Tullio Jappelli

Financial Literacy
An International Comparison

Discussion Paper 09/2010-064

 Financial Literacy: An International Comparison1

Tullio Jappelli

Abstract

This paper uses international panel data on 55 countries from 1995 to 2008, merging
indicators of financial literacy with a large set of macroeconomic and institutional variables.
Results show that there is substantial heterogeneity of financial and economic competence
across countries, and that human capital indicators (PISA test scores and college attendance)
are positively correlated with financial literacy. Furthermore, inhabitants of countries with
more generous social security systems are generally less literate, lending support to the
hypothesis that the incentives to acquire financial literacy are related to the amount of
resources available for private accumulation.

JEL Classification: E2, D8, G1

1
Department of Economics, University of Naples Federico II, Via Cinzia 45, 80146 Napoli, Italy (email:
[email protected]). I am grateful for helpful comments from an anonymous referee, Dimitris Christelis,
Elsa Fornero, Luigi Guiso, Michalis Haliassos and participants at the SAVE Conference on Economic and
Psychological Aspects of Households’ Saving Behaviour: Old-age Provision, Financial Literacy and the
Financial Crisis (Mannheim, June 29-30, 2009), and to the Italian Ministry of Education for financial support.
Households have interacted with financial markets in the last 20 years, much more so than in
the past, and also have been exposed to increased financial risk as a consequence of financial
market liberalization and policy reforms aimed at promoting retirement savings through
private pension funds and individual retirement accounts. Although to different extents, these
trends are affecting all countries and all dimensions of economic transactions, from payment
needs, as witnessed by the growth of the credit card industry, portfolio investments, and
borrowing in the mortgage and consumer credit markets. Many of these activities, however,
are entered into by uninformed individuals.
The recent crisis has amplified the risks that people face when they lack the financial
sophistication required to absorb financial shocks. Other things equal, differences in financial
literacy create the potential for significant distributional consequences of a financial crisis,
because unsophisticated investors are more exposed to financial market fluctuations than
investors that are able to manage and diversify risks. The risks are especially severe for
individuals whose pensions depend on stock market developments and for the elderly whose
assets are based on decisions made in the past and whose margins for adjustment are smaller.
Some recent financial economic studies have made considerable progress in measuring
financial literacy. Economists have tended to measure literacy through a rough self-
assessment of respondents’ financial sophistication; however, there is a second generation of
studies based on detailed and more reliable questions on finance. These surveys have
established convincingly that a large proportion of the adult population knows very little
about finance and that many individuals are unfamiliar with even the most basic economic
concepts, such as risk diversification, inflation, interest compounding, and mortgage and other
debt instruments (Lusardi, 2008). There is also substantial evidence that financial literacy
differs widely across households and tends to be rather limited in the less educated, poorer
demographic groups. What makes this evidence even more worrying is that many people are
not even aware of their ignorance.
Although considerable progress has been made on measuring financial literacy, its
determinants, the effectiveness of financial education and the consequences of financial
literacy for households’ financial decisions are not well understood. This paper adopts an
international comparative perspective, which involves merging indicators of financial literacy
with a wide set of macroeconomic and institutional variables. The purpose of the analysis is to

1
study the factors that are more likely to explain international differences in literacy using
cross-country and time-variable indicators.
To study cross-country differences in financial literacy, the ideal dataset would include
an assessment of financial knowledge and skills, such as is provided by OECD-PISA for 15-
year olds for math or science. In the absence of such detailed (and expensive) data, we rely on
the IMD World Competitiveness Yearbook (WCY), which compiles summary indicators of
financial literacy for 1995 to 2008. The indicators are computed based on interviews with
senior business leaders in 55 countries; the WCY aggregates their responses by country to
provide an overall score for financial literacy. The data show that financial literacy varies
substantially across countries, from the lowest scores in some Latin American and former
socialist countries to high values in the Scandinavian countries and East Asia.
Regression analysis indicates that PISA test scores and educational achievement are
positively associated with financial literacy. On the other hand, countries with high mandated
savings in the form of social security contributions and resulting more limited resources for
private wealth accumulation, show lower levels of financial literacy. The results are robust to
the presence of other macroeconomic and institutional variables and country-fixed effects.
These findings are consistent with standard human capital models where households’
knowledge depends on cognitive abilities and the incentives to acquire information, which, in
turn, are related directly to the size of financial markets (Delevande et al., 2008).
The paper is organized as follows. Sections 1 and 2 respectively, discuss the importance
of financial literacy and review the existing international evidence. Section 3 describes the
data used in the paper, and Section 4 reports the cross-section and panel regressions. Section 5
concludes.

1. Why is Finacial Literacy Important?

Financial literacy is increasingly important for households’ decisions about on how to invest
wealth and how much to borrow in financial markets. Literacy also has far-reaching
consequences for the stability of the overall economy.

2
1.1. The Asset Side

On the asset side, financial literacy is important because financial products have become
extremely complex. Even for simple products, such as savings accounts and government
bonds, there are usually several options and several different contracts, which makes choice
more difficult. Furthermore, due to financial market innovations and deregulation, since the
end of the 1980s, the number of financial products that is available has increased considerably,
with many new options in terms of investment in equities and bonds. In many countries,
households are more exposed to financial risks as a consequence of greater stock market
participation and policy shifts aimed at promoting retirement/pension arrangements through
individual retirement accounts and private pension funds.
Several empirical studies have found that lack of financial literacy is associated with
poor risk diversification, inefficient portfolio allocations and low levels of savings. Banks and
Oldfield (2007) look at numerical ability and other dimensions of cognitive function in a
sample of older adults in England (the English Longitudinal Study of Ageing ) and find that
numeracy levels are strongly correlated with measures of retirement saving and investment
portfolios, understanding of pension arrangements, and perceived financial security. In
subsequent work, Banks et al. (2009) look at the extent to which differences in numeracy and
broader cognitive ability predict subsequent trajectories for key economic outcomes such as
wealth, retirement income and key dimensions of retirement expectations.
Christelis et al. (2010) study the relation between cognitive abilities and stockholding
based on the Survey of Health, Assets, Retirement, and Expectations (SHARE), and find that
the propensity to invest directly and indirectly in stocks (through mutual funds and retirement
accounts) is strongly associated with mathematical ability, verbal fluency, and recall skills. In
a related paper, Ardle et al. (2009) find that numeracy, measured through the accuracy of
responses to three simple mathematical questions, is a strong predictor of total wealth,
financial wealth, and the fraction of wealth held in stocks. Smith et al. (2009) extend the
evidence studying the relationship between household wealth and the cognitive status of both
spouses.
Alessie et al. (2008) study the relation between financial sophistication and wealth
relying on specific measures of financial literacy available in a special module of the Dutch
DNB Household Survey. The module contains basic questions on the ability to perform

3
simple calculations and to understand compound interest, inflation, and money illusion, and
more advanced questions on stock market functioning, characteristics of stocks, mutual funds
and bonds, equity premiums, and the benefits of diversification. They find that financial
sophistication is associated with higher wealth, higher probability to invest in the stock
market and higher propensity to plan for retirement.2
Guiso and Jappelli (2008) relate financial literacy to portfolio diversification by Italian
investors. They use the 2007 Unicredit Customer Survey (UCS), which has detailed indicators
of investors’ portfolio choice, financial literacy and demographic characteristics. Financial
literacy is strongly correlated to the degree of portfolio diversification, even controlling for
other socioeconomic characteristics and proxies for risk aversion. The authors compare
objective measures of financial literacy obtained through specific questions on finance, with
investors’ self-assessment of financial knowledge, and find only a weak relation between the
two measures: 50 percent of those with poor financial literacy report above average
confidence on financial matters, while 15 percent of investors who score well on literacy
confess to knowing little about finance.
In the context of developing countries, Cole et al. (2009) analyze the relation between
financial literacy and participation in formal financial markets. Using survey data on India
and Indonesia, they show that financial literacy is a powerful predictor of demand for
financial services. However, in a field experiment where randomly selected unbanked
households were offered finance education, they find that, with the exception of completely
uneducated and financially illiterate households, the program had no effect on the likelihood
of opening a bank savings account. Hastings and Tejeda-Ashton (2008) use survey responses
and the results of an experiment involving participants in Mexico’s privatized social security
system, to examine how financial literacy impacts on workers’ choice behavior and how
simplifying the information related to management fees may increase measures of price
elasticity sensitivity among the financially illiterate. They find that how information is
presented to workers can have a substantial impact on the optimal fees that firms can charge
in the marketplace.
One of the limitations of all the studies cited is that the incentive to become financially
literate depends on the level of wealth and the portfolio allocation, which give rise to an

2
The study also addresses the endogeneity between financial literacy and wealth. To account for the fact that
wealth, portfolio management and planning activities independently exert an effect on financial literacy, the
study uses economic education as an instrument for financial information.

4
endogeneity bias. Two papers that address this crucial issue provide conflicting results.
Christiansen et al. (2008) use a large register-based panel data set containing detailed
information on Danish investors’ educational attainment as well as financial and
socioeconomic variables. The authors show that stockholdings increases if individuals have
completed an economic education program and if an economist moves into the household. To
sort out the double causality between portfolio choice and the decision to become an
economist, Christiansen et al. (2008) use improved access to education due to a new
university as an instrument for economic education. The instrumental variable estimates
suggest that causation runs from economic education to stock market participation. The
endogeneity issue is also addressed by Cole and Kartini Shastri (2009) who show that
financial literacy education mandated by US state governments does not have an effect on
financial market participation. They show that participation rates among those who graduated
before it became compulsory (and therefore were not exposed to financial literacy education)
are identical to the rate for those graduates who were exposed to this program.

1.2. The Debt Side

On the debt side, borrowing in mortgage markets, ownership of credit cards, and
consumer credit have increased in almost all OECD countries. To evaluate the information
available on different loan possibilities, choose among different credit instruments, and
identify predatory lending necessitate “a minimum level of financial literacy and skills to
distinguish between products” (OECD, 2005, p. 65). The recent crisis shows that poor
financial literacy can affect not only the choice of individual investors and borrowers, but can
be an aggravating factor in a recession because household debt plays a central role in the
balance sheets of banks and other financial intermediaries. Using cross-country data on
household debt, and panel data on arrears, Jappelli et al. (2008) find that household
indebtedness is associated with increased ‘financial fragility’, as measured by the sensitivity
of household arrears and insolvencies to the amount of lending and to macroeconomic shocks.
Despite its importance and potentially damaging consequences, the debt side of
financial literacy has received less research attention than the asset side. Lusardi and Tufano
(2009) analyze a national sample of Americans with respect to their debt literacy, financial
experience, and judgment about the level of their indebtedness. They measure debt literacy
through a set of questions testing the respondents’ knowledge of fundamental concepts related

5
to debt, and find that there is illiteracy in all segments of the population, but especially
women and the elderly. The paper finds a strong relationship between debt literacy and both
financial experience and debt load, and finds also that individuals with lower levels of debt
literacy tend to transact in high-cost ways (incurring fees and using high cost borrowing).
This finding lends support to the claim that low levels of financial literacy have contributed
to debt buildup, which, in some countries (e.g. the US and Germany), has been accompanied
by an increased number of insolvencies and bankruptcies.

1.3. The Macro Side

Financial literacy also contributes to the good workings of markets and policies. First,
lack of financial literacy may create more favorable conditions for deceitful financial
practices and unfair competition in financial markets, and be a serious impediment to
effective financial intermediation. In contrast, as stressed at the 2006 Meeting of the G8
Finance Ministers, ‘well-informed and educated financial consumers lead to better financial
markets where rogue products are forced from the marketplace and confidence is raised’ (G8,
2006). When households are well informed, they can also discipline policy makers, so that
‘better-informed citizenry makes for better economic policy-making’ (Mishkin, 2008), a point
stressed by Bernanke, who argues that improving financial literacy is a way to restore
confidence in the economy: the Federal Reserve's mission of conducting monetary policy
and maintaining a stable financial system depends upon the participation and support of an
educated public. As the Fed pursues the monetary policy objectives that have been set out by
Congress (price stability, maximum employment, and moderate long-term interest rates) it is
essential that the public understands our objectives and our actions. Educating the public
about the reasoning behind our decisions helps build confidence in our economic system -
another critical factor in keeping our economy running smoothly (Bernanke, 2006).

2. International Evidence

Despite the importance of financial literacy for households’ decisions and the proper
functioning of financial markets, the evidence on the importance of literacy and the

6
effectiveness of financial education is focused primarily on the US.3 There are surveys of
other regions, but they are not comparable in either focus or method. A recent OECD report
lists surveys in 12 countries that provide one or more indicators of financial literacy (OECD,
2005). 4 These surveys rely on two approaches to measuring literacy. One is to test
respondents on their knowledge and understanding of financial terms and their ability to apply
financial concepts to particular situations: usually available for the US, Italy, Korea and the
Netherlands. The other approach is to ask respondents to self-assess their financial
understanding and ability to deal with financial matters: these indicators are available for the
UK, Japan, Australia and some other countries. Outside the OECD, two recent surveys elicit
measures of financial literacy in India and Indonesia (Cole et al., 2009).
Although these surveys differ in terms of respondents targeted, the approach to
measuring financial literacy, and the methodology, we can identify some common findings.
First, many countries exhibit a rather low level of literacy; second, financial literacy is
correlated with education (as measured by school or college attendance). 5 Also, where
comparisons are possible, respondents generally report knowing more about financial matters
than is actually the case. Finally, surveys show that financial literacy tends to be associated
with higher income and wealth. However, this should not be interpreted as a causal link
running from literacy to wealth, because the incentives to learn about finance are directly
related to the level of resources. Indeed, as mentioned in Section 1, sorting out the causality
between financial literacy and portfolio outcomes is a major challenge.
In principle, to enable cross-country comparison, a single questionnaire should be
administered to a random sample of the population in each country and the data integrated
with economic, demographic, and institutional variables. However, such an approach (like the
PISA test of educational achievement among 15-year olds) would require substantial

3
Lusardi (2009) surveys the empirical evidence and finds mixed support for the effectiveness of these programs,
partly because attendance at seminars is voluntary, and partly because it is difficult to disentangle the
consequences of an increase in financial education from peer and community effects in raising savings (Duflo
and Saez, 2003). Willis’s (2008) view of the potentially negative role of financial education stresses that for
some consumers financial education programs increase confidence without improving ability, thus leading to
worse decisions.
4
According to the OECD report, in only 5 countries (Australia, Japan, Korea, the US and the UK) is there
detailed information on methodology, results, questions asked, and target groups.
5
Guiso and Jappelli (2005) document that in the 1995 and 1998 Bank of Italy Surveys of Household Income and
Wealth (SHIW) a significant proportion of households was not aware even of the existence of many financial
instruments. The paper also explores the determinants of awareness, and finds that the probability that survey
respondents are aware of the existence of stocks, mutual funds, and investment accounts is positively correlated
with education, household resources, long-term banking relations, and proxies for social interaction.

7
resources and coordination efforts. Christelis et al. (2010) make an attempt in this direction by
analyzing indicators of cognitive abilities (including some related to financial literacy) in the
11 countries covered by SHARE. Their analysis of a sample of respondents aged 50 and over,
shows that cognitive abilities tend to be higher in Northern Europe, decline with age, and be
positively associated with a college education.6 However, there is wide variation of cognitive
abilities within each country, age, and education group. Given the small number of countries
they cover, explaining cross-country differences using SHARE data is clearly not feasible.
In the absence of international microeconomic data, the present paper relies on a survey
of business leaders, which data have so far not been used to study financial literacy or its
determinants. There are two advantages to using this dataset. First, it provides consistent
international comparison of financial literacy in 55 countries over the 1995-2008 period,
allowing to relate literacy to macroeconomic and institutional variables within a panel
framework. Second, experts with an international dimension are less subject to the fact that
individuals in different countries might use different response scales. The limitations to these
data are that they are only available in aggregate form, which does not allow analysis of
specific socioeconomic groups.

3. Indicator of Economic Literacy

Since 1995, the IMD WCY has published an indicator of financial literacy. The indicator is
computed from a survey of senior business leaders who represent a cross-section of the
business community in the countries examined, and merged with data drawn from
international organizations. The sample distribution reflects a breakdown of industry by
sectors (manufacturing, services, and primary), and the sample size is proportional to each
country’s GDP. The survey questions are targeted to top and middle managers, nationals or
expatriates, located in local and foreign enterprises in the country in question, who generally
have an international experience and outlook. The surveys are administered in January for
completion and return by March of the same year. The overall size of the survey is about
4,000 business leaders, and 55 countries.

6
The cross-country variability of cognitive skills among people with similar levels of education is not a unique
feature of SHARE data. For instance, the PISA survey finds a significant North-South gradient in mathematics,
science and verbal skills among young (under 15 years) Europeans at the same school grade

8
 The financial literacy question asks respondents to evaluate, on a 0-10 scale, the
sentence: ‘Economic literacy among the population is generally high’. Dropping missing
values for some countries, we constructed a panel of 55 countries for 1995 to 2008: 14 in Asia,
7 in Latin America, 15 in the EU, 12 former socialist countries, and 7 other countries (South
Africa, US, New Zealand, Norway, Canada, Switzerland, Australia). The survey also includes
an ‘Education in finance’ question (available only from 1999 to 2008), which asks for an
evaluation of the statement: ‘Education in finance does meet the needs of the business
economy’. Clearly the financial literacy indicator is more closely related to the level of
literacy of the population at large. However, the correlation coefficient for the two indicators
is 0.81, and the main results of the paper are unaffected if we use the indicator for education
in finance.7

INSERT FIGURE 1

Figure 1 plots the distribution of financial literacy for the world, and highlights large
international differences from a score of less than 3 for South Africa, Venezuela, Peru,
Mexico, and Croatia, to values above 7 for Ireland, Finland, and Singapore. Most of the
former socialist countries show low literacy scores. This points to an interpretation that the
history of financial market developments matters, and that a relatively low development of
stock and credit markets is associated with a low level of literacy.
One caveat related to using the WCY index is that it indirectly elicits the level of
financial literacy in each country based on managers’ and country experts’ responses, rather
than a standardized survey of individuals. The only dataset that has some comparability with
WCY is SHARE, which provides detailed information on cognitive abilities (including a few
questions related to financial literacy) at the individual level, for 11 European countries.8 In
SHARE the variable closest to financial literacy measures the ability to perform basic
numerical operations and understanding of basic financial principles.
Specifically, SHARE respondents are asked: (1) to find 10 percent of a number; (2) to
compute the cost of a good that sells at half price; (3) to compute the cost of a new car based
on knowing the cost of a used car and that the used car is two-thirds of a new car; (4) to find
the value of an account balance after two years of an annual interest rate of 10 percent. On the
7
Table 5 reports fixed effects estimates using this alternative indicator of literacy.
8
It should be noted that SHARE is a representative of the population aged 50+, and not of the population at large.

9
basis of these questions, Dewey and Prince (2005) construct an indicator based on these
questions. The indicator ranges from 1 to 5 and is a function of the number of questions
answered correctly; its construction and the actual questions are provided in the Appendix to
this paper.
Although the SHARE variable is not the ideal measure of financial literacy because it
includes only a few economic concepts, there is evidence that knowledge about numerical
problems is related to financial outcomes. Ardle et al. (2009) suggest that more numerate
individuals are more adept at complex decision-making including financial decisions, and also
appear to be more patient and thus more likely to have saved and invested in the past.
Examining the results from a 25-item test of financial knowledge in the Cognitive Economics
Survey, Delevande et al. (2008) find that the number series score has a strong and significant
effect on the financial test score - as does educational attainment and number of economics
courses the respondent has attended.

INSERT FIGURE 2

In the context of the present study, the SHARE numeracy variable is quite useful
because it is available for 11 countries for which IMD data are also available. Figure 2 plots
the WCY and SHARE indicators, showing that the two series are strongly positively
correlated (correlation coefficient is 0.79). In both surveys, Italy and Spain have the lowest
scores, and Sweden, Switzerland, and Denmark the highest. Despite the very different survey
design, countries are well aligned, which increases confidence in the WCY literacy indicator
being a reasonable proxy for financial literacy. The comparison is useful also because the
scale of the WCY indicator is not directly interpretable. Figure 2 shows that two points
change in the WCY indicator (the distance between Italy and Belgium, or between France and
Sweden) is associated with a one point change in the SHARE numeracy indicator. It would be
rather arbitrary, however, to interpret the WCY indicator as a function of the number of
correctly asked questions in each country, as in SHARE. Therefore, in the regressions
analysis we shall standardize the WCY indicator and the independent variables to have mean
zero and a standard deviation of one.
We can make some comparisons between WCY and other measures of financial
literacy also using the Cole et al. (2009) survey responses. In Indonesia and India two survey

10
measures of financial literacy are obtained through the responses to three questions adapted
from Lusardi and Mitchell (2007), which makes the comparable with the US. 9 Measured
financial literacy in India and Indonesia is substantially lower than in the US, which is in line
with WCY ranking.

4. Descriptive Analysis

The most natural framework to study the determinants of financial literacy is to consider that
people accumulate financial knowledge combining ability and effort according to a human
capital production function similar to Cunha and Heckman (2007). Applying this framework
to the context of financial literacy, Delevande, Rohwedder, and Willis (2008) and Willis
(2009) suggest that the incentives to acquire financial knowledge depend on the level of
private resources: while increased knowledge raises the expected return from each dollar
invested, the total value of the investment depends on the number of dollars to which the
improved return is applied. Thus, incentives to acquire financial knowledge are greater for
individuals with higher levels of resources available for investment.10 Similarly, investment
will be greater among people with lower costs or greater efficiency in acquiring additional
knowledge because of their greater ability or because of their greater financial knowledge
obtained through formal education.
To apply this framework to our cross-country data, we relate financial literacy to
measures of human capital, social interactions, and resources. Similar to cognitive abilities,
we use the PISA test scores (available for 1995, 2000, 2003 and 2006 for a maximum of 44
countries)11 and formal education, measured by college enrolment rates and health conditions

9
The questions are: (1) Suppose you borrow 100,000 rupiahs from a money lender at an interest rate of 2% per
month, with no repayment for 3 three months. After 3 months, do you owe less than 102,000 rupiahs, exactly
102,000 rupiahs, or more than 102,000 rupiahs? (2) If you have 100,000 rupiahs in a savings account earning 1%
interest per annum, and prices for goods and services rise 2% over a 1-year period, using the money in the
account, can you buy more than, less than, or the same amount of goods in 1 year as you could today? (3) Is it
riskier to plant multiple crops or one crop?
10
van den Berg et al. (2010), using a Dutch longitudinal database, find that cognitive functioning of elderly
individuals may be affected by negative economic shocks such as job loss or the reduction of pension benefits,
and by events such as the loss of a child or partner or the onset of a serious chronic condition.
11
PISA is available for 27 countries in 1998, 28 in 2000, 33 in 2003 and 44 in 2006. For each country, we take
the average value where more than one observation is available.

11
(proxied by life expectancy).12 Countries differ also in the opportunities to exploit cognitive
abilities. We thus consider technological infrastructures (internet connections or computers
per capita) and social interactions (proxied by the fraction of urban population) to measure
how abilities can be combined to obtain additional financial knowledge.13 Finally, to proxy
for the resources available for financial investments we use the generosity of the social
security system (measured by the social security contributions rate), GDP per capita (PPP
adjusted), and an indicator of financial development (the GDP ratio of stock market
capitalization and private credit). In the cross-section analysis, each of these variables is
averaged over the 1995-2008 period, and merged with institutional and legal indicators
available in the World Bank Doing Business dataset (degree of contract enforcement, judicial
efficiency, legal origin of the country, quality of credit information sharing). Statistics for the
1995-2008 sample are reported in Table 1.

INSERT TABLE 1

The most informative and reliable indicator of cognitive ability is provided by the PISA
scores (Hanushek and Woessman, 2008). Figure 3 shows that there is a strong positive
association between economic and mathematical abilities, and that the effects are potentially
large. For example, in countries where the PISA score is less than 400, the indicator of
financial literacy does not exceed 4, while in almost all countries with math scores above 500
the indicator for financial literacy is higher than 6. There is also a positive correlation
between financial literacy and the fraction of the adult population with college education: in
countries with college achievement rates below 25 percent financial literacy is less than 5 and
that where college achievement exceeds 40 percent, literacy is above 6.

INSERT FIGURE 3

The share of the urban population is also positively correlated to literacy (see Figure 4).
Countries where most of the population is concentrated in large cities (e.g. Australia,

12
Ardle et al. (2009) mention that factors associated with lower cognitive performance include low
socioeconomic status, birth complications, and poor early nutrition.
13
As noted in a recent OECD report, access through internet to many financial products has reduced transaction
costs, but also increased the likelihood that consumers will encounter sophisticated financial assets (OECD,
2005).

12
Belgium, Hong Kong, Israel) also feature relatively high literacy, lending support to the idea
that more intense social interactions are associated with higher literacy.14

INSERT FIGURE 4

Financial literacy is positively correlated to economic development (measured by GDP

per capita) and financial development (the GDP ratio of stock market capitalization plus
private credit, the most widely used indicator of financial development (Beck et al., 2009).
These correlations may be interpreted as financial development raising the volume of saving
and the incentives to learn about finance. However, such a conclusion would be premature,
because the two-way correlation may be driven simply by financial development being
correlated to education, GDP per capita and other determinants of literacy. Furthermore,
financial literacy can affect financial development, for at least three reasons. First, higher
literacy leads to more efficient allocation of savings and higher per dollar returns, attracting
more investment and growth in the country. Second, higher literacy might induce greater
stock market participation and financial market depth, as shown in Christelis et al. (2010) and
Alessie et al. (2007). Third, as Bernanke (2006) claims, financial literacy might build
confidence in the market economy, discipline financial intermediaries, and create a better
policy environment for growth.
The social security payroll tax rate is directly related to the amount of mandated saving
in the form of social security contributions and, therefore, is also an indicator of the resources
available for private accumulation, particularly in the form of private pension funds, life
insurance, and other retirement savings vehicles. Since payroll tax is set by the government, it
can be fairly safely assumed that it is not affected by the level of financial literacy. This rules
out that governments decide to introduce less generous socials security benefits in countries
where people do not have sufficient financial literacy to manage their own pension
investment well. This assumption is reasonable, as recent social security reforms (in Sweden,
Italy, Germany and elsewhere) were implemented to address population aging and fiscal
crisis, rather than to raise the level of financial literacy of the country.

14
Hong et al. (2004) show that stock market participation is higher among more socially connected individuals.
A related line of research points out that trust is an important determinant of economic exchange and financial
transactions. Guiso et al. (2004) find that, other things being equal, the proportion of stockholders is higher in
Italian provinces with relatively high social trust. People who are more active socially might be more inclined to
trust, making the effects of sociability and trust difficult to distinguish empirically.

13
 INSERT FIGURE 5

Figure 5 shows that contribution rates are negatively correlated to financial literacy:
countries where contribution rates are lower than 20 percent (Chile and New Zealand) score
above 6 for financial literacy, while countries with higher contribution rates (Greece, Spain
and Portugal) show relatively low literacy levels. Since the contribution rate is more likely to
be exogenous with respect to financial literacy, in the literacy regressions we rely primarily
on contribution rate to proxy for private resources. However, we also provide OLS and
instrumental variables estimates based on financial development and GDP per capita.

5. Regression Analysis

Table 2 shows the sample average of the variables of interest for each country and reports a
first set of regressions for the 46 countries with non-missing observations. The baseline
specification includes the math score in PISA, the social security contribution rate, and the
share of urban population, which arguably are exogenous variables. To ease the interpretation
of the results, in the regression analysis each of the variables is standardized to have mean
zero and a standard deviation of one.
The coefficients of each of the three variables are precisely estimated, and have the
expected sign. The math score and the share of urban population are positively associated
with literacy, while the coefficient of the social security tax rate is negative, lending support
to the hypothesis that incentives to acquire financial literacy are higher when savings
mandated by government are lower. In terms of economic significance, an increase of one
standard deviation in math score (equivalent to moving from Ireland to Korea) is associated to
an increase in literacy of 0.54 standard deviations. Likewise, an increase of one standard
deviation of urbanization (equivalent to moving from Finland to France) is associated to an
increase of 0.26 standard deviations of literacy; and an increase in one standard deviation of

14
the social security contribution rate (for instance, moving from Germany to Italy) is
associated with a reduction of 0.31 standard deviations of literacy.

INSERT TABLE 2

Column 2 of Table 2 extends the baseline specification introducing a dummy for former
socialist countries. The coefficient of this dummy is negative and statistically different at the
5 percent level, in line with the fact that for historical and institutional reasons the populations
in these countries are less familiar with portfolio management and financial decisions. In
particular, in former socialist countries literacy is 0.55 standard deviations lower than in other
countries. The coefficients of other regional dummies (for Latin America, Asia and other
European countries) are not statistically different from zero. The last two columns in Table 2
repeat the estimation using a robust estimation method to check for the effect of influential
observations; the results are unaffected.

INSERT TABLE 3

In what follows, we provide a number of checks for the correlations in Table 2,

expanding the variables used in the estimation, controlling for the potential endogeneity of
some of the variables, and reporting fixed effects estimates at the level of individual countries.
Table 3 presents expanded baseline regression that include the additional variables: science
score in PISA, secondary school enrolment rate, college achievement, life expectancy,
internet connections, log of per capita GDP and GDP growth. None of these coefficients is
statistically different from zero at standard confidence levels. The results of the baseline
estimates are unaffected, with the exception of the coefficient of the share of urban population
which is less precisely estimated.
We next report regression estimates for the relation between financial literacy and three
proxies for financial market deepening: GDP ratio of stock market capitalization, GDP ratio
of private credit, and the sum of these two items (financial development). Table 4 shows that,
other things being equal, the measures of financial market deepening are not correlated with
literacy. Given the potential endogeneity of financial market development with respect to

15
literacy, in the last two columns of Table 4 we supplement the OLS estimates with
instrumental variable regressions.

INSERT TABLE 4

We draw on the large literature on the legal and institutional determinants of financial
development to obtain our instruments, and use dummies for the legal origin of the country
and an index of the strength of investors’ protection (a combined indicator of transparency of
transactions, director liability, and shareholders’ ability to sue officers and directors for
misconduct).15 The Sargan test does not reject the hypothesis of valid instruments. The F-tests
of the exclusion of the instrument set in the first-stage regression were statistically significant
at the 1% level (5% in case of private credit). Since such a diagnostic has limitations when
there is more than one endogenous regressor (as in column 3), we compute the partial R
squared (Shea, 1997).16 In both specifications, the instrumental variable estimates confirm
lack of significance of the indicators of financial development; the lack of correlation is not
affected by the particular set of instruments used.17
The cross-sectional estimates can be criticized for excluding too many country-level
characteristics that potentially are related to financial literacy and for which the regressions in
Table 3 do not control. To address this issue, we exploit the panel structure of the sample;
Table 5 reports the fixed effects estimates. We drop the dummy for former socialist countries
(which is absorbed by the country fixed effects) and to the baseline regression add a cyclical
indicator (GDP growth rate) and - in column (2) - proxies for financial development. Since for
many countries only one set of PISA scores is available, we replace them with the fraction of
individuals with college education. Both regressions confirm a positive and significant

15
The variables are drawn from the World Bank Doing Business database. La Porta et al. (1997) argue
forcefully that legal origin of the country and investors’ protection are strong determinant of the depth of its
financial markets. Djankov et al. (2007) and Brown et al. (2009) find that information sharing among lenders is
associated with improved availability and lower costs of credit.
16
Shea’s partial R-squared is a test of the individual explanatory power of the instruments, accounting for
correlation among the instruments. The results obtained indicate that there is enough separate variation in the
instruments.
17
We also experiment with an index of Creditor Rights as a measure of creditor legal protection built using the
methodology proposed by La Porta et al. (1997). Higher values of this index imply that secured lenders are
better protected in the case that a borrower defaults. As a measure of actual creditor protection, we include the
variable Time to Enforce Payment, which measures the (log of the) number of days it takes for a creditor to
secure an outstanding payment through the courts if a debtor defaults. Finally, we include among the instruments
the variable Information Sharing in Credit Markets because Djankov et al. (2007) and Brown et al. (2009) find
that information sharing among lenders is associated with improved availability and lower costs of credit.

16
association between share of urban population and financial literacy. The coefficient of the
social security contribution rate is statistically different from zero (at the 5 percent level) only
in the second specification. The magnitude of the coefficients is smaller than in the OLS
regressions in Table 2: a one standard deviation increase in the contribution rate is associated
with a reduction of 0.15 standard deviations of literacy.

INSERT TABLE 5

In the panel estimates the coefficients of the school attainment variable are not
statistically different from zero. This is expected, given that the level of education is a slowly
changing variable whose effect is hard to pinpoint in a relatively short panel.18 Finally, it
should be noted that since the institutional determinants of financial market deepening are
constant over time or change only slowly, in the panel regressions there is not enough
variability in the instruments to provide reliable IV estimates.
In columns (3) and (4) of Table 5 we repeat the estimation using education in finance as
an alternative indicator of financial literacy (recall that the variable is available only from
1999 to 2008). The estimates show that education in finance is negatively associated with the
social security contribution rate, confirming the results in columns 1 and 2. Instead, the
coefficients of urban population, college achievement and financial development are not
statistically different from zero.

6. Conclusions

Many surveys have shown that investors have poor financial literacy. These surveys are
targeted at different population groups around the world, and elicit financial literacy in very
different ways, from self-assessment to detailed questions aimed at understanding whether
individuals are familiar with basic economic concepts, portfolio management, and specific
financial products. The data used in this paper offer a comprehensive assessment of literacy
across the world based on a survey of executives in 55 countries, in 1995-2008. The
advantage of the dataset is strict international comparability, which allows financial literacy

18
In the panel estimates the coefficients of internet connection and life expectancy are never statistically
different from zero.

17
to be related to the quantity and quality of human capital, technological infrastructure,
economic, and financial development. The drawback to it is that the survey respondents are a
selected group of managers and country-experts, and that data are only available in aggregate
form, preventing analysis of specific socioeconomic groups.
The descriptive analysis shows that literacy varies quite substantially among countries,
and the regression analysis shows that its level depends on educational achievement, social
interactions (as proxied by the share of urban population), and mandated savings in the form
of social security contributions. The contribution rate is used as an (inverse) proxy for
financial market deepening to minimize the risk of reverse causation between financial
literacy and financial development. The findings can be rationalized using a standard human
capital model, where financial knowledge depends on cognitive ability, and incentives to
accumulate knowledge are directly related to the level of household resources invested in
financial markets, and particularly in pension funds.
The paper has two implications for policy. First, the international comparison suggests
that financial literacy improves with the drivers of human capital and financial market reform,
both of which change slowly over time. Second, social security reforms associated with
financial market deepening (e.g., the creation of private pension funds), by raising the
incentive to acquire financial knowledge, eventually will lead also to improvements in
financial literacy.

18
 References

Alessie, R., Lusardi A. and Van Rooij M. (2007), ‘Financial literacy and stock market
participation’, NBER Working Paper No. 13565.

Alessie, R., Lusardi A. and Van Rooij M. (2008), ‘Financial literacy, retirement planning, and
household wealth’, Working Paper, Dartmouth College.

Ardle, J., Smith J.P. and Willis R.J., (2009), “Cognition and economic outcomes in the Health
and Retirement Survey’, NBER Working Paper No. 15266.

Banks, J. and Oldfield Z. (2007), ‘Understanding pensions: cognitive function, numeracy and
retirement saving’, Fiscal Studies, vol. 28(2), pp. 143-70.

Banks, J., O’Dea C. and Oldfield Z. (2009), ‘Cognitive function, numeracy and retirement
saving trajectories’, Working Paper, Institute for Fiscal Studies.

Beck, T., Demirgüç-Kunt A. and Levine R.E. (2009), ‘A new database on financial
development and structure (updated May 2009)’, Working Paper, World Bank Policy
Research.

Bernanke, B. (2006), ‘A message from Chairman Bernanke’, Federal Reserve Bank of Dallas,
July.

Brown, M., Jappelli T. and Pagano M. (2009), ‘Information sharing and credit: firm-level
evidence from transition countries’, Journal of Financial Intermediation, vol. 18(2), pp.
151-72.

Christelis, D., Jappelli T. and Padula M. (2010), ‘Cognitive abilities and portfolio choice’,
European Economic Review, vol. 54 (1), pp. 18-39.

Christiansen, C., Schröter Joensen J. and Rangvid J. (2008), ‘Are economists more likely to
hold stocks?’, Review of Finance, vol. 12(3), pp. 465-96.

Cole, S. and Kartini Shastri G. (2009), ‘Smart money: the effect of education, cognitive
ability and financial literacy on financial market participation’, Harvard Business School
Working Paper No. 09-071.

Cole, S., Sampson T. and Zia B. (2009), ‘Money or knowledge? What drives the demand for
financial services in developing countries?’, Harvard Business School Working Paper,
No. 09-117.

Cunha, F. and Heckman J. (2007), ‘The technology of skill formation’, American Economic
Review. Papers and Proceedings, vol. 97(2), pp. 31-47.

Delevande, A., Rohwedder, S. and Willis R.J. (2008), ‘Preparation for retirement, financial
literacy and cognitive resources’, Michigan Retirement Research Center Working Paper
2008-190.

19
Dewey, M.E. and Prince M.J. (2005), ‘Cognitive function’, in (A. Börsch-Supan, A.
Brugiavini, H. Jürges, J. Mackenbach, J. Siegriest and G. Weber, eds.), Health, Aging
and Retirement in Europe: First Results from the Survey of Health, Aging and
Retirement in Europe. Mannheim: Mannheim Research Institute for the Economics of
Aging.

Djankov, S., McLiesh C. and Shleifer A. (2007), ‘Private credit in 129 countries’, Journal of
Financial Economics, vol. 84(2), pp. 299-329.

Duflo, E. and Saez E. (2003), ‘The role of information and social interactions in retirement
plan decisions: evidence from a randomized experiment’, Quarterly Journal of
Economics, vol. 118(3), pp. 815–42.

G8 (2006), ‘Improving financial literacy. Summary of discussion’, Moscow, April 18-19,

http://www.g8finance.ru/en/documents/index.php?id19=58.

Guiso, L. and Jappelli T. (2005), ‘Awareness and stock market participation’, Review of
Finance, vol. 9(4), pp. 537-67.

Guiso, L. and Jappelli T. (2008), ‘Financial literacy and portfolio diversification’, CSEF
Working Paper 212.

Guiso, L., Sapienza P. and Zingales L. (2004), ‘The role of social capital in financial
development’, American Economic Review vol. 94(3), pp. 526-56.

Hanushek, E.A. and Woessmann L. (2008), ‘The role of cognitive skills in economic
development’, Journal of Economic Literature vol. 46(3), pp. 607-68.

Hastings, J.S. and Tejeda-Ashton L. (2008), ‘Financial literacy, information, and demand
elasticity: survey and experimental evidence from Mexico,’ NBER Working Paper No.
14538.

Hong, H.G., Kubik J.D. and Stein J.C. (2004), “Social interaction and stock market
participation’, Journal of Finance, vol. 59(1), pp. 137-63.

Jappelli, T., Pagano M. and Di Maggio M. (2008), ‘Households’ indebtedness and financial
fragility’, CSEF Working Paper No. 208.

La Porta, R., Lopez-de-Silanes F., Shleifer A. and Vishny R.W. (1997), ‘Legal determinants
of external finance’, Journal of Finance, vol. 52(3), pp. 1131-50.

Lusardi, A. (2008), ‘Household saving behavior: the role of literacy, information and financial
education programs’, NBER Working Paper No. 13824.

Lusardi, A. and Mitchell O.S. (2007), ‘Baby boomers retirement security: the role of planning,
financial literacy and housing wealth’, Journal of Monetary Economics, vol. 54(1), pp.
205-224.

20
Lusardi, A. and Tufano P. (2009), ‘Debt literacy, financial experiences, and
overindebtedness’, NBER Working Paper No. 14808.

Mishkin, F. (2008), ‘The importance of economic education and financial literacy’, Speech
before the Federal Reserve Board at the Third National Summit on Economic and
Financial Literacy, Washington, D.C.

Organisation for Economic Co-Operation and Development (2005), Improving Financial

Literacy: Analysis of Issues and Policies. Paris: OECD.

Shea, J. (1997), ‘Instrument relevance in multivariate linear models: a simple measure’,

Review of Economics & Statistics 79(2), PP. 348-52.

Smith, J.P., McArdle J.J. and Willis R.J. (2009), ‘Financial decision making and cognition in
a family context’, Working Paper, Rand Corporation.

van den Berg, G.J., Deeg D.J., Lindeboom M. and Portrait F. (2010), ‘The role of early-life
conditions in the cognitive decline due to adverse events later in life’, IZA Discussion
Papers 4780, Institute for the Study of Labor (IZA).

Willis, L.E. (2008), “Against financial literacy education’, Research Paper No. 2008-10,
Public Law and Legal Theory Research Paper Series,

Willis, R.J. (2009), “Disentangling cognitive function and financial literacy: implications for
financial retirement security research’, paper presented at the Conference on Financial
Literacy in Times of Turmoil and Retirement Insecurity. Brookings Institution,
Washington D.C., March 20.

21
 Appendix

A1. The World Competitiveness Yearbook (WCY)

The IMD World Competitiveness Yearbook (WCY) is a comprehensive annual report on the
competitiveness of nations available for 1995 to 2008. The WCY includes 329 variables on the
following topics:

 Economic Performance (82 variables), covering domestic economy, international trade,

international investment, employment, and prices.
 Government Efficiency (70 variables), covering public finance, fiscal policy, institutional
framework, business legislation, and societal framework.
 Business Efficiency (67 variables), covering productivity, labor markets, finance, management
practices and attitudes, and values.
 Infrastructure (110 variables), covering basic infrastructure, technological infrastructure,
scientific infrastructure, health and environment, and education.

The WCY uses different types of data to measure quantitative and qualitative issues separately.
Statistical indicators are acquired from international, national and regional organizations, private
institutions and a network of 54 partner institutes worldwide. These statistics are referred to in the
WCY as Hard Data and include 245 variables. The other variables are drawn from the annual
Executive Opinion Survey and are referred to in the WCY as Survey Data.

The Executive Opinion Survey was designed to quantify issues that are not easily measured, for
example management practices, labor relations, corruption, environmental concerns, and quality of
life. The Executive Opinion Survey is sent to executives in top and middle management in all of the
economies covered by the WCY. The sample of respondents covers a cross-section of the business
community in each economic sector: primary, manufacturing, and services, based on their contribution
to the GDP of the economy. The survey respondents are nationals or expatriates, located in local and
foreign enterprises in a country and who, in general, have an international dimension. The surveys are
sent in January and are returned in April of each year. From the last Opinion Survey, WCY indicators
were based on 3,960 responses from 57 countries.

The following variables are drawn from the WCY database: financial literacy, life expectancy,
fraction of urban population, internet and computers per capita, GDP per capita.

A2. The OECD-PISA Survey

The OECD Programme for International Student Assessment (PISA - www.pisa.oecd.org) is a regular
survey of 15-year olds which assesses aspects of their preparedness for adult life.

Mathematical Literacy: the capacity to identify, to understand, and to engage in mathematics and
make well-founded judgments about the role of mathematics, needed in current and future private life,
occupational life, social life with peers and relatives, and life as a constructive, concerned, and
reflective citizen.

Scientific Literacy: the capacity to use scientific knowledge, to identify questions and to draw
evidence-based conclusions in order to understand and contribute to decisions about the natural world
and the changes wrought on it by human activity

22
A3. Macroeconomic Variables

Education: Secondary and tertiary enrolment rates (Source: OECD, Education at a Glance).
Financial development: Stock market capitalization relative to GDP and private credit relative to GDP
(Source: Beck, Demirgüç-Kunt, and Levine, 2009).

A4. Institutional Variables

The following variables are drawn from the Doing Business database, available at the World Bank
web site.

Legal Origin. Identifies the legal origin of the company law or commercial code of each country.
Source: La Porta, Lopez-de-Silane, Schleifer and Vishny (1997).

Investor Protection: Measures the strength of minority shareholder protection against directors’
misuse of corporate assets for personal gain. The indicators distinguish among three dimensions of
investor protection: transparency of related-party transactions (extent of disclosure index); liability for
self-dealing (extent of director liability index); and shareholders’ ability to sue officers and directors
for misconduct (ease of shareholder suit index). The data are from a survey of corporate lawyers and
are based on securities regulations, company law, and court rules of evidence transparency of
transactions, director liability index, and shareholders’ ability to sue officers and directors for
misconduct

Information Sharing Index. The depth of credit information index measures the scope, accessibility
and quality of credit information available through either public or private credit registries. The index
ranges from 0 to 6, with higher values indicating the availability of more credit information, from
either a public registry or a private bureau. A score of 1 is assigned for each of the following 6 features
of the public registry or the private credit bureau (or both):

 positive and negative credit information

 data on firms and individuals
 data from retailers, trade creditors, utility companies, and financial institutions
 at least 2 years of historical data
 data on loans below 1% of per capita incomes
 borrowers have the right to access their data in the largest registry in the economy.

Enforcing Contracts. Indicators on enforcing contracts measure the efficiency of the judicial system
at resolving a commercial dispute. The data are built following a step-by-step evolution of commercial
sale disputes in the local courts. Data are collected from codes of civil procedure and other court
regulations and surveys completed by local litigation lawyers (and, in 25% of the countries, also by
judges).The value of a claim equals 200% of the national per capita income. Time to collect the claim
is recorded in calendar days, counted from the moment the plaintiff files the lawsuit in court until
payment. Costs are recorded as a percentage of the claim.

23
4. Numeracy Indicator in SHARE

The (abridged) questions on numeracy are as follows. Possible answers are shown in a card while the
interviewer is instructed not to read them out to the respondent:

1. If the chance of getting a disease is 10 per cent, how many people out of one thousand would
be expected to get the disease? The possible answers are 100, 10, 90, 900 and another answer.
2. In a sale, a shop is selling all items at half price. Before the sale a sofa costs 300 euro. How
much will it cost in the sale? The possible answers are 150, 600 and another answer.
3. A second hand car dealer is selling a car for 6,000 euro. This is two-thirds of what it costs
new. How much did the car cost new? The possible answers are 9,000, 4,000, 8,000, 12,000,
18,000 and another answer.
4. Let’s say you have 2,000 euro in a saving account. The account earns ten per cent interest
each year. How much would you have in the account at the end two years? The possible
answers are 2,420, 2,020, 2,040, 2,100, 2,200, 2,400 and another answer.

The numeracy indicator is a function of the number of questions answered correctly, and ranges from
1 to 5. If a person answers (1) correctly she is then asked (3) and if she answers correctly again she is
asked (4). Answering (1) correctly results in a score of 3, answering (3) correctly but not (4) results in
a score of 4 while answering (4) correctly results in a score of 5. On the other hand if she answers (1)
incorrectly she is directed to (2). If she answers (2) correctly she gets a score of 2 while if she answers
(2) incorrectly she gets a score of 1.

24
 Table 1
Descriptive Statistics

Mean Median Standard Minimum Maximum

deviation

Financial literacy 4.87 4.73 1.44 2.06 7.50

Education in finance 5.66 5.81 1.24 3.54 7.93
Math score in PISA survey 481.94 495.88 55.06 353.17 561.82
Science score in Pisa survey 486.09 497.62 47.57 385.11 550.10
Secondary school enrolment rate (%) 83.04 87.83 12.20 54.74 99.13
Tertiary education (%) 26.31 26.62 11.97 4.93 49.38
Social security contribution rate (%) 19.30 17.48 12.10 1.08 49.88
Share of urban population (%) 70.60 70.01 15.86 27.85 99.46
Life expectancy 74.35 75.67 5.07 56.08 81.16
Stock market capitalization (% of GDP) 65.25 44.97 52.53 7.41 229.90
Private credit (% of GDP) 75.61 74.99 44.02 12.41 167.83
Financial development (% of GDP) 142.58 138.49 89.79 21.66 389.73
Log per capita GDP 9.20 9.28 1.18 6.31 11.07
GDP Growth 0.07 0.06 0.04 -0.02 0.21
Old Europe 0.27 0.00 0.45 0.00 1.00
Latin America 0.13 0.00 0.34 0.00 1.00
Asia 0.24 0.00 0.43 0.00 1.00
Former socialist countries 0.24 0.00 0.43 0.00 1.00
Other countries 0.13 0.00 0.34 0.00 1.00
English origin 0.22 0.00 0.42 0.00 1.00
French origin 0.33 0.00 0.47 0.00 1.00
Investor protection index 10.58 10.33 3.25 4.33 18.33

Note. Variables are averaged over the 1995-2008 period, except for education in finance, available from 1999 to
2008. The sample includes the following countries. Asia: China, Hong Kong, India, Indonesia, Israel, Japan,
Jordan, Korea, Malaysia, Philippines, Singapore, Taiwan, Thailand, Turkey. Latin America: Argentina, Brazil,
Chile, Colombia, Mexico, Peru, Venezuela; New Europe: Bulgaria, Croatia, Czech Republic, Estonia, Hungary,
Lithuania, Poland, Romania, Russia, Slovak Republic, Slovenia, Ukraine; Old Europe: Austria, Belgium,
Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, Sweden,
United Kingdom; Other countries: Australia, Canada, New Zealand, Norway, South Africa, Switzerland, United
States.

25
 Table 2
Regressions for Financial Literacy – Baseline Estimates

OLS Robust regression

(1) (2) (3) (4)

Math score in Pisa 0.541 0.562 0.525 0.551

(0.084)*** (0.080)*** (0.089)*** (0.087)***
Social security contribution rate (%) -0.312 -0.209 -0.317 -0.222
(0.082)*** (0.087)** (0.087)*** (0.095)**
Share of urban population (%) 0.260 0.180 0.282 0.202
(0.095)*** (0.095)* (0.101)*** (0.104)*
Former socialist countries -0.551 -0.524
(0.214)** (0.235)**
Constant 0.116 0.248 0.136 0.254
(0.082) (0.093)** (0.087) (0.102)**

Observations 46 46 46 46
R-squared 0.65 0.70 0.63 0.66

Note. The dependent variable is the index of financial literacy. Left-hand-side and right-hand-side variables
(except for the dummy for former socialist countries) have been standardized to have mean zero and standard
deviation equal to one. The robust estimation method is implemented using the rreg robust estimation command
in Stata, which performs an initial OLS regression, calculates the Cook's distance, eliminates the gross outliers
for which the Cook's distance exceeds 1, and then performs iterations based on Huber weights. Standard errors
are reported in parenthesis. One star indicates significance at the 10 percent level, two stars at the 5 percent level,
three stars at the 1 percent level.

26
 Table 3
Regressions for Financial Literacy – Additional Variables

(1) (2) (3) (4) (5)

Math score in Pisa 0.656 0.527 0.545 0.427 0.492

(0.292)** (0.127)*** (0.108)*** (0.133)*** (0.136)***
Social security contribution rate -0.207 -0.182 -0.211 -0.189 -0.224
(0.088)** (0.096)* (0.088)** (0.088)** (0.091)**
Share of urban population (%) 0.182 0.165 0.174 0.150 0.160
(0.096)* (0.100) (0.099)* (0.097) (0.101)
Former socialist countries -0.537 -0.533 -0.521 -0.507 -0.369
(0.221)** (0.218)** (0.250)** (0.216)** (0.385)
Science score in Pisa -0.098
(0.292)
Secondary school enrolment rate -0.056
(0.163)
College achievement 0.129
(0.111)
Life expectancy 0.038
(0.159)
Internet connections per 100 people 0.189
(0.149)
Log per capita GDP 0.113
(0.182)
GDP growth -0.040
(0.129)
Constant 0.244 0.249 0.234 0.202 0.184
(0.095)** (0.095)** (0.112)** (0.099)** (0.137)

Observations 46 46 46 46 46
R-squared 0.70 0.71 0.70 0.71 0.70

Note. The dependent variable is the index of financial literacy. Left-hand-side and right-hand-side variables
(except for the dummy for former socialist countries) have been standardized to have mean zero and standard
deviation equal to one. All regressions are estimated by OLS. Standard errors are reported in parenthesis. One
star indicates significance at the 10 percent level, two stars at the 5 percent level, three stars at the 1 percent level.

27
 Table 4
Regressions for Financial Literacy: The Role of Financial Development

OLS IV
(1) (2) (3) (4)

Math score in Pisa 0.554 0.507 0.649 0.512

(0.112)*** (0.103)*** (0.198)*** (0.130)***
Social security contribution rate -0.255 -0.234 -0.318 -0.238
(0.102)** (0.099)** (0.154)** (0.125)*
Share of urban population (%) 0.238 0.255 0.234 0.257
(0.101)** (0.099)** (0.106)** (0.102)**
Stock market capitalization (% of GDP) 0.205 0.299
(0.138) (0.268)
Private credit (% of GDP) -0.061 -0.273
(0.151) (0.393)
Financial development (% of GDP) 0.158 0.148
(0.125) (0.227)
Constant 0.174 0.160 0.199 0.160
(0.086)** (0.084)* (0.092)** (0.079)**

Observations 42 42 42 42
R-squared 0.68 0.67 0.66 0.67
Sargan test (p-value) 0.65 0.94
Shea’s partial R squared: stock market capitalization 0.24
Shea’s partial R squared: private credit 0.13
F-test of the exclusion of the instruments 4.22

Note. The dependent variable is the index of financial literacy. Left-hand-side and right-hand-side variables
have been standardized to have mean zero and standard deviation equal to one. In the IV estimates the
instruments are: a dummy for English origin, a dummy for French origin, and an index of the strength of
investors’ protection (based on the average of transparency of transactions, director liability index, and
shareholders’ ability to sue officers and directors for misconduct). Standard errors are reported in parenthesis.
One star indicates significance at the 10 percent level, two stars at the 5 percent level, three stars at the 1 percent
level.

28
 Table 5
Regressions for Financial Literacy and Education in Finance.
Fixed Effects Panel Estimates

Financial literacy Education in finance

(1) (2) (3) (4)

College achievement 0.026 -0.018 0.149 0.064

(0.062) (0.073) (0.091) (0.106)
Social security contribution rate -0.101 -0.148 -0.276 -0.414
(0.068) (0.070)** (0.147)* (0.166)**
Share of urban population (%) 0.681 0.678 -0.076 -0.096
(0.310)** (0.313)** (0.507) (0.510)
GDP growth 0.134 0.112 0.232 0.225
(0.060)** (0.066)* (0.089)*** (0.099)**
Financial development 0.125 -0.000
(0.067)* (0.101)
Constant 0.046 0.077 0.121 0.212
(0.065) (0.052) (0.109) (0.087)**

Observations 367 324 314 277

Number of countries 46 41 46 41
R-squared 0.06 0.07 0.07 0.07

Note. The table reports fixed effects panel estimates. The dependent variables are the index of financial literacy
and the index of education in finance. Left-hand-side and right-hand-side variables have been standardized to
have mean zero and standard deviation equal to one. Standard errors are reported in parenthesis. One star
indicates significance at the 10 percent level, two stars at the 5 percent level, three stars at the 1 percent level.

29
 Fig. 1

Financial Literacy Around the World

Venezuela
Peru
Mexico
Croatia
Russia
Ukraine
Romania
Portugal
Brazil
Colombia
China
Indonesia
Poland
Italy
Lithuania
Bulgaria
India
Spain
Thailand
Slovenia
Turkey
France
Greece
Slovak Republic
United Kingdom
Hungary
Argentina
Czech Republic
Philippines
Jordan
Chile
Germany
United States
Luxembourg
Belgium
New Zealand
Estonia
Austria
Malaysia
Korea
Norway
Canada
Sweden
Taiwan
Israel
Netherlands
Switzerland
Japan
Denmark
Australia
Hong Kong
Ireland
Finland
Singapore

0 2 4 6 8
Financial literacy

30
 Fig. 2

Comparison between WCY and SHARE Indicators of Financial Literacy

Denmark Switzerland
Sweden
1
Financial literacy, WCY

Austria
Belgium
Germany
0

Greece
France
Spain

Italy
-1
-2

2.5 3 3.5 4 4.5

Numeracy indicator in SHARE

31
 Fig. 3

Financial Literacy and Math Score in the PISA Survey

8

Ireland Finland
Australia Hong Kong
Denmark Switzerland Japan
Israel Netherlands
Sweden Taiwan
Norway Canada
6

Korea
Financial literacy

Austria
Estonia
United States New
Luxembourg Zealand
Belgium
Germany
Chile
Jordan
Czech Republic
Argentina
Hungary Kingdom
Turkey Greece Slovak United
Republic
France
Thailand Spain Slovenia
4

Bulgaria
Italy Lithuania
Indonesia Poland
Colombia
Brazil Romania Portugal
Croatia Russia
Mexico
2

350 400 450 500 550

Math score in Pisa

32
 Fig. 4

Financial Literacy and Share of Urban Population

8

Singapore

Finland
Ireland
Australia Hong Kong
Japan
Switzerland Denmark
Netherlands Israel
Taiwan Sweden
Canada
Norway
6

Malaysia Korea
Financial literacy

Austria
Estonia
United New Zealand Belgium
Luxembourg
States
Germany
Chile
Jordan
Philippines Czech Republic
Argentina
Hungary United Kingdom
Slovak Republic
Greece Turkey France
Thailand Slovenia Spain
India
4

Bulgaria
Lithuania
Italy
Poland
China Indonesia Colombia
Portugal
Romania Brazil
UkraineRussia
Croatia Mexico
Peru Venezuela
2

20 40 60 80 100
Urban population (%)

33
 Fig. 5

Financial Literacy and Social Security Contribution Rate

Singapore

Ireland Finland
Hong Kong Australia
Denmark Switzerland Japan
IsraelTaiwan Netherlands
Sweden
Canada Norway
6

Korea
Malaysia
Financial literacy

Austria Estonia
New ZealandUnited Luxembourg
States Belgium
Germany
Chile
Jordan
Philippines Czech Republic
Argentina
United Kingdom Hungary
Turkey Greece Slovak Republic France
Thailand Slovenia Spain
India
4

Bulgaria
Italy Lithuania
Indonesia Poland
Colombia China
Portugal Brazil
RomaniaUkraine
Croatia Russia
Mexico
Peru Venezuela
2

0 10 20 30 40 50
Social security contribution rate

34