Applied Economics

ISSN: 0003-6846 (Print) 1466-4283 (Online) Journal homepage: http://www.tandfonline.com/loi/raec20

The effect of financial education training on the

financial literacy of Spanish students in PISA

José Manuel Cordero & Francisco Pedraja

To cite this article: José Manuel Cordero & Francisco Pedraja (2018): The effect of financial
education training on the financial literacy of Spanish students in PISA, Applied Economics, DOI:
10.1080/00036846.2018.1528336

To link to this article: https://doi.org/10.1080/00036846.2018.1528336

Published online: 13 Oct 2018.

Submit your article to this journal

Article views: 25

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=raec20
APPLIED ECONOMICS
https://doi.org/10.1080/00036846.2018.1528336

The effect of financial education training on the financial literacy of Spanish

students in PISA
José Manuel Cordero and Francisco Pedraja
Departamento de Economía, Universidad de Extremadura, Badajoz, Spain

ABSTRACT KEYWORDS
The aim of this paper is to analyze the effect of financial education training on Spanish secondary Education; financial literacy;
students. To do this, we rely on data from PISA 2012. This included an assessment of students’ PISA; difference-in-
financial literacy for the first time. In order to identify the causal effect of financial education differences
courses, we employ a difference-in-differences (DiD) approach to compare the outcomes of JEL CLASSIFICATION
students receiving and not receiving education on financial concepts for two different subjects I21; H52; C13
(financial literacy and reading comprehension). Our results suggest that financial education
programs only have an impact if they are taught as part of other subjects, i.e. by means of a
cross-curricular approach.

I. Introduction be applied in a generalized way, using different

teaching methods to courses designed for adults
The outbreak of the international financial and eco-
that normally target specific segments of the popula-
nomic crisis in 2008 led to a process of collective
tion primarily for the purpose of correcting bad
reflection on the social and ethical implications of
money management habits (Lusardi and Mitchell
financial transactions and, particularly, the inability
2014).
of many citizens to make sound financial invest-
By now, many countries have developed differ-
ments (Gerardi, Goette, and Meier 2010). However,
ent national initiatives designed to adopt financial
many countries had already launched initiatives to
education (FE) in their education systems to
improve their population’s financial literacy even
ensure access for the entire school-age population.
before the advent of the crisis (Appleyard and
FE has taken different forms depending on the
Rowlingson 2013). Many of these programs date
countries and different models may even coexist
back to the 2005 OECD Council Resolution recom-
within the same country. This leads to major
mending member states to educate their citizens
deviations between schools and regions.1 They
with respect to financial issues at all stages of their
range from special-purpose courses taught by tea-
life, not only for their own personal wellbeing but
chers from outside the teaching institutions (nor-
also for the good of the economy and society as a
mally specialists from financial institutions) to
whole (Gnan, Silgoner, and Weber 2007; Stango and
special finance-related subjects or contents that
Zinman 2009; Lusardi and Mitchell 2007).
are adopted within subject syllabuses as part of
This paper focuses on initiatives of this kind tar-
the school curriculum.
geting young people, since they have proved to have
In Spain, until very recently, there has been no
more limited financial knowledge (Mandell 2008;
specific subject with economic or financial content
Shim et al. 2010; Lusardi and Mitchell 2007).
in either primary or compulsory secondary educa-
Moreover, they will have to deal with increasingly
tion. Actually, when the PISA 2012 survey was
complex financial products, services and markets
conducted, financial contents were only taught in
when they reach adulthood (OECD 2014a). Such
some schools adopting a cross-curricular approach
programs play a preventive role, thus they should

CONTACT José Manuel Cordero [email protected] Departamento de Economía, Universidad de Extremadura, Badajoz, Spain
1
In recent years, a number of OECD researchers have developed several papers giving an overview of the level of development of FE strategies in the
international arena (Grifoni and Messy 2012; Atkinson and Messy, 2013; Messy and Monticone 2012, 2016).
© 2018 Informa UK Limited, trading as Taylor & Francis Group
2 J. M. CORDERO AND F. PEDRAJA

in courses such as Social Sciences, Geography or to be taught as part of social science, addressing
History, i.e. adopting a more historical or geogra- basic concepts concerning the value of money,
phical point of view. In fact, Spain was the country saving, personal budgeting and responsible and
with the lowest percentage of schools at which this sustainable consumption. On the other hand, sec-
type of education was available in 2012, as we can ondary education includes specific subjects
see at Figure 1, which shows the percentage of addressing FE like economics and introductory
schools offering some sort of FE as part of the entrepreneurship and business activity, as well as
school curriculum in the 18 countries that parti- adopting a cross-curricular approach whereby
cipated in the test on financial literacy in PISA financial contents are included in other subject
2012.2 syllabuses.
Several measures have recently been taken to In this paper, we attempt to examine whether
reverse this situation. The first was the develop- providing some form of financial education is an
ment of a pilot FE program called Finanzas para effective practice for improving young students’
todos (Finance for All), coordinated by the Bank financial literacy. Likewise, we also explore the
of Spain and the Spanish National Stock Market effectiveness of different types of teaching strate-
Commission in 2010. This program started in gies. For this purpose, we use data from the
during the academic year 2010/2011 (and contin- OECD’s Programme for International Student
ued also in 2011/2012) for 15 year-old students Assessment (PISA), which included a module on
and involved the cross-curricular integration of financial literacy for the first time in 2012. The
contents related to financial issues into social assessment provides comparable information with
science, mathematics and civic education taught regard to the financial literacy of 15-year-old stu-
in the third year of compulsory secondary educa- dents by testing how well they apply their knowl-
tion (ESO) as well as the implementation of some edge in everyday life situations. The dataset also
specific courses about financial concepts in some includes extensive information about individual
specific regions across the Spanish territory characteristics, socioeconomic background and
(Andalusia, Castile and Leon, Madrid, Murcia school contexts. In this manner, the analysis can
and Navarre). Later, with the implementation of account for these factors.
the LOMCE (Basic Law 8/2013, of 9 December, One of the main concerns about using observa-
for the Improvement of Educational Quality), tional data from PISA is that omitted variables and
financial education became part of the primary selection bias are hard to account for when assessing
and secondary education curriculum in the school the relationship between financial education pro-
year 2014/2015. In primary education, FE started grams and financial outcomes (Fox, Bartholomae,

90
80
70
60
50
40
30
20
10
0

Figure 1. Schools offering FE in countries participating in PISA 2012.

Source: OECD (2014a, p.148).

2
This information was taken from a questionnaire completed by the principals of the participant teaching institutions who were asked whether FE was
available at their schools.
 APPLIED ECONOMICS 3

and Lee 2005). For instance, the lack of information understanding of the effects attributable to finan-
about students’ cognitive abilities, which could pos- cial education programs (Fox, Bartholomae, and
sibly be correlated with other potential explanatory Lee 2005). Financial education refers to the pro-
variables, as well as having an influence on financial cess of providing individuals with information or
learning, could bias the estimation of the causal instruction to improve their understanding of
effect of financial education. To avoid this problem, financial products, develop their skills with regard
our estimation strategy involves using a difference- to their awareness of risk and opportunities, mak-
in-differences (DiD) approach comparing the results ing informed choices and taking effective actions
achieved by the same students for two different for their financial wellbeing (OECD 2005).
subjects (differences between financial literacy and Financial literacy can be interpreted as knowledge
reading comprehension), as suggested by Jürges, of the financial system, the ability to understand
Schneider, and Büchel (2005). The main advantage key financial concepts related to the management
of this approach is that we use each student as his or of money, loans and investment in different assets
her control group in order to control for most of the or both (Hung, Parker, and Yoong 2009; Xu and
heterogeneity at the individual level represented by Zia 2012). Finally, financial outcomes refer to the
innate abilities. The key assumption required to skills and behaviors regarding how people deal
identify the causal effect is that the difference in with financial matters such as saving or participat-
both outcome variables should be identical at ing in the stock market (Hastings, Madrian, and
schools not offering financial education. Therefore, Skimmyhorn 2013).
the difference in excess of the financial literacy test at Recent literature has shown that FE has positive
schools offering financial education courses should effects on the development of certain behaviors,
reflect its impact. such as saving for retirement (Lusardi and
The rest of the paper is structured as follows. Mitchell 2007; Cole, Sampson, and Zia 2011; Xu
Section 2 provides a brief literature review about and Zia 2012) or contracting mortgages and loans
the importance of financial education and its with lower interest rates and commissions (Disney
effects on student outcomes. Section 3 explains and Gathergood 2013; Lusardi and Tufano 2015).
the adopted estimation strategy. Section 4 pro- Additionally, there is evidence demonstrating that
vides a description of the dataset and the variables the acquisition of financial knowledge is asso-
considered in our empirical analysis, and Section 5 ciated with a wide range of behaviors such as
reports the main results. Finally, Section 6 outlines participation in the stock market, portfolio diver-
our main conclusions. sification or the tendency to avoid over-indebted-
ness (Guiso and Jappelli 2002; Christelis, Jappelli,
and Padula 2010; Van Rooij, Lusardi, and Alessie
II. Literature review
2011; Lusardi and Tufano 2015).
As more and more programs and initiatives aimed Nevertheless, a recent extensive meta-analysis
at improving the population’s financial literacy are conducted by Fernandes, Lynch Jr, and Netemeyer
set up, literature has grown up around the evalua- (2014), clearly separating studies focused on correla-
tion of their effectiveness (McCormick 2009). The tion (e.g. using OLS) from others using causal
assessment of those programs is a complex task, designs (experimental design or natural experi-
since it requires access to information about the ments), concludes that financial education interven-
characteristics of individuals that have and have not tions have hardly any aggregate effect on improving
participated in such initiatives in order to ascertain financial literacy. Actually, they found much larger
their real impact. In fact, as Lusardi and Mitchell effects in non-experimental studies applying a less
(2014) claim, there are few empirical studies report- rigorous econometric method. The findings of the
ing meaningful evaluations of the impact of FE literature review carried out by Miller et al. (2015)
programs on the behavior of individuals. are similar, since most of the analyzed papers have
Before reviewing this literature, it is important positive outcomes for financial education, but many
to make a distinction between some intercon- suffer from selection bias or other econometric con-
nected concepts in order to gain a better cerns. Their explanation for these results is that the
4 J. M. CORDERO AND F. PEDRAJA

inflated effect detected in the analyzed studies might Other papers refer to South America. For example,
mask a problem of omitted-variable bias, since there Bruhn et al. (2013) studied the case of several
are some underlying factors that have not been taken Brazilian schools that voluntarily decided to
into account (e.g. innate abilities or interest in finan- teach a FE course with a very positive effect on
cial matters) and might contribute to both higher student knowledge. Finally, some FE programs
levels of financial literacy and better financial out- established in much poorer countries have been
comes. Some previous papers have tried to deal with evaluated in recent years. For instance, Berry,
this issue. However, we were unable to identify a Karlan, and Pradhan (2015) also identified posi-
convincing identification strategy with the available tive effects for two different FE programs deployed
data, and thus there are still some unresolved issues. in primary and secondary education in Ghana,
As already discussed, in this paper we focus on and Jamison, Karlan, and Zinman (2014) observed
the analysis of the effectiveness of educational in Uganda that the deployment of a 10-week
programs targeting secondary education students. course on different issues related to FE had a
Therefore, we are concerned with exploring the favorable impact on student behavior, whereas
relationship between FE and the acquisition of the alternative of giving them a bank account to
financial knowledge or financial culture. With manage had no significant effect.
regard to this issue, almost all the studies evaluat- The only evaluation of the effectiveness of FE
ing the effectiveness of such programs, most of programs in Spain refers to the Finance for All
which refer to the United States of America pilot program at a sample of teaching institutions
(Bernheim, Garrett, and Maki 2001; Tennyson within the Autonomous Community of Madrid,
and Nguyen 2001), reveal that they have a positive which was found to be quite effective (Hospido,
impact on students’ financial knowledge (Swinton Villanueva, and Zamarro 2015). In particular, after
et al. 2007; Batty, Collins, and Odders-White correcting for the differences in the composition
2015), although its relevance varies depending on of the students attending the schools that taught
the type of course taught and the characteristics of FE contents with respect to schools that did not,3
the students attending the course (Walstad 2013). the results in the test of financial knowledge
However, other papers suggest that the effect of achieved by the former were a third of standard
such courses is negligible (Peng et al. 2007; deviation greater than for the latter.
Mandell and Klein 2009).
Following the trend initiated in the United
III. Empirical strategy
States, where high school financial education man-
dates have been enacted in many states over the The evidence provided by the studies that aim to
past fifty years, other countries have created finan- evaluate the effectiveness of FE programs is based
cial education programs in recent years and on experiments where there is some homogeneity
assessed their effectiveness. For example, an with respect to the characteristics of the schools
experimental program for adopting FE in the that do and do not teach the programs. This
Italian school curriculum had a positive effect on reduces the possibility of selection bias of the
the students’ knowledge, even one year after com- sample between the treatment and control groups
pleting the program (Romagnoli and Trifilidis (Duflo, Glennerster, and Kremer 2007). The
2013). Becchetti, Caiazza, and Coviello (2013) golden rule of experimental evaluation therefore
reached a similar conclusion when analyzing a holds (Collins and O’Rourke 2010), and the out-
16-hour course deployed at another set of schools comes achieved by individuals that have and have
in Italy. With respect to Germany, Lührmann, not received financial education are comparable.
Serra-Garcia, and Winter (2015) also found that However, experimental information is seldom
teaching a specific course on FE in secondary available, thus the impact of these programs is
schools helped to improve knowledge of finance. normally evaluated based on the analysis of

3
To guarantee the homogeneity between the treatment group (students that attended the course) and the control group (students that did not attend the
course), the authors applied matching techniques to pair students with similar observable characteristics belonging to each sample.
 APPLIED ECONOMICS 5

observed data, where, as there is no previous ran- difference between the results of students before
domization, the individuals in the two groups that and after participating in the program at schools
are compared may have different characteristics. where FE is available (treatment group) with the
In these cases, the validity of the evaluations is difference in the results achieved at equivalent
questionable unless suitable econometric methods times by individuals from schools where no such
are employed to correct the common problem of program exists (control group) (Schlotter,
endogeneity in data (Fox, Bartholomae, and Lee Schwerdt, and Woessmann 2011). The main
2005; Lyons et al. 2006; Willis 2011).4 advantage of using this technique is that it limits
This problem may arise on different grounds, any bias caused by there being non-observable
such as the simultaneity between the dependent differences between the evaluated subjects.
and independent variables or the omission of a However, data about the performance of stu-
key variable from the analysis. For instance, some dents before and after receiving FE training are
unobserved characteristics of students such as not available in our case. Therefore, we have
intelligence, ability, interest in financial matters adapted this method to an alternative framework
or previous experiences with money might be where we observe the performance of the same
relevant factors in determining financial learning. individuals in different subjects. This strategy was
Additionally, we have to consider that the assign- originally employed by Jürges, Schneider, and
ment of students across schools might not be Büchel (2005) to identify the causal effect of cen-
random when comparing schools that offer FE tral exams on student performance in Germany
courses with schools that do not. For example, using TIMSS data. Other studies have used similar
children from families with greater economic and models based on student fixed effects to estimate
cultural capital are more likely to attend schools the impact of teacher characteristics or practices
with better resources, where this type of financial on student performance (Dee 2005, 2005;
courses are more likely to be implemented. In Schwerdt and Wuppermann 2011¸ Bietenbeck
short, some mechanism has to be established to 2014) or the influence of instruction time on aca-
prevent these problems and guarantee the validity demic achievement (Rivkin and Schiman 2015). In
of the results in order to be able to identify the fact, a recent study by Cordero, Cristobal, and
effect of FE. Santín (2017) analyzing the papers that have
One common option to prevent the problem of applied causal inference techniques on data from
data endogeneity is to use an instrumental variable international databases highlights that this meth-
related to the explanatory variable under analysis odological approach is the most frequent in the
in order to generate an exogenous variation on the literature, as it is the strategy calling for the fewest
results. Bucher-Koenen and Lusardi (2011), assumptions.
Fornero and Monticone (2011) and Van Rooij, The underlying assumption of our estimation
Lusardi, and Alessie (2011), (2012)) adopted this strategy is based on the fact that the treatment, i.e.
approach. However, the identification of an the provision of financial training by the school,
instrument that meets the requirements of this has an influence on only one dimension of student
econometric approach is seldom straightforward. performance represented by the scores in the
In this paper we use a difference-in-differences financial literacy test. Therefore, the control
(DiD) approach in order to address such potential group should be represented by students attending
sources of selection bias. This methodology is schools where this course is not available. Since
usually applied when panel data are available. It PISA provides test results for two additional com-
is thus possible to observe individuals in treatment petences, mathematics and reading, we can esti-
and control groups at two different points in time, mate DiD by subject. Later, we take advantage of
i.e. before and after implementing the evaluated the fact that we have information on the outcomes
program (Becchetti, Caiazza, and Coviello 2013). of the same student for another competence,
The strategy adopted consists of comparing the namely reading comprehension, and we can thus

4
For a more detailed analysis of this issue, see Hill, Griffiths, and Lim (2008).
6 J. M. CORDERO AND F. PEDRAJA

estimate a DiD equation making a distinction the expected value of the difference between both
between the two subjects. In particular, we error terms is null:
selected reading comprehension outcomes for


comparison in order to avoid a potential source E εFi  εRi ¼ 0 (4)
of endogeneity between achievement in mathe-
matics and financial literacy, since most of the This assumption would not hold if the character-
questions in the financial literacy test include alge- istics of students attending schools offering FE
braic calculations (see OECD 2013 for details). courses are potentially different from the pupils
In this framework, the estimation strategy con- of schools that do not offer such courses, i.e. if
sists of separating the sample into students attend- there is a self-selection bias into treatment. We do
ing schools offering a FE course and students not think that this problem occurs with our data-
attending schools where no such course was avail- set since parents are unlikely to decide between
able. The key assumption required to identify the schools depending on whether or not they offer
causal effect is that the difference in the two out- FE courses. In addition, in our dataset we can
come variables would be identical in the absence identify that all schools were FE courses were
of treatment. Thus an excess difference in the implemented belonged to the five regions that
financial literacy test at schools offering FE decided to participate in the pilot program imple-
courses should reflect the causal effect that we mented in courses 2010/2011 and 2011/2012.
are looking for. Therefore, the decision to participate was not
Formally, our estimator can be described as taken by schools, but by the corresponding regio-
follows. We consider two different regressions to nal government, which subsequently selected ran-
explain the results in reading (R) and financial domly some schools to implement the program
literacy (F): within their territory.
A second aim of this research is to examine
yRi ¼ μi þ Xi β þ εRi (1) whether the manner in which the financial knowl-
yFi ¼ μi þ Xi γ þ FEi δ þ εFi (2) edge is taught may have an influence on the
financial literacy of students. In particular, we
where μi is any individual specific characteristic check whether the fact that these concepts are
(e.g. cognitive abilities), Xi represents a vector of taught as part of a specific course, either extra-
covariates that might affect the performance in curricularly or by means of lessons taught within
reading comprehension and financial literacy dif- school hours, or are taught adopting a cross-cur-
ferently, FEi is a dummy variable for the availabil- ricular approach, that is, as part of other subjects
ity of FE courses and εKi are error terms. The DiD like mathematics or social science, has any effect.
method basically consists of a subtraction of the To evaluate these aspects, we estimate a multiple
equations: treatment model in which additional dichotomous
variables built based on the information provided
Di ¼ yFi  yRi ¼ Xi ðγ  βÞ þ Ci δ þ εFi  εRi (3)
by the principals of the evaluated teaching institu-
where δ is our parameter of interest. The key tions are added to Equation 3.
advantage of this approach is that the use of dif- Both the model presented in Equation 3 and the
ferences removes the intrinsic characteristics of multiple treatment models described above were
each individual (μi ) from the equation. Thus we estimated using the traditional least squares model
are able to control for most of the heterogeneity (OLS), which estimates parameters associated with
represented by innate ability or previous experi- variables that reflect the average value of the effect
ence at the individual level. This means that each of the respective variable on the conditional expec-
student is serving as his or her control group. This tation of the dependent variable. Additionally, as a
equation is estimated using the traditional least robustness check, we also estimated regressions
squares method. The interpretation of parameter using the semi-parametric quantile regression
^
δ as the causal effect of FE courses on financial method (Koenker and Basset 1978). This approach
literacy performance relies on the assumption that allows us to estimate different lines of regression
 APPLIED ECONOMICS 7

using all the available information, thereby consid- for a total of nine million 15-year-olds. However,
ering the heterogeneity of the effect of the explana- our research focuses on the Spanish database, com-
tory variables on different sections of the posed of a total of 1050 students from 170 schools.
distribution of the dependent variable (Koenker The number of students whose financial literacy
and Hallock 2001). was evaluated does not match the total number of
students participating in the main PISA test (35 per
school). To be exact, only eight students from each
IV. Data and variables
selected school participated in the financial knowl-
Our empirical study is based on the data provided by edge test. These pupils also answered the questions
the well-known PISA (Programme for International on mathematics and reading comprehension.
Student Assessment) survey, which evaluates the Additionally, these students answered a question-
mathematics, reading and science knowledge of 15- naire on different issues related to their motivations,
year-old students. Financial literacy was added as a their family background and their learning strate-
new competence in 2012. Within PISA, this compe- gies. Although the volume of available information
tence is conceived as students’ ability to apply is very broad, in our empirical analysis we opted to
knowledge and skills in some key areas and analyze, select only some factors that had been reported in
reason and communicate effectively as they state, the background literature to have an important
solve and interpret problems in different situations. impact on educational achievement, like student
PISA focuses on young people’s ability to use their gender, immigrant status, preschool attendance,
knowledge and skills to respond to real-life chal- parents’ educational level and job qualifications,
lenges rather than just their mastery of a specific and number of books available in the household.
curriculum content. Although the content varies On the other hand, school principals completed
across countries, financial literacy usually includes a questionnaire on school resources, learning
categories such as money and transactions, planning environment and the presence of FE. Based on
and managing finance, risk and rewards, and an this information, we can find out if and how FE
understanding of the financial landscape. These is taught at the school and who is responsible for
categories are illustrated by means of several open teaching it.5 Our main focus is placed on the
and multiple-choice questions. The PISA test devel- information provided by the principals to the
oped by the OECD is designed to obtain, based on question of whether FE is available for 15 year-
the responses to these questions, five plausible values old students at their school.6 We can use this
extracted at random from the distribution of the information to construct our main variable of
results to approximate the level of students’ financial interest (FE availability).7 Likewise, school princi-
knowledge (Wu 2005). pals also report how financial education is taught,
All 65 countries participating in PISA 2012 were i.e. whether it is taught as a separate subject (FE
given the option to evaluate students’ knowledge in course) or as part of other subjects (FE cross-cur-
this field, although only 18 decided to participate in ricular). Thus we have defined two dummy vari-
the financial knowledge test (Australia, the Flemish ables according to this information. To build these
Community of Belgium, Colombia, Croatia, Czech variables, we took into account information sup-
Republic, Estonia, France, Israel, Italy, Latvia, New plied by the principals on the teaching hours spent
Zealand, Poland, Russia, Shanghai-China, Slovak on each type of course.8
Republic, Slovenia, Spain and United States). The We also selected some variables representative
sample was composed of 29,000 students accounting of the characteristics of the teaching institution,

5
The information available in this respect refers exclusively to the whether the training is given by teachers from the school or from external institutions.
Unfortunately, no information is available on teaching staff’s expertise with respect to financial issues.
6
The possible responses are: (a) FE is not available; (b) FE has been available for less than two years; (c) FE has been available for two years or more.
7
We collapsed information about responses b and c into a single option (FE availability), thus we can construct a binary variable whose value is one if FE is
available and 0 if it is not.
8
The original information provided by school principals refers to the number of hours per year, divided into five categories (not at all, 1–4, 5–19, 20–49 and
more than 50). To build the dichotomous variables, we followed the criterion of assigning a value of zero if the principals marked options a or b (we regard
a training of less than five hours per year to be practically non-existent) and a value of one if they marked options c, d or e.
8 J. M. CORDERO AND F. PEDRAJA

such as school type (public or government-subsi- Table 2. Descriptive statistics.

dized/private), two indicators representative of Mean SD Min Max
Difference between competences 0.6621 75.3491 −275.47 256.68
school principal leadership skills (LEADCOM Gender 0.4724 0.4995 0 1
and LEADINST),9 another two that approximate Age 15.8540 0.2833 15.33 16.33
Immigrant 0.0933 0.2910 0 1
the level of decision-making responsibility on the No preschooling 0.0609 0.2394 0 1
curriculum and evaluation (RESPCUR) and MaternalEduc 0.3819 0.4861 0 1
PaternalEduc 0.4016 0.4901 0 1
resource allocation (RESRES)10 and, finally, an Books <25 0.2229 0.4164 0 1
indicator that approximates the peer effect, repre- Books>200 0.2771 0.4478 0 1
Private 0.0962 0.2950 0 1
sented by the mean of the ESCS variable of stu- Mean ESCS −0.1289 0.6229 −3.33 1.35
dents from the same school.11 Table 1 defines all LEADCOM −0.4249 0.8462 −2.87 2.84
LEADINST −0.6266 1.0813 −3.85 2.14
the variables included in our empirical study and RESPCUR −0.4453 0.7036 −1.23 1.44
RESPRES −0.3718 0.6867 −0.79 2.71
Table 2 displays the descriptive statistics for these FE availability 0.4971 0.5002 0 1
variables in the Spanish sample. FE course 0.0695 0.2551 0 1
FE cross-curricular 0.4362 0.4961 0 1
Besides variable selection, we should note that the
dataset needed to be manipulated for the purposes of
empirical analysis in order to avoid the usual pro-
blems derived from missing values for some principals in their questionnaire responses. This
variables.12 We apply a multiple imputation method procedure was applied after detecting several cases
which consists of filling the missing values using an in which the principals did not supply information
iterative chained equations process (Schafer 1999; on FE availability at the school but gave positive
Royston, 2009). Moreover, for our main variables responses to questions like who taught FE or
of interest (FE availability, FE course and FE cross- whether financial education was part of the content
curricular), we applied an additional procedure of of other subjects (cross-curricular education). As a
imputation using the information supplied by result, we identify that almost half of the sample

Table 1. Variable description.

DESCRIPTION
Dependent variable
Difference between two Differences between reading comprehension and financial knowledge for each plausible value
competences
Control variables (at individual and school level)
Gender Dichotomous variable whose value is 1 if the student is a girl
Age Continuous variable reflecting the age of the student considering years, months and days from the date of birth to the
date of the test
Immigrant Dichotomous variable whose value is 1 if the student is a first-generation immigrant
No preschooling Dichotomous variable whose value is 1 if the student did not attend preschool
MaternalEduc Dichotomous variable whose value is 1 if the student’s mother has higher education
PaternalEduc Dichotomous variable whose value is 1 if the student’s father has higher education
Books < 25 Dichotomous variable whose value is 1 if there are fewer than 25 books in the household
Books> 200 Dichotomous variable whose value is 1 if there are more than 200 books in the household
Private Dichotomous variable whose value is 1 if the school is a private or government-subsidized independent school
Mean ESCS Mean ESCS variable of all the students from the same school
LEADCOM Compound index representing the principal’s ability to communicate the school’s goals and curriculum development
LEADINST Compound index representing the principal’s leadership with regard to educational issues
RESPCUR Compound index representing the school managers’ responsibility for resource allocation
RESPRES Compound index representing the school managers’ decision-making responsibility on school curriculum and
evaluation
Variables specifically related to FE
FE availability Dichotomous variable whose value is 1 if the student receives some sort of FE at the school
FE course Dichotomous variable whose value is 1 if FE is taught by means of a specific course (at least five hours)
FE cross-curricular Dichotomous variable whose value is 1 if FE is taught by means of a cross-curricular course (at least five hours)

9
Both indicators were built by PISA specialists taking the responses provided by the principals to a number of questions asking about the frequency with
which they perform a series of activities as a baseline (see OECD 2014b, pp. 343–346 for a more detailed description).
10
These indicators were also built by PISA specialists taking the responses of principals to several questions structured similarly to the above as a baseline
(see OECD 2014b, p. 310).
11
The ESCS variable is a synthetic index built by PISA specialists based on information supplied by students on several issues related to their socioeconomic
environment like their parents’ educational level and job qualifications or possessions related to culture in the household.
12
These values were as much as 18% in some cases.
 APPLIED ECONOMICS 9

provides some kind of financial education training, this instruction. Therefore, we need to check
being the cross-curricular approach the most com- whether the characteristics of the students attend-
mon option among them. Likewise, if we observe the ing schools where FE courses are taught are dif-
distribution of observations across regions (Table 3), ferent from schools where no such courses are
we detect that the availability of financial education available, that is, if there is a self-selection bias
is more intensive in regions that decided to partici- into the treatment. For that purpose, we calculate
pate in the ‘Finance for All’ pilot program. In fact, all the mean differences between the two subsamples
the observations belonging to schools teaching spe- for a set of representative variables at both indivi-
cific FE courses are concentrated in those five dual level (including gender, age, immigrant sta-
regions (Andalusia, Castile and Leon, Madrid, tus, preschool attendance, parents’ educational
Murcia and Navarre). level and number of books in the household)
As mentioned previously, our main aim is to and school level (attendance of a private/govern-
analyze whether FE availability at the teaching ment-subsidized school and the mean socioeco-
institution leads to an improvement in students’ nomic level of students of the school as a proxy
financial literacy. To do this, it is necessary to take of the peer-effect).
into account the type of students enrolled in As the values reported in Table 4 show, the
schools in order to properly evaluate the effect of students in the two subsamples are quite similarly
this type of training and rule out possible selection distributed, and no significant differences can be
bias between schools that do and do not provide found for any of the considered variables. As a

Table 3. Distribution of observations across regions.

FE availability FE cross-curricular FE course
TOTAL NO YES YES YES
Region Obs. Obs. % Obs. % Obs. % Obs. %
Andalusia 182 60 32.97% 122 67.03% 102 56.04% 20 10.99%
Aragon 32 25 78.13% 7 21.88% 7 21.88% 0 0.00%
Asturias 29 29 100.00% 0 0.00% 0 0.00% 0 0.00%
Blearic Islands 18 18 100.00% 0 0.00% 0 0.00% 0 0.00%
Cantabria 23 12 52.17% 11 47.83% 11 47.83% 0 0.00%
Castile and Leon 46 14 30.43% 32 69.57% 24 52.17% 8 17.39%
Catalonia 140 73 52.14% 67 47.86% 67 47.86% 0 0.00%
Extremadura 27 19 70.37% 8 29.63% 8 29.63% 0 0.00%
Galicia 46 26 56.52% 20 43.48% 20 43.48% 0 0.00%
La Rioja 17 10 58.82% 7 41.18% 7 41.18% 0 0.00%
Madrid 119 41 34.45% 78 65.55% 48 40.34% 30 25.21%
Murcia 31 13 41.94% 18 58.06% 11 35.48% 7 22.58%
Navarre 32 0 0.00% 32 100.00% 24 75.00% 8 25.00%
Basque Country 62 41 66.13% 21 33.87% 21 33.87% 0 0.00%
Rest of Spain 246 147 59.76% 99 40.24% 99 40.24% 0 0.00%
Total 1,050 528 50.29% 522 49.71% 458 43.62% 73 6.95%

Table 4. Differences between the two subsamples according to FE availability.

With FE Without FE
Variable Mean Std. Dev. Mean Std. Dev. Mean difference t-test (mean difference)
PVmat 485.07 3.8308 485.75 3.6676 −0.6821 0.1286
PVlect 483.37 4.2873 481.17 3.9695 2.2019 0.9571
PVfin 489.66 3.7598 485.57 3.6785 3.9857 1.1471
Gender 0.4827 0.5001 0.4621 0.4992 0.0206 0.6692
Age 15.859 0.2926 15.848 0.2740 0.0105 0.6015
Immigrant 0.0977 0.2972 0.0890 0.2850 0.0086 0.4834
No preschooling 0.0689 0.2536 0.0530 0.2243 0.0159 1.0787
MaternalEduc 0.3716 0.4837 0.3920 0.4886 −0.0204 0.6797
PaternalEduc 0.4157 0.4933 0.3844 0.4869 0.0312 1.0327
Books < 25 0.2356 0.4248 0.2102 0.4078 0.0254 0.9881
Books> 200 0.2567 0.4372 0.2973 0.4575 −0.0406 1.4713
Private 0.0900 0.2865 0.1022 0.3032 −0.0122 0.6718
Mean ESCS −0.1255 0.6274 −0.1322 0.6189 −0.0067 0.1742
Observations 522 528
10 J. M. CORDERO AND F. PEDRAJA

result, both subsamples are comparable with each schools could decide to use both alternatives. The
other. There are no significant differences between sample weights provided by PISA dataset were
the results for students in the mathematics or applied when estimating these models. This guaran-
reading comprehension tests either. Finally, we tees that each student within the sample is properly
find that students attending schools where FE is represented. Additionally, we accounted for the fact
available present slightly better results in financial that students are grouped by schools, applying the
literacy, although the differences are not statisti- appropriate adjustment to calculate the standard
cally significant. errors using the bootstrap method, and the informa-
tion provided by the five plausible values that the
PISA database includes, where the dependent vari-
V. Results
able is the difference between the value for financial
Table 5 displays the results of estimating Equation 3 literacy and reading comprehension. Table 5 shows
using OLS for all the observations available consid- the mean values of the coefficients and standard
ering three alternative models. Model 1 only ana- deviations for the five estimates conducted as
lyses the effect of the availability of FE in the school, recommended by OECD in their reports (see
whereas Models 2 and 3 include two additional OECD 2014b).
dichotomous variables that represent the two alter- According to the information shown in Table 5,
native ways of implementing FE (through a specific the transmission of knowledge related to FE at the
course or adopting a cross-curricular approach) teaching institution does not have any effect on
separately. Finally, Model 4 includes both dichoto- student knowledge acquisition. The result is consis-
mous variables together since it is possible that some tent with the evidence reported in other empirical

Table 5. Results applying OLS.

Model 1 Model 2 Model 3 Model 4
FE availability −5.924 FE availability 0.469 FE availability −7.193 FE availability −4.524
(4.162) (4.346) (4.184) (6.42)
FE course −38.82*** FE cross-curricular 45.09*** FE course −8.317
(9.335) (6.519) (15.40)
FE cross-curricular 44.28***
(16.16)
Gender −22.94*** Gender −21.69*** Gender −22.91*** Gender −21.67***
(4.140) (4.093) (4.004) (3.997)
Age 12.06* Age 10.23 Age 11.96* Age 10.15
(6.637) (7.569) (7.099) (7.682)
Immigrant 5.212 Immigrant 4.230 Immigrant 4.846 Immigrant 4.161
(6.852) (7.301) (7.283) (7.534)
No Preschooling −29.49*** No Preschooling −37.20*** No Preschooling −28.62*** No Preschooling −36.72***
(8.379) (7.766) (7.010) (8.559)
MaternalEduc −14.79*** MaternalEduc −15.78*** MaternalEduc −15.14*** MaternalEduc −15.86***
(4.201) (6.019) (4.309) (5.126)
PaternalEduc 8.107* PaternalEduc 10.65* PaternalEduc 8.210** PaternalEduc 10.86**
(4.812) (5.976) (3.848) (5.154)
Books < 25 −6.325 Books < 25 −3.853 Books < 25 −6.247 Books < 25 −3.898
(4.188) (5.659) (5.673) (5.018)
Books > 200 −3.924 Books > 200 −2.311 Books > 200 −3.985 Books > 200 −2.546
(3.924) (5.619) (4.755) (5.203)
Priv/GSI 7.630 Priv/GSI 10.93 Priv/GSI 7.006 Priv/GSI 10.05
(7.304) (8.552) (6.231) (7.959)
MeanESCS 6.837** MeanESCS 7.673* MeanESCS 8.149** MeanESCS 8.597**
(3.025) (4.465) (3.292) (3.597)
LEADINST −4.460** LEADINST −4.892** LEADINST −4.857*** LEADINST −4.371*
(1.769) (2.317) (1.857) (2.370)
LEADCOM 5.415** LEADCOM 4.391 LEADCOM 5.875** LEADCOM 3.296
(2.658) (2.873) (2.709) (3.100)
RESPRES −5.140 RESPRES −2.023 RESPRES −3.989 RESPRES −3.107
(3.427) (3.685) (3.452) (3.390)
RESPCUR 8.206*** RESPCUR 7.079** RESPCUR 6.896** RESPCUR 6.558**
(2.869) (2.755) (3.268) (3.143)
Constant −168.5 Constant −141.7 Constant −166.8 Constant −141.3
(104.6) (118.5) (113.3) (121.9)
The standard errors are shown between parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1
 APPLIED ECONOMICS 11

studies referring to the United States (Mandell 2008; shows the results of the median regression method
Willis 2008; Cole, Paulson, and Shastry 2016). (0.50) for Models 1, 2, 3 and 4.13 The values of the
Likewise, if we focus on the results of the three estimated parameters for the main variables of
multiple treatment models, we also observe that interest confirm that FE availability at the teaching
the availability of FE is ineffective irrespective of institution is not statistically significant. Likewise,
the type of training adopted. The most striking we also detect a positive effect for cross-curricular
result is that cross-curricular training significantly training and a negative effect for specific training
contributes to the improvement of financial literacy, that disappears if this variable is included together
while the availability of a specific FE training course with the other alternative of implementation.
has a negative effect, although it vanishes when it is Apart from analyzing the effectiveness of FE, it is
included in the regression together with variable worthwhile interpreting the information provided
representing cross-curricular training. This result by the parameters associated with the control vari-
has major implications with respect to how training ables included in the models. Thus, for example, we
on such competences should be instrumented. find that there are significant differences by gender
To supplement this analysis, we also present the (in favor of boys), by age (with OLS only and weak
results of estimating the model using the quantile significance) and between students that have and
regression method. In particular, we estimated the have not attended preschool, but not between
model for three quartiles (0.25, 0.50 and 0.75), native and immigrant students. With regard to
although, for reasons of space, Table 6 only the indicators representing family background, we

Table 6. Results of the estimations using mean quantile regression.

Model 1 Model 2 Model 3 Model 4
FE availability −5.034 FE availability 1.742 FE availability −4.817 FE availability −5.137
(5.085) (5.573) (5.214) (25.58)
FE course −44.70*** FE cross-curricular 40.66*** FE course 2.888
(10.54) (10.29) (20.36)
FE cross-curricular 56.53**
(24.10)
Gender −22.44*** Gender −27.84*** Gender −22.92*** Gender −24.25***
(6.268) (7.263) (5.722) (8.219)
Age 6.644 Age −1.729 Age 9.555 Age 10.63
(11.23) (11.22) (10.27) (15.32)
Immigrant −9.556 Immigrant −10.30 Immigrant −9.925 Immigrant −19.62
(9.242) (8.706) (8.635) (13.81)
No Preschooling −20.47* No Preschooling −30.57*** No Preschooling −22.43* No Preschooling −20.41
(11.62) (10.79) (11.56) (21.20)
MaternalEduc −12.23* MaternalEduc −15.62** MaternalEduc −13.05** MaternalEduc −15.72*
(6.580) (6.792) (6.054) (8.604)
PaternalEduc 3.057 PaternalEduc 6.583 PaternalEduc 2.639 PaternalEduc 12.74
(6.322) (8.148) (6.303) (8.696)
Books <25 −2.321 Books <25 −1.617 Books <25 −1.493 Books <25 5.707
(7.507) (7.699) (6.627) (10.78)
Books > 200 −6.928 Books > 200 −6.570 Books > 200 −6.378 Books > 200 −12.79
(6.007) (7.746) (6.799) (9.689)
Priv/GSI 12.38 Priv/GSI 7.085 Priv/GSI 11.62 Priv/GSI 17.28*
(11.20) (11.52) (8.665) (10.11)
MeanESCS 2.697 MeanESCS 5.616 MeanESCS 4.263 MeanESCS −0.868
(4.356) (4.387) (4.498) (7.807)
LEADINST −3.438 LEADINST −5.828* LEADINST −4.363 LEADINST −0.678
(3.061) (3.014) (3.326) (3.166)
LEADCOM 3.426 LEADCOM 6.717 LEADCOM 4.902 LEADCOM −1.745
(3.289) (4.515) (4.083) (4.794)
RESPRES 0.639 RESPRES 4.288 RESPRES 2.605 RESPRES −0.0839
(5.360) (5.330) (6.043) (5.169)
RESPCUR 4.403 RESPCUR 2.222 RESPCUR 2.375 RESPCUR 8.326
(4.178) (5.301) (4.884) (5.714)
Constant −86.43 Constant 49.70 Constant −132.2 Constant −102.7
(178.3) (179.9) (163.4) (242.1)
The standard errors are shown between parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1

13
The significance of the parameters does not vary with respect to the main variables of interest. The results for the other quantile estimations are available
upon request.
12 J. M. CORDERO AND F. PEDRAJA

find that the parameter that has a larger significant This result is very interesting bearing in mind
effect is mother’s educational level, although this that, since the LOMCE was passed in 2013, FE has
effect is negative, which implies that this is a factor been adopted on a cross-curricular basis as part of
that accounts largely for the reading comprehen- the primary and secondary school curriculum. In
sion results. Another interesting result is the non- any case, we should not overlook the fact that the
significance of variables representing the number of implementation of FE can have a series of direct
books in the household, especially because this is and opportunity costs, which should be examined
usually one of the most important factors for from the cost-benefit analysis viewpoint. In this
explaining educational achievement in other sub- respect, note that there are less costly options,
jects like mathematics or reading comprehension such as the reinforcement of mathematics educa-
(Evans et al. 2010). tion, which could lead to more significant
With regard to school variables, we find some improvements in financial literacy (see Cole,
deviations between the results of the two estima- Paulson, and Shastry 2016). Even so, more specific
tions. Thus, we found that the composition of the complementary training on financial matters
school students, approximated by means of the would still be necessary to acquaint students with
mean socioeconomic level of peers, is statistically the broad spectrum of activities that are part of
significantly related to the dependent variable in financial literacy.
the estimation by OLS. Likewise, we do not find Note that our analysis has some limitations that
significant differences between public and private/ could partly explain the results. The main shortcom-
government-subsidized schools. However, both ing is that we do not have individualized informa-
aspects are not significant in the quantile estima- tion about the specific training that each student has
tion. Finally, the two variables used to account for received or their level of achievement. This poses a
principals’ leadership skills have opposite effects in serious threat to the validity of the results. On top of
OLS estimations, which suggests that one has more this, the empirical analysis did not take into account
influence on the reading comprehension education the quality of the teachers that teach FE, even though
(LEADINST) and the other on financial literacy this is a factor that, according to several researchers,
(LEADCOM). Likewise, we find that there is a may have an influence on the success of this type of
positive and statistically significant relationship initiatives (Way and Holden 2009; Walstad et al.,
with the variable representing the responsibility of 2010). In this respect, Mandell (2008) points out that
local managers for curriculum management and better materials and, especially, better teacher train-
evaluation (RESPCUR) according to the OLS ing are required for FE programs to be effective.
estimation. Unfortunately, the PISA 2012 database does not
provide information on this major issue.
Neither did the analysis take into account that
VI. Concluding remarks
the students have rather heterogeneous experience
In this paper, we reported the evaluation of the in money-related issues. Although the PISA data-
effect of adopting financial education into the base does provide this information, it is only avail-
school curriculum using the information provided able for half of the evaluated students, as a result
by the PISA 2012 database, which for the first time of which it cannot be adopted as control variables
evaluated the Spanish students’ knowledge of in our model because we would run into serious
competences related to money management and trouble regarding missing values. With regard to
different financial products. According to the this issue, some authors have suggested that FE
results of the proposed empirical analysis, the programs targeting individuals with particular
availability of financial education in the school characteristics are usually more effective than gen-
does not have any effects per se, although their eral-purpose courses (Lusardi, Michaud, and
implementation using a cross-curricular approach, Mitchell 2013). This could explain why FE pro-
that is, as part of the curriculum of other subjects, grams proved to be ineffective in Spain.
can make a significant contribution to the acquisi- Finally, note that the number of schools providing
tion of financial literacy. this type of education should grow over the coming
 APPLIED ECONOMICS 13

years with the gradual adoption of financial educa- Behavior, and Attitudes.” Journal of Consumer Affairs 49
tion in the school curriculum in both primary and (1): 69–96.
secondary education, scheduled after the adoption of Becchetti, L., S. Caiazza, and D. Coviello. 2013. “Financial
Education and Investment Attitudes in High Schools:
the LOMCE. Therefore, it will be interesting to rea- Evidence from a Randomized Experiment.” Applied
nalyze the effectiveness of these initiatives in the near Financial Economics 23 (10): 817–836.
future to check whether these measures have helped Bernheim, B. D., D. M. Garrett, and D. M. Maki. 2001.
to improve Spanish students’ financial literature. “Education and Saving: The Long-Term Effects of High
Unfortunately, the 2015 PISA wave provided hardly School Financial Curriculum Mandates.” Journal of Public
any information with regard to the deployment of Economics 80 (3): 435–465.
Berry, J., D. Karlan, and M. Pradhan 2015. The Impact of
financial education in schools, and it is therefore not Financial Education for Youth in Ghana. National Bureau
possible to conduct a comparable study using those of Economic Research, Paper No. 21068.
data. We trust that future waves will again gather the Bietenbeck, J. 2014. “Teaching Practices and Cognitive
data necessary to be able to reexamine the effective- Skills.” Labour Economics 30: 143–153.
ness of these initiatives in a context where many Bruhn, M., L. de Souza Leão, A. Legovini, R. Marchetti, and
more teaching institutions offer financial education. B. Zia 2013. Financial Education and Behavior Formation:
Large-Scale Experimental Evidence from Brazil.
Presentation, World Bank, Washington, DC.
Bucher-Koenen, T., and A. Lusardi. 2011. “Financial Literacy
Acknowledgments
and Retirement Planning in Germany.” Journal of Pension
The authors would like to express their gratitude to the Economics and Finance 10 (04): 565–584.
Savings Banks Foundation (Fundación de las Cajas de Christelis, D., T. Jappelli, and M. Padula. 2010. “Cognitive
Ahorros –FUNCAS-) and the Spanish Ministry for Abilities and Portfolio Choice.” European Economic
Economy and Competitiveness for supporting this research Review 54 (1): 18–38.
through grant ECO2017-83759-P. Cole, S., A. Paulson, and G. K. Shastry. 2016. “High School
Curriculum and Financial Outcomes: The Impact of
Mandated Personal Finance and Mathematics Courses.”
Disclosure statement Journal of Human Resources 51 (3): 656–698.
Cole, S., T. Sampson, and B. Zia. 2011. “Prices or Knowledge?
No potential conflict of interest was reported by the authors. What Drives Demand for Financial Services in Emerging
Markets?” The Journal of Finance 66 (6): 1933–1967.
Collins, J. M., and C. M. O’Rourke. 2010. “Financial
Funding Education and Counseling - Still Holding Promise.”
Journal of Consumer Affairs 44 (3): 483–498.
This work was supported by the Ministerio de Economia, Cordero, J. M., V. Cristobal, and D. Santín. 2017. “Causal
Industria y Competitividad [ECO2017-83759-P] and the Inference on Education Policies: A Survey of Empirical
Saving Banks Foundation (Fundación de las Cajas de Studies Using PISA, TIMSS and PIRLS.” Journal of
Ahorros –FUNCAS-). Economic Surveys en prensa. doi:10.1111/joes.12217.
Dee, T. S. 2005. “Teachers and the Gender Gaps in Student
Achievement.” Journal of Human Resources 42 (3): 528–554.
ORCID Disney, R., and J. Gathergood. 2013. “Financial Literacy and
José Manuel Cordero http://orcid.org/0000-0001-8783- Consumer Credit Portfolios.” Journal of Banking and
6748 Finance 37 (7): 2246–2254.
Duflo, E., R. Glennerster, and M. Kremer. 2007. “Using
Randomization in Development Economics Research: A
References Toolkit.” Handbook of Development Economics 4: 3895–3962.
Evans, M., J. Kelley, J. Sikora, and D. Treiman. 2010. “Family
Appleyard, L., and K. Rowlingson. 2013. “Children and Financial Scholarly Culture and Educational Success Books and
Education: Challenges for Developing Financial Capability in Schooling in 27 Nations.” Research in Social Stratification
the Classroom.” Social Policy and Society 12 (04): 507–520. and Mobility 28 (2): 171–197.
Atkinson, A., and F. A. Messy 2013. Promoting Financial Fernandes, D., J. G. Lynch Jr, and R. G. Netemeyer. 2014.
Inclusion through Financial Education. OECD Working “Financial Literacy, Financial Education, and Downstream
Papers on Finance, Insurance and Private Pensions, (34), 1. Financial Behaviors.” Management Science 60 (8): 1861–1883.
Batty, M., J. M. Collins, and E. Odders-White. 2015. Fornero, E., and C. Monticone. 2011. “Financial Literacy and
“Experimental Evidence on the Effects of Financial Pension Plan Participation in Italy.” Journal of Pension
Education on Elementary School Students’ Knowledge, Economics and Finance 10 (04): 547–564.
14 J. M. CORDERO AND F. PEDRAJA

Fox, J., S. Bartholomae, and J. Lee. 2005. “Building the Case Education and Program Evaluation.” Journal of Consumer
for Financial Education.” Journal of Consumer Affairs 39 Affairs 40 (2): 208–235.
(1): 195–214. Mandell, L. 2008. “Financial Education in High School.” In
Gerardi, K., L. Goette, and S. Meier 2010. Financial Literacy Overcoming the Saving Slump: How to Increase the
and Subprime Mortgage Delinquency: Evidence from a Effectiveness of Financial Education and Saving Programs,
Survey Matched to Administrative Data. Federal Reserve edited by A. Lusardi, 257–279. Chicago and London:
Bank of Atlanta, Working Paper Series No. 2010–10. University of Chicago Press.
Gnan, E., M. A. Silgoner, and B. Weber. 2007. “Economic and Mandell, L., and L. S. Klein. 2009. “The Impact of Financial
Financial Education: Concepts, Goals and Measurement.” Literacy Education on Subsequent Financial Behavior.”
Monetary Policy & the Economy Quarterly 3 (1): 28–49. Journal of Financial Counseling and Planning 20 (1): 15-24.
Grifoni, A., and F. Messy 2012. Current Status of National McCormick, M. H. 2009. “The Effectiveness of Youth
Strategies for Financial Education: A Comparative Analysis Financial Education: A Review of the Literature.” Journal
and Relevant Practices, OECD Working Papers on Finance, of Financial Counseling and Planning 20 (1): 70–83.
Insurance and Private Pensions, No. 16, OECD Publishing. Messy, F., and C. Monticone 2012. The Status of Financial
Guiso, L., and T. Jappelli. 2002. “Household Portfolios in Education in Africa, OECD Working Papers on Finance,
Italy.” In Household Portfolios, 251–289, MIT press. Insurance and Private Pensions, No. 25, OECD Publishing,
Hastings, J. S., B. C. Madrian, and W. L. Skimmyhorn. 2013. Paris.
“Financial Literacy, Financial Education, and Economic Messy, F., and C. Monticone 2016. Financial Education
Outcomes.” Annual Review of Economics 5 (1): 347–373. Policies in Asia and the Pacific, OECD Working Papers
Hill, R. C., W. E. Griffiths, and G. C. Lim. 2008. Principles of on Finance, Insurance and Private Pensions, No. 40, OECD
Econometrics. 4th ed. Hoboken, NJ: Wiley. Publishing, Paris.
Hospido, L., E. Villanueva, and G. Zamarro 2015. Finance for Miller, M., J. Reichelstein, C. Salas, and B. Zia. 2015. “Can
All: The Impact of Financial Literacy Training in You Help Someone Become Financially Capable? a Meta-
Compulsory Secondary Education in Spain, Banco de analysis Of The Literature.” The World Bank Research
España, Working Paper 1502/2015. Observer 30 (2): 220-246.
Hung, A., A. M. Parker, and J. Yoong 2009. Defining and OECD. 2014a. “PISA 2012 Results: Students and Money:
Measuring Financial Literacy. RAND Working Paper 708. Financial Literacy Skills for the 21st Century (Volume
Jamison, J. C., D. Karlan, and J. Zinman 2014. Financial VI) pp. 1-204.” In Pisa. Paris: OECD Publishing.
Education and Access to Savings Accounts: OECD. 2014b. PISA 2012 Technical Report, OECD
Complements or Substitutes? Evidence from Ugandan Publishing, Paris.
Youth Clubs. National Bureau of Economic Research, OECD. 2005. Improving Financial Literacy: Analysis of Issues
Paper No. 20135. and Policies. Paris: OECD Publishing.
Jürges, H., K. Schneider, and F. Büchel. 2005. “The Effect of OECD. 2013. Advancing National Strategies for Financial
Central Exit Examinations on Student Achievement: Quasi- Education. Paris: OECD Publishing.
Experimental Evidence from TIMSS Germany.” Journal of Peng, T., S. Bartholomae, J. Fox, and G. Cravener. 2007. “The
the European Economic Association 3 (5): 1134–1155. Impact of Personal Finance Education Delivered in High
Koenker, R., and G. Basset. 1978. “Regression Quantiles.” School and College Courses.” Journal of Family and
Econometrica 46 (1): 33–50. Economic Issues 28 (2): 265–284.
Koenker, R., and K. F. Hallock. 2001. “Quantile Regression.” Rivkin, S. G., and J. C. Schiman. 2015. “Instruction Time,
Journal of Economic Perspectives 15 (4): 143–156. Classroom Quality, and Academic Achievement.” The
Lührmann, M., M. Serra-Garcia, and J. Winter. 2015. Economic Journal 125: 425–448.
“Teaching Teenagers in Finance: Does It Work?” Journal Romagnoli, A., and M. Trifilidis 2013. Does Financial
of Banking & Finance 54: 160–174. Education at School Work? Evidence from Italy.
Lusardi, A., and O. S. Mitchell. 2007. “Financial Literacy and Evidence from Italy. Bank of Italy Occasional Paper, (155).
Retirement Planning in the United States.” Journal of Royston, P. 2009. “Multiple Imputation Of Missing Values:
Pension Economics and Finance 10 (04): 509–525. Further Update Of Ice, with an Emphasis on Categorical
Lusardi, A., and O. S. Mitchell. 2014. “The Economic Variables.” Stata Journal 9 (3): 466–477.
Importance of Financial Literacy: Theory and Evidence.” Schafer, J. L. 1999. “Multiple Imputation: A Primer.”
Journal of Economic Literature 52 (1): 5–44. Statistical Methods in Medical Research 8 (1): 3–15.
Lusardi, A., and P. Tufano. 2015. “Debt Literacy, Financial Schlotter, M., G. Schwerdt, and L. Woessmann. 2011.
Experiences, and Overindebtedness.” Journal of Pension “Econometric Methods for Causal Evaluation of
Economics and Finance 14 (04): 332–368. Education Policies and Practices: A Non-Technical
Lusardi, A., P. C. Michaud, and O. S. Mitchell 2013. Optimal Guide.” Education Economics 19 (2): 109–137.
Financial Knowledge and Wealth Inequality. National Schwerdt, G., and A. C. Wuppermann. 2011. “Is Traditional
Bureau of Economic Research, Paper No. 18669. Teaching Really All that Bad? A Within-Student Between-
Lyons, A., L. Palmer, K. Jayaratne, and E. Scherpf. 2006. “Are Subject Approach.” Economics of Education Review 30 (2):
We Making the Grade? A National Overview of Financial 365–379.
 APPLIED ECONOMICS 15

Shim, S., B. L. Barber, N. A. Card, J. J. Xiao, and J. Walstad, W. B. 2013. “Economic Understanding in US High
Serido. 2010. “Financial Socialization of First-Year School Courses.” The American Economic Review 103 (3):
College Students: The Roles of Parents, Work, and 659–663.
Education.” Journal of Youth and Adolescence 39 (12): Walstad, W.B., K. Rebeck, and R.A. MacDonald. 2010. “The
1457–1470. Effects Of Financial Education on The Financial
Stango, V., and J. Zinman. 2009. “Exponential Growth Bias Knowledge Of High School Students.” Journal of
and Household Finance.” The Journal of Finance 64 (6): Consumer Affairs 44 (2): 336-357.
2807–2849. Way, W. L., and K. C. Holden. 2009. “2009 Outstanding AFCPE
Swinton, J. R., T. W. DeBerry, B. Scafidi, and H. C. Woodard. Conference Paper Teachers’ Background and Capacity to
2007. “The Impact of Financial Education Workshops for Teach Personal Finance: Results of a National Study.”
Teachers on Students’ Economic Achievement.” The Journal of Financial Counseling and Planning 20 (2): 64.
Journal of Consumer Education 24: 63–77. Willis, L. E. 2008. “Against Financial Literacy Education.”
Tennyson, S., and C. Nguyen. 2001. “State Curriculum Iowa Law Review 94: 08–10.
Mandates and Student Knowledge of Personal Finance.” Willis, L. E. 2011. “The Financial Education Fallacy.”
Journal of Consumer Affairs 35 (Winter): 241–262. American Economic Review 101 (3): 429–434.
Van Rooij, M., A. Lusardi, and R. Alessie. 2011. “Financial Wu, M. 2005. “The Role of Plausible Values in Large Scale
Literacy and Stock Market Participation.” Journal of Surveys.” Studies in Educational Evaluation 31: 114–128.
Financial Economics 101 (2): 449–472. Xu, L., and B. Zia 2012. Financial Literacy around the World:
Van Rooij, M. C. J., A. Lusardi, and R. J. M. Alessie. 2012. An Overview of the Evidence with Practical Suggestions
“Financial Literacy, Retirement Planning and Household for the Way Forward. World Bank Policy Research
Wealth.” The Economic Journal 122 (560): 449–478. Working Paper, 6107.