Financial Literacy and Portfolio Diversification

Abstract

We use a survey of individual investors disclosed by the Portuguese Securities Commission (CMVM) in May 2005 to study the impact of investors' levels of financial literacy on portfolio diversification. We consider distinct aspects of financial literacy, and control for socioeconomic and behavioral differences among individual groups of investors. Our results suggest that investors' educational levels and their financial knowledge have a positive impact on investor diversification. The information sources used by retail investors to gather information on markets and financial products also have a significant impact on the number of different assets included in a portfolio.

Full text

Financial Literacy and Portfolio Diversification
Margarida Abreu
∗
CISEP - ISEG - Technical University of Lisbon
Rua Miguel Lupi, 20
1200-725 Lisboa - PORTUGAL
Victor Mendes
♦
CMVM – Portuguese Securities Commission
Av. da Liberdade, 252
1056-801 Lisboa - PORTUGAL
Abstract
We use a survey of individual investors disclosed by the Portuguese securities
commission (CMVM) in May 2005 to study the impact of investors’ levels of
financial literacy on portfolio diversification. We consider distinct aspects of financial
literacy, and control for socioeconomic and behavioral differences among individual
groups of investors. Our results suggest that investors’ educational levels, their
financial knowledge and the information sources used by retail investors to gather
information on markets and financial products have a significant impact on the
number of different assets included in a portfolio.
JEL: G11, I20, G32, J24
Keywords: Educational Finance, Resource allocation
∗
Corresponding author. Tel: +351213955745; Fax: +351213967309; Email address:
mabreu@iseg.utl.pt
♦
The views stated in this paper are those of the author and are not necessarily those of the Portuguese
Securities Commission.

1. Introduction
Perhaps the one aspect of the data on portfolio choice that is most challenging to
traditional theories is the apparent lack of diversification in the financial portfolio of
individual investors.
Contrary to what is required by standard theory, investors hold very undiversified
portfolios made up of a limited number of assets (Calvet, Campbell and Sodini, 2006;
Goetzmann and Kumar, 2001 and 2005). In fact, most investors hold nearly all of
their wealth in domestic assets (French and Poterba, 1991), concentrate their portfolio
in the equity market (Barber and Odean, 2000; Barber, Heath and Odean, 2003;
Goetzmann and Kumar, 2001) and select these stocks mostly on the basis of
geographical or professional proximity (Coval and Moskowitz, 1999; Curcuru et al.,
2005).
The first studies of investor behavior are credited to Merton (1969) and Samuelson
(1969). The finance theory built upon their work assumes that the investor’s portfolio
is optimally chosen, in such a way as to maximize its expected utility and choose
between assets with differentiated expected returns and risks, subject to a wealth
constraint. An important result of this approach is that, while the portfolio return
equals the weighted average return of the assets that form it, the portfolio risk is not,
in general, equal to the average of the risks of the assets included in it. As such,
adding differentiated assets to a portfolio makes it possible, in general, to reduce
portfolio risk in comparison with the average risk of the assets which compose the
portfolio. This is due to risk diversification. The portfolio risk is not solely dependent
upon the risks of its assets in isolation; it also depends on the way in which their
returns are affected (similarly or not) by the events that change them. Portfolio risk is
therefore dependent on the covariance (or correlation) between the rates of return of

the assets comprising the portfolio. With the exception of those situations where the
rates of return are perfectly and positively correlated, diversification makes it possible
to reduce the risk of a financial investment. Therefore, rational and informed investors
are supposed to have diversified portfolios, irrespective of their levels of risk
aversion.
In the framework of traditional finance theory, important contributions have been
made towards explaining why a rational investor may hold a less diversified portfolio
than might be considered optimal. In fact, in the traditional utility maximizing
framework, factors such as transaction costs, institutional constraints or customs
barriers can explain some (but not all) aspects of the observed cross-sectional
variation in family portfolio holdings. Transaction costs are important because the
fees charged per financial transaction frequently do not depend on the amount of the
transaction. They therefore strongly limit the expected return of an investment and
may hinder diversification (Guiso, Haliassos and Jappelli, 2002). On the other hand,
the fiscal regime governing capital gains that are dependent upon the nature of the
financial asset also contributes towards portfolio under-diversification (Campbell,
2006). In an empirical study for the USA, Rendleman and Shackelford (2003)
conclude, for example, that the American fiscal regime favors diversification. Similar
conclusions emerge from diverse studies on European household portfolios (Banks
and Tanner 2002; Guiso and Jappelli, 2002; Eymann and Borsch-Supan, 2002).
Notice, for example, the fiscal benefits granted to retirement saving plans or, in
previous years, to investment in Portuguese stocks. Finally, institutional and customs
barriers can, for instance, restrict access to foreign markets, thus limiting the
international diversification of portfolios (French and Poterba, 1991).

However some financial puzzles remain which cannot be explained by the traditional
utility maximizing framework. Why is individual portfolio turnover so high (Glaser,
Noth and Weber, 2004)? Why are individual portfolios so heterogeneous (Guiso,
Haliassos and Japelli, 2002)? Why is participation so low in some specific assets
(Curcuru et al., 2005)?
Recently, in the literature on portfolio diversification, some attention has been paid to
information as a key factor determining individual under-diversification. Investors
may not be informed of the existence and characteristics of all available assets and
invest only in the few that they are aware of or familiar with (Grinblatt and Keloharju,
2001). They may make a selective choice of investment in information, investing only
in a certain group of assets about which they think they have superior information, in
order to maximize their investment in information (Coval and Moskowitz, 1999 and
2001; Hau, 2001), or they may become “specialists” (Nieuwerburgh and Veldkamp,
2004).
Simultaneously, a group of alternative explanations draws attention to the fact that the
attitude towards diversification varies systematically among groups of agents with
different characteristics (Shiller, Kon-Y and Tsutsui, 1991). Portfolio diversification
may vary according to differences in investors’ age (DaSilva and Giannikos, 2004),
competence (Graham, Harvey and Huang, 2005), wealth (Bertaut, 1998; Kumar,
2005), trading experience (Nicolasi, Peng and Zhu, 2004), occupancy (Christiansen,
Joensen and Rangvid, 2005) or the environment in which they live (Goetzmann,
Massa and Simonov, 2004).
This paper combines the last two approaches. Our starting point is that individual
investors may experience greater difficulties in gathering and digesting all pertinent
financial information. Less informed individual investors are worse positioned for

making appropriate financial decisions, possibly achieving a less efficient wealth
allocation as a consequence.
The basic question we seek to answer in this work is to what extent differences in
investors’ portfolio diversification behavior can be explained by differences in their
levels of financial literacy. We also control for socioeconomic differences and
behavioral differences among groups of investors and consider three distinct aspects
of financial literacy: specific financial knowledge, the investors’ educational level
(used as a proxy for their ability to use gathered information) and the sources of
information commonly used by investors as the basis for their financial choices.
For that purpose, we use a very complete database disclosed by the Portuguese
securities commission (CMVM) in May 2005, which includes the results of a survey
aimed at all Portuguese citizens with at least one bank account. Each questionnaire
contains not only socio-economic information about investors, but also reveals the
type and importance of the assets held, investors’ experience, their knowledge of
financial markets and the sources of information used in the decision-making process.
Anticipating some of our conclusions, we can state that, in Portugal, individual
investors reveal remarkable deficiencies at the level of both their general education
and their specific knowledge of financial markets and products. This situation may
have important consequences insofar as both aspects of financial literacy impact on
investors’ behavior. In fact, the educational level, specific knowledge of the market
and information sources used by retail investors to gather information on markets and
financial products have a clear impact on portfolio composition, at least as far as the
number of assets it is composed of is concerned.
This paper makes five contributions. Firstly, we add to the existing literature on
portfolio choice, showing that the influence of investors’ information on financial

diversification has three different dimensions. These three relevant dimensions are
investors’ financial knowledge about products and financial market rules, their
educational level and the sources of information that they commonly use as the basis
for their financial choices.
Secondly, by using a survey instead of actual portfolio data, we contribute towards a
better understanding of the financial behavior of individual investors, the motivations
behind their preferences, their fears and their stimuli. Consequently, this paper
contributes with (non-experimental) empirical arguments, helping to overcome the
lack of data available to confirm that the behavior motivating non-expected utilities is
determinant in individual investors’ financial behavior (Polkovnichenko, 2004).
Thirdly, our results provide some potentially useful evidence on investors’ behavior in
small financial markets. Most studies focus on well-developed financial markets and
very little is known about investor profile, motivations and behavior in smaller and
less-developed markets. By using data from a survey of Portuguese individual
investors we contribute towards bridging this gap.
Fourthly, our results are of major importance from a regulatory perspective. They
justify and reinforce the efforts made by regulatory authorities to increase disclosure
of financial information and to take action with the aim of increasing investors’
financial literacy, thereby reinforcing the efficiency of financial markets. Insufficient
information on the part of investors, for example, cannot be disconnected from issues
such as the importance of unperformed credit liabilities or banking panic or the
emergence of fraudulent financial activities (such as the cases of Enron in the US or
Afinsa and Forum Filatélico in Portugal and Spain).

Finally, understanding the choices made by investors will shed light on the impact
that public policy choices have on the portfolio allocations of households. Examples
include the privatization of Social Security and the taxation of capital income. The
potential effects of these policies greatly depend on the predicted impact on the
portfolio choices of households (Curcuru et al., 2005).
The remainder of the paper is structured as follows: in the next section, we discuss the
measures of financial literacy we use in our work; section 3 presents the model;
section 4 discusses the data, describing both the construction of variables and the
estimation procedure used; section 5 reports the empirical results. A brief conclusion
follows.
2. Measuring Financial Literacy
The impact of education on asset composition has been analyzed in terms of portfolio
theory (Mayers, 1972) and empirically studied (Bradley and Graham, 1988).
However, despite the growing interest and concern regarding the financial education
1
of individual investors, very little is known about how financial literacy affects
investors’ portfolio diversification and which financial literacy features best explain
investors’ behavior. Moreover, there is still no single agreed definition of financial
literacy and even less is there a widely acceptable measure for it
2
. In this work, a
definition is adopted that is similar to the one used by the FunnieMae Foundation
(Vitt et al., 2000): financial literacy is the ability to obtain information, analyze,
manage and communicate about one’s personal financial situation as it affects one’s
material well-being. This concept reflects the ability to collect the relevant
1
OECD has recently published a major study of financial education at the international level (OECD,
2005).
2
See Dougherty (2003) for a discussion of measures of literacy.

information, but also that of differentiating between different financial options,
discussing monetary and financial issues, planning and answering competently to
events affecting daily financial decisions, including those linked to global economic
trends.
As such, we consider three dimensions of financial literacy. The most direct one is the
individual knowledge about financial markets, the way they function, the products
negotiated and their characteristics. Informed investors who are aware of financial
opportunities, choices and consequences are better positioned to increase their well-
being, thus becoming more responsible and financially more self-sufficient.
But the perspective of “investor’s competence” is also very important (Graham,
Harvey and Huang, 2005). In a time of growing economic uncertainty and ever more
complex financial markets, investors are constantly required to make decisions based
on ambiguous and subjective conditions. Moreover, investors’ competence is also
important insofar as decisions taken today will affect their and their families’ future.
Consequently, their general level of education is also fundamental for obtaining a
correct perception of financial information and available opportunities, as well as
being crucial in the decision-making process.
Finally, a certain dimension is linked to the quantity and quality of financial
information that an investor may gather on a regular basis. From this point of view,
the quality of the sources of information commonly used by investors is also a crucial
element. Better-advised individual investors, with more and better information about
general financial issues and products, market agents and new market opportunities,
are better positioned to take superior financial decisions, achieving, as a consequence,
a more efficient and timely wealth allocation.

In order to measure the financial literacy of individual investors under this broad
definition, we consider three dimensions of information: the financial knowledge
revealed by investors in their answers to concrete questions about the financial
market, the investors’ educational level (a proxy for investors’ ability to use the
information gathered), and the sources of information commonly used by investors as
the basis for their financial choices.
3. Model
The central theme of this work is the analysis of the importance of financial literacy
for investors’ behavior, through the analysis of its impact on portfolio diversification.
One key variable in this study is the level of financial knowledge revealed by
individual investors (INFOR). This variable was constructed on the basis of questions
directly geared towards evaluating their level of knowledge about markets and
Portuguese financial products (this methodology is more clearly explained in section
4 below). We seek to apprehend the general level of financial literacy of Portuguese
investors and understand the extent to which it is possible in the community of
Portuguese investors to identify socio-economic groups that can be differentiated on
the basis of their level of financial education. Therefore, a group of socio-economic
factors are identified and we seek to analyze which of these are significantly related
with the investors’ level of financial knowledge. Besides investors’ level of academic
education, we also considered, their gender, age, marital status, household size, level
of income, area of residence, habitat, social status, and occupation/profession.
3
3
See section 4 on the definition of these variables.

Having this goal in mind, a model is defined in which the dependent variable is
INFOR and the explanatory variables are the socio-economic ones
4
. The aim of this
regression is not to establish a cause-effect link between these variables but rather to
use multivariate statistical techniques to properly identify the socio-economic
characteristics of better-informed investors. So, the model being used is:
[1]
ii l l
il
I
NFOR c X EDU
αθ
=+ +
∑∑
where:
i
X
represents the socio-economic control variables, and
l
E
DU
represents the investors’ educational level.
The diversification variable (PORTFDIVST) was constructed bearing in mind both
the diversity of assets in a portfolio (deposits, treasury bills, stocks, bonds, investment
funds or derivatives) and the number of different assets within the same category
(details available in section 4), corresponding basically to what is known as the 1/N
rule for asset allocation. The 1/N diversification rule simply means that investors
allocate 1/N of their wealth to each of the N risky assets available for investment.
The number of different types of assets combined with the number of different assets
of the same type within a portfolio is not the perfect measure for identifying the
degree of diversification. As a matter of fact, two individuals may both hold the same
number of stocks in their portfolios, but one may hold stocks with low correlations
and the other may hold strongly correlated stocks confined to a single industry – the
volatility of these portfolios will certainly differ. In short, this diversification measure
4
Graham, Harvey and Huang (2005) use a similar procedure to model a variable relating to investors’
self-evaluation of their literacy.

is frequently seen as overstating the level of portfolio diversification (Blume,
Crockett, and Friend, 1974; Goetzmann and Kumar, 2005; Vissing-Jorgensen, 1999).
Recent research, however, concludes that this apparently naïve 1/N allocation rule is
far from being an inefficient strategy and therefore the number of stocks in a portfolio
can be seen as a useful heuristic method for identifying the degree of diversification.
According to the findings of DeMiguel, Garlappi and Uppal (2005), when compared
with several static and dynamic models of optimal asset allocation, this rule appears to
have a higher out-of-sample Sharp ratio
5
, a higher certainty-equivalent return
6
and a
lower turnover than optimal asset allocation strategies.
In order to evaluate the importance of financial literacy on individual investors’
portfolio diversification behavior, we considered the above-mentioned three
dimensions of information: the financial literacy (the INFOR variable, already
discussed), the investors’ educational level (the EDUNIV and EDLB variables) and
the sources of information used by investors (INFS1 to INFS6).
Besides these central variables, we controlled for other factors possibly conditioning
investors’ behavior. The investors’ choice of which assets to have in a portfolio can
be influenced by the quality of the market and its different segments. It is necessary,
for instance, for the various types of assets to be negotiated in well-developed, liquid
markets if they are to become effective investment alternatives. Thus, besides the
socio-economic variables already discussed (used as control variables), we also
included as variables designed to explain portfolio diversification the investors’
individual evaluation of the overall quality (MARKEVL) and the overall risk level
(MARKRISK) of the Portuguese market.
5
The Sharp ratio measures the portfolio's excess return relative to the total variability of the portfolio.
6
The certainty-equivalent return is the zero risk return an investor would trade for a given larger return
with an associated
risk.

It is also important to control for some of the specific characteristics of individual
investors. We expect more experienced investors, those more used to financial
matters, to be more skilled at handling a diversified portfolio than investors who are
newcomers to the market. Market experience, represented by the TREX1 to TREX4
variables, was therefore introduced as an explanatory variable. The higher the index
is, the longer is investors’ participation in the market. Being in the market longer
means that, in general, investors have enjoyed success (otherwise they would have
left) and the greater their success, the greater the confidence of investors in their own
abilities and the greater their portfolio diversification: when investors feel confident in
their ability to understand the risks and benefits linked to financial investments, they
are able to invest in new assets, thus diversifying their portfolios (Graham, Harvey
and Huang, 2005).
Investor profile, given by individual investment styles, may also be relevant in their
choice of financial assets. Variables INVT1 to INVT3 were introduced in order to
differentiate investors according to the average length of time (increasing with the
index) that they keep an asset in their portfolio. Investors that are more active, those
that more frequently review the assets that they hold, i.e. those whose assets are held
for shorter periods, are considered by traditional finance theory to be those that accept
a greater risk, so that this could be linked to a higher concentration of assets in the
portfolio (cf. Campbell and Viceira, 1999; Brennan, Schwartz and Lagnado, 1997).
This idea derives from the concept of dynamic diversification, according to which,
generally speaking, above-average returns tend to compensate for below-average
returns if the investment time span is long. On the other hand, and looking at
behavioral finance, the investment time span of individuals affects their own
perception of risk and, consequently, their portfolio diversification. Investors that

reveal a myopic refusal to accept losses are more prone to accepting risks (and
therefore, diversify less) if they evaluate their portfolio less frequently (Thaler et al.,
1997). The fact that individuals are more sensitive to losses than to gains (disposition
effect), together with the cognitive error of equating low probability with null
probability (mental accounting), are additional arguments that converge to conclude
that investors accept more risk (diversify less) if they revise their portfolio less often.
Thus, individual investors’ portfolio diversification was analyzed through the
following model:
[2]
1ii j j l l k k
ij l k
PORTDIVST c X X INFORRES EDU INFS
αβγ θ δ
=+ + + + +
∑∑ ∑ ∑
where:
i
X
represents socio-economic control variables;
j
X
represents other control variables;
INFORRES
represents investors’ financial knowledge;
l
E
DU
represents investors’ educational level: the EDUNIV and EDLB variables;
k
I
NFS
represents the sources of information commonly used by investors: INFS1,
INFS2, INFS3, INFS4, INFS5, INFS6.
4. The database and the variables
CMVM has undertaken several surveys to identify the profiles of individual
Portuguese investors. The original databases of two of those surveys were publicly
disclosed in May 2005. The present study uses the 2000 database, which is the most
recent one.

The 2000 survey was addressed to Portuguese citizens, residing on the Portuguese
mainland, in the Azores and in Madeira, aged over 18 and with at least one bank
account. CMVM, BVLP and Interbolsa workers were excluded. A total of 15039
contacts were made between 2 October and 22 December 2000, stratified by region
and habitat. The direct interview technique was used. The contacts made allowed for
the identification of 1559 investors in securities
7
. All these investors were then
interviewed using a structured questionnaire. Overall, 1268 investors completed the
questionnaire (although some investors may not have answered all questions). Those
questioned were individuals who were responsible or co-responsible for family
investment decisions.
Each questionnaire is composed of 4 parts. The first contains information of a socio-
economic nature: gender, marital status, age, educational level, profession, income
and place of residence. The second includes information relating to the nature, type
and importance of the assets held, experience and type of investor (short, medium or
long-term). The third part refers to investors’ information about markets and their
agents, sources of information used and their evaluation of various aspects related to
the Portuguese securities market. Lastly, the fourth part refers to information about
investors’ behavior: frequency of transactions and information gathering, investors’
concerns about the securities market and their criteria for selecting assets.
Notwithstanding its importance in terms of investors’ literacy, investors’ educational
level in general is not a good proxy for assessing their level of education regarding
financial matters. In fact, unless their educational training is specifically directed to
financial matters, most of the Portuguese curriculum does not include issues related to
7
An investor in securities is one holding one or more of the following assets: shares, bonds, units in
collective investment schemes, títulos de participação (participation certificates) and derivatives.

securities markets, their agents and instruments. It is therefore necessary to find a
better proxy for investors’ literacy.
The survey makes it possible to build up such a proxy. Three of the questions are
particularly useful for attaining this goal. These are question 7, question 11 (combined
with 11A) and question 13. In question 7, investors are asked to name companies with
shares or bonds listed, up to a maximum number of 5. Responses to this answer are
marked from 0 to 5, with 0 meaning that investors fail to mention the name of any
company and 5 meaning that they refer to the name of 5 companies with shares or
bonds listed. In question 11A (and in question 11) investors are asked whether they
know any of the following entities: BVLP, Interbolsa, CMVM, Credit Institutions,
Dealers. Again, the answers are marked from 0 to 5, with 0 meaning that investors are
unaware of these entities and 5 meaning that they know them all. Finally, question 13
is as follows: “If you wish to file a complaint about a financial intermediary, an issuer
or any other entity related with the securities markets, to whom would you address
it?” Answers are marked with 5, if CMVM is mentioned, and with 0 if any other
entity is mentioned, or if no entity at all is identified.
The INFOR variable is the simple arithmetical average of the answers obtained to
questions 7, 11A and 13 and will be used as the proxy for the financial knowledge of
individual investors. INFOR varies between 0 and 5, higher values mean a better
understanding of financial markets.
Table 1: INFOR Variable
INFOR
Number of observations
Frequency
Cumulated frequency
0 30 2.37 2.37
0.333 23 1.81 4.18
0.667 84 6.62 10.8
1 133 10.49 21.29
1.333 159 12.54 33.83

1.667 199 15.69 49.53
2 173 13.64 63.17
2.333 107 8.44 71.61
2.667 65 5.13 76.74
3 53 4.18 80.91
3.333 45 3.55 84.46
3.667 53 4.18 88.64
4 62 4.89 93.53
4.333 43 3.39 96.92
4.667 32 2.52 99.45
5 7 0.55 100
Total 1268 100 100
A very significant number of investors (21.29%) have very limited specific
knowledge (INFOR<=1) and only 11.36% of investors have sound knowledge
(INFOR >=4) – Table 1. About two thirds of investors (71.61%) exhibit what we can
class as negative knowledge (INFOR<2.5).
As previously said, a better understanding of securities markets can result in lower
risk-taking, through portfolio diversification. In the absence of any direct knowledge
of what the investor portfolio comprises precisely (both in terms of the assets included
and their respective weight), one must find a suitable proxy for diversification.
Questions 1, 2 and 2b) are useful for achieving this aim. In question 1, investors are
asked to identity their assets (real estate, bank deposits, savings certificates, treasury
bills, securities or other assets). In question 2, the securities are identified (shares,
bonds, units in collective investment schemes, títulos de participação (participation
certificates) and derivatives). In question 2b), investors claiming to hold shares are
asked to identify the names of the issuers of such shares. Assuming that each asset
and/or security and/or issuer contributes to portfolio diversification (and therefore
similarly reduces the risk), the PORTDIVST variable is used as a proxy for
diversification. Thus, for each investor, PORTDIVST equals the number of assets,

plus the number of securities plus issuers they hold in their portfolio
8
. PORTFDIVST
ranges from 1 to 11, meaning that the most diversified portfolio has 11
assets/securities/issuers.
On average, each portfolio comprises 2.6 assets/securities/issuer (the median is 2), but
a significant number of investors hold only one security/issuer. This contrasts with the
findings of Barber and Oden (2000) for the United States: a typical American investor
has a portfolio of 4 issuers (the median being 3)
9
. This means that most Portuguese
investors’ portfolios are under-diversified.
10
The MARKEVL variable summarizes investors’ evaluation of some of the
characteristics of the Portuguese securities market that may influence their behavior.
Question 9 of the survey requests investors to classify the securities markets in
Portugal, on a scale ranging from 1 (low) to 7 (high), in relation to the following
characteristics: “easy access”, “liquidity” and “level of development”. MARKEVL
was calculated as the simple arithmetical average of the scores attributed to these
three characteristics. MARKEVL ranges from 1 to 7, with higher values meaning a
better evaluation. The sample mean of this variable is 3.97, and the standard deviation
is 0.913.
Another market characteristic that may also influence investors’ behavior towards
diversifying their portfolios is their evaluation of the market risk. Thus, if the market
8
As such, it is assumed that each different asset/security/issuer makes the same contribution to
diversification.
9
Barber and Oden (2000) only report different shares of different issuers, whereas we report the total
number of assets, thus including shares and other securities.
10
Nonetheless, under-diversification does not always mean sub-optimal behavior. Economies of scale
in obtaining and dealing with information may lead investors to specialize in a certain number of assets
or in assets that are closely correlated (Nieuwerburgh and Veldkamp, 2004). But, as their knowledge of
these assets increases, these become less risky to that investor. This is a case of conflict between the
benefits of specialization and those of diversification, where a less diversified portfolio does not
necessarily mean a sub-optimal choice. The rationality of such behavior is supported by studies such as
the one undertaken by Ivkovic, Sialm and Weisbenner (2004), where, on average, those families that
concentrate their financial investments in a small number of assets achieve better results than families
that greatly diversify their portfolios.

is considered too risky, this may rationally lead investors who are less prone to
accepting risks to greatly diversify their portfolios. The MARKRISK variable is based
on question 17: “How would you classify the risk of the Portuguese securities market
on a scale from 1 (very low) to 7 (very high)?”. This variable thus evolves from 1 to
7, its sample mean is 4.68 and the standard deviation is 1.24 (1208 respondents).
Three additional characteristics are considered in our analysis: the information
sources used by the investor, investors’ experience in the securities market and their
respective investment styles. As far as the information sources which investors
commonly resorted to when wishing to obtain information about the securities market
were concerned, the following were mentioned: 1) bank/account manager; 2)
friends/family/colleagues; 3) specialized newspapers; 4) other written press; 5)
television/radio; 6) stock exchange bulletin of quotations; 7) none. The variables from
INFS1 to INFS7 are, therefore, 7 dummy variables, taking the value of 1 when
investors stated that they had used source j (j=1, …, 7) to obtain information about
the securities market.
In order to measure experience, the following question was used: “how long have you
been investing in the securities market?”. The answers were marked as: i) less than a
year, ii) between 1 and 2 years; iii) between 2 and 5 years; and iv) 5 years or more.
Therefore, TREX1, TREX2, TREX3 and TREX4 are dummy variables, taking the
value of 1 if the investor has been investing for less than 1 year, 2 for between 1 and 2
years, 3 for between 2 and 5 years, or 4 for over 5 years, respectively.
Finally, in terms of their investment style, investors were classified as follows: 1) very
short-term, when holding their assets for a maximum period of one month; 2) short-
term, when holding their assets from one month to one year; 3) medium-term, when
holding their assets from 1 to 3 years; and 4) long-term, when holding their assets for

more than 3 years. Accordingly, the variables INVT1, INVT2, INVT3 and INVT4 are
binary variables, taking the value of 1 to 4 respectively, depending on whether the
respondent was a very-short term, short-term, medium-term or long-term investor.
The socio-economic variables are the following:
1. GENDER: binary variable, equal to 1 if the investor is a male or 0 if female;
2. AGE: investor’s age, in years;
3. MARRIED: binary variable, equal to 1 if married or living in a de facto union;
4. FAMSIZE: number of persons in the household;
5. Maximum educational level. This variable was considered under 3 categories:
EDUNIV=1, if the maximum educational level is an intermediate or university
degree; EDLB=1, if the maximum educational level is the 9th or 12th grade; and
EDLSEC=1, if the maximum educational level is below the 9th grade;
6. Net annual household income. Five categories were considered: INC1=1, if net
annual household income is below 14964 €; INC2=1, if equal to or above 14964
€, but below 24938 €; INC3=1, if equal to or above 24938 €, but below 37410 €;
INC4=1, if equal to or above 37410 €, but below 49880 €; and INC5=1, if net
annual household income is above 49880 €;
7. Investor’s area of residence. Six geographical locations were considered:
RCOAST=1, if the investor lives in the northern or central coastal area (except for
Porto and Lisbon); RPORTO=1, if living in the Porto metropolitan area;
RLISBON=1, if living in the Lisbon metropolitan area; RINT=1, if living inland;
RALG=1, if living in the Algarve; and RISL=1, if living in the Azores or Madeira
islands;
8. Investor’s occupation. Five categories were considered: OCCEO=1, if the investor
is the owner/boss; OCHE=1, if the investor is a senior or middle manager, or if the

investor’s profession is a technical, scientific or artistic one; OCIND=1, if the
investor is a liberal professional or an independent worker; OCDEP=1, if the
investor is an office clerk, semi-skilled or unskilled worker; and OCINACT=1, if
the investor is inactive (student or unemployed);
9. Investor’s habitat. The survey considers three types of habitat, scored as follows:
H1=1, if the investor lives in a location of up to 4999 inhabitants; H2=1, if the
investor lives in a location of between 5000 and 19999 inhabitants; H3=1, if the
investor lives in a location of 20000 or more inhabitants;
10. Investor’s social status. The survey considers five categories: STA1=1, if the
investor has a type A status (the lowest); STA2=1, if the investor has a type B
status; STA3=1, if the investor has a type C status; STA4=1, if the investor has a
type D status; and STA5=5, if the investor has a type E status (the highest).
The sample is quite diversified (see Table 2). In fact, the number of investors in each
category is reasonably high. For example, 672 investors claimed to look for
information about the market with their bank or account manager, and 149 stated that
they did not use any source of information whatsoever. On the other hand, 130
investors claimed to have been investing for less than one year, and 285 said that they
had had their assets for more than three years.
Table 2: The sample variables
Mean Standard
Deviation
Total Answers
INFS1 0.542 0.498 672 1240
INFS2 0.152 0.360 189 1240
INFS3 0.268 0.443 332 1240
INFS4 0.202 0.401 250 1240
INFS5 0.456 0.498 565 1240
INFS6 0.114 0.318 141 1240
INFS7 0.120 0.094 149 1240
MARKEVL 3.969 0.913 na 1065

MARKRISK 4.684 1.237 na 1208
TREX1 0.105 0.307 130 1234
TREX2 0.254 0.436 314 1234
TREX3 0.442 0.497 545 1234
TREX4 0.199 0.399 245 1234
INVT1 0.056 0.230 69 1236
INVT2 0.241 0.428 298 1236
INVT3 0.472 0.499 584 1236
INVT4 0.231 0.421 285 1236
GENDER 0.689 0.463 874 1268
AGE 41.587 14.282 na 1263
MARRIED 0.715 0.452 906 1268
FAMSIZE 3.091 1.215 na 1264
EDUNIV 0.326 0.469 412 1263
EDLB 0.470 0.499 594 1263
EDLSEC 0.203 0.400 257 1263
INC1 0.421 0.494 421 1000
INC2 0.324 0.468 324 1000
INC3 0.121 0.326 121 1000
INC4 0.075 0.264 75 1000
INC5 0.059 0.236 59 1000
RCOAST 0.347 0.476 440 1268
RINT 0.263 0.441 334 1268
RPORTO 0.110 0.314 140 1268
RLISBON 0.222 0.416 282 1268
RALG 0.032 0.175 40 1268
RISL 0.025 0.157 32 1268
OCCEO 0.212 0.409 269 1267
OCDEP 0.253 0.435 320 1267
OCHE 0.328 0.470 415 1267
OCINACT 0.108 0.311 137 1267
OCIND 0.099 0.299 126 1267
Notes: 1. The sum of the variables INFS1 to INFS7 is more than 1240 since investors could select more
than one information source; 2. The number of respondents in each group of variables is not the same,
because not all investors answered all the questions.
In short, the survey contains information referring to 1268 investors, although they
may not have answered all questions. As is the case in most surveys, the questions
relating to the household income level were the ones that received fewest answers (in
this case, only 1000 responses).
5. Results
5.1. Investors’ financial knowledge

As already mentioned, the INFOR variable is used in this study as an indicator of the
specific level of information shown by investors about matters of a financial nature.
The values obtained for this variable are shown in Table 1; the conclusion is that most
investors have highly insufficient knowledge about the market and financial products.
In order to find out whether there is a typical investor’s profile for the different levels
of financial knowledge, Model [1] was estimated by the ordinary least squares
method. The estimation results are shown in Table 3. It is interesting to note that:
a) Men are better informed than women;
b) Married investors or those living in a de facto union are less well-informed;
c) The level of information is at its highest in investors of around 44 years of age;
d) Household size is not relevant;
e) Investors’ financial knowledge is greater if they have completed an intermediate
or university course of education;
f) The higher the average income level the better the level of information;
g) Investors located in the coastal area, in the Porto metropolitan area and in the
islands have greater knowledge than those located in the Lisbon metropolitan area
and in the other regions;
h) Neither habitat nor status are relevant;
i) Liberal professionals and non-specialised employees have a higher level of
information.
The influence of these variables is particularly evident in the quartile of better
informed investors
11
(i.e. in the 4th quartile of the INFOR variable). In the first 3
quartiles, there are no significant differences to be pointed out.
11
These results are not reported.

Table 3: Model [1] estimation results
Dependent Variable: INFOR
Estimation Method: OLS
Number of observations: 993
R
2
=0.1808
Variable Coefficient t-stat
C 0.149 0.342
GENDER 0.523 6.622 ***
AGE 0.046 2.921 ***
AGE*AGE -0.001 -3.061 ***
MARRIED -0.269 -2.884 ***
FAMSIZE 0.044 1.328
EDUNIV 0.776 3.889 ***
EDLB 0.424 2.749 ***
INC2 0.019 0.235
INC3 0.271 2.318 **
INC4 0.132 0.898
INC5 -0.063 -0.379
RCOAST 0.240 2.192 **
RINT -0.084 -0.714
RPORTO 0.412 3.125 ***
RALG -0.563 -1.620
RISL 0.565 2.627 ***
H2 -0.106 -0.902
H3 -0.146 -1.430
STA2 -0.060 -0.565
STA3 0.170 1.034
STA4 0.042 0.172
STA5 -0.122 -0.379
OCCEO 0.106 0.556
OCHE 0.136 0.817
OCIND 0.528 3.103 ***
OCDEP 0.288 2.049 **
Notes: ***, **: significant at the 1% and 5% significance levels, respectively.
5.2. The importance of financial literacy for portfolio diversification
Model [2] was used to discover the importance of financial literacy for the portfolio
diversification of individual investors. The dependent variable (PORTFDIVST)
assumes only non-negative integer values, as it is a count model. The estimation was
therefore made by the QML (quasi maximum likelihood) method, assuming an
exponential distribution, and with the variance-covariance matrix being estimated
through the Huber/White method
12
.
12
EViews 5.0 was the software used.

Considering the characterisation of the INFOR variable and in order to avoid
multicollinearity, Model [1] estimation residuals (called INFORRES) were used in
Model [2]. By construction, INFORRES is orthogonal to the socio-economic
variables.
Table 4: Model [2] estimation results
Dependent Variable: PORTFDIVST
Estimation Method: QML - Exponential Count (Quadratic hill-climbing)
Number of observations: 778
QML (Huber/White) standard errors & covariance
Variable Coefficient z-stat
C 0.042 0.138
GENDER 0.117 2.582 ***
AGE -0.004 -0.398
AGE*AGE 0.000 0.694
MARRIED 0.054 1.031
FAMSIZE -0.001 -0.075
EDUNIV 0.216 1.929 *
EDLB 0.142 1.690 *
INC2 0.032 0.671
INC3 0.142 2.105 **
INC4 0.015 0.194
INC5 0.179 1.865 *
RCOAST -0.036 -0.506
RINT 0.005 0.068
RPORTO 0.171 2.370 **
RALG -0.172 -1.050
RISL -0.042 -0.385
H2 -0.146 -1.972 **
H3 -0.148 -2.461 **
STA2 0.038 0.606
STA3 0.025 0.263
STA4 0.097 0.736
STA5 0.170 0.901
OCCEO 0.272 2.370 **
OCHE 0.078 0.750
OCIND 0.217 2.072 **
OCDEP 0.090 1.061
INFOR_RES 0.048 2.484 **
INFS1 -0.042 -1.011
INFS2 0.184 3.657 ***
INFS3 0.197 4.353 ***
INFS4 0.046 0.997
INFS5 0.073 1.704 *
INFS6 0.151 2.793 ***
MARKEVL 0.054 2.438 **
MARKRISK -0.016 -0.920
TREX2 0.146 2.117 **
TREX3 0.342 4.719 ***

TREX4 0.518 5.791 ***
INVT1 -0.112 -1.112
INVT2 -0.100 -1.494
INVT3 -0.036 -0.622
Adjusted R-squared 0.220 LR statistic (41 df) 75.859
Obs: ***, **, *: significant at the 1%, 5% and 10% significance levels, respectively.
The most important result is that, in view of all the aspects considered, financial
literacy does indeed matter. Specific financial knowledge shown by investors about
matters relating to the securities markets, measured by the INFORRES variable,
contributes towards an increase in the number of assets comprising the investor’s
portfolio, at least at the 95% confidence level. Features relating to the educational
level of investors are also important for understanding their attitude towards
diversification. Therefore, at least at the 90% confidence level, it can be concluded
that the higher the investor’s educational level, the greater the number of assets
included in the portfolio. This means that the level of education contributes towards
increasing diversification and therefore decreasing the portfolio risk. In other words,
both at a general level of education and a specific level of expertise, the investor’s
financial literacy is a positive factor in decreasing risk.
Finally, as far as the sources of information are concerned, we conclude that these
sources can be grouped into two. On the one hand are the sources that contribute
towards an increase in diversification. This group includes specialised newspapers
(INFS3) and the stock exchange bulletin of quotations (INFS6), but also the advice of
friends, colleagues and family (INFS2) and, residually (with only 10% significance),
radio and television (INFS5). The other group is comprised of non-relevant sources,
such as the other written press (INFS4) and advice from the bank/account manager
(INFS1). In other words, collecting information from the bank/account manager on

issues related with the securities market does not seem to significantly contribute
towards an increase in portfolio diversification. Both those who do not collect any
information (INFS7) and those who collect information from the account manager
behave similarly in terms of diversification, having on average less diversified
portfolios.
As far as market characteristics are concerned, “access, liquidity and level of
development” (the MARKEVL variable) also contribute towards increasing portfolio
diversification. Two possibilities appear in opposition to one another here. A more
easily accessed, more liquid and more highly developed market attracts more
investors, who invest in more assets. But those very same characteristics can also lead
investors to engage in riskier behavior. In fact, a more easily accessed, more liquid
and more highly developed market may give investors greater confidence and
encourage them to diversify less, as they remain aware of the possibility of easily
quitting the market in a crisis situation. Our results suggest that the first effect is
stronger than the second one.
Curiously, investors’ evaluation of the risk of the Portuguese securities market
(MARKRISK) is not a determinant factor of diversification. In fact, the MARKRISK
coefficient does not prove to be statistically significant, even being negative. In other
words, the fact that only 4% of investors classified the Portuguese market risk as low
(MARKRISK = 1 or 2) whereas 25% classified it as high (MARKRISK = 6 or 7) does
not seem to have had an impact on the composition of investors’ portfolios.
In contrast to the above findings, investors’ experience (TREX) was seen to
contribute towards diversification. Therefore, it can be concluded that the greater the
investors’ experience the higher the number of assets in their portfolios. Respondents
who had invested in the securities market for more than 5 years were those who had

more diversified portfolios. It was not possible to obtain information relating to the
amount (and volume) of transactions effectively carried out by investors in order to
infer whether the reported experience effect resulted from a greater actual experience
in transactions or from the acquisition of assets at different moments in time, which
are then “stored” as the investor does not need liquidity. Consequently, this remains
an issue that requires further research.
The investors’ style of investment (the average time investors keep assets in their
portfolios) does not seem to have an impact on diversification. Finally, as far as socio-
economic variables are concerned, men diversify more than women, intermediate
(INC3) and high (INC5) income levels are associated with greater diversification,
investors from the Porto metropolitan area diversify more than the others, as do
investors located in habitats with up to 5000 inhabitants, together with CEOs and
liberal professionals.
6. Conclusions
Investors’ financial information and financial literacy have lately been emerging as
particularly relevant factors for explaining investor behavior. In this paper, we sought
to identify those factors that influence the level of financial knowledge of Portuguese
individual investors and to investigate the relationship between financial literacy and
the behavior of agents, by focusing on portfolio diversification.
The results reported lead us to conclude that there is a general problem of a low level
of information amongst individual Portuguese investors. In fact, two out of three
investors show that they have an insufficient level of specific knowledge about
financial issues. It can also be concluded that married men of around 44 years of age,
with an intermediate or higher educational level, living in the coastal area or in the

Porto metropolitan area and who are liberal professionals, are those with a higher
level of information.
Our results also show that the portfolios of Portuguese investors are generally under-
diversified: the average number of assets in the portfolio is 2.6, and a significant
number of investors hold only one security. This finding is consistent with the
findings of other studies for European countries (Eymann and Borsch-Supan, 2002),
and reveals an even greater problem of under-diversification than the US evidence
shows (Barber and Oden, 2000).
Our results show that financial literacy matters as far as diversification behavior is
concerned. It is possible to infer that both the level of specific financial knowledge
and the general educational level of investors have an influence on the number of
different assets that comprise their portfolios.
Finally, the sources of information used by investors are also significant in
determining the composition of their portfolios. As a matter of fact, many individual
investors seek help from their bank/account managers, trusting in their advice.
However, in terms of the number of different assets that comprise their portfolios,
these investors behave similarly to those who do not seek any information at all,
which is in itself a surprising result. The great importance of the relationships that
retail investors have with the market through their account managers means that these
professionals need to be suitably trained. This allows them to provide investors with
clear, simple and accurate information, helping them to make a reasoned and well-
informed decision. In addition, information ought to be given to investors when they
most need it. This usually happens when investors are confronted with the need to
take a decision, i.e. when the information is relevant and immediately applicable.

From a regulatory point of view, these results are of major importance. On the one
hand, understanding the choices made by investors will shed light on the impact that
public policies can have on the portfolio allocation of households. This is not only
important for the taxation of capital income, but it is also particularly vital at a time
when European countries are undertaking reforms to introduce a more privatized form
of social security and health care. The potential effects of these policies greatly
depend on their expected impact on the portfolio choices of households (Curcuru et
al., 2005).
On the other hand, the significance of the level of individual investors’ financial
literacy (not only for themselves, but also for the better functioning of the markets and
for society in general) justifies and reinforces the efforts made by regulatory
authorities to increase the disclosure of financial information. In fact, if individual
investors behave in a less diversified manner when they have lower levels of general
and specific knowledge, the more important it becomes to promote and develop
programs to improve the financial literacy of (present and future) investors. Such
programs should take into account the socio-economic characteristics of those at
whom they are aimed and specific programs should be designed to meet the
requirements of their target audience. Circulation of information is fundamental, since
results show that information sources have an influence on investors’ behavior.
Young people should be regarded as a priority target audience, since the constant
innovation in financial markets, the increasing complexity of such markets and the
different investors’ needs in view of their age, require each person to be familiar with
issues of a financial nature from an early age. As their mission is specifically to
protect investors, regulators can play a fundamental role in this field. Take, for

example, the French AMF
13
, which recently created an Institute for Public Financial
Literacy (Institut Pour L’Éducation Financière du Public). Such a move clearly
underlines the pertinence of this subject.
13
The Autorité dés Marchés Financiers is the supervisory and regulatory authority of the French
securities market.

Appendix: Variable Definition
Variable
Definition
PORTDIVST
proxy for
diversification
Number of different assets, plus the
number of securities, plus the number of
issuers in the portfolio. PORTFDIVST
varies between 1 and 11, meaning that the
most diversified portfolio has 11
assets/securities/issuers.
Financial literacy variables
INFOR
proxy for the financial
knowledge of
individual investors
created by the authors, varies between 0
and 5. Higher values mean a better
understanding of financial markets.
INFORRES
proxy for the financial
knowledge of
individual investors
created by the authors. Higher values
mean a better understanding of financial
markets.
EDUNIV
dummy variable, equal to 1 if the
maximum educational level is an
intermediate or university degree
EDLB
dummy variable, equal to 1 if the
maximum educational level is the 9
th
or
the 12
th
grade
Educational
level
This variable has
been considered
under 3 categories
EDLSEC
dummy variable, equal to 1 if the
maximum educational level is below the
9
th
grade
INFS1
dummy variable, equal to 1 when the
investor states having used advice from
the account/bank manager to obtain
information about the securities market
INFS2
dummy variable, equal to 1 when the
investor states having used advice from
friends/family/colleagues to obtain
information about the securities market
INFS3
dummy variable, equal to 1 when the
investor states having used specialized
newspapers to obtain information about
the securities market
INFS4
dummy variable, equal to 1 when the
investor states having used other written
press to obtain information about the
securities market
Sources of
information
This variable has
been considered
under 7 categories
INFS5
dummy variable, equal to 1 when the
investor states having used
television/radio to obtain information
about the securities market

INFS6
dummy variable, equal to 1 when the
investor states having used stock
exchange bulletin of quotations to obtain
information about the securities market
INFS7
dummy variable, equal to 1 when the
investor states not having used any source
to obtain information about the securities
market
Socio-economic variables
GENDER
binary variable, equal to 1 if the investor
is a male
AGE investor’s age, in years
MARRIED
binary variable, equal to 1 if married or
living in a de facto union
FAMSIZE number of persons in the household
INC1
binary variable, equal to 1 if net annual
household income below 14964 €
INC2
binary variable, equal to 1 if equal to or
above 14964 € but below 24938 €
INC3
binary variable, equal to 1 if equal to or
above 24938 € but below 37410 €
INC4
binary variable, equal to 1 if equal to or
above 37410 € but below 49880 €
Net annual
household
income
Five categories
have been
considered
INC5
binary variable, equal to 1 if the net
annual household income is above 49880
€
RCOAST
binary variable, equal to 1 if the investor
lives in the northern or central coastal area
(except for Porto and Lisbon)
RPORTO
binary variable, equal to 1 if living in the
Porto metropolitan area
RLISBON
binary variable, equal to 1 if living in the
Lisbon metropolitan area
RINT binary variable, equal to 1 if living inland
RALG
binary variable, equal to 1 if living in the
Algarve
Area of
residence
Six geographical
locations have been
considered
RISL
binary variable, equal to 1 if living in the
Azores or Madeira islands
OCCEO
binary variable, equal to 1, if the investor
is the owner/boss
OCHE
binary variable, equal to 1 if the investor
is a senior or middle manager, or if the
investor’s profession is a technical,
scientific or artistic one
Occupation
Five categories
have been
considered
OCIND
binary variable, equal to 1 if the investor
is a liberal professional or an independent
worker

OCDEP
binary variable, equal to 1 if the investor
is an office clerk, semi-skilled or unskilled
worker
OCINACT
binary variable, equal to 1 if the investor
is inactive (student or unemployed)
H1
binary variable, equal to 1 if the investor
lives in a location of up to 4999
inhabitants
H2
binary variable, equal to 1 if the investor
lives in a location of between 5000 and
19999 inhabitants
Habitat
Three types of
habitat
H3
binary variable, equal to 1 if the investor
lives in a location of 20000 or more
inhabitants
STA1
binary variable, equal to 1 if the investor
has a type A status (the lowest)
STA2
binary variable, equal to 1 if the investor
has a type B status
STA3
binary variable, equal to 1 if the investor
has a type C status
STA4
binary variable, equal to 1 if the investor
has a type D status
Social status
The questionnaire
considers five
categories
STA5
binary variable, equal to 1 if the investor
has a type E status (the highest)
Other control variables
MARKEVL
Investor’s evaluation
of access, liquidity
and the level of
development of the
Portuguese securities
market
Calculated as the simple arithmetical
average of the scores attributed to these
three characteristics. MARKEVL ranges
from 1 to 7, higher values mean a better
market evaluation
MARKRISK
Investor’s evaluation
of the market risk
Varies from 1 to 7, higher values mean a
better market risk evaluation
TREX1
dummy variable, equal to 1 if the investor
has been investing for less than 1 year
TREX2
dummy variable, equal to 1 if the investor
has been investing for between 1 and 2
years
TREX3
dummy variable, equal to 1 if the investor
has been investing for between 2 and 5
years
Trade
Experience
Four categories
have been
considered
TREX4
dummy variable, equal to 1 if the investor
has been investing for over 5 years
Investment style
Four categories
hb
INVT1
binary variable, equal to 1 if it is a very-
short term investor: holds the assets for a
maximum period of 1 month

INVT2
binary variable, equal to 1 if it is a short
term investor: holds the assets from one
month to one year
INVT3
binary variable, equal to 1 if it is a
medium term investor: holds the assets
from one to three years
INVT4
binary variable, equal to 1 if it is a long
term investor: holds the assets for more
than 3 years

References
Banks, James, Sarah Tanner (2002), “Household Portfolios in United Kingdom”, in
Household Portfolios, Edited by Luigo Guiso, Michael Haliassos and Tullio Jappelli,
MIT Press, ISBN 0-262-07221-1, 219-250.
Barber, Brad e Terrance Odean (2000), “Trading is Hazardous to Your Wealth: The
Common Stock Investment Performance of Individual Investors”, Journal of Finance
55, 773-806.
Barber, Brad, Chip Heath e Terrance Odean (2003), “Good Reasons Sell: Reason-
Based Choice Among Group and Individual Investors in the Stock Market”,
Management Science, Vol. 49, No. 12, December 2003, 1636–1652.
Bertaut, Carol C. (1998), “Stockholding Behavior of U.S. Households: Evidence from
the 1983-1989 Survey of Consumer Finances”, The Review of Economics and
Statistics, vol.80, nº2, 263-275.
Blume, M. E., J. Crockett and I. Friend (1974), “Stock Ownership in the United
States: Characteristics and Trends”, Survey of Current Business, 54(11), 16-40.
Bradley, Michael G., John W. Graham (1988), “Education and Asset Composition”,
Economics of Education Review, vol.7, nº2, 209-220.
Brennan, Michael, Eduardo Schwartz e Ronald Lagnado (1997), “Strategic Asset
Allocation”, Journal of Economic Dynamics and Control, 21, 1377-1403.
Calvet, Laurent E., John Y. Campbell, Paolo Sodini (2006), “Down or Out: Assessing
The Welfare Costs of Household Investment Mistakes”, NBER Working Paper
nº12030, February.
Campbell, John (2006), “Household Finance”, NBER Working Paper 12149, March.
Campbell, John Y., Luis Viceira (1999), “Consumption and Portfolio Decision When
Expected Returns Are Time Variying”, Quarterly Journal of Economics 114, 433-495.
Christiansen, C., J. Schröter Joensen, J. Rangvid (2005), “Do More Economists Hold
Stock?”, Centre for Analytical Finance Working Paper Series nº206, April, University
of Aarhus, Aarhus School of Business.
Coval, Joshua D., Tobias J. Moskowitz (1999), “Home Bias at Home: Local Equity
Preference in Domestic Portfolios", Journal of Finance, 1999, 54(6), 2045-73.

Coval, Joshua D., Tobias J. Moskowitz (2001), “The Geography of Investment:
Informed Trading and Asset Prices", Journal of Political Economy, 2001, 109(4), 811-
41.
Curcuru, Stephanie, John Heaton, Deborah Lucas, Damien Moore (2005),
“Heterogeneity and Portfolio Choice: Theory and Evidence”, in Handbook of
Financial Econometrics, Edited by Yacine Ait-Sahalia and Lars Peter Hansen, North-
Holland, ISBN-13: 978-0-444-50897-3.
DaSilva, Amadeu, Christos Giannikos (2004), “Higher Risk Aversion in Older
Agents: Its Asset Pricing Implications”, Paper presented at the Financial Management
Association 2005 Conference, Chicago.
DeMiguel, Victor, Lorenzo Garlappi, Raman Uppal (2005), “How Inefficient is the
1/N Asset-Allocation Strategy?”, CEPR Working Paper 5142, July.
Dougherty, Christopher (2003), “Numeracy, Literacy, and Earnings: Evidence from
the National Longitudinal Survey of Youth”, Economics of Education Review, 22,
511-521.
Eymann, Angelika, Axel Borsch-Supan (2002), “Household Portfolios in Germany”,
in Household Portfolios, Edited by Luigo Guiso, Michael Haliassos and Tullio
Jappelli, MIT Press, ISBN 0-262-07221-1, 291-340.
French, Kenneth, R., James M. Poterba (1991), “Investor Diversification and
International Equity Markets”, The American Economic Review, May, 81, 2.
Glaser, Markus, Markus Nöth, and Martin Weber (2004), “Behavioral Finance”, in
Blackwell Handbook of Judgment and Decision Making, Edited by Derek J. Koehler
and Nigel Harvey, cap.26, ISBN: 9781405107464.
Goetzmann, William N., Alok Kumar (2001), “Equity Portfolio Diversification”,
NBER Working Paper 8686, December.
Goetzmann, William N., Alok Kumar (2005), “Why Do Individual Investors Hold
Under-Diversified Portfolios?”, Society for Economic Dynamics in its series 2005
Meeting Papers, nº 77.

Goetzmann, William N., Massimo Massa, Andrei Simonov (2004), “Portfolio
Diversification, Proximity Investment and City Agglomeration”, CEPR Discussion
Paper nº4786, December.
Graham, John R., Campbell R. Harvey, Hai Huang (2005), “Investor Competence,
Trading Frequency, and Home Bias”, NBER Working Paper 11426, June.
Grinblatt and Keloharju, 2001, “How Distance, Language, and Culture Influence
Stock Holdings and Trades”, Journal of Finance, 56, 1053-1073.
Guiso, Luigi, Tullio Jappelli (2002), “Household Portfolios in Italy”, in Household
Portfolios, Edited by Luigo Guiso, Michael Haliassos and Tullio Jappelli, MIT Press,
ISBN 0-262-07221-1, 251-290.
Guiso, Luigi, Michael Haliassos, Tullio Jappelli (2002),
Household Portfolios, MIT
Press, ISBN 0-262-07221-1.
Hau, Harald (2001), “Location matters: An examination of Trading Profits”, Journal
of Finance, 56 (5), October, 1959-1983.
Ivkovic, Zoran, Clemens Sialm, Scott Weisbenner (2004), “Portfolio Concentration
and Performance of Individual Investors”, NBER Working Paper No. 10675, August.
Kumar, Alok (2005), “Who Gambles in The Stock Market?”, Paper presented at the
European Finance Association 2005 Conference, Moscow.
Mayers, D. (1972), “Nonmarketable Assets and Capital Market Equilibrium Under
Uncertainty”, in Studies in Theory of Capital Markets, Edited by Michael Jensen,
Praeger, New York, 119-149.
Merton, Robert. (1969), “Lifetime Portfolio Selection Under Uncertainty: The
Continuous-Time Case”, Review of Economics and Statistic, vol. 51, 247-257.
Nicolosi, Gina, Liang Peng, Ning Zhu (2004), “Do Individual Investors Learn From
Their Trading Experience?”, Yale ICF Working Paper nº03-32, March.
Nieuwerburgh, Stijn Van, Laura Veldkamp (2004), “Information Acquisition and
Portfolio Underdiversification”, Working Paper, Stern School of Business, New York
University, September.
OECD (2005), Improving Financial Literacy, OECD, 177 p., ISBN: 9264012567.

Polkovnichenko, Valery (2004), “Household Portfolio Diversification: A Case for
Rank Dependent Preferences”, The Federal Reserve Bank of Minneapolis, August
2004.
Rendleman, Richard J., Douglas A. Shackelford (2003), “Diversification and the
Taxation of Capital Gains and Losses”, NBER Working Paper 9674, May.
Samuelson, P.A. (1969), “Lifetime Portfolio Selection by Dynamic Stochastic
Programming”, Review of Economics and Statistics, vol. 51, 239-46.
Shiller, Robert J., Fumiko Kon-Y, Yoshiro Tsutsui (1991), “Speculative Behavior in
the Stock Markets: Evidence from the United States and Japan”, NBER Working
Paper 3613, February.
Thaler, Richard, Amos Tversky, Daniel Kahneman, Alan Schewartz (1997), “The
Effect of Myopia and Loss Aversion on Risk Taking: An Experimental Test”,
Quarterly Journal of Economics, 112, 647-661.
Vissing-Jorgensen, Annette (1999), Discussion from the NBER Macroeconomics
Annual 1999 of the paper "Stock Prices and Fundamentals" by John Heaton and
Deborah Lucas.
Vitt, Lois A., Carol Anderson, Jamie Kent, Deanna M. Lyter, Jurg K. Siegenthaler,
Jeremy Ward (2000), Personal Finance and the Rush to Competence: Financial
Literacy Education in the U.S., Institute for Socio-Financial Studies, FunnieMae
Foundation.