Financial education and student financial literacy:

A cross-country analysis using PISA 2012 data

José Manuel Cordero1*, María Gil-Izquierdo2, Francisco Pedraja-Chaparro1

1 Universidad de Extremadura, Av. Elvas sn, 06006, Badajoz, Spain2 Universidad Autónoma de Madrid, C/ F. Tomás y Valiente, 528049, Madrid,

Spain

Abstract

The aim of this research is to explore whether the availability of training on basic financial concepts at schools contributes to improve the abilities of students to apply the knowledge and skills that they learn to real-life situations involving financial issues and decision making. To do this, we exploit the rich set of comparative data about the countries participating in the PISA 2012 financial literacy assessment. Our empirical analysis is based on multilevel (hierarchical) regression modeling including country fixed-effects. Our results suggest that the availability of specific financial training is positively and significantly related with students´ financial literacy, regardless of the strategy applied to teach financial concepts, although its influence appears to be relatively small when we account for the potential presence of significant differences among countries. In addition, we find that students receiving courses taught by specialists from private institutions and non-governmental organizations obtain better results than students who receive financial education training from their teachers.

Keywords: Education policy; Cross-country analysis; Financial literacy; PISA; Large-scale assessments; Multilevel regressions.

*Corresponding author´s e-mail: jmcordero@unex.ex

1

1. Introduction

Interest in financial literacy has increased worldwide based on the belief that its

improvement will empower people to make better financial decisions (Hilgert et al.,

2003; Borden et al., 2008). As a result, financial education programs figure prominently

in the national public policy agenda of most countries (Lusardi and Mitchell, 2011;

Appleyard and Rowlingson, 2013). Likewise, some international institutions such as the

World Bank and the Organization for Economic Cooperation and Development

(OECD) with its International Network on Financial Education (INFE) have made

efforts to promote financial literacy, coordinate national programs and provide guidance

on the direction that such strategies should take in the future (OECD, 2012).

Even though most financial decisions are made by adults, all the above

international bodies appear to be in agreement that it is crucial to begin cultivating

financial literacy at school in the hope that students will develop the skills needed to

successfully manage their finances in adulthood when they will be exposed to

increasingly complex financial products and services. Moreover, there is evidence to

show that young people’s levels of financial literacy are consistently lower than for

other demographics (Mandell, 2008; Shim et al., 2010). Therefore, the improvement of

students’ financial knowledge is essential so that they can participate in modern society

as well as being beneficial for the economy and society as a whole (Lusardi et al.,

2010).

As a result, many countries are developing plans to introduce contents related to

financial education (FE) in their school curriculum, especially for low-income or lesser

educated populations (Kozup and Hogarth, 2008). This should give the entire school-

age population equal access to financial education. To do so, they are adopting different

strategies. These strategies range from a well-developed framework to basic pilot

programs to test the introduction of financial competences in the curriculum. However,

the most common option is to merely provide a form of financial education by means of

a cross-curricular approach, i.e., linking financial concepts with some other learning

areas. Moreover, schools may adopt a flexible approach to the integration of financial

education into the curriculum, and teachers may also decide whether or not to include

aspects of financial literacy within their subjects. Therefore, there are a lot of

2

differences across territories and also among schools within the same territory (Grifoni

and Messy, 2012; Atkinson and Messy, 2013).

In this paper, we attempt to exploit this heterogeneity in the international context

in order to determine whether training of students in financial concepts contributes to

enhance their financial literacy, since the existence of larger variations in school and

population characteristics generally improves the prospects of detecting the influence of

specific factors on student outcomes (Hanushek and Woessman, 2014). To do this, we

adopt a multilevel regression approach, which allows us to avoid potential problems of

bias in the estimations derived from classic methods, such as OLS regression, due to the

existence of correlation between the values of student variables aggregated at school

level or schools belonging to the same country (Hox, 2002).

In our empirical analysis we use data from the OECD’s Programme for

International Student Assessment (PISA), which included a module on financial literacy

for the first time in 2012. In particular, students from 18 countries participated in this

optional PISA assessment. The assessment provides comparable information with

regard to the financial competences of 15-year-olds worldwide by testing their

knowledge as applicable in everyday life situations rather than the reproduction of

knowledge (OECD, 2013). The availability of such comparable data across countries is

essential for understanding how well prepared young people are to deal with new and

changing financial environments. Moreover, the dataset includes extensive information

about individual characteristics, socio-economic background and school contexts. In

this manner, the analysis can account for these factors.

This research falls within the scope of recent literature focusing on the assessment

of the effectiveness of financial education programs (McCormick, 2009; Xu and Zia,

2012; Lusardi and Mitchell, 2014). Most existing papers regarding this topic refer to the

specific context of the United States where high school financial education mandates

have been enacted in many states over the past fifty years (Bernheim et al. 2001).

Nevertheless, there are also some interesting initiatives in other developed and

developing countries (e.g., Romagnoli and Trifilidis, 2013; Becchetti, Caiazza and

Coviello, 2013; Berry, Karlan and Pradhan, 2018; Lührmann, Serra-Garcia and Winter,

2015). All the above studies focus on specific programs with different characteristics

3

implemented in highly heterogeneous contexts. Therefore, the evidence about their

effectiveness is mixed (Walstad, 2013), since several authors are rather skeptical about

the effective contributions of financial education (Willis, 2008; Fernandes et al., 2014;

Cole et al., 2016), while many others have found a positive correlation between its

implementation and educational outcomes (Varcoe et al. 2005; Walstad et al., 2010;

Batty et al., 2015).

The scant previous evidence using a cross-country approach opens up

opportunities as well as challenges, since very few studies have used international data

to analyze differences in financial literacy across countries. For instance, Jappelli (2010)

uses international panel data on 55 countries in order to explore the macroeconomic and

institutional variables that are more likely to explain international differences in literacy

across countries. Likewise, Nicolini et al. (2013) also use a similar approach to collect

data about only four countries and construct a financial literacy index based on the

number of correct answers to similar questions in different national surveys.

In this article we attempt to take advantage of a common measure of financial

literacy for students from different countries, as well as data about diverse initiatives

retrieved by means of the same data collection process. Thus, we can examine whether

receiving some form of training about financial concepts contributes to improve young

students’ financial knowledge. Furthermore, given that the PISA dataset also provides

additional information about how financial education is implemented at each school, we

also explore in which extent different types of teaching strategies affect the results in an

attempt to provide evidence about one of the most actively debated issues in both the

scientific community and among educators.

The rest of the paper is structured as follows. Section 2 summarizes the situation

of financial education in countries participating in the financial literacy test in PISA

2012. Section 3 provides a description of the dataset and the variables considered in our

empirical analysis. Section 4 explains the methodology employed in our study and

Section 5 reports the main results. Finally, Section 6 outlines our main conclusions.

4

2. Financial education in countries participating in the PISA 2012 financial

literacy test

As pointed out previously, the awareness of the importance of financial education

has led many countries to develop an increasing number of national strategies for

financial education. Such strategies represent a systematic approach to reinforce the

financial literacy of their citizens (OECD, 2015). Those strategies started mostly in

developed economies such as the United States, Japan, the United Kingdom, Australia,

New Zealand, the Netherlands or Singapore. However, since the beginning of the

economic crisis, such initiatives have spread to other countries with varying economic,

financial and socio-demographic contexts (Grifoni and Messy, 2012).

One of the main challenges of such strategies is to include financial competences

in primary and secondary school education programs to improve financial awareness

from early ages. For example, several US states have adopted mandates to include

financial education in the curriculum of high school students, while Australia has had a

financial education mandate since 2011. However, only a few countries have so far

established a well-developed framework for introducing financial competences into

their education systems. Given that the main focus of this research is to analyze the role

of financial training on students’ knowledge about financial issues, our starting point is

to examine the proportion of students attending financial education courses in each

country. Figure 1 shows this information based on the responses provided by the

principals of schools participating in the PISA survey. Although the average percentage

of students attending financial education courses is relatively high (70%), we observe

that there is a considerable variation across countries, ranging from percentages above

80% in Australia, Belgium, the United States or New Zealand to less than a half in

Slovenia, Poland, Croatia, Italy or Spain.

Table A1 in the Appendix summarizes the different financial education set-ups in

the curricular design of all the countries participating in the PISA 2012 financial literacy

test. It also includes information about some pilot programs implemented in a number

of countries to incorporate financial competences into the curriculum before they launch

a national strategy. For instance, the Spanish and Italian central banks and a number of

ministries promoted several experimental programs with the aim of incorporating

5

financial education into school curricula. In contrast, such programs were mainly

implemented by private financial institutions in Colombia.

Figure 1. Availability of FE in schools by countries

SVNPOLHRVITAESPISR

QCNFRAESTRUSCOL

OECDLVASVKNZL

USACZEBEL

AUS

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%36%

42%43%

46%48%

50%50%

52%58%

67%69%70%

77%79%

82%83%83%

85%85%

No Yes

Source: Own elaboration from PISA 2012

Based on the content of Table A1, as well as the information provided by school

principals, we can take a step further and explore different ways of including financial

education in the curriculum. Firstly, note importantly that financial education courses

are not compulsory in most countries. Exceptions are frequently represented by schools

located in specific regions or states where financial education is established as a

compulsory subject (e.g. the United States). As a result, the proportion of schools that

can be included in this category is relatively low in most countries, as indicated in

Figure 21.

With regard to the manner in which financial education courses are incorporated

into the curriculum, the most common option is the cross-curricular approach, i.e.,

financial concepts are included as a part of other subjects such as mathematics,

humanities or social sciences, whereas it is less common for financial education to be

1. The main exception is the Czech Republic where financial education has been compulsory at upper secondary school level since 2009 and at lower secondary school since 2013.

6

taught as a separate subject. Notice again that financial education might be included at

different levels of the educational system. Thus we have found that there are several

countries where financial education concepts are studied in primary education (Latvia,

the Czech Republic, Shanghai-China, Estonia or Australia), whereas they are taught

during compulsory secondary education in other education systems (the Flemish

Community of Belgium, the Slovak Republic, Israel, Italy or Poland).

Figure 2. Financial education as a compulsory subject by countries

ESPNZLISREST

HRVQCNPOLSVNFRAAUSITA

COLLVAUSARUSBELSVKCZE

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

3%9%

12%12%

20%23%

25%27%

29%29%29%30%

34%41%41%

52%59%

73%

NO YES

Source: Own elaboration from PISA 2012

Nevertheless, our empirical analysis focuses on the strategies implemented in

lower secondary schools since our data source is the information provided by school

principals participating in PISA as explained below. According to this information,

summarized as country averages in Figures 3 and 4, we can identify some countries

where the cross-curricular approach is clearly the main alternative (e.g., the Slovak

Republic, the Czech Republic or Estonia) and others were the preferred option is to

teach financial education as a separate subject (e.g., the United States or New Zealand),

although a combination of both strategies is also a common practice (e.g., Shangai-

China, Colombia or the Russian Federation).

7

Figure 3. Financial education using a cross-curricular approach by countries

HRVPOLITANZLFRAESP

QCNRUSAUSLVAISR

COLUSASVNESTBELCZESVK

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

10%13%

16%19%

26%28%

30%33%

36%38%39%

40%43%

50%64%66%

77%86%

NO YES

Source: Own elaboration from PISA 2012

Figure 4. Financial education taught as a separate subject by countries

ESPESTLVAPOLHRVITA

SVNISR

SVKFRACZEBEL

AUSCOL

QCNRUSNZL

USA

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

7%9%

11%14%

19%19%

21%23%24%

26%26%

28%30%

38%40%

44%48%

59%

NO YES

Source: Own elaboration from PISA 2012

3. Data and variables

8

For the first time, PISA 2012 conducted an assessment of the financial knowledge

of 15-year-old students around the world. This was an optional assessment for countries

and economies. Eighteen countries and economies participated in the assessment of

financial literacy. They include 13 OECD countries and economies: Australia, the

Flemish Community of Belgium, the Czech Republic, Estonia, France, Israel, Italy,

New Zealand, Poland, the Slovak Republic, Slovenia, Spain and the United States; and

five partner countries and economies: Colombia, Croatia, Latvia, the Russian Federation

and Shanghai-China. Around 29,000 students completed the financial literacy

assessment in 2012.

The students participating in the financial literacy assessment were recruited and

assessed separately from and in addition to the other pupils participating in the core

PISA assessment (35 per school). In particular, eight additional 15-year-old students

were selected randomly from students enrolled in each participating school to take the

financial literacy assessment. The test comprised four 30-minute clusters of test material

which each student had a total of two hours to complete. Each booklet included two

clusters of financial literacy items (a total of 40 questions) that they had to complete in

60 minutes. These questions cover different contents (e.g. identify different ways to pay

for items, calculate correct change, work out which of two consumer items of different

sizes would give better value for money according to their needs and circumstances,

understand that money can be borrowed or lent and the reasons for paying or receiving

interest, identify which providers are trustworthy, etc.) and adopt a wide array of

alternatives (open-constructed response, constructed response and multiple-choice) that

usually requires to perform some simple mathematical calculations. In general terms,

questions and answers are quite short and direct, thus they only require a basic

proficiency in reading literacy. Moreover, each booklet also includes one cluster of

mathematics test items and one cluster of reading items very similar to the core

assessment in PISA. Therefore, data about three different domains (financial literacy,

mathematics and reading) is available for this smaller sample of students.

The financial literacy assessment includes three different dimensions: contents,

processes and contexts. The content categories comprise the areas of knowledge and

understanding that are essential in order to perform a particular financial task. They

include money and transactions, planning and management of finances, risk and reward

9

and financial landscape. The process categories refer to cognitive processes and

describe students’ ability to recognize and apply key concepts in the domain, as well as

to understand, evaluate and suggest solutions. Finally, the contexts represent the

situations in which financial knowledge, skills and understanding are applied. The focus

may be on the individual, family or peer group, the community or the school or even on

a global scale.

PISA does not evaluate cognitive abilities or skills through using one single score

but each student receives different scores (plausible values) that represent the range of

abilities that a student might reasonably have (see OECD, 2014 for details).

Specifically, the dataset provides measures on student achievement based upon pupils’

responses to different test booklets, each of which includes only a limited number of test

questions. Using a complex process based on item response theory model, the survey

organizers produce test scores for financial literacy taking into account the difficulty of

each test question2. Plausible values can therefore be defined as random values drawn

from this distribution of proficiency estimates (Mislevy et al., 1992; Wu, 2005). In our

analysis we use the five available plausible values for financial literacy as our

dependent variables.

The PISA dataset also includes a wide range of variables on student background,

learning experiences and attitudes drawn from the student questionnaire, as well as data

about school resources and policies completed by school principals. Among them, we

select a set of student background variables that have been most frequently identified as

influential factors in previous literature, such as gender, age, being an immigrant,

attending pre-school, parents´ level of education or the number of books at home.

Likewise, we consider several control variables at school level (being a private school,

location in a rural area and the average socio-economic status index (ESCS3) of students

in the school as a proxy of the peer effect.

More importantly, principals from all the participating schools provide the same

information about financial education at the school in their responses to the school

questionnaire. This is the main focus of this empirical research. In particular, data 2. See Von Davier and Sinharay (2013) for further details.3 This is an indicator of the economic, social and cultural status of students created by PISA

analysts from three variables related to family background from students’ questionnaire: the highest educational level of either of the student’s parents, the highest occupational status of either of the student’s parents and an index of educational possessions with respect to household economy.

10

include a specific question about whether or not there is availability of financial

education for 15 year-old students. We can use this information to construct our main

variable of interest. Likewise, school principals also report how financial education

courses are taught, including whether they are compulsory for students, whether they

are taught as a separate subject or by means of a cross-curricular approach, i.e., as part

of other subjects and who provides financial education (teachers or people from

different institutions –private, public or non-government organizations–). Since we are

also interested in studying different ways of implementing financial education courses

at schools, we have defined several dummy variables according to this information.

Moreover, several questions about students’ experience with money matters were

included at the end of the financial literacy test booklets. The questionnaire covered

multiple aspects such as having access to financial products or their sources of money.

Unfortunately, these questions were partly implemented by students4, thus these

variables contain a substantial proportion of missing values. In order to account for this

relevant information, we have aggregated the available information about these issues at

country level. Specifically, we retrieved three variables representing the percentages of

students holding a bank account, having money coming from gifts from friends or

relatives and receiving money from an allowance (without having to do chores)5.

Moreover, we have also collected additional information at country level about the

gross domestic product per capita and some proxy variables representing the level of

financial development such as the number of commercial bank branches or the total

value of stocks traded. These data were collected from the World Bank´s Indicators

database for the year 2012.

Table 1 contains the definition of all the variables considered in our empirical

analysis and Table 2 shows the descriptive statistics of all the variables classified in four

blocks: those related to students, the ones referred to the school, the specific variables

related to financial education and the covariates at country level.

Table 1. Variable description

DESCRIPTION4 This questionnaire was split into four parts or booklets. Each part was given to a quarter of the

students. Consequently, not all the students answered all the questions.5 See OECD (2014, pp. 99-109) for details.

11

Dependent variable

PV1FLIT 1st plausible value (PV) in financial literacyPV2FLIT 2nd plausible value (PV) in financial literacyPV3FLIT 3rd plausible value (PV) in financial literacyPV4FLIT 4th plausible value (PV) in financial literacyPV5FLIT 5th plausible value (PV) in financial literacyCovariates at student levelGender Dummy variable (DM) that takes value 1 if the student is

a girlAge Age of the studentImmigrant DM: value 1 if the student was born in other countryPre-

primary_NODM: 1 if the student has not attended pre-primary school

Motheredu DM: 1 if student´s mother has an university degreeFatheredu DM: 1 if student´s father has an university degreeBooks25 DM: 1 if there are less than 25 books at homeBooks200 DM: 1 if there are more than 200 books at homeCovariates at school levelPrivate DM: 1 if the school is privateRural DM: 1 if the school is placed in a village or small town.ESCS mean Average value of the ESCS index at school levelSpecific variables related to financial education

FE available DM: 1 when Financial education is available in the school

FE compulsory DM: 1 when Financial education is compulsory in the school

FE separate DM: 1 if FE is taught as a separate subjectFE cross DM: 1 if FE is taught as a cross-curricular subjectFE priv DM: 1 if FE is provided by people from private sector

institutions (e.g. banks, insurance company)FE pub DM: 1 if FE is provided by people from public

institutions (e.g. ministry of finance, reserve bank)FE NGO DM: 1 if FE is provided by people from non-government

organizationsCovariates at country levelAccount Percentage of students holding a bank account

Gifts Percentage of students having money coming from gifts from friends or relatives

Allowance Percentage of students receiving money from an allowance without having to do chores

GDPpc Gross domestic product per capitaBanks Number of commercial branches in the countryStocks Total value of stocks traded (% GDP)

Table 2. Descriptive statistics

Mean Std. Deviation

Min. Max.

12

Dependent variablePV1FLIT 491.94 101.22 2.71 921.03PV2FLIT 491.82 101.47 5.79 873.48PV3FLIT 491.32 101.40 17.07 860.80PV4FLIT 491.52 101.69 31.61 850.55PV5FLIT 491.63 101.63 11.10 879.08Covariates at student level0.4699Gender 0.50 0.50 0.00 1.00Age 15.78 0.29 15.25 16.33Immigrant 0.04 0.20 0.00 1.00Preprim_N

O0.07 0.25 0.00 1.00

Motheredu 0.39 0.49 0.00 1.00Fatheredu 0.36 0.48 0.00 1.00Books25 0.33 0.47 0.00 1.00Books200 0.19 0.40 0.00 1.00Covariates at school levelPrivate 0.05 0.21 0.00 1.00Rural 0.25 0.43 0.00 1.00ESCS mean -0.08 0.65 -3.55 1.88Specific variables related to financial educationFE

available0.67 0.47 0.00 1.00

FE compulsory

0.30 0.46 0.00 1.00FE separate 0.26 0.44 0.00 1.00FE cross 0.35 0.48 0.00 1.00FE priv 0.12 0.32 0.00 1.00FE pub 0.05 0.22 0.00 1.00FE NGO 0.10 0.30 0.00 1.00Covariates at country levelAccount 0.55 0.23 0.20 0.93Gifts 0.83 0.09 0.61 0.93Allowance 0.47 0.18 0.31 0.84GDPpc 31570.8

517768.64 7885.00 67678.00

Banks 43.85 17.61 17.34 88.20Stocks 44.97 54.20 0.46 264.50

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 problems

derived from missing values in some variables. In our case, we apply a multiple

imputation method which consists of filling the missing values using an iterative

chained equations process (Schaffer, 1999; Royston, 2009). This method uses all the

13

available variables in the model to estimate unobserved data according to the particular

characteristics of each variable. In addition to this procedure, we apply another

imputation approach to complete information about our core variable, the availability of

financial education courses, based on the responses that school principals give to other

related questions. We enacted this procedure after detecting several cases where

principals indicate that financial courses were not available, but then they answer to

other related questions (e.g., how financial education courses are taught) indicating how

financial education is provided in the school. For items where this contradiction was

observed, we filled missing data using the responses given to related questions. If we

were unable to complete missing values using this procedure, we followed a list wise

deletion method. This led to a slight reduction in the size of the original dataset6.

4. Methodology

Given the hierarchical nature of data (students nested into schools, schools nested

into countries), we adopt a multilevel (or hierarchical) regression approach (Goldstein,

1995). This model allows us to avoid potential problems of bias in the estimations

derived from classic methods, such as OLS regression, due to the existence of

correlation between the values of the school variables of pupils from the same school

(Hox, 2002). In particular, we adopt a three-level approach in order to account not only

for divergences among schools, but also across countries.

Therefore, we assume that there are njk students nested within each of j = 1,…, Jk

schools, in turn nested within each of k = 1,…, K countries. At level 1, the outcome Yijk

for case i within level-2 unit j and level-3 unit k is represented with the following

expression:

Y ijk=π0 jk+∑p=1

P

π pjk a pjk+eijk (1)

The pjk are level-1 coefficients, with the corresponding level-1 predictors; e ijk is

the level-1 random effect, with the assumption that e ijk≈N (0 , σ2 ) . At level 2, the pjk

6 The original dataset was reduced by only 1,253 observations, which is equivalent to less than 5%.

14

coefficients at level 1 are treated as outcomes to be predicted. Thus, we have the

following expression:

π pjk=β p 0 k+∑q=1

Q p

β pqk X qjk+r pjk(2)

The pqk are level-2 coefficients, the Xqjk level-2 predictors and rpjk is the level-2

random effect. Taken as a vector, the r's are assumed to have a multivariate normal

distribution with a mean vector of 0 and a covariance matrix with maximum dimension

(P + 1) x (P + 1). At level 3, the pqk coefficients at level 2 are treated as outcomes to be

predicted:

β pqk=β pq 0+∑s=1

SPQ

γ pqs W sk+u pqk (3)

The pqs are level-3 coefficients, the Wsk are level-2 predictors, and upqk is the

level-3 random effect. Taken as a vector, the u's are assumed to have a multivariate

normal distribution with a mean vector of 0 and a covariance matrix with maximum

dimension∑p=0

P(QP+1 )x∑p=0

Q(QP+1) .

Throughout the following empirical analysis, we make the appropriate adjustment

to the estimated standard errors (bootstrapping standard errors by cluster)7. Likewise,

we have also applied the sampling weights included in PISA to correct for non-response

bias, while also scaling the sample up to the size of the national population (See

Rutkowski et al., 2010 for details).

5. Results

7 Estimates are bootstrapped by cluster (schools) using 50 replications to calculate approximate standard errors (see OECD, 2013 for details).

15

This section contains the main results obtained by estimating the multilevel model

explained above to examine the determinants of test scores in financial literacy. For that

purpose we have employed the HLM 7 software (Raudenbush et al., 2011), because it

allows us to estimate the parameters using the five available plausible values and

computing the average sampling variance correctly (Willms and Smith, 2005). First of

all, we have estimated regressions including all the covariates at student and school

level and the variable representing the availability of financial education in the school,

since this variable constitutes our main focus of interest. According to the results

reported in Table 3, where we present estimations without and with country fixed

effects (Models 1a and Model 1b, respectively), we find a positive association between

this variable and students´ financial literacy, although the value of the parameter is

relatively small8.

TABLE 3. HLM estimates (I)

Model 1a Model 1b

Model 2a

Model 2b

Gender -8.446*** -7.893***

-8.406***

-7.865***

(1.147) (1.094) (1.155) (1.101)AGE 14.81*** 17.53**

*14.69***

17.43***

(2.024) (1.959) (2.035) (1.967)Immigrant -15.04*** -

14.05***-

15.36***-

14.46***(3.026) (3.027) (3.053) (3.049)

Preprim_no -22.06*** -19.49***

-22.35***

-19.53***

(2.431) (2.374) (2.435) (2.374)Motheredu -1.728 -0.163 -1.708 -0.0648

(1.453) (1.387) (1.476) (1.405)Fatheredu -0.268 -0.626 -0.243 -0.531

(1.439) (1.435) (1.447) (1.441)Books25 -36.53*** -

34.34***-

36.61***-

34.47***(1.411) (1.358) (1.428) (1.374)

Books200 21.07*** 21.26***

21.03***

21.23***

(1.558) (1.486) (1.560) (1.490)Private 12.59*** 4.654* 12.91**

*4.826*

(3.090) (2.825) (3.108) (2.837)Rural -4.833*** -1.406 - -1.679

8 We focus on the estimation of the model with fixed effects, since if we disregard the nesting of observations within countries we would be ignoring the fact that individuals within the same country share unobserved characteristics (See Bryan and Jenkins, 2013 for details).

16

5.029***(1.407) (1.322) (1.416) (1.333)

ESCS mean 51.86*** 50.04***

51.68***

49.97***

(1.063) (1.188) (1.070) (1.202)FE available 9.932*** 3.059** 9.809**

*3.221**

(1.339) (1.283) (1.470) (1.408)FE

compulsory-1.127 1.164

(1.465) (1.391)Constant 264.1*** 238.8**

*266.3***

240.6***

(32.10) (32.39) (32.26) (32.50)Observations 24,889 24,889 24,889 24,889Country FE NO YES NO YESStandard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

While our focus is clearly on the influence of financial education availability,

other parameters in the estimation are also noteworthy. Note that the majority of

individual variables are significantly associated with the dependent variable.

Specifically, test scores in financial literacy are clearly better for boys, older students,

native pupils and those who attended pre-primary school. Likewise, we find a strong

relationship with the number of books at home. With regard to the covariates at school

level, we find that there is a positive and significant relationship for private schools and

the socioeconomic characteristics of schoolmates.

Apart from exploring the influence of having financial education available in the

school, we are also interested in examining how financial education is taught at schools.

First, we incorporate the variable FE compulsory to our model, thus we can test whether

the configuration of financial education as mandatory might affect the level of

knowledge acquired by students in this competence. The values of the estimated

parameters associated with this variable indicate that its influence is not significant,

independently of considering (or not) country fixed effects (Models 2a and 2b in Table

3). This result contradicts previous evidence about mandated financial education

courses in the specific context of the United States (Tennyson and Nguyen, 2001).

Second, we include in the model different strategies to teach financial education

concepts, i.e. using a cross-curricular approach or as a separate subject. The results of

the estimated parameters reported in Table 4 for alternative models 3 and 3b (without

17

and with country fixed) suggest that having FE available in the school is more relevant

than the specific strategy applied, since we found no statistically significant relationship

for both variables. Nevertheless, when we take into account who provides financial

education training in the school, it seems that courses delivered by people coming from

private institutions and non-governmental organizations are more effective than training

conducted by school teachers, considering the estimated coefficients shown in Table 4

(Models 4a and 4b).

TABLE 4. HLM estimates (II)

Model 3a Model 3b

Model 4a

Model 4b

Gender -8.403*** -7.855***

-8.502***

-7.901***

(1.155) (1.102) (1.155) (1.102)AGE 14.73*** 17.43**

*14.57***

17.33***

(2.036) (1.967) (2.037) (1.969)Immigrant -15.36*** -

14.43***-

15.66***-

14.45***(3.054) (3.055) (3.051) (3.051)

Preprim_no -22.36*** -19.54***

-22.16***

-19.59***

(2.434) (2.373) (2.432) (2.372)Motheredu -1.697 -0.0654 -1.669 -0.0523

(1.475) (1.406) (1.474) (1.406)Fatheredu -0.203 -0.534 -0.0396 -0.590

(1.448) (1.441) (1.449) (1.444)Books25 -36.58*** -

34.46***-

36.58***-

34.46***(1.430) (1.375) (1.428) (1.373)

Books200 21.01*** 21.24***

21.05***

21.24***

(1.559) (1.489) (1.559) (1.489)Private 13.00*** 4.794* 13.01**

*4.925*

(3.119) (2.840) (3.127) (2.841)Rural -5.089*** -1.658 -

5.146***-1.578

(1.422) (1.334) (1.423) (1.336)ESCS mean 51.65*** 49.95**

*51.25***

49.91***

(1.071) (1.203) (1.070) (1.202)FE available 10.27*** 3.324** 8.863**

*1.813

(1.626) (1.228) (1.667) (1.550)FE

compulsory-0.858 1.171 -0.886 1.122

(1.490) (1.420) (1.492) (1.420)

18

FE separate -1.306 0.431 -1.903 -0.116(1.576) (1.496) (1.588) (1.500)

FE cross -0.159 -0.885 -1.020 -1.382(1.361) (1.353) (1.370) (1.354)

FE priv 10.15***

3.292*

(2.091) (1.989)FE pub -5.136* -2.306

(3.068) (2.842)FE ONG 2.876 5.389**

(2.352) (2.162)Constant 265.7*** 240.8**

*268.4***

242.3***

(32.27) (32.48) (32.29) (32.51)Observations 24,889 24,889 24,889 24,889Country FE NO YES NO YESStandard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

The final step of our empirical analysis consists of including several additional

variables at country level in a three-level hierarchical model. Table 5 reports these

estimates including, in a first step, only variables representing students’ experience with

money matters (Model 5) and, subsequently, adding also economic indicators (Model

6). In order to save space, we do not report the estimated parameters for the covariates

at student and school level because they were very similar to the values displayed in

Tables 3 and 4.

TABLE 5. HLM estimates including variables at country level

Model 5 Model 6Covariates at student level X XCovariates at school level X XFE available 3.107** 3.202*

(1.510) (1.573)FE compulsory 1.108 0.741

(1.420) (1.494)FE separate -0.0915 0.248

(1.500) (1.578)FE cross -1.408 -1.687

(1.354) (1.517)FE priv 3.294* 2.837

(1.990) (2.189)FE pub -2.288 -1.094

(2.842) (3.117)FE ONG 5.383** 6.991***

(2.162) (2.389)account 27.96 -19.97

19

(37.27) (32.48)gifts 48.95 37.65

(103.9) (32.92)Allowance 102.3** 131.2**

(47.74) (64.56)GDPpc 0.000552

(0.000547)Banks 0.0206

(0.307)Stocks -0.134

(0.108)Constant 129.2 109.2*

(81.22) (61.16)Observations 24,889 24,889Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

Among all these variables, only the one representing the percentage of student

receiving money from an allowance is significantly (and positively) associated with

financial literacy test scores. A potential explanation for this widespread absence of

significant variables among country indicators is the small number of countries

available in our dataset (18). In this sense, some authors suggest that at least 25

countries are required in order to derive reliable estimates in three-level models (Bryan

and Jenkins, 2013; Stegmueller, 2013).

6. Concluding remarks

This paper provides empirical evidence about the influence of financial education

training at schools as a mechanism for improving young students’ knowledge of

financial issues. For that purpose, we have exploited the information provided by the

financial literacy test conducted by students from 18 countries participating in PISA

2012. This was the first initiative that offered such comparable data in an international

framework together with a rich dataset about the organization of financial education at

schools.

Our empirical findings suggest that the availability of specific financial training is

positive and significantly related with students´ knowledge about financial contents and

processes, irrespective of how this training is implemented. This result is robust to the

consideration of country fixed effects, although the influence is clearly smaller when we

account for the potential presence of significant differences among countries. This

indicates that there are national and cultural differences that policymakers should

20

consider when developing financial literacy assessment tools for their respective

countries, as pointed out by Nicolini and Chatterjee (2013).

In addition, our results suggest that students receiving courses taught by

specialists from private institutions and non-governmental organizations obtain better

results than students who receive training provided by the teachers of their school. Our

interpretation of this result is that those specialists place a greater emphasis on the study

of the specific financial concepts included in the PISA assessments than teachers, who

in some cases may lack of an appropriate knowledge about some specific financial

instruments. This is common due to the fact that the establishment of financial

education courses neither requires nor promotes teacher training in the field (Way and

Holden, 2009). Therefore, the training of teachers to develop and implement financial

literacy programs appears to be a key aspect in order to enhance the effectiveness of

financial education, since there is empirical evidence demonstrating that they exert a

decisive influence on student achievement (Hattie, 2009). In fact, some previous studies

have identified some successful initiatives regarding teaching financial education in

different countries (e.g. Koh, 2016; O´Neill and Hensley, 2016).

Despite these interesting results, there remain some issues that require further

research, such as determining the number of teaching hours required in order to obtain

meaningful results, examining the different effects of financial education courses

depending on whether they are taught during primary or secondary education or

considering potential displacements caused by the incorporation of financial education

into the school curriculum. Unfortunately, the PISA dataset does not include enough

reliable information about these aspects, although the growing development of

initiatives and pilot programs involving financial education in multiple countries should

allow researchers to make significant progress in developing empirical evidence about

these issues in the near future.

References

Appleyard, L., & Rowlingson, K. (2013). Children and financial education: challenges for developing financial capability in the classroom. Social Policy and Society, 12(4), 507-520.

21

Atkinson, A., & Messy, F. A. (2013). Promoting financial inclusion through financial education: OECD/INFE Evidence, Policies and Practice, OECD Working Papers on Finance, Insurance and Private Pensions, No. 34, OECD Publishing.

Batty, M., Collins, J. M. & Odders‐White, E. (2015). Experimental Evidence on the Effects of Financial Education on Elementary School Students’ Knowledge, Behavior, and Attitudes, Journal of Consumer Affairs, 49(1), 69-96.

Berry, J., Karlan, D., & Pradhan, M. (2018). The impact of financial education for youth in Ghana. World Development, 102, 71-89.

Bernheim, B. D., Garrett, D. M., & Maki, D. M. (2001). Education and saving:: The long-term effects of high school financial curriculum mandates. Journal of Public Economics, 80(3), 435-465.

Borden, L. M., Lee, S. A., Serido, J. & Collins, D. (2008). Changing college students’ financial knowledge, attitudes, and behavior through seminar participation. Journal of Family and Economic Issues, 29(1), 23-40.

Bryan, M. L. & Jenkins, S. P. (2013). Regression analysis of country effects using multilevel data: a cautionary tale, Institute for Social and Economic Research, Working Paper No. 2013-14.

Cole, S., Sampson, T. & Zia, B. (2011). Prices or Knowledge? What Drives Demand for Financial Services in Emerging Markets. Journal of Finance, 66(6), 1933-1967.

Cole, S., Paulson, A., & Shastry, G. K. (2016). High school curriculum and financial outcomes: The impact of mandated personal finance and mathematics courses. Journal of Human Resources, 51(3), 656-698.

Fernandes, D., Lynch Jr, J. G., & Netemeyer, R. G. (2014). Financial literacy, financial education, and downstream financial behaviors. Management Science, 60(8), 1861-1883.

Ganzeboom, H. B., De Graaf, P. M., & Treiman, D. J. (1992). A standard international socio-economic index of occupational status. Social Science Research, 21(1), 1-56.

Goldstein H (1995). Multilevel statistical models (2nd edition). New York: Wiley Publishers.

Grifoni, A. and Messy, F.A. (2012). Current Status of National Strategies for Financial Education: A Comparative Analysis and Relevant Practices, OECD Working Papers on Finance, Insurance and Private Pensions, No. 16, OECD Publishing.

Hanushek, E.A. & Woessmann, L. (2014). Institutional Structures of the Education Systemand Student Achievement: A Review of Cross-country Economic Research, in Strietholt, R., Bos, W., Gustafsson, J.E. & Rosen, M. (eds.). Educational

22

Policy Evaluation Through International Comparative Assessments, Waxmann Verlag, New York, 145-175.

Hattie, J. (2009). Visible Learning. A Synthesis of Over 800 Meta-Analyses Relating to Achievement. Oxon: Routledge.

Hilgert, M. A., Hogarth, J. M., & Beverly, S. G. (2003). Household financial management: The connection between knowledge and behavior. Federal Reserve Bulletin, 106, 309-322.

Hox, J. (2002). Multilevel Analysis. Techniques and Applications. Mahwah, Lawrence Erlbaum Associates.

Jappelli, T. (2010). Economic literacy: An international comparison. The Economic Journal, 120(548), 429-451.

Koh, N. K. (2016). Approaches to Teaching Financial Literacy: Evidence-Based Practices in Singapore Schools. In Aprea, C., Wuttke, E., Breuer, K., Koh, N.K., Davies, P.D., Greimel-Fuhrmann, B. & Lopus, J.S. (eds.). International Handbook of Financial Literacy, Springer, 499-513.

Kozup, J. & Hogarth, J.M. (2008). Financial Literacy, Public Policy, and Consumers’ Self‐Protection—More Questions, Fewer Answers. Journal of Consumer Affairs, 42(2), 127-136.

Lusardi, A. & Mitchell, O.S. (2011). Financial Literacy and Planning: Implications for Retirement Wellbeing, in A. Lusardi and O. S. Mitchell (eds.), Financial Literacy: Implications for Retirement Security and the Financial Marketplace, Oxford University Press.

Lusardi, A. & Mitchell, O.S. (2014). The Economic Importance of Financial Literacy: Theory and Evidence. Journal of Economic Literature, 52(1), 5-44.

Lusardi, A., Mitchell, O.S. & Curto, V. (2010). Financial Literacy Among the Young. Journal of Consumer Affairs, 44(2), 358-380.

Mandell, L. (2008). Financial literacy of high school students, in Xao (ed.). Handbook of consumer finance research, Springer, New York, 163-183.

McCormick, M.H. (2009). The Effectiveness of Youth Financial Education: A Review of the Literature. Journal of Financial Counseling and Planning, 20(1), 1-14.

Mislevy, R. J., Beaton, A. E., Kaplan, B., & Sheehan, K. M. (1992). Estimating population characteristics from sparse matrix samples of item responses. Journal of Educational Measurement, 29(2), 133-161.

Nicolini, G., Cude, B. J., & Chatterjee, S. (2013). Financial literacy: A comparative study across four countries. International Journal of Consumer Studies, 37(6), 689-705.

23

OCDE (2012). High-Level Principles on National Strategies for Financial Education, OECD Publishing.

OECD (2013). Advancing National Strategies for Financial Education, OECD Publishing.

OECD (2014). PISA 2012 Results: Students and Money Financial Literacy Skills for the 21st Century (Volume VI), OECD Publishing.

OECD (2015). National Strategies for Financial Education: OCDE/INFE Policy Handbook, OECD Publishing.

O’Neill, B. M. & Hensley, B.J. (2016). Building the Financial Education Capacity of Teachers: Does It Make a Difference?. In Aprea, C., Wuttke, E., Breuer, K., Koh, N.K., Davies, P.D., Greimel-Fuhrmann, B. & Lopus, J.S. (eds.). International Handbook of Financial Literacy, Springer, 639-654.

Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., Congdon, R. & du Toit, M. (2011). HLM statistical software: Version 7. Computer software.

Romagnoli, A. & Trifilidis, M. (2013). Does Financial Education at School Work? Evidence from Italy. Bank of Italy Occasional Paper, 155.

Royston, P. (2009). Multiple Imputation of Missing Values: Further Update of Ice, with an Emphasis on Categorical Variables. Stata Journal, 9(3): 466–477.

Rutkowski, L., González, E., Joncas, M. & von Davier, M. (2010). International Large-Scale Assessment Data: Issues in Secondary Analysis and Reporting. Educational Researcher, 39 (2), 142-151.

Schafer, J. L. (1999). Multiple Imputation: A Primer. Statistical Methods in Medical Research, 8(1): 3-15.

Shim, S., Barber, B. L., Card, N. A., Xiao, J. J., & Serido, J. (2010). Financial socialization of first-year college students: The roles of parents, work, and education, Journal of Youth and Adolescence, 39 (12), 1457-1470.

Stegmueller, D. (2013). How many countries for multilevel modeling? A comparison of frequentist and Bayesian approaches. American Journal of Political Science, 57(3), 748-761.

Tennyson, S. & Nguyen, C. (2001). State curriculum mandates and student knowledge of personal finance. Journal of Consumer Affairs, 35(2), 241-262.

Varcoe, K.P., Martin, A., Devitto, Z. & Go, C. (2005). Using a Financial Education Curriculum for Teens. Financial Counseling and Planning, 16(1), 63–71.

Von Davier, M. & Sinharay, S. (2013). Analytics in international large-scale assessments: Item response theory and population models, in Rutkowski, L., Von Davier, M. & Rutkowski, D. (eds.). Handbook of international large-scale assessment:

24

Background, technical issues, and methods of data analysis, 155-174, CRS Press, London.

Walstad, W.B. (2013). Economic Understanding in US High School Courses, American Economic Review, 103(3), 659-663.

Walstad, W.B., Rebeck, K. & MacDonald, R.A. (2010). The Effects of Financial Education on the Financial Knowledge of High School Students. Journal of Consumer Affairs 44(2), 336-357.

Way, W. L. & Holden, K. (2009). Outstanding AFCPE® Conference Paper Teachers’ Background and Capacity to Teach Personal Finance: Results of a National Study. Journal of Financial Counseling and Planning, 20(2): 64-78.

Willis, L. E. (2008). Evidence and Ideology in Assessing the Effectiveness of Financial Literacy Education. Loyola—LA Legal Studies Paper No. 2008-6.

Willms, J.D. y Smith, T. (2005): A Manual for Conducting Analyses with Data from TIMMS and PISA, Report prepared for UNESCO Institute for Statistics.

Wu, M. (2005). The role of plausible values in large-scale surveys. Studies in Educational Evaluation, 31(2-3), 114-128.

Xu, L. & Zia, B. (2012). Financial Literacy Around the World: An Overview of the Evidence with Practical Suggestions for the Way Forward. World Bank Policy Research Working Paper, 6107.

25

APPENDIX

Table A1. Situation of FE in countries participating in PISA 2012 financial literacy

AustraliaFE is guided by the National Financial Literacy Strategy, a collaborative multi-agency strategy coordinated by the Australian

Securities and Investments Commission. The introduction of financial education in Australian schools started in 2005. Financial literacy topics were integrated into school subjects, including mathematics, English and science.

Belgium (Flemish

Community)

Learning outcomes for secondary schools (that came into effect in 2010-11) cover typical FE topics, such as budgeting, alongside economics topics, such as labour, goods and services, welfare and poverty. They are mandatory in all secondary schools while schools can decide how and in which subjects these cross-curricular competencies should be integrated.

Colombia*

Law 1450 of 2011 (Plan Nacional de Desarrollo 2010-2014), mandated the Ministry of Education to define the set of basic financial and economic abilities that the Colombian school curricula should include. This mandate is being implemented by Decree 457 of 2014. Through this, the government created a multiagency system to coordinate public and private financial education initiatives within the framework of a national strategy for economic and financial education. In the meanwhile, several pilot programs have been implemented by various institutions such as “Schools for entrepreneurship”, launched by BBVA Colombia in 2012 or “Financial Education”, promoted by Bancolombia in 2013.

Czech Republic

The introduction of financial education in schools started in 2007 with the definition of the financial literacy standards established by the Ministry of Finance. Subsequently, FE has been made mandatory at upper secondary school level since 2009 and at primary and lower secondary school level since 2013. FE is usually incorporated into various subjects such as Social Studies, Civics, Citizenship Education, Mathematics, or is carried out in some form of project-based learning within the curriculum of these courses.

EstoniaA National Strategy for Financial Literacy started in 2010 and a seven-year national program was launched in 2013. In primary

and lower secondary school, monetary and finance-related topics are incorporated in social studies, home economics and mathematics. In general upper-scondary schools financial issues are taught in social studies and human studies under compulsory or elective courses.

France At the beginning of the 2010 school year, the Ministry of Education reformed the way economics was taught in secondary schools, making the subject compulsory for 15/16 year old students of the general and scientific tracks of French high-schools.

Israel*

Teaching FE in schools started in 2010 with a gradually expanding pilot program for 10 th-graders (15- and 16-year-old students). The program deals with a variety of issues, including budget management, the role and importance of banks, loans, savings, investments, and consumer rights. The Ministry of Education is also considering integrating some aspects of financial education in other existing subjects such as mathematics, and Homeland, Society and Citizenship.

Italy*

The Bank of Italy and the Italian Ministry of Education implemented an experimental program to incorporate FE into school curricula through a cross-curricular approach starting in the school year 2008/2009. The program was piloted in selected primary and secondary schools in the first year and then has been extended to a larger number of schools in the following years. Teachers who decide to participate receive training and pedagogical resources from Bank of Italy´s officials.

26

LatviaCross-curricular; financial education is integrated into various subjects of the General Curriculum in primary schools. The basics

of economics are an optional subject at general secondary schools. Less than a half are learning economics, finance and business basics at school.

New Zealand

Financial literacy was included in the New Zealand Curriculum in 2009 following a cross-curricular approach. FE provides a context for linking learning areas, such as social sciences, mathematics and statistics, English, business studies, health and technology. Likewise, it also provides a relevant context for strengthening literacy and numeracy skills and understandings, developing the key competencies, and exploring values. Given the self-governing structure of the school system in the country, the school principals have to decide how and to what extent to integrate FE courses.

Poland In secondary schools, the subject “Introduction to Management of Firms” is compulsory during three courses, 2 hours per week.

Shanghai-China

Some FE topics have been integrated into the existing national curriculum since the 1970s including knowledge about the (socialist) economic system, about budgeting and money management, basic financial services and risks and benefits of financial products. Schools have some autonomy in teaching FE with respect to the national curriculum, thus they can decide to teach additional ones. In the Pudong New Area of Shanghai-China, a regional curriculum called Finance and Money has been delivered since 2001 in 116 primary and lower secondary schools.

Spain*

In the first three courses of Compulsory Secondary Education there is not any specific course related to FE. Some financial contents may be treated under a cross-sectional approach, in courses such as Social Sciences, Geography or History, but from a more historical o geographical point of view. A pilot program was put into place in for the first time during the academic year 2010/2011 for eleven graders (15 years old). The program involved 3,000 students and 70 teachers from 32 schools. The new education law (Organic Law 872013) introduced some contents of financial education as part of Social Sciences area of knowledge.

Russian Federation

In 2011 the Russian government launched a comprehensive five-year nationwide project to support FE and consumer protection. The project targets low-income and vulnerable social groups as well as young people, including school and university students. As part of this project, Russia is preparing its National Strategy for FE to provide a vision and a common framework for the further development of financial literacy policies and programs in Russia.

Slovak Republic

Independently of the track in secondary education, all schools can include in their curriculum optional subjects (maximum load 30%), such as management and entrepreneurship. In all the tracks in secondary education there is an available non compulsory course related to supporting entrepreneurship and management for young people.

Slovenia Although there is not a compulsory course related to Economics, all schools can offer between 30 and 50% of their curriculum related to financial contents.

USA Most states have integrated compulsory FE in their curriculum. There are differences across states in whether schools are mandated to offer courses in economics and/or personal finance.

* Countries applying pilot programs to introduce FE in the curriculum.

27