social Studies 6 4 15 40 65
Total 49 62 195 523 829

Panel C. School district characteristics

Mean
Standard
deviation

Percentile

Min 25 50 75 Max

Population 31,522.95 35,083 6,208 11,382 18,412 37,567 148,678
Poverty 9.6% 5.9% 3.3% 5.9% 8.2% 11.3% 25.8%
Pct_college 29.5% 14.8% 11.2% 16.3% 29.8% 41.2% 57.8%

Note: Number of students across different grade level and different favorite subjects for the whole sample

(Panel A) and the test sample (Panel B). Panel C shows the school district characteristics: Population is the pop-

ulation in the school district. Poverty is the poverty rate in the school district. Pct_college is the percentage of

residents who have attained bachelor degree or higher.

320 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340



continued to the test sample with math and science as the favorite subjects both at 27.5%.

Favorite subject is included in our survey to determine whether academic interest area plays

a significant role in financial literacy educational outreach.

Appendixes A and B show the pre- and post-survey questions addressed by program

participants in order to assess financial knowledge (both objective and subjective ques-

tions), financial behavior, and self-esteem. Lusardi and Mitchell (2014) point out that a

substantial mismatch exists between individual’s self-assessed (subjective) financial

knowledge and their actual knowledge. The survey design is consistent with Hastings,

Madrian, and Skimmyhorn (2013), who argue that financial literacy should focus on com-

petences that individuals need. The organization of the survey is constructed in a way

that assesses the four major keys to financial success: financial self-esteem, perceived

(subjective) financial knowledge, financial behavior, and objective financial knowledge

with numeracy. The financial behavior and subjective financial knowledge test financial

self-esteem and perceived financial knowledge by gauging the student’s self-reported

understanding.

Questions are broadly categorized into two types: financial knowledge (subjective

and objective) and financial behavior. The only difference between pre- and post-survey

questions are in the objective financial knowledge questions—the same concepts are

tested with different questions. Objective financial knowledge questions are asked in a

manner that contains a right or wrong answer. Each objective financial knowledge ques-

tion contains at least one wrong answer and a choice of “I Don’t Know.” The questions

are based upon five major categories of financial literacy: risk diversification, compound

interest, credit, numeracy (interest), and inflation. The questions are analyzed using two

methods: correctness and willingness to answer. The first method of correctness assigns

a 1 for each correct answer and 0 for any other answer. The second method to measure

willingness assigns a 1 for an answer of I Don’t Know and 0 for any other answer. The

second method is used to measure financial self-esteem as measured by the amount of

questions answered with I Don’t Know. Improved self-esteem occurs as students

become less likely to answer I Don’t Know and instead select an answer that could be

correct or incorrect, showing greater confidence after completing financial literacy

education.

The survey consists of 21 overall questions: four financial behavior, six objective finan-

cial knowledge, and 11 subjective financial knowledge. Financial behavior and subjective fi-

nancial knowledge questions are rated on a 5-point scale ranging from 1 = strongly disagree

to 5 = strongly agree. The four financial behavior questions are “I like to save money more

than I like to spend it,” “I have a checking and/or a savings account,” “I have conversations

with my parents regarding personal finance,” and “I think it is important to contribute to a

retirement plan (ex: Roth IRA, 401k, etc.)”. Subjective financial knowledge questions

involve perceived understanding of financial concepts, and include questions such as “I

understand how to establish a financial plan,” “I think financial literacy is important for my

future,” and “I understand the process by which my parents/guardians make financial deci-

sions.” The final six objective knowledge questions test financial self-esteem and objective

financial knowledge by assessing correctness of answers and willingness to select an answer

other than I Don’t Know.

G. Filbeck et al. / Financial Services Review 28 (2020) 315–340 321



5. Test results

First, t test and path analysis are used to analyze the pre-survey results of the full sample.

Following the pre-survey analysis, we run t test and regressions to compare pre- and post-

survey results within the test sample.

5.1. Pre-survey results

For the t test of the pre-survey responses, two student characteristics are present: gender

and GPA. Each characteristic divides the full sample into two groups. Gender is broken into

a subgroup for females and males. The median GPA of the whole sample divides the stu-

dents into a higher GPA or lower GPA group.

The response differences between groups are shown in Table 2. The average responses

are compared with the individual subgroups of gender and GPA to determine if the charac-

teristics exhibit a statistically significant effect. The gender characteristic identifies the effect

of a student being a female versus a male (gender), as well as the effect of a student having a

high GPA versus low GPA. The data shows no statistically significant effect of GPA on sub-

jective financial knowledge. However, consistent with Cupák et al. (2018), female students

tend to have lower subjective financial knowledge compared with male students (statistically

significant at the 1% level). For financial behavior, students with a high GPA tend to be bet-

ter financially behaved (statistically significant at the 1% level). In objective financial

knowledge, females tend to score lower in correctness, while students with high GPAs tend

to score higher in correctness (both significant at the 1% level). Similarly, for self-esteem,

females tend to answer I Don’t Know more often, while students with higher GPAs are less

likely to select I Don’t Know (both statistically significant at the 1% level).

5.2. Pearson correlation and path analysis

Table 3 reports Pearson correlations between financial behavior and self-esteem (IDK),

subjective, and objective financial knowledge. Financial behavior, objective and subjective

knowledge are significantly correlated with each other. IDK is significantly negatively corre-

lated with the other three measures of financial literacy, which indicates a positive correla-

tion of self-esteem level and the other three measures of financial literacy. Naturally, due to

the setup of the self-esteem measure, we find a �0.77 correlation between IDK answers and

objective knowledge.

Next, following Tang and Baker (2016), we use path models to analyze the relationship

between characteristics and four financial literacy measures. We choose path analysis, as it

forces us to specify relationships among all of the independent variables. This results in a

model showing causal mechanisms, through which independent variables produce both

direct and indirect effects on a dependent variable. All causal relationships between varia-

bles must go in one direction only, for path models. To identify the impact, we assign the de-

pendent variable as Behavior (Subjective, IDK), which is the total score for financial

behavior questions (subjective questions, objective questions with I Don’t Know answers).

322 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340



T
ab
le

2
D
if
fe
re
n
ce
s
b
as
ed

o
n
st
u
d
en
t
ch
ar
ac
te
ri
st
ic
s:
P
re
-s
u
rv
ey

re
su
lt
s

A
v
er
ag
e

re
sp
o
n
se

F
em

al
e

H
ig
h
er

G
P
A

P
an
el
A
.
S
u
b
je
ct
iv
e
fi
n
an
ci
al
k
n
o
w
le
d
g
e
q
u
es
ti
o
n
s

2
.
I
u
n
d
er
st
an
d
h
o
w
to

es
ta
b
li
sh

a
fi
n
an
ci
al
p
la
n
.

3
.0
3
9

�0
.2
0
4
*
*
*

�0
.0
2
3

3
.
I
th
in
k
fi
n
an
ci
al
li
te
ra
cy

is
im

p
o
rt
an
t
fo
r
m
y
fu
tu
re
.

4
.3
6
5

0
.0
2
1

0
.1
8
6
*
*
*

6
.
I
u
n
d
er
st
an
d
th
e
p
ro
ce
ss

b
y
w
h
ic
h
m
y
p
ar
en
ts
/g
u
ar
d
ia
n
s
m
ak
e
fi
n
an
ci
al
d
ec
is
io
n
s.

3
.3
5
8

�0
.1
4
3
*
*
*

�0
.0
9
3
*

7
.
I
k
n
o
w
h
o
w
to

d
et
er
m
in
e
th
e
ap
p
ro
p
ri
at
e
to
ta
l
co
st
s
as
so
ci
at
ed

w
it
h
th
e
co
ll
eg
es
/u
n
iv
er
si
ti
es

I
am

in
te
re
st
ed

in
at
te
n
d
in
g
.

3
.0
8
6

�0
.0
6
6

0
.1
4
4
*
*

8
.
I
u
n
d
er
st
an
d
th
e
p
ro
ce
ss

b
y
w
h
ic
h
lo
an

re
p
ay
m
en
ts
ta
k
e
p
la
ce

in
cl
u
d
in
g
th
e
im

p
ac
t
o
f
in
te
re
st
,
d
el
in
q
u
en
cy

an
d
d
ef
au
lt
.

2
.7
4
3

�0
.2
5
9
*
*
*

�0
.1
6
0
*
*

9
.
I
u
n
d
er
st
an
d
th
e
p
ro
ce
ss

b
y
w
h
ic
h
cr
ed
it
ca
rd

ch
ar
g
es

an
d
re
p
ay
m
en
t
sc
h
ed
u
le
s
ca
n
im

p
ac
t
th
e
le
v
el
o
f

fi
n
an
ci
al
d
eb
t
le
v
el
s.

3
.4
2
8

0
.0
0
1

�0
.0
5
3

1
0
.
W
h
en

it
co
m
es

to
p
u
rc
h
as
in
g
a
ca
r,
I
k
n
o
w
h
o
w
to

d
et
er
m
in
e
h
o
w
m
u
ch

o
f
a
ca
r
I
ca
n
af
fo
rd
.

3
.2
8
7

�0
.2
8
4
*
*
*

�0
.2
5
9
*
*
*

1
1
.
I
u
n
d
er
st
an
d
h
o
w
to

ev
al
u
at
e
th
e
co
st
-b
en
efi
t
an
al
y
si
s
o
f
tr
ai
n
in
g
fo
r
th
e
jo
b
I
w
o
u
ld

li
k
e
to

p
er
fo
rm

af
te
r

co
m
p
le
ti
n
g
sc
h
o
o
l.

3
.0
1
5

�0
.2
1
9
*
*
*

�0
.1
2
3
*
*

1
2
.
I
k
n
o
w
w
h
at
a
R
o
th

IR
A
is
an
d
h
o
w
it
w
o
rk
s
fr
o
m

a
ta
x
at
io
n
st
an
d
p
o
in
t.

1
.9
5
1

�0
.1
8
1
*
*
*

�0
.0
9
5
*

1
3
.
I
k
n
o
w
h
o
w
to

cr
ea
te
a
sa
v
in
g
s
p
la
n
b
as
ed

o
n
th
e
ab
il
it
y
to

es
ti
m
at
e
m
o
n
th
ly

li
v
in
g
ex
p
en
se
s.

3
.1
4
8

�0
.1
6
7
*
*
*

0
.0
0
5

1
4
.
I
k
n
o
w
h
o
w
to

p
la
n
fi
n
an
ci
al
ly

fo
r
re
ti
re
m
en
t.

2
.6
2
1

�0
.2
8
3
*
*
*

�0
.0
5
0

T
o
ta
l
sc
o
re

fo
r
su
b
je
ct
iv
e
fi
n
an
ci
al
k
n
o
w
le
d
g
e

3
3
.8
9
3

�1
.6
2
2
*
*
*

�0
.3
8
1

P
an
el
B
.
F
in
an
ci
al
b
eh
av
io
r
q
u
es
ti
o
n
s

1
.
I
li
k
e
to

sa
v
e
m
o
n
ey

m
o
re

th
an

I
li
k
e
to

sp
en
d
it
.

3
.5
0
5

�0
.0
1
3
*
*

0
.2
1
9
*
*
*

4
.
I
h
av
e
a
ch
ec
k
in
g
an
d
/o
r
a
sa
v
in
g
s
ac
co
u
n
t.

4
.3
8
1

0
.0
3
4

0
.1
8
6
*
*
*

5
.
I
h
av
e
co
n
v
er
sa
ti
o
n
s
w
it
h
m
y
p
ar
en
ts
re
g
ar
d
in
g
p
er
so
n
al
fi
n
an
ce
.

3
.4
1
8

0
.0
1
2

0
.2
8
3
*
*
*

1
5
.
I
th
in
k
it
is
im

p
o
rt
an
t
to

co
n
tr
ib
u
te
to

a
re
ti
re
m
en
t
p
la
n
(e
x
:
R
o
th

IR
A
,
4
0
1
k
,
et
c.
).

4
.0
8
3

�0
.0
1
0

0
.3
1
8
*
*
*

T
o
ta
l
sc
o
re

fo
r
fi
n
an
ci
al
b
eh
av
io
r

1
4
.7
4
6

0
.1
2
2

1
.2
3
1
*
*
*

P
an
el
C
.
O
b
je
ct
iv
e
q
u
es
ti
o
n
s
(C
o
rr
ec
t
A
n
sw

er
s)
a

1
.
Is
it
sa
fe
r
to

p
u
t
y
o
u
r
m
o
n
ey

in
to

o
n
e
in
v
es
tm

en
t
o
r
p
u
t
y
o
u
r
m
o
n
ey

in
to

m
u
lt
ip
le
in
v
es
tm

en
ts
?

0
.6
1
9

�0
.1
2
4
*
*
*

0
.0
9
4
*
*
*

2
.
If
y
o
u
in
v
es
t
$
1
0
0
in

a
R
o
th

IR
A
an
d
ea
rn

1
0
%

p
er

y
ea
r
fo
r
3
y
ea
rs
,
h
o
w
m
u
ch

w
o
u
ld

it
b
e
w
o
rt
h
at
th
e

en
d
o
f
th
re
e
y
ea
rs
.

0
.2
8
9

�0
.1
0
8
*
*
*

0
.0
7
3
*
*
*

3
.
If
y
o
u
u
se

a
cr
ed
it
ca
rd

in
Ja
n
u
ar
y
fo
r
a
to
ta
l
o
f
$
3
0
0
,
w
h
ic
h
p
ay
m
en
t
o
p
ti
o
n
w
il
l
re
su
lt
in

th
e
lo
w
es
t
am

o
u
n
t

o
f
o
v
er
al
l
in
te
re
st
p
ai
d
.

0
.4
9
3

�0
.0
7
5
*
*
*

0
.1
3
5
*
*
*

4
.
S
u
p
p
o
se

y
o
u
d
ec
id
e
to

b
u
y
a
B
M
W

fo
r
$
5
0
,0
0
0
.
If
y
o
u
ta
k
e
o
u
t
an

au
to

lo
an

fo
r
5
y
ea
rs
w
it
h
5
%

in
te
re
st
,

h
o
w
m
u
ch

to
ta
l
w
il
l
y
o
u
p
ay

p
er

y
ea
r?

0
.4
2
9

�0
.1
4
2
*
*
*

0
.1
6
7
*
*
*

5
.
In

th
e
fu
tu
re
,
th
e
co
st
o
f
th
in
g
s
y
o
u
b
u
y
d
o
u
b
le
s
A
N
D
y
o
u
r
in
co
m
e
al
so

d
o
u
b
le
s.
H
o
w
m
u
ch

w
il
l
y
o
u
b
e

ab
le
to

b
u
y
in

th
e
fu
tu
re

in
co
m
p
ar
is
o
n
to

to
d
ay
?

0
.5
5
7

�0
.0
6
6
*
*
*

0
.1
7
3
*
*
*

6
.
S
u
p
p
o
se

y
o
u
h
av
e
$
3
0
,0
0
0
in

st
u
d
en
t
lo
an
s.
W
h
ic
h
p
ay
m
en
t
o
p
ti
o
n
w
o
u
ld

re
su
lt
in

th
e
lo
w
es
t
am

o
u
n
t
o
f

o
v
er
al
l
in
te
re
st
p
ai
d
?

0
.5
3
0

�0
.0
5
4
*
*

0
.2
3
5
*
*
*

T
o
ta
l
sc
o
re

fo
r
o
b
je
ct
iv
e
q
u
es
ti
o
n
s
(C
o
rr
ec
t
A
n
sw

er
s)

2
.9
1
6

�0
.5
7
0
*
*
*

0
.8
7
7
*
*
*

(c
o
n
ti
n
u
ed

o
n
n
ex
t
p
a
g
e)

G. Filbeck et al. / Financial Services Review 28 (2020) 315–340 323



T
ab
le

2
(C
o
n
ti
n
u
ed
)

A
v
er
ag
e

re
sp
o
n
se

F
em

al
e

H
ig
h
er

G
P
A

P
an
el
D
.
O
b
je
ct
iv
e
q
u
es
ti
o
n
s
("
I
D
o
n
’t
K
n
o
w
"
A
n
sw

er
s)
b

1
.
Is
it
sa
fe
r
to

p
u
t
y
o
u
r
m
o
n
ey

in
to

o
n
e
in
v
es
tm

en
t
o
r
p
u
t
y
o
u
r
m
o
n
ey

in
to

m
u
lt
ip
le
in
v
es
tm

en
ts
?

0
.2
6
3

0
.1
2
1
*
*
*

�0
.0
1
2

2
.
If
y
o
u
in
v
es
t
$
1
0
0
in

a
R
o
th

IR
A
an
d
ea
rn

1
0
%

p
er

y
ea
r
fo
r
3
y
ea
rs
,
h
o
w
m
u
ch

w
o
u
ld

it
b
e
w
o
rt
h
at
th
e

en
d
o
f
th
re
e
y
ea
rs
.

0
.3
4
5

0
.1
7
6
*
*
*

�0
.0
8
6
*
*
*

3
.
If
y
o
u
u
se

a
cr
ed
it
ca
rd

in
Ja
n
u
ar
y
fo
r
a
to
ta
l
o
f
$
3
0
0
,
w
h
ic
h
p
ay
m
en
t
o
p
ti
o
n
w
il
l
re
su
lt
in

th
e
lo
w
es
t

am
o
u
n
t
o
f
o
v
er
al
l
in
te
re
st
p
ai
d
.

0
.3
4
0

0
.0
4
0
*

�0
.0
6
5
*
*
*

4
.
S
u
p
p
o
se

y
o
u
d
ec
id
e
to

b
u
y
a
B
M
W

fo
r
$
5
0
,0
0
0
.
If
y
o
u
ta
k
e
o
u
t
an

au
to

lo
an

fo
r
5
y
ea
rs
w
it
h
5
%

in
te
re
st
,

h
o
w
m
u
ch

to
ta
l
w
il
l
y
o
u
p
ay

p
er

y
ea
r?

0
.2
9
5

0
.1
3
7
*
*
*

�0
.0
6
4
*
*
*

5
.
In

th
e
fu
tu
re
,
th
e
co
st
o
f
th
in
g
s
y
o
u
b
u
y
d
o
u
b
le
s
A
N
D
y
o
u
r
in
co
m
e
al
so

d
o
u
b
le
s.
H
o
w
m
u
ch

w
il
l
y
o
u
b
e

ab
le
to

b
u
y
in

th
e
fu
tu
re

in
co
m
p
ar
is
o
n
to

to
d
ay
?

0
.1
8
6

0
.0
5
2
*
*
*

�0
.0
7
1
*
*
*

6
.
S
u
p
p
o
se

y
o
u
h
av
e
$
3
0
,0
0
0
in

st
u
d
en
t
lo
an
s.
W
h
ic
h
p
ay
m
en
t
o
p
ti
o
n
w
o
u
ld

re
su
lt
in

th
e
lo
w
es
t
am

o
u
n
t
o
f

o
v
er
al
l
in
te
re
st
p
ai
d
?

0
.2
8
7

0
.0
7
4
*
*
*

�0
.1
2
3
*
*
*

T
o
ta
l
sc
o
re

fo
r
o
b
je
ct
iv
e
q
u
es
ti
o
n
s
("
I
D
o
n
’t
K
n
o
w
"
A
n
sw

er
s)

1
.7
1
7

0
.5
9
9
*
*
*

�0
.4
2
1
*
*
*

N
o
te
:
S
h
o
w
s
th
e
d
if
fe
re
n
ce
s
o
f
p
re
-s
u
rv
ey

st
u
d
en
t
re
sp
o
n
se
s
o
n
fi
n
an
ci
al

k
n
o
w
le
d
g
e
an
d
fi
n
an
ci
al

b
eh
av
io
ra
l
q
u
es
ti
o
n
s
ac
ro
ss

d
if
fe
re
n
t
g
en
d
er

an
d
G
P
A

fo
r
th
e
w
h
o
le

sa
m
p
le
.
G
en
d
er

is
b
ro
k
en

in
to

a
S
u
b
-g
ro
u
p
fo
r
fe
m
al
es

an
d
m
al
es
.
T
h
e
m
ed
ia
n
G
P
A

o
f
th
e
w
h
o
le

sa
m
p
le

d
iv
id
es

th
e
st
u
d
en
ts

as
a
h
ig
h
er

G
P
A
o
r
lo
w
er

G
P
A
g
ro
u
p
.
T
h
e
g
en
d
er

ch
ar
ac
te
ri
st
ic

id
en
ti
fi
es

th
e
ef
fe
ct

o
f
a
st
u
d
en
t
b
ei
n
g
a
fe
m
al
e
v
er
su
s
a
m
al
e,
as

w
el
l
as

th
e
ef
fe
ct

o
f
a
st
u
d
en
t
h
av
in
g

a
h
ig
h
G
P
A
v
er
su
s
lo
w
G
P
A
.

*
*
*
,
*
*
,
*
in
d
ic
at
e
st
at
is
ti
ca
l
si
g
n
ifi
ca
n
ce

at
0
.0
1
,
0
.0
5
,
an
d
0
.1
0
le
v
el
,
re
sp
ec
ti
v
el
y
.

a
Im

p
ro
v
em

en
t
is
in
d
ic
at
ed

b
y
a
p
o
si
ti
v
e
d
if
fe
re
n
ce

an
d
t-
st
at
(m

o
re

p
eo
p
le
se
le
ct
in
g
th
e
co
rr
ec
t
an
sw

er
).

b
Im

p
ro
v
em

en
t
is
in
d
ic
at
ed

b
y
a
n
eg
at
iv
e
d
if
fe
re
n
ce

an
d
t-
st
at
(l
es
s
p
eo
p
le
se
le
ct
in
g
“I

D
o
n
’t
K
n
o
w
”)
.

324 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340



Three path models are used to test the effect of financial behavior scores (Behavior), subjec-

tive financial knowledge scores (Subjective), and I Don’t Know answers in the objective

questions (IDK). In all three path models, we include gender (Female), grade level

(Sophomore, Junior, and Senior), favorite subject (English, Math, and Science), GPA and fa-

vorite learning method (Learning by doing [LBD], Listening, Discussing, and Visual) as in-

dependent variables. Control variables LBD, Listening, Discussing, and Visual allow for

direct testing of instructional preferences of students which may impact success of the finan-

cial literacy program (Amagir, Groot, van den Brink, & Wilschut 2018). Detailed definitions

of variables are listed in Appendix C.

Also, to explore whether the school district of participating classes has any effect on the

pre-survey results, we include three school district characteristics as control variables.

Specifically, for each participating school, we collect its school district census data from

https://censusreporter.org/. We exclude private schools, virtual, finance knowledge learning

centers, and chartered schools from the whole sample as these schools are hard to determine

their school districts. For each public high school, we collect its school district data on its

population, poverty rate and percentage of residents who attained bachelor or higher

degrees. Also, because the average number of students who participated in the pre-surveys

from each school is 20 with a median value of one (ranges from 1 to 225 students per

school), we removed school districts with less than 20 student participants. This process

reduces our whole sample from 1,613 to 1,441 pre-surveys with available school district

data. Summary statistics of school district data are reported in Panel C of Table 1.

Table 4 illustrates the results from these models. Path (1) uses Behavior as the dependent

variable. The results show students of higher grade level (sophomore, junior, and senior),

with high subjective financial knowledge, higher level of self-esteem, as well as a high

GPA, have a higher probability to be better behaved financially. Both the subjective financial

knowledge coefficient and self-esteem measure (IDK) are statistically significant at the 1%

level, implying that a student who possesses more subjective knowledge and higher self-

esteem are more likely to exhibit better financial behavior. On school district characteristics,

Table 3 Shows the correlation coefficients between financial behavior and self-esteem (IDK answers), objec-

tive and subjective financial knowledge

Subjective Behavior Objective IDK answers

Subjective
Corr 1.000
p-value

Behavior
Corr 0.334*** 1.000
p-value <0.0001

Objective
Corr 0.277*** 0.245*** 1.000
p-value <0.0001 <0.0001

IDK answers
Corr �0.361*** �0.249*** �0.771*** 1.000
p-value <0.0001 <0.0001 <0.0001

***, **, * indicate statistical significance at 0.01, 0.05 and 0.10 level, respectively.

G. Filbeck et al. / Financial Services Review 28 (2020) 315–340 325



results show school districts with higher poverty rate are more likely to have worse financial

behavior (statistically significant at the 1% level).

The second model (Path 2) uses Subjective (representing the measure for subjective

knowledge) as the dependent variable. The results show female students with low GPAs are

more likely to have lower subjective financial knowledge. Additionally, students who prefer

the learning styles of learning by doing (LBD), listening, or discussing are more likely to ex-

hibit greater subjective financial knowledge (all statistically significant at the 5% level). The

negative coefficient of IDK (statistically significant at the 1% level) shows that a student

who has higher level of self-esteem is also more likely to be subjectively knowledgeable in

finance. The insignificant coefficient of Objective (representing the measure for objective

knowledge) shows the positive linkage between subjective and objective financial knowl-

edge, which is consistent to the results of Tang and Baker (2016) who find disconnect

between subjective and objective knowledge when comparing actual and perceived financial

Table 4 Regressions on student characteristics: Pre-survey results

Path (1) Path (2) Path (3)

Dep. Var.: Behavior Dep. Var.: Subjective Dep. Var.: IDK

Coefficient t-Value Coefficient t-Value Coefficient t-Value

IDK �0.093 �3.53*** �0.330 �13.43***
Objective 0.010 0.38 0.032 1.18
Subjective 0.296 11.91***
Female 0.042 1.59 �0.035 �1.28 0.172 6.38***
Sophomore 0.062 1.82* �0.037 �1.04 �0.024 �0.65
Junior 0.205 4.22*** �0.066 �1.31 �0.043 �0.82
Senior 0.294 5.66*** �0.074 �1.36 �0.133 �2.37**
GPA 0.150 5.72*** �0.046 �1.68* �0.142 �5.18***
English �0.024 �0.80 0.039 1.27 0.036 1.12
Math �0.015 �0.51 0.015 0.49 �0.025 �0.78
Science �0.031 �1.04 0.020 0.64 �0.027 �0.83
LBD 0.043 1.70* 0.066 2.52** �0.015 �0.55
Listening �0.005 �0.20 0.066 2.49** �0.060 �2.18**
Discussing �0.005 �0.20 0.059 2.29** �0.008 �0.28
Visual 0.047 1.86* �0.029 �1.09 0.048 1.77*
Log(Population) 0.019 0.72 0.079 2.91*** �0.038 �1.34
Poverty �0.130 �3.72*** 0.034 0.92 0.065 1.74*
Pct_College 0.041 1.22 0.008 0.24 �0.029 �0.80
Note: Table 4 Shows the regression results on student characteristics of the whole sample. Behavior (objec-

tive, subjective, IDK) is the total score for financial behavior questions (subjective questions, objective questions

with correct answers, objective questions with “I Don’t Know” answers). Female is a dummy variable that is

equal to 1 if the student is a female student and 0 otherwise. Sophomore (junior, senior) is a dummy variable

that is equal to 1 if the student is a sophomore (junior, senior), and 0 otherwise. English (math, science) is a

dummy variable which is equal to 1 if the student’s favorite subject is English (math, science), and 0 otherwise.

LBD (listening, discussing, visual) is a dummy variable that is equal to 1 if the student chooses learning by

doing (listening, discussing with peers, features visual support) as favorite instruction method, and 0 otherwise.

GPA is a student’s grade point average. Log(population) is the log of the population in the school district.

Poverty is the poverty rate in the school district. Pct_college is the percentage of residents who have attained

bachelor degree or higher.

***, **, * indicate statistical significance at 0.01, 0.05, and 0.10 level, respectively.

326 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340



knowledge. Students in larger school district are more likely to be more subjectively knowl-

edgeable in finance (statistically significant at the 1% level). One possible explanation for

this finding is that students who live in larger school districts may be exposed to more finan-

cial knowledge/concepts.

The third model (Path 3) uses IDK (representing the measure of self-esteem) as the de-

pendent variable. Female students exhibit lower self-esteem scores, while junior or senior

students who prefer learning by listening are more likely to have higher self-esteem. On

school district characteristics, results show school districts with high poverty rate are more

likely to have lower level of self-esteem (statistically significant at the 10% level).

Overall, our results are consistent with Tang and Baker (2016) and suggest that self-

esteem plays a statistically significant role in each of the remaining three variables being

studied: objective knowledge, subjective knowledge, and financial behavior.

The results in Table 4 need to be interpreted with caution due to the possible endogeneity

issues among four financial literacy measures and omitted variable problem. We cannot com-

pletely rule out the broader theoretical concern of a reverse causality among these variables

although our Hausman test statistics for endogeneity cannot reject the null hypothesis of no mea-

surement error. As for omitted variable bias issue, some variables such as subjective knowledge

and self-esteem measure may potentially cause omitted variable bias. Future studies can test and

expand upon our results by incorporating additional control variables when data are available.

5.3. Post-survey results

We compare the results of the pre- and post-survey using our test sample of 829 matched

students. We define improvement in several ways. For subjective financial knowledge and fi-

nancial behavior, we define gains as the post-survey scores minus the pre-survey response

scores. For objective financial knowledge questions, we define gains in financial knowledge

as the difference between the post-survey scores and the pre-survey response. To gauge fi-

nancial self-esteem, we define confidence gains as a decrease in the responses of I Don’t

Know in the post-survey minus the pre-survey. We run univariate tests on the gains in our

test samples and subsamples.

Table 5 illustrates the t test results by question and overall score for each of the four char-

acteristics measured: subjective financial knowledge, financial behavior, objective financial

knowledge, and financial self-esteem. The results show a profound, statistically significant

improvement across all areas measured, with 24 of 25 questions showing total score

improvement as statistically significant at the 1% level. As a result, the data shows financial

literacy educational efforts can lead to better student results in financial literacy understand-

ing and behavior.

Within subjective financial knowledge, all questions result in statistically significant

improvement at the 1% level. The biggest gains come from understanding of Roth IRA

(gain of 1.682) and retirement (1.401). Seven of the 12 questions result in a greater than

25% improvement. The biggest gain in financial behavior derives from conversations with

parents on personal finance (gain of 0.033). These results closely mirror the research of

Filbeck and Zhao (2018), who find the largest growth within the same three questions.

G. Filbeck et al. / Financial Services Review 28 (2020) 315–340 327



T
ab
le

5
T
-t
es
t
re
su
lt
s
b
et
w
ee
n
p
re
-
an
d
p
o
st
-s
u
rv
ey

P
re

P
o
st

D
if
f

T
-s
ta
t

P
an
el
A
.
F
in
an
ci
al
su
b
je
ct
iv
e
k
n
o
w
le
d
g
e
q
u
es
ti
o
n
s

2
.
I
u
n
d
er
st
an
d
h
o
w
to

es
ta
b
li
sh

a
fi
n
an
ci
al
p
la
n
.

3
.0
4
7

4
.0
0
0

0
.9
5
3

2
1
.6
2
*
*
*

3
.
I
th
in
k
fi
n
an
ci
al
li
te
ra
cy

is
im

p
o
rt
an
t
fo
r
m
y
fu
tu
re
.

4
.3
9
1

4
.6
9
2

0
.2
8
0

1
0
.3
0
*
*
*

6
.
I
u
n
d
er
st
an
d
th
e
p
ro
ce
ss

b
y
w
h
ic
h
m
y
p
ar
en
ts
/g
u
ar
d
ia
n
s
m
ak
e
fi
n
an
ci
al
d
ec
is
io
n
s.

3
.3
6
9

3
.8
3
2

0
.4
6
3

1
1
.3
2
*
*
*

7
.
I
k
n
o
w
h
o
w
to

d
et
er
m
in
e
th
e
ap
p
ro
p
ri
at
e
to
ta
l
co
st
s
as
so
ci
at
ed

w
it
h
th
e
co
ll
eg
es
/u
n
iv
er
si
ti
es

I
am

in
te
re
st
ed

in
at
te
n
d
in
g
.

3
.1
1
9

3
.8
6
5

0
.7
4
5

1
6
.3
5
*
*
*

8
.
I
u
n
d
er
st
an
d
th
e
p
ro
ce
ss

b
y
w
h
ic
h
lo
an

re
p
ay
m
en
ts
ta
k
e
p
la
ce

in
cl
u
d
in
g
th
e
im

p
ac
t
o
f
in
te
re
st
,
d
el
in
q
u
en
cy

an
d
d
ef
au
lt
.

2
.7
2
3

3
.8
3
2

1
.1
0
9

2
3
.7
6
*
*
*

9
.
I
u
n
d
er
st
an
d
th
e
p
ro
ce
ss

b
y
w
h
ic
h
cr
ed
it
ca
rd

ch
ar
g
es

an
d
re
p
ay
m
en
t
sc
h
ed
u
le
s
ca
n
im

p
ac
t
th
e
le
v
el
o
f

fi
n
an
ci
al
d
eb
t
le
v
el
s.

3
.4
1
8

4
.1
9
9

0
.7
7
7

1
7
.0
5
*
*
*

1
0
.
W
h
en

it
co
m
es

to
p
u
rc
h
as
in
g
a
ca
r,
I
k
n
o
w
h
o
w
to

d
et
er
m
in
e
h
o
w
m
u
ch

o
f
a
ca
r
I
ca
n
af
fo
rd
.

3
.2
5
8

4
.1
5
0

0
.8
9
3

1
9
.8
3
*
*
*

1
1
.
I
u
n
d
er
st
an
d
h
o
w
to

ev
al
u
at
e
th
e
co
st
-b
en
efi
t
an
al
y
si
s
o
f
tr
ai
n
in
g
fo
r
th
e
jo
b
I
w
o
u
ld

li
k
e
to

p
er
fo
rm

af
te
r

co
m
p
le
ti
n
g
sc
h
o
o
l.

2
.9
9
8

3
.8
7
7

0
.8
7
8

1
9
.2
5
*
*
*

1
2
.
I
k
n
o
w
w
h
at
a
R
o
th

IR
A
is
an
d
h
o
w
it
w
o
rk
s
fr
o
m

a
ta
x
at
io
n
st
an
d
p
o
in
t.

1
.8
7
5

3
.5
5
5

1
.6
8
2

3
2
.3
5
*
*
*

1
3
.
I
k
n
o
w
h
o
w
to

cr
ea
te
a
sa
v
in
g
s
p
la
n
b
as
ed

o
n
th
e
ab
il
it
y
to

es
ti
m
at
e
m
o
n
th
ly

li
v
in
g
ex
p
en
se
s.

3
.1
1
0

4
.1
4
8

1
.0
3
6

2
2
.4
1
*
*
*

1
4
.
I
k
n
o
w
h
o
w
to

p
la
n
fi
n
an
ci
al
ly

fo
r
re
ti
re
m
en
t.

2
.5
5
5

3
.9
5
0

1
.4
0
1

2
8
.2
8
*
*
*

T
o
ta
l
sc
o
re

fo
r
fi
n
an
ci
al
su
b
je
ct
iv
e
k
n
o
w
le
d
g
e

3
3
.7
8
0

4
3
.6
7
7

9
.8
9
7

2
9
.1
7
*
*
*

P
an
el
B
.
F
in
an
ci
al
b
eh
av
io
r
q
u
es
ti
o
n
s

1
.
I
li
k
e
to

sa
v
e
m
o
n
ey

m
o
re

th
an

I
li
k
e
to

sp
en
d
it
.

3
.5
3
1

3
.8
0
9

0
.2
7
9

6
.9
7
*
*
*

4
.
I
h
av
e
a
ch
ec
k
in
g
an
d
/o
r
a
sa
v
in
g
s
ac
co
u
n
t.

4
.3
8
0

4
.2
6
8

0
.0
4
3

1
.0
6

5
.
I
h
av
e
co
n
v
er
sa
ti
o
n
s
w
it
h
m
y
p
ar
en
ts
re
g
ar
d
in
g
p
er
so
n
al
fi
n
an
ce
.

3
.4
6
8

3
.7
9
9

0
.3
3
0

7
.5
5
*
*
*

1
5
.
I
th
in
k
it
is
im

p
o
rt
an
t
to

co
n
tr
ib
u
te
to

a
re
ti
re
m
en
t
p
la
n
(e
x
:
R
o
th

IR
A
,
4
0
1
k
,
et
c.
)

4
.0
6
0

4
.4
8
1

0
.4
2
1

1
0
.7
8
*
*
*

T
o
ta
l
sc
o
re

fo
r
fi
n
an
ci
al
b
eh
av
io
r

1
4
.7
9
5

1
6
.3
3
4

1
.5
3
9

1
3
.4
1
*
*
*

P
an
el
C
.
O
b
je
ct
iv
e
q
u
es
ti
o
n
s
(C
o
rr
ec
t
A
n
sw

er
s)
a

1
.
Is
it
sa
fe
r
to

p
u
t
y
o
u
r
m
o
n
ey

in
to

o
n
e
in
v
es
tm

en
t
o
r
p
u
t
y
o
u
r
m
o
n
ey

in
to

m
u
lt
ip
le
in
v
es
tm

en
ts
?

0
.2
5
0

0
.6
9
3

0
.0
5
3

2
.5
3
*
*

2
.
If
y
o
u
in
v
es
t
$
1
0
0
in

a
R
o
th

IR
A
an
d
ea
rn

5
%

p
er

y
ea
r
fo
r
3
y
ea
rs
,
h
o
w
m
u
ch

w
o
u
ld

it
b
e
w
o
rt
h
at
th
e
en
d

o
f
3
y
ea
rs
.

0
.3
6
7

0
.4
8
7

0
.2
0
2

1
0
.0
3
*
*
*

3
.
If
y
o
u
u
se

a
cr
ed
it
ca
rd

in
Ja
n
u
ar
y
fo
r
a
to
ta
l
o
f
$
5
0
0
,
w
h
ic
h
p
ay
m
en
t
o
p
ti
o
n
w
il
l
re
su
lt
in

th
e
lo
w
es
t

am
o
u
n
t
o
f
o
v
er
al
l
in
te
re
st
p
ai
d
.

0
.3
5
1

0
.7
2
9

0
.2
4
5

1
2
.5
5
*
*
*

4
.
4
.
S
u
p
p
o
se

y
o
u
d
ec
id
e
to

b
u
y
an

A
u
d
i
fo
r
$
5
0
,0
0
0
.
If
y
o
u
ta
k
e
o
u
t
an

au
to

lo
an

fo
r
5
y
ea
rs
w
it
h
5
%

in
te
re
st
,

h
o
w
m
u
ch

to
ta
l
w
il
l
y
o
u
p
ay

p
er

y
ea
r?

0
.3
0
6

0
.5
5
5

0
.1
3
1

6
.1
8
*
*
*

5
.
In

th
e
fu
tu
re
,
th
e
co
st
o
f
th
in
g
s
y
o
u
b
u
y
d
o
u
b
le
s
B
U
T
y
o
u
r
in
co
m
e
re
m
ai
n
s
th
e
sa
m
e.
H
o
w
m
u
ch

w
il
l
y
o
u
b
e

ab
le
to

b
u
y
in

th
e
fu
tu
re

in
co
m
p
ar
is
o
n
to

to
d
ay
?

0
.1
9
0

0
.7
6
6

0
.1
9
4

9
.9
2
*
*
*

6
.
S
u
p
p
o
se

y
o
u
h
av
e
$
4
0
,0
0
0
in

st
u
d
en
t
d
eb
t.
W
h
ic
h
p
ay
m
en
t
o
p
ti
o
n
w
il
l
re
su
lt
in

th
e
lo
w
es
t
am

o
u
n
t
o
f
o
v
er
al
l

in
te
re
st
p
ai
d
?

0
.3
0
6

0
.7
1
4

0
.1
8
2

9
.0
1
*
*
*

T
o
ta
l
sc
o
re

fo
r
o
b
je
ct
iv
e
q
u
es
ti
o
n
s
(C
o
rr
ec
t
A
n
sw

er
s)

2
.9
3
6

3
.9
4
4

1
.0
0
7

1
5
.8
8
*
*
*

(c
o
n
ti
n
u
ed

o
n
n
ex
t
p
a
g
e)

328 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340



T
ab
le

5
(C
o
n
ti
n
u
ed
)

P
re

P
o
st

D
if
f

T
-s
ta
t

P
an
el
D
.
O
b
je
ct
iv
e
q
u
es
ti
o
n
s
(“
I
D
o
n
’t
K
n
o
w
”
A
n
sw

er
s)
b

1
.
Is
it
sa
fe
r
to

p
u
t
y
o
u
r
m
o
n
ey

in
to

o
n
e
in
v
es
tm

en
t
o
r
p
u
t
y
o
u
r
m
o
n
ey

in
to

m
u
lt
ip
le
in
v
es
tm

en
ts
?

0
.6
4
0

0
.0
5
2

�0
.1
9
8
�1

2
.4
2
*
*
*

2
.
If
y
o
u
in
v
es
t
$
1
0
0
in

a
R
o
th

IR
A
an
d
ea
rn

5
%

p
er

y
ea
r
fo
r
3
y
ea
rs
,
h
o
w
m
u
ch

w
o
u
ld

it
b
e
w
o
rt
h
at
th
e
en
d

o
f
3
y
ea
rs
.

0
.2
8
6

0
.1
3
8

�0
.2
2
9
�1

2
.5
0
*
*
*

3
.
If
y
o
u
u
se

a
cr
ed
it
ca
rd

in
Ja
n
u
ar
y
fo
r
a
to
ta
l
o
f
$
5
0
0
,
w
h
ic
h
p
ay
m
en
t
o
p
ti
o
n
w
il
l
re
su
lt
in

th
e
lo
w
es
t
am

o
u
n
t

o
f
o
v
er
al
l
in
te
re
st
p
ai
d
.

0
.4
8
4

0
.0
7
8

�0
.2
7
3
�1

5
.6
9
*
*
*

4
.
S
u
p
p
o
se

y
o
u
d
ec
id
e
to

b
u
y
an

A
u
d
i
fo
r
$
5
0
,0
0
0
.
If
y
o
u
ta
k
e
o
u
t
an

au
to

lo
an

fo
r
5
y
ea
rs
w
it
h
5
%

in
te
re
st
,

h
o
w
m
u
ch

to
ta
l
w
il
l
y
o
u
p
ay

p
er

y
ea
r?

0
.4
2
4

0
.1
1
0

�0
.1
9
6
�1

0
.7
6
*
*
*

5
.
In

th
e
fu
tu
re
,
th
e
co
st
o
f
th
in
g
s
y
o
u
b
u
y
d
o
u
b
le
s
B
U
T
y
o
u
r
in
co
m
e
re
m
ai
n
s
th
e
sa
m
e.
H
o
w
m
u
ch

w
il
l
y
o
u

b
e
ab
le
to

b
u
y
in

th
e
fu
tu
re

in
co
m
p
ar
is
o
n
to

to
d
ay
?

0
.5
7
1

0
.0
7
2

�0
.1
1
8
�7

.9
3
*
*
*

6
.
S
u
p
p
o
se

y
o
u
h
av
e
$
4
0
,0
0
0
in

st
u
d
en
t
d
eb
t.
W
h
ic
h
p
ay
m
en
t
o
p
ti
o
n
w
il
l
re
su
lt
in

th
e
lo
w
es
t
am

o
u
n
t
o
f

o
v
er
al
l
in
te
re
st
p
ai
d
?

0
.5
3
2

0
.1
0
4

�0
.2
0
2
�1

1
.6
4
*
*
*

T
o
ta
l
sc
o
re

fo
r
o
b
je
ct
iv
e
q
u
es
ti
o
n
s
(“
I
D
o
n
’t
K
n
o
w
”
an
sw

er
s)

1
.7
7
0

0
.5
5
5

�1
.2
1
5
�1

8
.8
1
*
*
*

N
o
te
:
T
ab
le

5
S
h
o
w
s
th
e
t-
te
st
re
su
lt
s
o
f
st
u
d
en
t
re
sp
o
n
se
s
to

fi
n
an
ci
al

b
eh
av
io
r
an
d
k
n
o
w
le
d
g
e
q
u
es
ti
o
n
s
b
ef
o
re

an
d
af
te
r
th
e
fi
n
an
ci
al

li
te
ra
cy

ed
u
ca
-

ti
o
n
al
ef
fo
rt
s
fo
r
th
e
te
st
sa
m
p
le
.

*
*
*
,
*
*
,
*
in
d
ic
at
e
st
at
is
ti
ca
l
si
g
n
ifi
ca
n
ce

at
0
.0
1
,
0
.0
5
,
an
d
0
.1
0
le
v
el
,
re
sp
ec
ti
v
el
y
.

a
Im

p
ro
v
em

en
t
is
in
d
ic
at
ed

b
y
a
p
o
si
ti
v
e
d
if
fe
re
n
ce

an
d
t-
st
at
(m

o
re

p
eo
p
le
se
le
ct
in
g
th
e
co
rr
ec
t
an
sw

er
).

b
Im

p
ro
v
em

en
t
is
in
d
ic
at
ed

b
y
a
n
eg
at
iv
e
d
if
fe
re
n
ce

an
d
t-
st
at
(l
es
s
p
eo
p
le
se
le
ct
in
g
“I

D
o
n
’t
K
n
o
w
”)
.

G. Filbeck et al. / Financial Services Review 28 (2020) 315–340 329



Additionally, both objective financial knowledge and self-esteem exhibit statistically sig-

nificant improvements across the board. The biggest gains from objective financial knowl-

edge are credit (gain of 0.245) and compounding interest (gain of 0.202). The biggest

improvements in self-esteem also stem from credit (0.273 improvement) and compounding

interest (0.229 improvement). The results show a link between confidence to answer a ques-

tion (self-esteem) and correctness (objective financial knowledge).

The t test analyzes hypotheses related to the objective financial knowledge questions.

Students experienced a positive gain in correct responses of 1.007 (statistically significant at

the 1% level), which represents an improvement of 34%. Furthermore, the mean total score

for the test sample increased to almost four. At a statistically significance level of 1%, we

reject null hypothesis and conclude that students are more likely to be more financially

knowledgeable after completing financial literacy education.

Additionally, students experience an increase in financial self-esteem, as measured by the

amount of questions answered with I Don’t Know. Students experienced an improvement in

the number of I Don’t Know responses of 1.215 (statistically significant at the 1% level),

representing an improvement in financial self-esteem to answer the question. As a result, we

reject null hypothesis and conclude that students are less likely to answer I Don’t Know and

have greater confidence after completing financial literacy education.

Table 6 T-test results between pre- and post-survey for different subsamples

Financial behavior Subjective questions Objective questions

Pre Post Diff Pre Post Diff Pre Post Diff

Panel A. Subsamples by gender
Male 14.797 16.246 1.449*** 34.492 43.535 9.043*** 3.173 4.041 0.868***
Female 14.792 16.431 1.640*** 32.987 43.835 10.848*** 2.673 3.835 1.162***

Panel B. Subsamples by grade level
Freshman 14.143 15.918 1.776*** 35.694 44.082 8.388*** 2.878 3.980 1.102***
Sophomore 14.565 16.387 1.823*** 34.903 43.774 8.871*** 2.887 3.952 1.065***
Junior 14.340 16.294 1.954*** 33.010 44.340 11.330*** 2.782 3.970 1.188***
Senior 15.053 16.381 1.328*** 33.758 43.379 9.621*** 3.006 3.930 0.924***

Panel C. Subsamples by GPA
Lower than median GPA 14.251 16.028 1.777*** 33.749 43.375 9.625*** 2.568 3.699 1.131***
Higher than median
GPA

15.389 16.668 1.279*** 33.814 44.008 10.193*** 3.339 4.211 0.872***

Panel D. Subsamples by preferred instructions methods
Learning by doing 14.956 16.491 1.535*** 33.738 44.002 10.263*** 2.969 4.044 1.075***
Listening 14.752 16.238 1.486*** 34.227 43.467 9.240*** 2.934 3.826 0.891***
Discussing with peers 14.756 16.261 1.505*** 34.185 43.519 9.333*** 2.919 3.878 0.959***
Features visual support 14.968 16.440 1.472*** 33.680 43.984 10.305*** 2.975 4.049 1.074***
Interactive with websites 14.667 16.262 1.596*** 34.191 44.018 9.827*** 3.218 4.040 0.822***

Panel E. Subsamples by favorite subjects
English 14.494 16.165 1.671*** 33.890 43.616 9.726*** 2.451 3.506 1.055***
Math 15.140 16.654 1.513*** 33.675 44.333 10.658*** 3.175 4.053 0.877***
Science 14.596 16.136 1.539*** 33.645 43.193 9.548*** 3.061 4.013 0.952***
Social studies 14.866 16.373 1.507*** 33.876 43.569 9.694*** 2.919 4.086 1.167***

Note: Shows the t-test results of student responses to financial behavior, subjective and objective questions

before and after the financial literacy educational efforts for different subsamples.

***, **, * indicate statistical significance at 0.01, 0.05 and 0.10 level, respectively.

330 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340



Table 6 reports the t test results for different subgroups, which show statistically signifi-

cant improvement across all subgroups at the 1% level, indicating a significant improvement

after completing the financial literacy program. Within the objective knowledge category,

female students and students with low GPAs exhibit greater improvement, a positive sign in

learning potential, as these characteristics are more likely to lead to lower initial financial lit-

eracy (Tables 2 and 4 findings). Consistent with Gerrans and Heaney (2019), female students

experience a greater improvement in financial behavior, financial knowledge, and objective

knowledge in comparison to their male peers.

Next, we run regression analysis to examine how student characteristics and other control

variables affect their knowledge and behavior gains. Table 7 reports the regression results.

Panel A reports the regression results on student characteristics and school district character-

istics after controlling for fixed effects of classes. Specifically, the class fixed effects allow

the class dummy variable to differ for each class and control for the variations across classes.

Because we include school district characteristics as control variables, and these data are the

same in the same school district, we cluster the standard errors at the school district level.

The dependent variable is Diff_BEHAV (Diff_SUBJ, Diff_OBJ, Diff_IDK), which is the

difference between the students’ pre- and post-study scores (post- minus pre-) for the finan-

cial behavior (subjective, objective) questions. All the other variables are defined the same

as in Table 4 and listed in Appendix C. We also add Pre_BEHAV, Pre_SUBJ Pre_OBJ, and

Pre_IDK in the regressions to test whether student gains in financial knowledge and behavior

are affected by their presurvey knowledge and behavior. A negative coefficient for

Diff_IDK indicates an improvement in self-esteem, as it means that students answer I Don’t

Know less in the post-survey and select more answers that are correct or incorrect. The

results show that students who are less knowledgeable or exhibit inferior financial behavior

gain most in the study. Similarly, students in school districts with higher poverty experience

a statistically significant gain in financial behavior. This finding is encouraging as Kaiser

and Menkhoff (2017) indicate that financial education is often less effective for lower (low

and lower-middle) income clients (economics) due to lack of relatability to topics such as

handling of debt, a fact also noted by Fernandes and Lynch (2014). Our differing results

may be attributable to the manner in which the curriculum from the CFA Society Pittsburgh

directly addresses relatability in the context of typical, high-school appropriate, smaller cost

purchases. Such an approach is consistent with Stolper and Walter (2017) as they point out

that the opportunity to relate financial literacy to various demographics in the context of

their spending behavioral is key to program success. Female students improve by a greater

amount in financial knowledge across both subjective and objective questions. Students in

school districts with a higher poverty improve more in financial behavior. Female students

are more likely to experience an increase in self-esteem (statistically significant at the 1%

level). The improvement is consistent with our findings in Table 6 and the findings of

Gerrans and Heaney (2019), which both find that female students benefit more from financial

literacy education.

Next, we include other control variables such as favorite subjects and learning style and

use fixed effect regressions controlling for classes differences. Because there are no school

district data included in these regressions, we cluster standard errors at the class level. Panel

B of Table 7 reports the regression results. We use the same dependent variables as in Panel

G. Filbeck et al. / Financial Services Review 28 (2020) 315–340 331



T
ab
le

7
S
h
o
w
s
th
e
re
g
re
ss
io
n
re
su
lt
s
o
f
th
e
te
st
sa
m
p
le

M
o
d
el
(1
)

M
o
d
el
(2
)

M
o
d
el
(3
)

M
o
d
el
(4
)

D
ep
.
V
ar
.:
D
if
f_
B
E
H
A
V

D
ep
.
V
ar
.:
D
if
f_
S
U
B
J

D
ep
.
V
ar
:
D
if
f_
O
B
J

D
ep
.
V
ar
.:
D
if
f_
ID

K

C
o
ef
fi
ci
en
t

Z
-s
ta
t

C
o
ef
fi
ci
en
t

Z
-s
ta
t

C
o
ef
fi
ci
en
t

Z
-s
ta
t

C
o
ef
fi
ci
en
t

Z
-s
ta
t

P
an
el
A
.
R
eg
re
ss
io
n
R
es
u
lt
s
af
te
r
co
n
tr
o
ll
in
g
fo
r
fi
x
ed

ef
fe
ct
o
f
cl
as
se
s,
w
it
h
st
an
d
ar
d
er
ro
rs
cl
u
st
er
ed

at
sc
h
o
o
l
d
is
tr
ic
t
le
v
el

In
te
rc
ep
t

2
.7
6
2

1
.2
7

1
4
.5
6

1
.2
0

3
.8
0
7

2
.4
8
*
*

�2
.3
0
9

�1
.5
4

F
em

al
e

0
.3
4
1

1
.5
7

1
.8
1
2

2
.1
0
*
*

0
.2
8
7

1
.6
6
*

�0
.4
3
5

�3
.0
1
*
*
*

S
o
p
h
o
m
o
re

0
.3
3
6

0
.7
3

1
.1
5
2

0
.4
1

0
.0
0
6

0
.0
2

�0
.2
1
6

�0
.5
4

Ju
n
io
r

0
.1
8
6

0
.5
5

3
.1
0
0

1
.4
0

0
.0
7
0

0
.2
8

�0
.1
8
3

�0
.5
5

S
en
io
r

�0
.4
3
1

�1
.5
5

1
.7
6
6

0
.8
2

�0
.0
4
7

�0
.1
5

�0
.0
2
9

�0
.0
8

G
P
A

�0
.3
1
5

�2
.0
2
*
*

0
.7
2
3

1
.3
9

�0
.0
9
6

�0
.9
3

0
.1
2
8

1
.1
9

L
o
g
(P
o
p
u
la
ti
o
n
)

�0
.1
5
3

�0
.7
9

�1
.1
4
7

�0
.7
7

�0
.2
4
5

�1
.5
5

0
.0
9
3

0
.5
3

P
o
v
er
ty

7
.3
6
1

1
.9
3
*

6
.2
2
3

0
.2
3

�0
.7
1
4

�0
.2
4

�0
.1
1
7

�0
.0
4

P
ct
_
C
o
ll
eg
e

2
.4
1
9

1
.2
3

4
.4
9
0

0
.3
3

�0
.0
2
8

�0
.0
2

0
.0
6
7

0
.0
4

P
an
el
B
.
R
eg
re
ss
io
n
re
su
lt
s
af
te
r
co
n
tr
o
ll
in
g
fo
r
fi
x
ed

ef
fe
ct
o
f
cl
as
se
s,
w
it
h
st
an
d
ar
d
er
ro
r
cl
u
st
er
ed

at
cl
as
s
le
v
el

In
te
rc
ep
t

1
0
.3
4
9

1
0
.4
0
*
*
*

3
5
.3
6
8

1
2
.0
2
*
*
*

2
.1
3
0

5
.9
6
*
*
*

0
.8
9
1

4
.0
3
*
*
*

P
re
_
B
E
H
A
V

�0
.6
8
8

�1
2
.8
0
*
*
*

P
re
_
S
U
B
J

�0
.8
3
6

�1
2
.8
3
*
*
*

P
re
_
O
B
J

�0
.6
4
7

�1
6
.2
3
*
*
*

P
re
_
ID

K
�0

.8
1
8

�2
5
.4
9
*
*
*

F
em

al
e

0
.1
5
1

0
.9
4

0
.6
6
0

0
.9
7

�0
.0
0
1

�0
.0
1

0
.0
6
8

1
.1
9

S
o
p
h
o
m
o
re

0
.5
3
8

1
.1
6

0
.0
7
4

0
.0
4

�0
.0
3
1

�0
.1
0

�0
.3
0
7

�1
.3
9

Ju
n
io
r

0
.3
6
0

0
.6
6

0
.8
4
6

0
.5
9

0
.0
9
9

0
.5
1

�0
.1
1
6

�0
.6
3

S
en
io
r

0
.2
5
4

0
.5
6

�0
.2
2
2

�0
.1
6

�0
.0
3
6

�0
.1
8

�0
.0
5
6

�0
.3
0

G
P
A

0
.2
5
6

1
.8
7

0
.4
4
4

1
.4
4

0
.1
9
9

2
.0
0
*
*

�0
.1
7
3

�3
.7
3
*
*
*

E
n
g
li
sh

�0
.0
7
1

�0
.3
6

�0
.2
6
0

�0
.3
3

�0
.3
8
7

�2
.8
8
*
*
*

0
.2
9
7

2
.0
8
*
*

M
at
h

0
.1
8
0

0
.5
9

0
.6
0
0

0
.6
7

�0
.1
2
1

�0
.7
2

0
.1
8
5

2
.1
3
*
*

S
ci
en
ce

�0
.1
5
0

�0
.6
3

�0
.3
5
6

�0
.4
4

�0
.0
9
4

�0
.5
7

0
.1
1
7

1
.0
5

L
B
D

0
.3
0
5

1
.3
7

1
.0
8
9

2
.1
0
*
*

0
.3
0
3

3
.0
0
*
*
*

�0
.1
8
5

�1
.7
9
*

L
is
te
n
in
g

�0
.1
4
6

�0
.7
4

�0
.3
6
6

�0
.5
5

�0
.1
8
8

�1
.9
6
*
*

0
.0
4
3

0
.6
2

D
is
cu
ss
in
g

�0
.1
1
1

�0
.4
5

�0
.3
5
1

�0
.5
1

�0
.1
0
5

�0
.7
8

0
.0
8
3

0
.9
9

V
is
u
al

0
.1
0
7

0
.6
1

0
.8
2
1

1
.4
5

0
.2
5
0

2
.2
4
*
*

�0
.1
1
6

�1
.3
9

N
o
te
:
P
an
el

a
re
p
o
rt
s
th
e
re
g
re
ss
io
n
re
su
lt
s
o
n
st
u
d
en
t
ch
ar
ac
te
ri
st
ic
s
an
d
sc
h
o
o
l
d
is
tr
ic
t
ch
ar
ac
te
ri
st
ic
s
af
te
r
co
n
tr
o
ll
in
g
fo
r
fi
x
ed

ef
fe
ct

o
f
cl
as
se
s,
w
it
h
st
an
d
ar
d
er
ro
rs

cl
u
st
er
ed

at
th
e
sc
h
o
o
l
d
is
tr
ic
t

le
v
el
.
P
an
el

B
re
p
o
rt
s
th
e
re
g
re
ss
io
n
re
su
lt
s
o
n
st
u
d
en
t
ch
ar
ac
te
ri
st
ic
s
af
te
r
co
n
tr
o
ll
in
g
fo
r
fi
x
ed

ef
fe
ct

o
f
cl
as
se
s,
w
it
h
cl
u
st
er
ed

st
an
d
ar
d
er
ro
rs
.
P
re
_
B
E
H
A
V
(p
re
_
S
U
B
J,
p
re
_
O
B
J,
p
re
_
ID

K
)
is
th
e
st
u
-

d
en
t’
s
to
ta
l
sc
o
re

fo
r
th
e
fi
n
an
ci
al

b
eh
av
io
r
(k
n
o
w
le
d
g
e,
o
b
je
ct
iv
e)

q
u
es
ti
o
n
s.
D
if
f_
B
E
H
A
V
(d
if
f_
S
U
B
J,
d
if
f_
O
B
J,
d
if
f_
ID

K
)
is
th
e
d
if
fe
re
n
ce

b
et
w
ee
n
th
e
st
u
d
en
t’
s
p
re
-
an
d
p
o
st
-s
u
rv
ey

sc
o
re
s
fo
r
th
e

fi
n
an
ci
al
b
eh
av
io
r
(k
n
o
w
le
d
g
e,
o
b
je
ct
iv
e)

q
u
es
ti
o
n
s.
F
em

al
e
is
a
d
u
m
m
y
v
ar
ia
b
le
w
h
ic
h
is
eq
u
al
to

1
if
th
e
st
u
d
en
t
is
a
fe
m
al
e
st
u
d
en
t,
an
d
0
o
th
er
w
is
e.
S
o
p
h
o
m
o
re

(j
u
n
io
r,
se
n
io
r)
is
a
d
u
m
m
y
v
ar
ia
b
le

w
h
ic
h
is
eq
u
al

to
1
if
th
e
st
u
d
en
t
is
a
so
p
h
o
m
o
re

(j
u
n
io
r,
se
n
io
r)
,
an
d
0
o
th
er
w
is
e.
E
n
g
li
sh

(m
at
h
,
sc
ie
n
ce
)
is
a
d
u
m
m
y
v
ar
ia
b
le

w
h
ic
h
is
eq
u
al

to
1
if
th
e
st
u
d
en
t’
s
fa
v
o
ri
te

su
b
je
ct

is
E
n
g
li
sh

(m
at
h
,
sc
i-

en
ce
),
an
d
0
o
th
er
w
is
e.
L
B
D

(l
is
te
n
in
g
,
d
is
cu
ss
in
g
,
v
is
u
al
)
is
a
d
u
m
m
y
v
ar
ia
b
le

w
h
ic
h
is
eq
u
al

to
1
if
th
e
st
u
d
en
t
ch
o
o
se
s
le
ar
n
in
g
b
y
d
o
in
g
(l
is
te
n
in
g
,
d
is
cu
ss
in
g
w
it
h
p
ee
rs
,
fe
at
u
re
s
v
is
u
al

su
p
p
o
rt
)
as

fa
v
o
ri
te
in
st
ru
ct
io
n
m
et
h
o
d
,
an
d
0
o
th
er
w
is
e.
G
P
A
is
a
st
u
d
en
t’
s
g
ra
d
e
p
o
in
t
av
er
ag
e.
L
o
g
(p
o
p
u
la
ti
o
n
)
is
th
e
lo
g
o
f
th
e
p
o
p
u
la
ti
o
n
in

th
e
sc
h
o
o
l
d
is
tr
ic
t.
P
o
v
er
ty

is
th
e
p
o
v
er
ty

ra
te

in
th
e
sc
h
o
o
l
d
is
tr
ic
t.

P
ct
_
co
ll
eg
e
is
th
e
p
er
ce
n
ta
g
e
o
f
re
si
d
en
ts
w
h
o
h
av
e
at
ta
in
ed

b
ac
h
el
o
r
d
eg
re
e
o
r
h
ig
h
er
.

*
*
*
,
*
*
,
*
in
d
ic
at
e
st
at
is
ti
ca
l
si
g
n
ifi
ca
n
ce

at
0
.0
1
,
0
.0
5
an
d
0
.1
0
le
v
el
,
re
sp
ec
ti
v
el
y
.

332 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340



A. We also add Pre_BEHAV, Pre_SUBJ Pre_OBJ, and Pre_IDK in the regressions to test

whether student gains in financial knowledge and behavior are affected by their pre-survey

knowledge and behavior. The results show that students who are less knowledgeable or ex-

hibit inferior financial behavior gain most in the study. Students whose favorite instruction

method is learning by doing tend to gain more in subjective financial knowledge. Students

with a high GPA, whose favorite subject is not English, whose favorite instruction method is

learning by doing or visual tend to gain more in objective financial knowledge. students with

lower self-esteem gain the largest amount of self-esteem. Additionally, students with higher

GPA and students that prefer learning by doing experience the largest self-esteem gain (stat-

istically significant at the 1% and 10% level, respectively).

6. Conclusions

This research study investigates the effectiveness of a high school financial literacy cam-

paign to significantly improve financial literacy in four areas: subjective financial knowl-

edge, financial behavior, objective financial knowledge and self-esteem. The financial

literacy campaigns within the study were launched by the CFA Society Pittsburgh based

upon the book The Missing Semester.

Initially, the result of the pre-survey, taken by students before beginning the financial edu-

cation program, are analyzed using a t test. The results show students with higher GPAs are

more likely to display better financial behavior and objective financial knowledge than stu-

dents with lower GPAs. Students with lower GPAs exhibit greater perceived knowledge

(subjective financial knowledge) in the concepts of loans and cost-benefit analysis; however,

the same students exhibit lower actual knowledge (objective financial knowledge) in the

same categories. This finding is consistent with previous literature that shows a disconnect

between actual and perceived knowledge, as well as a connection between poor financial

understanding and negative debt implications (e.g., higher debt and higher borrowing costs).

Similarly, male students are more likely to exhibit better objective financial knowledge,

while female students are more likely to exhibit lower financial self-esteem.

Subsequently, logistic regressions test the relationship of subjective financial knowledge,

financial behavior, and objective financial knowledge. The results further display the link

between gender and initial financial knowledge, as females score lower on pre-survey objec-

tive and subjective financial knowledge. In addition, higher GPA has a statistically signifi-

cant effect on better financial behavior and objective financial knowledge, but worse

subjective knowledge, reinforcing the findings about the disconnect between actual and per-

ceived knowledge. Furthermore, self-esteem is shown to play a statistically significant

impact on both actual and perceived knowledge. Students with higher self-esteem exhibit

higher financial behavior, subjective financial knowledge, and objective financial knowledge

(significant at the 1% level), signifying that self-esteem is an important part of the financial

literacy equation. Additionally, the initial positive link between objective and subjective fi-

nancial knowledge is eliminated when we added the additional regression variable, self-

esteem, which further emphasizing the importance of self-esteem. This finding is consistent

G. Filbeck et al. / Financial Services Review 28 (2020) 315–340 333



with Tang and Baker (2016), which introduces the importance of self-esteem on financial

behavior. Students in school districts with lower income levels exhibit lower financial

behavior and objective financial knowledge, while students in school districts with smaller

populations score higher in subjective financial knowledge.

To test the effectiveness of the financial literacy program, we then conducted a t test

between results of the pre- and post-survey, taken after completion of the course. The t test

analyzes the four major topic areas previously listed. Total scores for financial behavior,

subjective knowledge, objective knowledge, and self-esteem improve by 29.3%, 10.4%,

34.3%, and 68.6%, respectively. Students experience a statistically significant improvement

of an average 35.7% in all four topic areas at the 1% level. Within subjective financial

knowledge, the largest gains result from the concepts of Roth IRA and planning for retire-

ment. Financial behavior shows the largest improvement in having personal finance conver-

sations with parents. Objective financial knowledge and self-esteem improve the most for

the concepts of compound interest and credit.

Overall, the characteristics with the lowest pre-survey scores show the greatest improve-

ments in the post-survey scores. Within gender, females exhibit the highest improvement in

all four of the categories. Also, students with lower GPAs experience greater improvement

in financial behavior and objective knowledge, while students with higher GPAs improve

more in subjective knowledge. Students who prefer learning by doing and visual support ex-

perience the most improvement, while students whose favorite subject is English or social

studies experience the largest improvement.

Based upon the analysis, statistically significant gains in subjective financial knowl-

edge, financial behavior, objective financial knowledge, and financial self-esteem lead us

to the conclusion that the CFA Society Pittsburgh financial literacy program is successful

at increasing students’ chances of financial success. Therefore, the analysis shows the

program continues to be successful at attempting to confront the financial literacy crisis.

As the program continues to improve and expand, we look forward to expanding the

sample size and reach of the financial literacy efforts, especially to states where students

may receive no mandated financial education during high school. As financial analysts,

society will increasingly be looking to our profession to reduce the impact of financial

illiteracy and to make a positive difference in our fiduciary duties for investor education.

Our study has obvious limitations. First, we do not control for the manner in which

content is presented in the classroom—does it make a difference whether the program is

spread out over an entire system or conducted in longer sessions over a shorter time pe-

riod. We also do not control for the number of hours spent delivering the content. Future

research efforts will focus differences in program delivery. Additionally, we did not

introduce a true control group for comparison, as we did not want to jeopardize the main

purpose of our program, giving students the tools for a better financial future, just to pro-

vide a control group.

Our primary recommendation, based on the experience of CFA Society Pittsburgh, is

for financial professionals to consider taking a more active role in financial literacy out-

reach. Who better to lead these efforts than those who are trained to understand its

importance?

334 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340



Appendix A: Presurvey

State:
School:
Teacher:
Student ID:
Gender:
GPA:
Grade:

Favorite subject in school:

____ English
____ Math
____ Social Studies
____ Science

Questions:

1. I like to save money more than I like to spend it.

2. I understand how to establish a financial plan.

3. I think financial literacy is important for my future.

4. I have a checking and/or a savings account.

5. I have conversations with my parents regarding personal finance.

6. I understand the process by which my parents/guardians make financial decisions.

7. I know how to determine the appropriate total costs associated with the colleges/uni-

versities I am interested in attending.

8. I understand the process by which loan repayments take place including the impact

of interest, delinquency and default.

9. I understand the process by which credit card charges and repayment schedules can

impact the level of financial debt levels.

10. When it comes to purchasing a car, I know how to determine how much of a car I

can afford.

11. I understand how to evaluate the cost-benefit analysis of training for the job I would

like to perform after completing school.

12. I know what a Roth IRA is and how it works from a taxation standpoint.

13. I know how to create a savings plan based on the ability to estimate monthly living

expenses.

14. I know how to plan financially for retirement.

15. I think it is important to contribute to a retirement plan (ex: Roth IRA, 401k, etc.)

Learning Preferences: I am able to master material when instruction includes:

1. Learning by doing/manipulating objects

2. Listening

3. Discussing with peers

4. Features visual support (e.g., powerpoint slides)

5. Interactive with websites

G. Filbeck et al. / Financial Services Review 28 (2020) 315–340 335



Objective Questions:

1. Is it safer to put your money into one investment or put your money into multiple

investments?

a. One Investment

b. Multiple Investments

c. I Don’t Know*

2. If you invest $100 in a Roth IRA and earn 10% per year for 3 years, how much would

it be worth at the end of three years.

a. More than $130

b. Exactly $130

c. Less than $130

d. I Don’t Know*

3. If you use a credit card in January for a total of $300, which payment option will

result in the lowest amount of overall interest paid.

a. The Full Amount ($300)

b. The Minimum Payment Required

c. Paying nothing ($0)

d. I Don’t Know*

4. Suppose you decide to buy a BMW for $50,000. If you take out an auto loan for 5

years with 5% interest, how much total will you pay per year?

a. More than $10,000

b. Exactly $10,000

c. Less than $10,000

d. I Don’t Know*

5. In the future, the cost of things you buy doubles AND your income also doubles.

How much will you be able to buy in the future in comparison to today?

a. Less

b. The Same

c. More

d. I Don’t Know*

6. Suppose you have $30,000 in student loans. Which payment option would result in

the lowest amount of overall interest paid?

a. 10 years at $350 per month

336 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340



b. 15 years at $270 per month

c. 20 years at $230 per month

d. I Don’t Know*

*Survey respondents were required to answer the question, so I Don’t Know answer allows students to
select non correct/incorrect answer

Appendix B: Post-survey

State:
School:
Teacher:
Student ID:

Questions:

1. I like to save money more than I like to spend it.

2. I understand how to establish a financial plan.

3. I think financial literacy is important for my future.

4. I have a checking and/or a savings account.

5. I have conversations with my parents regarding personal finance.

6. I understand the process by which my parents/guardians make financial decisions.

7. I know how to determine the appropriate total costs associated with the colleges/uni-

versities I am interested in attending.

8. I understand the process by which loan repayments take place including the impact

of interest, delinquency and default.

9. I understand the process by which credit card charges and repayment schedules can

impact the level of financial debt levels.

10. When it comes to purchasing a car, I know how to determine how much of a car I

can afford.

11. I understand how to evaluate the cost-benefit analysis of training for the job I would

like to perform after completing school.

12. I know what a Roth IRA is and how it works from a taxation standpoint.

13. I know how to create a savings plan based on the ability to estimate monthly living

expenses.

14. I know how to plan financially for retirement.

15. I think it is important to contribute to a retirement plan (ex: Roth IRA, 401k, etc.)

Learning Preferences: I am able to master material when instruction includes:

1. Learning by doing/manipulating objects

2. Listening

3. Discussing with peers

4. Features visual support (e.g., powerpoint slides)

5. Interactive with websites

G. Filbeck et al. / Financial Services Review 28 (2020) 315–340 337



Objective Questions:

1. Which is less risky: Investing your money into one investment or multiple

investments?

a. One Investment

b. Multiple Investments

c. I Don’t Know*

2. If you invest $100 in a Roth IRA and earn 5% per year for 3 years, how much would

it be worth at the end of three years.

a. More than $115

b. Exactly $115

c. Less than $115

d. I Don’t Know*

3. If you use a credit card in January for a total of $500, which payment option will

result in the lowest amount of overall interest paid.

a. The Full Amount ($500)

b. The Minimum Payment Required

c. Paying nothing ($0)

d. I Don’t Know*

4. Suppose you decide to buy a Audi for $50,000. If you take out an auto loan for 5

years with 5% interest, how much total will you pay per year?

a. More than $10,000

b. Exactly $10,000

c. Less than $10,000

d. I Don’t Know*

5. In the future, the cost of things you buy doubles BUT your income remains the same.

How much will you be able to buy in the future in comparison to today?

a. Less

b. The Same

c. More

d. I Don’t Know*

6. Suppose you have $40,000 in student debt. Which payment option will result in the

lowest amount of overall interest paid?

a. 10 years at $450 per month

b. 15 years at $365 per month

338 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340



c. 20 years at $315 per month

d. I Don’t Know*

*Survey respondents were required to answer the question, so I Don’t Know answer allows students to
select non correct/incorrect answer

Appendix C: Variable definitions

Dependent variables
Subjective Total score for the financial subjective questions in the survey
Objective Total score for the financial objective questions in the survey with correct

answers
Behavior Total score for the financial behavior questions in the survey
IDK Total score for the financial objective questions in the survey with “I don’t

know” answers
Diff_Subj The difference between the students’ pre- and post-study scores (post minus pre)

for the financial subjective questions.
Diff_Obj The difference between the students’ pre- and post-study scores (post minus pre)

for the financial objective questions with correct answers.
Diff_Behav The difference between the students’ pre- and post-study scores (post minus pre)

for the financial behavior questions.
Diff_IDK The difference between the students’ pre- and post-study scores (post minus pre)

for the financial objective questions with “I don’t know” answers.
Independent variables

Gender:
Female A dummy variable which is equal to 1 if the student is a female, and 0 otherwise.

Grade level:
Sophomore A dummy variable which is equal to 1 if the student is a sophomore, and 0

otherwise.
Junior A dummy variable which is equal to 1 if the student is a junior, and 0 otherwise.
Senior A dummy variable which is equal to 1 if the student is a senior, and 0 otherwise.

Favorite subject:
English A dummy variable which is equal to 1 if the student’s favorite is English, and 0

otherwise.
Math A dummy variable which is equal to 1 if the student’s favorite is Math, and 0

otherwise.
Science A dummy variable which is equal to 1 if the student’s favorite is Science, and 0

otherwise.
Favorite learning style:

LBD A dummy variable which is equal to 1 if the student’s favorite learning style is
learning by doing (LBD), and 0 otherwise.

Listening A dummy variable which is equal to 1 if the student’s favorite learning style is
listening, and 0 otherwise.

Discussing A dummy variable which is equal to 1 if the student’s favorite learning style is
discussion, and 0 otherwise.

Visual A dummy variable which is equal to 1 if the student’s favorite learning style is
visualization, and 0 otherwise.

GPA:
GPA: A student’s grade point average (GPA).

School district characteristics:
Log(Population) The log of the population in the school district.
Poverty The poverty rate in the school district.
Pct_College The percentage of residents who have attained bachelor degree or higher.

G. Filbeck et al. / Financial Services Review 28 (2020) 315–340 339



References

Amagir, A., Groot, W., van den Brink, H., & Wilschut, A. (2018). A review of financial-literacy education pro-

grams for children and adolescents. Citizenship, Social and Economics Education, 17, 56-80.

Asaad, C. (2015). Financial literacy and financial behavior: Assessing knowledge and confidence. Financial

Services Review, 24, 101-117.

Berg, J., Dickhaut, J., & McCabe, K. (1995). Trust, reciprocity, and social history. Games and Economic

Behavior, 10, 122-142.

Cupák, A., Fessler, P., Schneebaum, A., & Silgoner, M. (2018). Decomposing gender gaps in financial literacy:

New international evidence. Economics Letters, 168, 102-106.

Fernandes, D., & Lynch, J. (2014). Financial literacy, financial education, and downstream financial behaviors.

Management Science, 60, 8, 1861-1883.

Filbeck, G., & Zhao, X. (2018). Financial literacy outreach: The effectiveness of outreach efforts in high schools.

Journal of Business and Economic Perspectives, 65, 87-117.

Gerrans, P., & Heaney, R. (2019). The impact of undergraduate personal finance education on individual finan-

cial literacy, attitudes, and intentions. Accounting & Finance, 59, 1, 177-217.

Green, D., & Riddell, W. (2012). Understanding Educational Impacts: The Role of Literacy and Numeracy

Skills. Working Paper: Institute of Labor Economics. Available at http://conference.iza.org/conference_files/

2015_Skill_Mismatch/riddell_w5670.pdf

Hastings, J., Madrian, B., & Skimmyhorn, W. (2013). Financial literacy, financial education, and economic out-

comes. Annual Review of Economics, 5, 347-373.

Kabala, M., & Natali, Jr., , G. (2012). The Missing Semester. Pittsburgh, PA: Authors.

Kaiser, T., & Menkhoff, L. (2017). Does financial education impact financial literacy and finance behavior, and

if so. The World Bank Economic Review, 31, 611-630.

Lin, J., Bumcrot, C., Ulicny, T., Lusardi, A., Mottola, G., Kieffer, C., & Walsh, G. (2016). Financial Capability

in the United States 2016. Available at http://www.usfinancialcapability.org/downloads/NFCS_2015_Report_

Natl_Findings.pdf

Lusardi, A., & Mitchell, O. (2014). The economic importance of financial literacy: The evidence. Journal of

Economic Literature, 52, 5-44.

McCannon, B., Asaad, C., & Wilson, M. (2016). Financial competence, overconfidence, and trusting invest-

ments: Results from an experiment. Journal of Economics and Finance, 40, 590-606.

Pelletier, J. (2015). 2015 National Report Card on State Efforts to Improve Financial Literacy in High School.

Available at https://www.champlain.edu/centers-of-excellence/center-for-financial-literacy/report-national-

high-school-financial-literacy/2015-national-report-card

Preston, A., & Wright, R. (2019). Understanding the gender gap in financial literacy: Evidence from Australia.

Economic Record, 95, 1-29.

Remund, D. (2010). Financial literacy explicated: The case for a clearer definition in an increasingly complex

economy. Journal of Consumer Affairs, 44, 276-295.

Stolper, O., & Walter, A. (2017). Financial literacy, finance advice, and financial behavior. Journal of Business

Economics, 87, 581-641.

Tang, N., & Baker, A. (2016). Self-esteem, financial knowledge and financial behavior. Journal of Economic

Psychology, 54, 164-176.

340 G. Filbeck et al. / Financial Services Review 28 (2020) 315–340