Does self-esteem influence financial behavior?

(1)

Does self-esteem influence financial

behavior?

Thomas Heuzeveldt

s1749536 t.f.a.heuzeveldt@student.rug.nl

MSc Finance

Faculty of Economics and Business

Supervisor: Dr. Artūras Juodis

Abstract

Alongside financial literacy, responsible financial behavior is being driven by an individual’s psychological traits. Having a high self-esteem might help individuals to participate, and keep an active role in the intimidating process of complex financial decision making. This study investigates the direct, and indirect effects through subjective financial literacy, of self-esteem on financial behavior, conditional on objective financial literacy.. Results of a path analysis on a Dutch representative dataset indicate self-esteem to directly relate to three out of eight measures of financial behavior. Indirect effects of self-esteem on financial behavior were found to be insignificant. An individual’s income and education are factors that have the most explanatory power for these relationships.

(2)

1

1. Introduction

Over the last two decades individuals are faced with an increasing array of financial decisions. Structural reforms of social security and pension, as well as market liberalization cause the decision making process to be transferred from governments and employers to individuals. Therefore individuals are more responsible for their own decisions to ensure long-term financial security (Van Rooij, Lusardi, and Alessi, 2011a). Furthermore individuals are more and more active in financial markets, where they face an increasing catalog of sophisticated complex financial products which are difficult to understand

Most individuals lack understanding of basic financial concepts, and financial illiteracy is widespread amongst them (Hilgert, Hogarth, and Beverly, 2003; Volpe, Chen, and Liu, 2006; Lusardi and Mitchell, 2007a). Consequently individuals tend to make sub-optimal decisions regarding their wealth. Lusardi and Tufano (2009), show that illiterate individuals borrow more expensively and take on excessive debt. Furthermore, financial illiteracy is found to limit the ability of individuals to plan for retirement, and to accumulate wealth (Lusardi and Mitchel, 2007b; Lusardi, 2008). To address these inefficiencies much research has been done to investigate the relationship between objective financial literacy and financial behavior (see, e.g., Hilgert, Hogarth, and Beverly, 2003; Van Rooij, Lusardi and Alessi, 2011b, 2012). Subsequently, other studies have been conducted aimed at improving financial literacy by educational programs (see, e.g., Lusardi and Mitchell, 2014; Sherraden and Grinstein-Weiss, 2015). These studies imply that improved financial literacy will result in more effective decision making. However Braunstein and Welch (2002) suggest that improved decision making doesn’t automatically follow from a higher financial literacy. So while improved financial literacy is important, the question remains how exactly it affects financial wellbeing (Robb and Woodyard, 2011).

(5)

4 Judge, Erez and Bono, 1998; Judge and Bono, 2001). One of the few studies that linked self-esteem to financial behavior is that of Yurichsin and Johnson (2004), where they linked self-esteem to compulsive buying behavior.

Because financial decision making is such a complex task, it can be intimidating for individuals to engage in responsible financial behavior. When encountering short-term financial set-backs, or distractions, individuals are expected to engage in sub-optimal long-term responsible financial behavior. So an individual has to be equipped with both financial literacy and a positive self-perception to effectively make financial decisions (Owens, 1993; Baumeister et al., 2003).

One can divide financial literacy in an objective and a subjective part. Objective literacy is considered as what a person knows, and subjective as what an individual thinks he knows. Studies have concluded that although an individual might have sufficient objective financial knowledge, he or she will not engage in efficient financial behavior because of low subjective financial knowledge (Parker et al., 2012; Hadar, Sood, and Fox, 2013; Asaad, 2015). Studies investigating indicators for this gap between objective and subjective literacy are almost non-existent.

Recently the first to study the relationship between self-esteem and financial behavior directly and indirectly through financial literacy, are Tang and Baker (2016). Using a national representative dataset of U.S. adults they find self-esteem to significantly relate to financial behavior after controlling for objective financial literacy and other socioeconomic factors. This link can be direct or indirect through subjective financial literacy. Therefore on might conclude that self-esteem seems to be an important factor in explaining financial behavior. This might have important implications for educational programs for improving financial decision making, where the focus mainly relies on improving objective financial literacy. Psychological traits such as self-esteem, which seem to have an impact on subjective financial literacy, should not be overlooked. Because research on this subject is still very limited, it is important to test whether the implied relationship between self-esteem and financial behavior can be confirmed for other population samples.

(6)

5 Furthermore I make use of a dataset representing a different population to test the validity of the relationship.

Therefore the core of this study is to determine whether this relationship between self-esteem and financial behavior holds when testing for other covariates, and a different population. This leads to the following research question:

Does self-esteem influence the financial behavior of Dutch individuals, directly or indirectly through subjective financial literacy?

Correlation results indicate self-esteem to correspond with responsible financial behavior for six out of eight measures. My results of a path analysis indicate a direct relationship between an individual’s self-esteem and one’s decision to invest in financial assets, have savings and the amount one saves.. Income and education are factors of most influence in these relationships. Furthermore I find no indication for self-esteem to indirectly affect financial behavior through subjective financial literacy. The remainder of this study is structured as follows: Section 2. provides a review of the current literate on this subject, leading to my research hypotheses. Section 3. describes the methodology of testing the research hypotheses. Section 4. describes the data acquiring process, and the construction of measures. Section 5. provides the results. Section 6. provides a discussion, limitations and recommendations. Section 7. gives the conclusion and implications of my study.

2. Literature review

2.1 Objective financial literacy and financial behavior

The current bodyof literature predominantly assumes a positive relation between objective financial literacy and responsible financial behavior, which might ultimately result in wealth accumulation for individuals.

(7)

6 levels at retirement. The same result can be found in other countries, for instance in the Netherlands (see, e.g., Van Rooij, Lusardi, and Alessie, 2009). Furthermore advanced financial literacy, and being able to perform calculations, seem to have the greatest impact on responsible retirement planning (Lusardi and Mitchell 2011b; Alessie, van Rooij, and Lusardi 2011; Fornero and Monticone, 2011). Knowledge for instance about risk diversification, asset valuation, investment fees and portfolio choice can be regarded as advanced financial literacy.

Secondly, financial literacy has a profound effect on the borrowing behavior of individuals. Lusardi and Tufano (2009) find debt illiterate American individuals to incur higher transaction costs, higher fees and make use of higher cost borrowing. Furthermore this lack of debt literacy leads individuals to take on excessive debt, and have an inability to correctly judge their debt position. This result is partly backed up by Mottola (2013), who conclude that financial illiterate individuals engage in more costly credit card behavior.

Stango and Zinmann (2009) examined the relation between exponential growth bias (the tendency to linearize exponential functions) and the borrowing behavior of household individuals. They conclude that an inability to estimate interest payments and future values leads to an increase in borrowing, and a decrease in wealth accumulation.

Decisions regarding mortgages are also affected by the amount of financial literacy. For instance Moore (2003) finds low literate individuals to be more likely to engage in high cost mortgages. Also individuals with lower income and less education (which can be regarded as factors related to financial illiteracy), are found to be less able to roll over their mortgages when interest rates increase (Campbell, 2006).

(8)

7 these factors, which leads to an increasing stock market participation (Van Rooij, Lusardi and Alessi, 2011). Therefore financial literacy increases the likelihood of individuals to invest in the stock market.

2.2. Direct effect of self-esteem on financial behavior

Self-Esteem is defined by Rosenberg et al. (1995) as an individual's positive or negative attitude toward his or herself as a totality. It has been widely examined in other psychological studies as a behavioral aspect in relation with various outcomes, such as job performance (Judge,Erez and Bono, 1998; Judge and Bono, 2011). However current literature lacks studies regarding self-esteem in relation to financial outcomes.

One of the few studies that linked self-esteem to financial behavior is that of Yurichsin and Johnson (2004). They used a sample of undergraduates to examine the link between amongst others variables self-esteem and compulsive buying behavior. Their results suggested that compulsive buying behavior is negatively related to self-esteem.

Tang and Baker (2016) expected to find a positive relation between self-esteem and financial behavior. They reasoned that responsible financial management is mainly about goal setting and achievement. Individuals with a high self-esteem are found to be better coping with goals. They achieve a higher amount of goals, experience a higher satisfaction when they make progress towards goals, are less negatively affected when they fail goal achievement, and display a higher amount of goal-pursuit behavior (Di Paula and Campbell, 2002).Furthermore individuals display a higher amount of persistence when they face challenging tasks or experience failures (Sommer and Baumeister, 2002).

(9)

8 savings outside of retirement accounts, investments in risky assets, retirement saving and credit management. The direct effects of self-esteem on financial behavior were found to be significant for saving outside of retirement accounts and investments in risky assets.

2.3. Indirect effects of self-esteem on financial behavior

Subjective financial literacy can be regarded as an important factor in explaining financial behavior. Individuals displaying higher levels of objective financial literacy are suggested to higher evaluate their own level of financial knowledge. Significant positive correlations of 0.16 and 0.14 have been found, which suggest this relationship (Xiao, Serido and Shim, 2014; Tang and Baker, 2016).

Asaad (2015) concludes that individuals with a high level of subjective literacy are more likely to display positive saving and borrowing behavior. Also when objective financial literacy is low, and subjective literacy is high, individuals are more likely to engage in risky financial behavior.

Similar results can be found in the work of Hadar, Sood and Fox (2013). In their studies individuals with a higher level subjective financial literacy were more likely to invest in risky assets and retirement savings plans. Most striking is the result that individuals allocated money to investment opportunities over which they had a relatively high subjective knowledge, regardless of their objective knowledge.

(10)

9 I will follow and build upon the work of Tang and Baker (2016) and therefore hypothesize that self-esteem will have a direct influence on responsible financial behavior, conditional on objective financial literacy. Also I will hypothesize that self-esteem influences the development, as well the self-perception of an individual’s subjective financial literacy, which in turn affects financial behavior. Therefore I will follow the model of Tang and Baker (2016), as depicted in figure 1, and hypothesize that:

H1. Self-esteem has a direct influence on financial behavior, conditional on objective financial literacy and other socio-economic factors.

H2. Self-esteem has an indirect influence on financial behavior, through subjective financial literacy, conditional on objective financial literacy and other socio-economic factors.

(11)

10

3. Data

3.1. Sample

The data used in this study is acquired from the Longitudinal Internet Studies for the Social Sciences, also known as the LISS panel. The panel consists of 4500 Dutch households, which comprises 7000 individuals. It is based on a true probability sample of households which is drawn from the population register by Statistics Netherlands. The LISS core study is a longitudinal study carried out yearly in the form of an online survey by LISS panel members aged 16 and older. The aim of the LISS core study is to capture changes in the lives of Dutch people, and their reactions to different live events and societal changes. It covers a wide variety of topics such as health, religion and politics, but also topics of interest for this study such as personality, and economic situations. Each set of questions is practically the same for each year. The LISS data website provides a SPSS or STATA dataset for each topic of the Core Study, a codebook containing the survey questions, a description of the methodology, and some summary statistics. Upon request the datasets can be acquired by researchers for further analysis. Furthermore a variety of other single wave survey modules regarding different topics are available, which can be combined with the Core Study.

For this study several datasets have been selected. To capture a measure of self-esteem, the sections about personality in the fourth and fifth wave of the Core Study have been used. This survey was administered to 6978 panel members in May, and repeated in June 2011 for non-responses. In total 5321 panel members completed the survey. In May 2012 the survey was repeated for the fifth wave. Panel members who responded in wave 4 received a shorter version of the survey, which did not cover the questions for measuring self-esteem. In wave 5, 1431 new responses have been acquired, and added to the responses of wave 4, which brought up the total to 6752 responses.

(12)

11 Questions concerning financial literacy are not included in the LISS Core Study. However a single wave survey has been conducted in August 2011, concerning a set of question about financial concepts. 6778 panel members have been selected, and 4858 completed the survey (response rate of 71.7%).

Some control variables such as age and position in households are incorporated in the Core Study. Other standard socio-economic control variables such as gender, education and income are available in the form of single datasets. These background variables are collected every month. Other more specific control variables regarding certain psychologic traits of panel members such as carefulness, or self-control are collected from various other single-wave studies.

After collecting, I’ve merged all studies into one dataset, leaving a total of 1987 observations.

3.2 Dependent variables

For the construction of responsible financial behavior, a couple of identifiers are common in the current literature. These are investments in risky assets, the amount of credit card debt, general savings, and savings for retirement (Hilgert, Hogarth, and Beverly, 2003). I’ve selected eight measurements which are expected to account for the major personal finance practices an individual faces.

3.2.1 Investments in risky assets

When measuring investments in risky assets it is common to look at ones investments in stocks and mutual funds. However the LISS Core Study does not differentiate between risky and less risky investments. It asks the respondents whether they invest in financial assets (growth funds, share funds, bonds, debentures, stocks, options, warrants, and so on), and the amount they invested. Therefore I will differ from common practice and consider investments in financial assets to be an indicator for financial behavior, regardless of the amount of risk attached to them.

(13)

12 second variable is the natural logarithm of the total amount invested in financial assets, conditional on having investments in financial assets.

3.2.2 Savings outside of retirement accounts

Savings outside of retirement accounts are constructed out of the following items: Banking accounts or giro accounts (current accounts), savings accounts, term deposit accounts, savings bonds or savings certificates, and bank savings schemes. Respondents are asked whether they possess one of these items, and the total amount they hold in these accounts.

The variable has savings is constructed with a value of 1 if the respondent has any of these savings accounts, and 0 if the respondents doesn’t. A variable is also created for the natural logarithm of the total amount saved, conditional on the respondent having savings.

3.2.3 Debt

Credit management is one of the factors in determining responsible financial behavior. In their study, Tang and Baker (2016) measured credit management in the form of credit card debt, and maxed out credit cards. When measuring credit management in a U.S. sample this is the correct procedure to get a reliable measure, since the U.S payment system is heavily reliant on credit card usage. However in the Netherlands credit card usage is much less common. Out of the 4013 responses in the economic situation core study, only 138 respondents say to have a negative credit card balance. Therefore as a single measure, credit card debt is not reliable enough to measure the extent to which Dutch individuals manage their credit in a responsible fashion.

(14)

13 measure whether respondents have debts, loans or credits, a dummy variable is created with a value of 1 if the respondents has, and 0 if the respondents haven’t.

3.2.4 Savings for retirement

Differences between the Netherlands and the United States also arise when looking at retirement savings. To get a reliable measure for retirement savings, Tang and Baker (2016) investigate the extent to which respondents participate in defined-contribution plans. With defined-defined-contribution plans, participants make active decisions in how much to contribute, and how contributions will be invested. In their sample, 85.74% eligible participants chose to participate in defined-contribution plans However. the Dutch pension system is predominantly a defined-benefit system. Employees are automatically enrolled in their firm’s retirement plans, and they are not able to make active decisions on how much they will contribute, and how their contributions will be invested (van Rooij, Kool and Prast, 2007).

A common way for Dutch individuals to supplement their defined-benefits pension is to participate in endowment insurance policies. Endowment insurance pays out a certain amount to the holder if that person is still alive at a certain date, or sooner when the person passes away (Bikker and van Leuvensteijn, 2008). Premiums paid to endowment insurance have certain tax benefits. Similar products are single-premium insurance policies and life-annuity insurance.

Therefore I created a variable for the natural logarithm of total amount saved in these three products, which serves as a proxy for defined-contribution plans. A dummy variable is created with a value of 1 if the respondent has one of these products and 0 if the respondent has not.

3.3 Explanatory variables

(15)

14 positively charged, and the other half negatively. The Liss Core Study lets the respondent answer on a scale of 1 to 7, where 1 stands for strongly disagreeing and 7 for strongly agreeing to the statements. I reversed the scores on the negatively charged statement. The self-esteem variable is created by adding the separate scores on all statements. High scores indicate a high self-esteem of the respondent. See the appendix for the exact wording of the statements.

Objective financial literacy is being measured by a set of financial questions. These questions cover the main topics an individual faces when making financial decisions. Box 2 in the appendix reports the exact wording of the questions. This practice is known to give a reliable measure for an individual’s basic financial knowledge, and is widely applied in current literature (van Rooij, Lusardi and Alessie, 2009).

The basic financial literacy index is normally constructed out of 5 questions. The single wave study I used, only provided 4 questions for the LISS Panel members. These questions covered the following topics; numeracy, inflation, risk and diversification, and the relationship between bond prices and interest rates. In other studies such as van Rooij, Lusardi and Alessie (2011a), the last two topics are classified as being indicators for advanced financial literacy. Their basic financial literacy index is complemented with questions regarding the time value of money, and the money illusion. Therefore the financial literacy index I constructed is a mixed bag of basic and advanced financial knowledge. I created 2 variables for each question, one for giving the correct answer, and one where the respondent did not know the answer to the question. I also created a variable for the total points the respondent scores on all the questions, ranging from 1-4. This total score will be the index for objective financial literacy.

(16)

15

3.3 Controls

First of all, I will control for basic variables such as age, gender, marital status, origin, education, having kids, and income.

Financial windfalls might affect the self-esteem of respondents, and therefore their financial behavior (Butler, 2014; Epley and Gneezy, 2007). Therefore I created a variable for the question whether respondents have experienced a bankruptcy.

The relationship between financial literacy and financial behavior might be influenced by a couple of personal traits. Careful individuals for instance might have a higher chance of holding stocks, or to save for a buffer in case of a financial windfall (van Rooij, Lusardi and Alessie, 2009). I use the answer to the statement: I think carefully about what I need to do to remain within my budget over the next few months, as a proxy for carefulness.

Patience is a factor which might influence an individual’s saving behavior. Less patient individuals might be inclined to enjoy the benefits of immediate spending instead of saving. I use information about the respondents whether they are heavy smokers or drinkers, as this can serve as a proxy for a short time preference (Fuchs, 1980).

Some individuals might be more inclined to save than others. Therefore I control for saving propensity. Respondents were asked in a single wave study in 2010 how important they find it to put a bit of money away for later on a scale from 1 to 6 (not important to very important), and how difficult they find it to spend money on a scale of 1 to 5 (very easy to very difficult). Both measures will be used to control for saving propensity.

(17)

16

3.4 Summary Statistics

The summary statistics of all variables are shown in table 1. In total 18.52% of the respondents have investments in financial assets, and have an average amount of € 65,207 invested. Of the respondents, 94.21% reported to have savings, with a mean amount of € 40,994 saved. A considerable less amount of respondents decided to save additionally for retirement, 6.56% reported to have endowment insurance with a mean amount of € 37,099 saved. Furthermore 84.76% of the respondent have any form of debt, and average around € 23,074.

Out of a maximum of 70, the self-esteem score amongst respondents averaged at 56.14. The objective financial literacy measure averaged at 2.38 out of 4, while respondents found themselves to have a relatively high financial knowledge with a mean subjective literacy score of 5.13 out of 7.

The final sample consists for 51.03% out of males. Ages range from 19 to 90 years old, with a mean of 56.78 years. 21.59% reports to be married and 74.08% have no children currently in their household. The biggest proportion of respondents said to have intermediate secondary education (25.92%) as their highest education, and higher vocational education (25.26%) follows closely. 8.9% reported university as their highest education. The origin of the respondents is for 88% Dutch, 2.72% first generation western, 2.92% first generation non-western, 4.53% second generation western, and 0.60% second generation non-western.

(18)

17 Table 1.

Summary statistics

Obs Mean Std. Dev. Min Max

Financial behavior Invest in financial assets 1,987 18.52% .39 0 1 Financial assets amount 255 € 65,207 € 155,241 € -200 € 1,400,000 Have savings 1,987 94.21% .23 0 1 Savings amount 1,337 € 40,994 € 238,233 €-150000 € 8135049 Have debt 1,98 84.76% .36 0 1 Debt amount 258 € 23,074 € 44,552 € 12 € 400,000 Have endowment insurance 1,987 6.56% .37 0 1 Endowment insurance amount 242 € 37,099 € 61,121 € 250 € 500,000 Explanatory variables Self-esteem 1,987 56.14 9.49 10 70 Objective financial literacy 1,987 2.38 1.04 0 4 Subjective financial literacy 1,987 5.13 1.16 1 7 Covariates Age 1,987 56.78 14.51 19 90 Male 1,987 51.03% .50 0 1

Net income per month 1,902 € 1,829 4,88 0 € 183,592

Has no children 1,987 74.08% .44 0 1

Married 1,987 21.59% .41 0 1

(19)

18 Primary school 1,987 8.56% .28 0 1 Intermediate secondary education 1,987 25.92% .44 0 1 Higher secondary education 1,987 8.86% .28 0 1 Intermediate vocational education 1,987 22.50% .42 0 1 Higher vocational education 1,987 25.26% .43 0 1 University 1,987 8.91% .28 0 1 Origin Dutch 1,987 88.37% .32 0 1 First generation western 1,987 2.72% .16 0 1

First generation non-western 1,987 2.92% .16 0 1 Second generation western 1,987 4.53% .21 0 1 Second generation non-western 1,987 0.60% .08 0 1 Saving propensity Importance to save 1,987 5.08 1.02 1 6 Difficulty to spend 1,987 2.58 0.84 1 5 Self-control 1,987 2.26 .87 1 5 Carefulness 1,987 4.65 1.33 1 6

(20)

19

4. Methodology

In order to assess the direct and indirect influences of self-esteem on financial behavior I will perform path analyses on the data. Path analysis is a powerful method because it facilitates a researcher to determine the direct and indirect effects of multiple dependent and independent variables simultaneously (Stage, Hasani and Nora, 2004). Although it’s an acceptable tool for making theoretical inferences about relationships based on correlations, one needs to be cautious in making assumptions about causal relationships between variables (Joo and Grable,2004). A direct effect occurs when an independent variable influences a dependent variable. When an independent variable influences a dependent variable through a mediating variable, it’s considered an indirect effect.

A series of eight path analyses will be performed, where each path model will correspond to the one of the eight measures of financial behavior. Invest in financial assets, have savings, have debt and have endowment insurance are dichotomous variables. These variables are specified to have a logit relationship with the explanatory variables. The other four variables, log(amount of financial assets), log(amount of savings), log(amount of debt) and log(amount of endowment insurance) are continuous and are specified to have a linear relationship with the explanatory variables.

To determine the path coefficients of my four continuous models I will perform structural equation modeling (SEM). SEM estimates a system of equations simultaneously. The regressions will take the form of the following equations:

𝑦𝑦

∗

= 𝛽𝛽

_{𝑦𝑦𝑦𝑦1}

𝑥𝑥

₁

+ 𝜀𝜀

₁

(1)

𝑦𝑦

∗

= 𝛽𝛽

_{𝑦𝑦𝑦𝑦.𝑧𝑧}

𝑥𝑥

₁

+ 𝛽𝛽

_{𝑦𝑦𝑧𝑧.𝑦𝑦1}

𝑧𝑧 + 𝑣𝑣

₁

(2)

𝑦𝑦

∗

= 𝛽𝛽

𝑥𝑥

₂

+ 𝜀𝜀

₂

(3)

(21)

20 Where:

𝑦𝑦

= Financial behavior.

𝑥𝑥

₁

= Self-esteem.

𝑥𝑥

₂ = Objective financial literacy.

𝑧𝑧

= Subjective financial literacy.

𝜀𝜀

1

𝜀𝜀

2

𝑣𝑣

1

𝑣𝑣

2 = Random error term.

The difference between the two beta coefficients for 𝑥𝑥₁ in formulas 1 and 2 describes the mediating effect of subjective financial literacy on the self-esteem – financial behavior relationship. This can be written as:

𝛿𝛿 = 𝛽𝛽

− 𝛽𝛽

_{𝑦𝑦𝑦𝑦.𝑧𝑧} (5) Consider the following linear model relating 𝑥𝑥₁ to 𝑧𝑧 :

𝑧𝑧 = 𝜃𝜃

_{𝑧𝑧𝑦𝑦1}

𝑥𝑥

₁

+ 𝑤𝑤

(6)

Where

𝜃𝜃

_{𝑧𝑧𝑦𝑦1} explains the effect of

𝑥𝑥

₁ on

𝑧𝑧

, and

𝑤𝑤

is a random error term independent of

𝑣𝑣

. According to Stolzenberg (1980), the properties of linear models and path analysis give the following result:

𝛿𝛿 = 𝛽𝛽

− 𝛽𝛽

_{𝑦𝑦𝑦𝑦.𝑧𝑧}

= 𝜃𝜃

× 𝛽𝛽

_{𝑦𝑦𝑧𝑧.𝑦𝑦1} (7) This is also known as the “difference between coefficients” and the “product of

coefficients” equivalence (Breen, Karlson and Holm, 2013).

The total effect of self-esteem on financial behavior can be decomposed in a direct and indirect effect through subjective financial knowledge in a following way:

(22)

21 The logit models for my four dichotomous variables are constructed using Generalized Structural Equation modeling (GSEM). The logit models will take the functional form of:

Logit [Pr(𝑦𝑦∗ > 0)] = 𝑏𝑏_{𝑦𝑦𝑦𝑦1.𝑧𝑧}

𝑥𝑥

₁

+ 𝑏𝑏

𝑧𝑧

=

𝛽𝛽

𝑦𝑦𝑦𝑦1.𝑧𝑧

𝜎𝜎

_𝐹𝐹1

𝑥𝑥

1

+

𝛽𝛽

𝑦𝑦𝑧𝑧.𝑦𝑦1

𝜎𝜎

𝐹𝐹1

𝑧𝑧

(11) Logit [Pr(𝑦𝑦∗ > 0)] = 𝑏𝑏_{𝑦𝑦𝑦𝑦1}

𝑥𝑥

₁

=

𝛽𝛽

𝑦𝑦𝑦𝑦1

𝜎𝜎

_𝑅𝑅 (12) Logit [Pr(𝑦𝑦∗_{> 0)] = 𝑏𝑏} 𝑦𝑦𝑦𝑦2.𝑧𝑧

𝑥𝑥

2

+ 𝑏𝑏

𝑧𝑧

=

𝛽𝛽

_{𝑦𝑦𝑦𝑦2.𝑧𝑧}

𝜎𝜎

_𝐹𝐹2

𝑥𝑥

2 +

𝛽𝛽

𝜎𝜎

_𝐹𝐹2

𝑧𝑧

(13) Logit [Pr(𝑦𝑦∗ > 0)] = 𝑏𝑏_{𝑦𝑦𝑦𝑦2}

𝑥𝑥

₂

=

𝛽𝛽

𝑦𝑦𝑦𝑦2

𝜎𝜎

𝑅𝑅2 (14)

Where 𝑦𝑦 =1 if 𝑦𝑦 *> τ , and 𝑦𝑦 =0 if otherwise. τ is a threshold, which is set to 0. The expected outcome of this indicator is the probability of observing 𝑦𝑦 =1, Pr(𝑦𝑦)=1. 𝜎𝜎_𝐹𝐹 and

𝜎𝜎

_𝑅𝑅 are scale parameters of the full and reduced model, with a variance of

𝜎𝜎

_𝜇𝜇=2

_𝜎𝜎

𝑒𝑒2

𝜋𝜋

2

/3

for the error term.

Because the variance of

𝑦𝑦

∗ cannot be estimated, the logit coefficients are similar to those of an underlying linear model divided by the scale parameter of that model. Hence the scale parameter or the underlying regression coefficient cannot be determined in logit models, only a ratio. To see the mediating effect of subjective financial literacy on the self-esteem – financial behavior relationship, I would like to see the difference between

𝛽𝛽

_{𝑦𝑦𝑦𝑦1} and

𝛽𝛽

_{𝑦𝑦𝑦𝑦1.𝑧𝑧}. However both coefficients are ratios measured on a different scale.

𝜎𝜎

_𝑅𝑅 >

𝜎𝜎

_𝐹𝐹, because of the inclusion of

𝑧𝑧

in the full model. Therefore:

𝑏𝑏

− 𝑏𝑏

=

𝛽𝛽

𝜎𝜎

_𝐹𝐹1

−

𝛽𝛽

(23)

22 Therefore I will also use the KHB method proposed by Breen, Karlson, and Holm (2013), which overcomes this problem. It provides a way to hold the scale and the error of the fit to the assumed logistic distribution constant. Again the “product of coefficients” method is applied to the full model. Therefore I use an auxiliary regression of z on x, which gives an expectation of:

𝐸𝐸(𝑧𝑧) = 𝜃𝜃

𝑥𝑥

(16)

Substituting and rearranging in equation 11 leads to the following equation:

Logit [Pr(

𝑦𝑦

∗ > 0)] =

𝛽𝛽

+𝛽𝛽

𝑦𝑦𝑧𝑧𝑦𝑦.1

𝜃𝜃

𝑧𝑧𝑦𝑦1

𝜎𝜎

𝑒𝑒

𝑥𝑥

1 (17) From here the total effect can be decomposed into:

Total effect

=

𝛽𝛽

𝑦𝑦𝑦𝑦1

𝜎𝜎

_𝑒𝑒 =

𝛽𝛽

+𝜃𝜃

× 𝛽𝛽

𝜎𝜎

_𝑒𝑒 (18) Direct effect

=

𝑏𝑏

=

𝛽𝛽

𝜎𝜎

𝑒𝑒 (19) Indirect effect =

𝜃𝜃

𝑏𝑏

=

𝜃𝜃

𝑧𝑧𝑦𝑦1

× 𝛽𝛽

𝜎𝜎

_𝑒𝑒 (20)

5. Results

5.1 Model fit

(24)

23 To test the model fit, I performed the likelihood ratio chi-squared test, the root mean squared error of approximation (RMSEA), the Tucker-Lewis index (TLI), and the comparative fit index (CFI). Results are displayed in table 11 in the appendix. The investments and endowment insurance models score a good fit on all 4 indices (Chi-square > 0.05, RMSEA score < 0.06 and p value> 0.05, CFI > 0.095, and TLI > 0.95). Furthermore the savings model scores a good fit on the CFI, an adequate fit on the RMSEA (score between 0.05 and 0.08, p > 0.05), and a poor fit on chi-square and TLI. The debt model scores a good fit on the CFI, and a poor fit on the remaining indices.

The remaining binary models are constructed using GSEM, which does not require multivariate normality, and doesn’t allow tests for model fit as with SEM.

5.2 Correlations

Table 2 shows the Pearson correlation and partial correlation results. Self-esteem positively correlates with the log odds of a respondent having investments in financial assets, savings, and endowment insurance. All coefficients of these binary measures are significant at a 0.05 level. The same result is present for the log amounts of saving, debt and endowment insurance. The correlations are positive and significant. Negative correlations are found for the log amount of investments in financial assets, as well as the log amount of savings. However both correlations are not significant. All correlations seem to be weak in nature with coefficients not larger than 0.3 or -0.3.. After controlling for covariates, none of the correlations between self-esteem and financial behavior remain significant.

Subjective financial literacy correlates positively with all measures of financial behavior. These relationships are significant for all measures, with the exception of the log amount of investments. Only the amount of savings remains significant after controlling for covariates

(25)

(26)

(27)

26

5.3. Path analysis

Tables 3, 5, 7 and 9 report the results of the path analysis for the observed variables. The main path shows the regression result of self-esteem, objective and subjective financial literacy, and the covariates on the eight financial behavior measures. Here one can make inferences about the direct relationships. The mediation path shows the regression results of self-esteem, objective financial literacy and covariates on subjective financial literacy. Combined with the main path, initial indirect relationships can be determined. Tables 4, 6, 8 and 10. show the results of the KHB tests, which I will use to determine the strength of the direct and indirect relationships between self-esteem and financial behavior.

5.3.1 Investing in financial assets

(28)

27 Table 3.

Results of path-analysis for the direct and indirect effects of self-esteem on investments in financial assets

Variables

Investment in financial assets (1) Invest in financial assets (2) Log(Amount invested in financial assets) Main path Dependent variable: Financial behavior Self-esteem .012 -.020 Objective financial Literacy .731* -.158 Subjective Financial Literacy -.0189 .056

Covariates yes yes

Mediation path Dependent variable: Subjective financial literacy Self-esteem .020* .018* Objective financial Literacy .18* .184*

Covariates yes yes

Observations 1902 252

(29)

28 Table 4.

KHB test results for total, direct and indirect effects of self-esteem on investments in financial assets

(1) Invest in financial assets (2) Log(Amount invested in financial assets) Total effect .012 -.019 Direct effect .012 -.020 Indirect effect -.0004 .0001 Observations 1902 252 * ρ < 0.05 5.3.2 Savings

(30)

29 Table 5.

Results of path-analysis for the direct and indirect effects of self-esteem on savings

Variables Savings (1) Have savings (2) Log (Savings amount) Main path Dependent variable: Financial behavior Self-esteem .013 -.003 Objective financial Literacy .387* .322* Subjective Financial Literacy -.121 .137*

Covariates yes yes

Mediation path Dependent variable: Subjective financial literacy Self-esteem .020* .025* Objective financial Literacy .189* .195*

Covariates yes yes

Observations 1877 1,272

(31)

30 Table 6.

KHB test results for total, direct and indirect effects of self-esteem on savings (1)Have savings (2)Log(Savings amount) Total effect .010 .0007 Direct effect .012 -.003 Indirect effect -.0002 .003 Observations 1902 1272 * ρ < 0.05 5.3.3 Debt

Table 7 reports no significant direct link between self-esteem and both measures of debt. The same result is found for subjective financial literacy. Therefore there is also no reason to suspect significant indirect effects. The KHB test results in table 8 are in line with these findings, no significant total, direct or indirect effects appear to be present.

(32)

31 Table 7.

Results of path-analysis for the direct and indirect effects of self-esteem on debt

Variables Debt (1) Have debt (2) Log(debt amount) Main path Dependent variable: Financial behavior Self-esteem -.011 .017 Objective financial Literacy -.079 .009 Subjective Financial Literacy .034 .095

Covariates yes yes

Mediation path Dependent variable: Subjective financial literacy Self-esteem -.020* .032 Objective financial Literacy .188* .175 *

Covariates yes yes

Observations 1902 255

(33)

32 Table 8.

KHB test results for total, direct and indirect effects of self-esteem on debt (1) Have debt (2) Log(Debt amount) Total effect -.010 .020 Direct effect -.011 .017 Indirect effect .0007 .003 Observations 1895 255 * ρ < 0.05 5.3.4 Endowment insurance

For the log amount of endowment insurance model I could not establish a fitted model when controlling for all covariates. Therefore I removed one of the origin variables (second generation non-western), which provided not much information with 12 observations and kept the model from being fitted.

(34)

33 Table 9.

Results of path-analysis for the direct and indirect effects of self-esteem on endowment insurance

(35)

34 Table 10.

KHB test results for total, direct and indirect effects of self-esteem on financial behavior (1) Have endowment insurance (2) Log(Endowment insurance amount) Total effect .010 .0003 Direct effect .009 .002 Indirect effect .001 -.002 Observations 1902 238 * ρ < 0.05 5.4. Omitting covariates

Section 5.2 shows the baseline results of the eight path models. All covariates are included, and are assumed to have an impact on financial behavior. This section shows the impact of omitting covariates from the regressions on the coefficients of the explanatory variables. For each model, each covariate measure is dropped one by one from every equation. I will focus on changes in the coefficients of self-esteem, to be able to make additional inferences about its influence on financial behavior. When coefficients show no or a very minor change (a change of 0.001), the covariate is assumed to have no influence on financial behavior in that particular model. Results can be found in tables 11 to 19 in the appendix.

5.4.1 Investing in financial assets

(36)

35 variables are the only relevant controls in this particular model, and perform a new regression, we see very different results. Self-esteem now has a direct influence on and individual having investments, with a coefficient of 0.017 (ρ value 0.024< 0.05). A new KHB test confirms this finding with a significant direct effect, as well as a total effect. An indirect effect remains absent, even when I regress against no controls. Similar results are found for the path model for the log amount of investments in financial assets. Age, education and income all show a minor impact on the self-esteem coefficient, as well as measures for a propensity to save (0.002 or -0.002). However new path regressions with only these variables as controls, or no controls at all, show no changes in the significance of any explanatory coefficient.

5.4.2 Savings

Table 14 shows a minor influence of age, having no kids, origin, and patience on the self-esteem coefficient when omitted from the path model of having savings (0.002 and -0.002). An individual’s income is shown to influence the model the most, with a moderate coefficient change of 0.04. This implicates self-esteem to have a lesser positive influence on an individual to have savings, when controlling for income. A path regression controlling for the above mentioned variables does not lead to a change in significance of the self-esteem coefficient. Although close, with a coefficient of 0.018, it’s ρ value of 0.088 exceeds 0.05. Leaving all covariates out of the equation does result in a significant direct effect (coefficient of 0.020, ρ<0.05), which is confirmed by results of a KHB test.

(37)

36 saves, with a coefficient of 0.13 and a ρ value of 0.016. Furthermore the total effect also becomes significant with a coefficient of 0.018 and a ρ value of 0.002.

5.4.3. Debt

The respondent’s odds of having debt seem not to be influenced by many covariates. I find age to influence the self-esteem coefficient to a little extent, when omitted (a change of 0.02). Self-control has the largest influence with a change of 0.05. Additional regressions controlling for these two, or no covariates, result in no significant changes for the self-esteem coefficient.

The log amount of debt follows the same pattern. A respondent’s origin and patience are of little influence (change of 0.002), while being married displays the largest influence (change of 0.003). Again no significant changes occur with additional regressions.

5.4.4 Endowment insurance

Self-esteem seemed to play no role in explaining an individual’s decision to have endowment insurance, as well as the amount saved. Education appears to be the only influence on self-esteem in the logit model, although small with a coefficient change of 0.002. But a regression controlling for this covariate only shows no significant differences compared to the baseline logit model.

(38)

37

6. Discussion

6.1 Self-esteem on financial behavior

My analysis shows mixed results regarding the relationship between self-esteem and financial behavior. First of all, self-esteem relates positively to most measures of financial behavior. Initial negative relationships are found for the amount an individual invests in financial assets, and whether an individual is in debt. After controlling for covariates, the relationship between self-esteem and the amount a respondent saves also appears to be negative.

Although not all correlations or regression coefficients are significant, these findings are mostly in line with the limited literature available for this subject. Responsible financial behavior is indicated by high amounts of investments in financial assets, general savings, and additional savings for retirement, as well as low amounts of debt. Tang and Baker (2016) find self-esteem to positively relate to the investment and savings measures, and to negatively relate to debt, which corresponds with responsible financial behavior. My correlation and path analysis results for the measures of the amount invested in financial assets, the amount of savings, and amount of debt, do not correspond with responsible financial behavior.

6.2 Self-esteem on subjective financial literacy

(39)

38

6.3 Direct and indirect effects

The baseline results of the eight path models indicate no direct effects of self-esteem on measures of financial behavior. Tang and Baker (2016) found self-esteem to directly influence an individual’s decision to invest in risky assets, and one’s amount of savings and credit card debt.

The significant correlation between self-esteem and having investments in financial assets does indicate the presence of a possible total and/or direct effect. However, partial correlations as well as the baseline path model results display no significant relationships. Both of these tests incorporate the full set of covariates, of which not all are relevant in explaining the relationship. When I omit the non-relevant from the baseline path model, the results indicate a significant positive direct relationship between self-esteem and investing in financial assets.

The same pattern can be observed for the measures of saving behavior. The baseline results do not indicate a presence of significant direct effects. Omitting non-relevant covariates from the model almost results in a significant direct effect of self-esteem on having savings. None of the covariates create a large coefficient change of self-esteem when being omitted from both baseline path models (a maximum of 0.004 is observed). When I make the assumption that no single covariate is relevant, and omit them from both savings models, I can observe significant positive direct effects between self-esteem and both measures of savings.

For the remaining measures of debt and endowment insurance, no clear cut direct effects of self-esteem appear to be present. Despite the insignificance, one needs to keep in mind the causality relationship between self-esteem and debt. Having high amounts of debt might have an enormous negative impact on an individual’s life, and might therefore be detrimental to this person’s self-esteem. So the 2011 measure of self-esteem might be affected by a respondent’s prior level of debt.

(40)

39 Having endowment insurance and the log amount of it, appear to be unrelated to self-esteem. This might have to do with the validity of the measure itself, as endowment insurance serves as a second best measure for an individual’s active choice to save for retirement in the Netherlands. By default Dutch individuals save in the form of a defined benefit plan, whereby endowment insurance serves as an addition to this amount. Only 329 (16.5%) of the 1987 respondents say to have endowment insurance. This stands in contrast to the sample of Tang and Baker (2016), whereby 85 percent of the sample said to contribute in defined contribution plans.

Only one out of eight models displays a direct effect of self-esteem on financial behavior, conditional on socioeconomic factors. I therefore say there is little evidence to accept hypothesis 1.

Self-esteem does not appear to be influencing financial behavior through subjective financial knowledge. Tables 3, 5, 7, and 9 do show self-esteem to relate to subjective financial literacy. However, all path models, with the exception of the log savings amount show subjective financial literacy not to significantly relate to financial behavior. This result is contradictory to current literature, as individuals with a higher level of self assessed financial knowledge tend to display a higher level of responsible financial behavior (Rob and Woodyard,2011; Tang and Baker,2016). As for the amount of savings model, I did find an initial indication for an indirect effect to be present, which would be in line with the findings of Tang and Baker (2016). However, the KHB test results could not confirm the significance of this finding.

As no indirect effects were found between self-esteem and financial behavior across all path-models, there is no evidence to accept hypothesis 2.

6.4 Covariates

(41)

40 Perhaps these differences are due to omitted variable bias. Two covariates which Tang and Baker (2016) incorporate in their path models, which are absent in mine are: risk tolerance and net worth. Although an individual’s risk tolerance and net worth might be important factors explaining financial behavior, inclusion of both measures would be unlikely to lower significance levels to < 0.05.

Furthermore my results show the relative impact of the covariates on self-esteem and subsequent financial behavior. Education and income contribute the most to a fluctuation of the self-esteem coefficients, when dropped from the regressions. They both have the biggest influence in both the investments models. Furthermore education is of most influence on both measures of endowment insurance, and income on having savings.

As mentioned above I also can observe changes in the significance of the total and direct effects of self-esteem on investing behavior when controlling for relevant covariates only. Education and income and being married are of such influence that a total significant effect can be observed for self-esteem on the decision to invest. The significance level is even lower when controlling for these two variables than for none (0.024 versus 0.035).

I see that self-control has strong negative influence on the amount an individual saves. This to be expected, as higher measures of self-control indicate a higher difficulty to control expenditures. When omitted from the model, the self-esteem coefficient changes from a negative to a positive sign. However when I use self-control as only covariate in this model, the coefficient is also positive, which indicates possible collinearity amongst covariates. Self-control has a relative strong significant correlation with age 0.150) and one of the saving propensity measures (-0.216).

Other covariates displaying influence to a lesser extent across all models are patience and age.

6.5 Limitations and recommendations

(42)

41 fit according to two out of four indices. The Chi square statistics p-value is under 0.05. However the sample sizes exceeds 200 which might result in an unfair rejection of the models, and the non-multivariate normality of the data might produce inaccurate chi-square test results. The Tucker Lewis Index of 0.398 displays an inadequate fit, while on the other hand the Root Mean Squared Error of Approximation shows an adequate fit of 0.078. Although there is no single best measure of model fit, all in all the savings model can be defined best as adequate in terms of model fit. As I found indications for an indirect effect of self-esteem on saving behavior, this effect might be confirmed when a better fitted model is used with additional covariates.

Therefore further recommendations can be made regarding possible omitting variable bias. Although the LISS panel dataset provided many possible measures to control for, not all covariates found in current literature could be accounted for. Certain psychological traits such as risk-tolerance or locust of control could be controlled for in future research.

Other limitations can be addressed to the fragmented nature of my dataset. Measures for self-esteem, financial behavior, financial knowledge and certain covariates originate from different points in time, which might be a cause for measurement error. Also the measures of investments are not ideal as an indicator for responsible financial behavior. The LISS core study does not differentiate between risky and less risky investments, which might have lead to an upward bias. Furthermore the amount of debt measure might not be representative for the entire population due to a large number of missing observations.

Although I merely make inferences about relationships between self-esteem and financial behavior, one needs to be cautious in interpreting causality. An individual’s self-esteem might be impacted by consequences of financial decisions in the past. Therefore it might be wise for future researchers to measure self-esteem before and after major financial decisions, to account for this reverse causality.

(43)

42

7. Conclusion

As individuals face an increasing responsibility toward financial decision making compared to two decades ago, there has been an increasing awareness towards financial literacy. As financial literacy is low amongst average households, policymakers focus on educational programs to improve financial literacy in order to facilitate an individual’s financial decision making. Such programs focus mainly on objective financial knowledge, while recent studies stress the importance of the role of psychological traits on financial behavior.

This study contributes to this relatively new branch of literature by investigating the simultaneous role of self-esteem and financial literacy on financial behavior, through subjective financial literacy as a mediator. I find that self-esteem positively relates to measures of an individual having investments in financial assets, having savings, the amount of debt, and both measures of endowment insurance. Self-esteem negatively relates to the amount an individual has invested in financial assets, and being in debt. Current literature on this subject (Tang and Baker,2016) states self-esteem to have a positive influence on investing and saving behavior, and a negative relationship with debt. My results contradict these findings for measures of the amount invested in financial assets as well as for the amount of debt.

Although my baseline model indicates no direct effects, when varying with covariates I do find indications of direct effects of self-esteem on three out of eight measures for financial behavior. Self-esteem positively and directly relates to whether an individual invests in financial assets, has savings, and the amount he or she saves

As for the role of subjective financial literacy as a mediator, I find there to be no significant indirect effects of self-esteem on financial behavior. Self-esteem does relate positively to subjective financial literacy across all models, however the significant link between subjective financial literacy and measures of financial behavior is missing.

An individual’s income and education are found to be the factors that explain these results the most. Other factors of influence, but to a lesser extent are: age, self-control, and patience.

(44)

(45)

44

References

Alessie, R., Van Rooij, M., & Lusardi, A. (2011). Financial literacy and retirement preparation in the Netherlands. Journal of Pension Economics and Finance, 10(04), 527-545.

Antonides, G., De Groot, I. M., & Van Raaij, W. F. (2011). Mental budgeting and the management of household finance. Journal of Economic Psychology, 32(4), 546-555. Asaad, C. T. (2015). Financial literacy and financial behavior: Assessing knowledge and confidence. Financial Services Review, 24, 101–117.

Baumeister, R. F., Campbell, J. D., Krueger, J. I., & Vohs, K. E. (2003). Does high self-esteem cause better performance, interpersonal success, happiness, or healthier lifestyles? Psychological Science in the Public Interest, 4, 1–44.

Bikker, J. A., & Van Leuvensteijn, M. (2008). Competition and efficiency in the Dutch life insurance industry. Applied Economics, 40(16), 2063-2084.

Braunstein, S., & Welch, C. (2002). Financial literacy: An overview of practice, research, and policy. Fed. Res. Bull., 88, 445.

Breen, R., Karlson, K. B., & Holm, A. (2013). Total, direct, and indirect effects in logit and probit models. Sociological Methods & Research, 42(2), 164-191.

Brown, S., & Taylor, K. (2014). Household finances and the Big Five personality traits. Journal of Economic Psychology, 45, 197–212

Butler, J. V. (2014). Inequality and relative ability beliefs. The Economic Journal. Campbell, J. Y. (2006). Household finance. The Journal of Finance, 61(4), 1553-1604.

Di Paula, A., & Campbell, J. D. (2002). Self-esteem and persistence in the face of failure. Journal of personality and social psychology, 83(3), 711.

Epley, N., & Gneezy, A. (2007). The framing of financial windfalls and implications for public policy. The Journal of Socio-Economics, 36(1), 36-47.

(46)

45 Fuchs, V., 1980, Time preference and health: An explorative study, NBER Working Paper,539.

Guiso, L., Haliassos, M., & Jappelli, T. (2002). Household portfolios. MIT press. Cambridge.

Guiso, L., & Jappelli, T. (2005). Awareness and stock market participation. Review of Finance, 9(4), 537-567.

Guiso, L., P. Sapienza, and L. Zingales, 2008, Trusting the stock market, Journal of Finance,63, 2557-2600

Hadar, L., Sood, S., & Fox, C. R. (2013). Subjective knowledge in consumer financial decisions. Journal of Marketing Research, 50(3), 303–316.

Haliassos, M., & Bertaut, C. C. (1995). Why do so few hold stocks?. The economic Journal, 1110-1129.

Hilgert, M. A., Hogarth, J. M., & Beverly, S. G. (2003). Household financial

management: The connection between knowledge and behavior. Fed. Res. Bull., 89, 309.

Joo, S. H., & Grable, J. E. (2004). An exploratory framework of the determinants of financial satisfaction. Journal of family and economic Issues, 25(1), 25-50.

Judge, T. A., & Bono, J. E. (2001). Relationship of core evaluations traits – self-esteem, generalized self-efficacy, locus of control, and emotional stability with job satisfaction and job performances: A meta-analysis. Journal of Applied Psychology, 86(1), 80–92.

Judge, T. A., Erez, A., & Bono, J. E. (1998). The power of being positive: The relation between positive self-concept and job performance. Human performance, 11(2-3), 167-187.

Karlson, K. B., Holm, A., & Breen, R. (2010). Comparing Regression Coefficients Between Models using Logit and Probit: A New Method. Retrieved July 07, 2014. Kourtidis, D., Šević, Ž., & Chatzoglou, P. (2011). Investors’ trading activity: A behavioural perspective and empirical results. The Journal of

(47)

46 Lusardi, A. (2008). Household saving behavior: The role of financial literacy,

information, and financial education programs (No. w13824). National Bureau of Economic Research.

Lusardi, A., & Mitchell, O. S. (2007a). Baby boomer retirement security: The roles of planning, financial literacy, and housing wealth. Journal of monetary

Economics, 54(1), 205-224.

Lusardi, A., & Mitchell, O. S. (2007b). Financial literacy and retirement preparedness: Evidence and implications for financial education. Business economics, 42(1), 35-44. Lusardi, A., & Mitchell, O. S. (2011a). Financial literacy and planning: Implications for retirement wellbeing. Oxford University Press , 2011, p p. 17 -39 .

Lusardi, A., & Mitchell, O. S. (2011b). Financial literacy and retirement planning in the United States. Journal of Pension Economics and Finance 10 (4): 509–25.

Lusardi, A., & Mitchell, O. S. (2014). The economic importance of financial literacy: Theory and evidence. Journal of Economic Literature, 52(1), 5-44.

Lusardi, A., & Tufano, P. (2009). Debt literacy, financial experiences, and

overindebtedness. National Bureau of Economic Research Working Paper 14808. Moore, D. L. (2003). Survey of financial literacy in Washington State: Knowledge, behavior, attitudes, and experiences. Washington State Department of Financial Institutions.

Mottola, G. R. (2013). In our best interest: Women, financial literacy, and credit card behavior. Numeracy, 6(2), 4.

Owens, T. J. (1993). Accentuate the positive – and the negative: Rethinking the use of self-esteem, self-deprecations and self-confidence. Social Psychology Quarterly, 56, 288–299.

Parker, A. M., De Bruin, W. B., Yoong, J., & Willis, R. (2012). Inappropriate

confidence and retirement planning: Four studies with a national sample. Journal of Behavioral Decision Making, 25, 382–389.

(48)

47 Rosenberg, M., Schooler, C., Schoenbach, C., & Rosernberg, F. (1995). Global self-esteem and specific self-self-esteem different concepts, different outcomes. American Sociological Review, 60(1), 141–156.

Sherraden, M. S., & Grinstein‐Weiss, M. I. C. H. A. L. (2015). Creating financial capability in the next generation: An introduction to the special issue. Journal of Consumer Affairs, 49(1), 1-12.

Sommer, K. L., & Baumeister, R. F. (2002). Self-evaluation, persistence, and

performance following implicit rejection: The role of trait self-esteem. Personality and Social Psychology Bulletin, 28(1), 926–938.

Stage, F. K., Carter, H. C., & Nora, A. (2004). Path analysis: An introduction and analysis of a decade of research. The Journal of Educational Research, 98(1), 5-13. Stango, V., & Zinman, J. (2009). Exponential growth bias and household finance. The Journal of Finance, 64(6), 2807-2849.

Stolzenberg, R. M. (1980). The measurement and decomposition of causal effects in nonlinear and nonadditive models. Sociological methodology, 11, 459-488.

Tang, N., & Baker, A. (2016). Self-esteem, financial knowledge and financial behavior. Journal of Economic Psychology, 54, 164-176.

Van Rooij, M. C., Kool, C. J., & Prast, H. M. (2007). Risk-return preferences in the pension domain: are people able to choose?. Journal of public economics, 91(3), 701-722.

Van Rooij, M., Lusardi, A., & Alessie, R. (2011a). Financial literacy and stock market participation. Journal of Financial Economics, 101(2), 449-472.

Van Rooij, M. C., Lusardi, A., & Alessie, R. J. (2011b). Financial literacy and

retirement planning in the Netherlands. Journal of Economic Psychology, 32(4), 593-608.

(49)

48 Volpe, R. P., Chen, H., & Liu, S. (2006). An analysis of the importance of personal finance topics and the level of knowledge possessed by working adults. Financial services review, 15(1), 81.

Xiao, J. J., Ahn, S. Y., Serido, J., & Shim, S. (2014). Earlier financial literacy and later financial behaviour of college students. International Journal of Consumer Studies, 38(6), 593-601.

(50)

49

Appendix.

Self-esteem statements.

Please indicate to what extent you agree or disagree with the statements below.

1. I feel that I’m a person of worth, at least on an equal plane with others 2. I feel that I have a number of good qualities

3. All in all, I am inclined to feel that I am a failure 4. I am able to do things as well as most other people 5. I feel I do not have much to be proud of

6. I take a positive attitude towards myself 7. On the whole, I am satisfied with myself 8. I wish I could have more respect for myself 9. I certainly feel useless at times

10. At times, I think I am no good at all

[ ]1 [ ]2 [ ]3 [ ]4 [ ]5 [ ]6 [ ] 7 1= totally disagree

7= totally agree

Financial literacy questions. Q1: Subjective Financial Literacy.

How would you score your understanding of financial matters (on a scale of 1 to 7, where 1 means ‘very poor’ and 7 means ‘very good’)?

Very Poor very good [ ]1 [ ]2 [ ]3 [ ]4 [ ]5 [ ]6 [ ] 7

Q2: Numeracy.

Suppose you have 100 euros on a savings account and the interest is 2% per year. How much do you think you will have on the savings account after five years,

(51)

50 [ ]1. More than 102 euros

[ ]2. Exactly 102 euros [ ]3. Less than 102 euros [ ]4. I don’t know

[ ]5. I would rather not say

Q3: Inflation.

Suppose that the interest on your savings account is 1% per year and that inflation amounts to 2% per year. After 1 year, would you be able to buy more, exactly the same, or less than you could today with the money on that account?

[ ]1. More than today

[ ]2. Exactly the same as today [ ]3. Less than today

[ ]4. I don’t know

Q4: Risk and Diversification.

A share in a company usually offers a more certain return than an investment fund that only invests in shares.

[ ]1. True [ ]2. Not true [ ]3. I don’t know

Q5: Interest Rates and Bond Prices.

If the interest rate goes up, what should happen to bond prices? [ ]1. They should increase

[ ]2. They should decrease [ ]3. They should stay the same [ ]4. I don’t know

(52)

51 Table 11.

(53)

52 Table 12.

Path analysis results for the direct effect of self-esteem on having investments in financial assets. Dependent variable: Having investments in financial assets Self-esteem ρ Baseline .012 0.161 Regression with omitted covariate Age .014 0.070 Education .015 0.062 Origin .011 0.152 Income .009 0.236

Regression with covariate

Education and Income

.017 0.024

(54)

53 Table 13.

Path analysis results for the direct effect of self-esteem on the log amount of financial assets.

Dependent variable: Log amount of financial assets Self-esteem ρ Baseline -.020 0.154 Regression with omitted covariate Age -.022 0.126 Education -.018 0.196 Income -.018 0.206 Propensity to save -.022 0.108

Age, Education Income, and

Propensity to save -.019 0.184

(55)

54 Table 14.

Path analysis results for the direct effect of self-esteem on having savings.

Dependent variable: Having savings Self-esteem ρ Baseline .013 0.267 Regression with omitted covariate Age .011 0.306 Income .017 0.105 No kids .015 0.183 Patience .011 0.320 Origin .011 0.303

Income .015 0.138

Does self-esteem influence financial behavior?