The impact of financial literacy on the investment
decision of American households
By
Lars Cuperus
University of Groningen Faculty of Economics and Business
Msc Finance June 2015
Supervisor: R. van Dalen
Abstract
This thesis examines the impact of financial literacy on stock market participation of older American households. The relationship is investigated for unique respondents for two different years, at the beginning of the financial crisis in 2008 and in 2011. In this way, the
financial literacy of unique individuals can also be investigated over time. Data from the Cognitive Economics Study from the University of Michigan is used. Descriptive statistics
1
1.Introduction
Households have an increasing responsibility for their financial decisions in retirement and health care in the United States and in the rest of the world. Market liberalization and reforms in the social security and pension plans have shifted decision making from the government to the individuals (Van Rooij et al., 2011). According to Lusardi and Mitchell (2011d), the reforms in the pension plans in the United States have worked as follows. In the 1980s, many workers relied on defined benefit (DB) pension plans. Over the years, most Americans have turned to defined contribution (DC) pension plans and Individual Retirement Accounts (IRA) to finance their retirement. The transition to the DC pension plan implies that employees have an increased responsibility to distribute their retirement wealth. Furthermore, the increase of DC pension plans means that individual employees are directly exposed to market risks, which was less evident in the DB pension plans.
Households do not only have to deal with the increased responsibility and market risk for their financial retirement well-being, but also with the recent financial crisis. Prior to the financial crisis, which started in 2008, it was relatively easy for people to take out mortgages and revolve credits (Jickling, 2009). Even if the terms and conditions for lending were not fully understood, households were able to fulfill their commitments in financial good times. When the financial crisis started, many households were not able anymore to pay their bills and arrived in high indebtedness.
All these developments have stimulated the research on financial literacy. It is namely shown that financial literacy skills enable individuals to navigate the financial world, make informed decisions about their money and minimize their changes of being misled on financial matters (Marcolin and Abraham, 2006). Consequently, financial literacy can help individuals to make the right financial decisions. Most of the research on financial literacy and financial decision-making investigates the impact of financial literacy on savings, retirement planning and portfolio choice. Existing studies show that those people who are financially illiterate are less likely to plan for retirement and save less wealth (Lusardi and Mitchell, 2007a), are more likely to have high-interest mortgages (Moore, 2003) and are lacking portfolio diversification (Guiso and Jappelli, 2005).
2 crisis, which started in 2008, was no exception. Households had to decide whether they would leave, enter, stay or still stay away from the stock market. Therefore, it is interesting to investigate what influence financial literacy had on stock market participation at the
beginning of the financial crisis and some years thereafter. In addition, it is appealing to investigate whether the financial literacy and stock market participation of households changed after the start of the crisis. When households improve their financial literacy level, they might be able to plan better for retirement, save more and increase the diversification of their portfolio.
This study has two aims. The first objective of this study is to investigate the relationship between financial literacy and stock market participation and whether this relationship changes in the aftermath of the unusually severe crisis. According to previous research (Van Rooij et al, 2011), it is expected that there is a positive relationship between financial literacy and stock market participation. The second objective is to examine whether the financial literacy of respondents increased during the crisis. Alessie et al. (2011) find that financial literacy levels stay the same during the last financial crisis.
It is important to specify what financial literacy is in order to measure it. Huston (2010) investigates 71 individual studies on financial literacy and find that there is no universal definition for financial literacy. Following the financial literacy conceptual framework of Appendix A, the author defined financial literacy as how good an individual can understand and use personal finance-related information.
4
2.Literature review
The literature review is divided in four parts. First, the literature describing the relationship between financial literacy and different types of financial decision-making is summarized. The second part provides a comprehensive review on the relationship between financial literacy and one specific type of financial decision-making, namely stock market participation. The third part gives an overview on the scarcely literature that measures financial literacy over time. Finally, the hypotheses are formulated.
2.1 Financial literacy and financial decision-making
Financial literacy is associated with many financial decisions ranging from obtaining loans and credits to portfolio choice in retirement planning. The most well-known paper on this latter subject, is from Lusardi and Mitchell (2005), who investigate the link between financial literacy and retirement planning of people who are 50 years or older, in the United States. They first point out that financially illiteracy is widespread and is specifically serious under woman, elderly and people with a low education. Other studies in other countries find similar results (Bernheim, 1998; Australia and New Zealand Banking Group, 2008). For instance, Fornero (2011), who investigates the financial literacy of Italian households, shows that most individuals of the population lack knowledge of basic financial concepts. The distribution of financial literacy among the Italian population illustrates that woman and individuals with low education perform most badly on financial literacy measures.
Lusardi and Mitchell (2005) also show that most older American households are not familiar with retirement planning. Less than one third of their respondents ever tried to set up a retirement plan. As a result, less than one-fifth of the respondents believe that they
succeeded in retirement planning. Overall, the authors find that financial literacy and
retirement planning are interconnected. The higher the financial literacy of the respondent, the higher the chance that he/she is likely to plan or succeeds in planning.
The relationship between financial literacy and financial decision-making is also confirmed by several studies on the loan market. According to Moore (2003), financial illiterate consumers are more likely to make bad mortgage product choices. Stango and Zinman (2010) show that consumers, who are unable to calculate the interest on a loan, borrow more, accumulate less wealth, and pay more for credit. Campbell (2006) finds that financially unsophisticated households tend to make significant mistakes; they are less likely to refinance their mortgages under beneficial situations.
5 Overall, these studies do not only specify that there is a relationship between financial literacy and financial decision making, but also validate that low financial literacy implies bad
financial decision making. Since a lot of research shows that there is a relationship between financial literacy and financial decision-making, it can be expected that this also holds for financial literacy and stock market participation.
2.2 Financial literacy and stock market participation
Capital market theory argues that every household should invest a part of their wealth in risky assets to get the risk premium (Merton, 1971; Samuelson, 1969) The absolute amount of assets hold in stocks should increase until retirement and decrease afterwards. However, empirical research discloses that a lot of households do not possess stocks at all. According to research of Grinblatt (2011), only about 50% of US households invests in stocks, directly or indirectly via mutual funds. For the Netherlands this number is even lower, only one-quarter of the Dutch households hold company stocks or mutual funds (Van Rooij et al., 2011). This problem is known as the stock market participation puzzle. An argument put forward by several authors, to explain why households do not invest in stocks, is the existence of participation costs (Haliassos and Bertaut, 1995; Vissing-Jorgensen, 2004). These costs can take different forms like fixed entry costs and transaction costs, but can also be interpreted as the amount of time to understand the functioning of stock markets. The paper by Andersen and Nielsen (2010) shows that fixed entry and participation costs in financial terms account for roughly one-third of non-participation in stocks. An interesting conclusion that these authors pose is that participation can also be influenced by behavioral biases and cognitive abilities. This implies that financial literacy can influence stock market participation.
Several recent studies specifically address measurements of intellect and stock market participation. Grinblatt et al. (2011) study the relationship of IQ and stock market
participation and Christellis et al. (2011) investigate the relationship between cognitive abilities and stock market participation in Europe. These papers show that IQ and cognitive abilities are strongly associated with stockholding.
6 One paper that specifically examines how financial decisions are influenced by the financial crisis is by Koenen and Ziegelmayer (2011) in Germany. The authors do not only find that financial illiterate individuals are less likely to invest in the stock market, but these individuals are also more likely to sell those assets which drop in value. Therefore, the high decline in the value of the stock market indexes over the world in 2008 can have severe consequences. The behavior of financially illiterate individuals might have potential long-term consequences if they choose not to participate (again) in the stock market.
The literature acknowledges that there can be various determinants of stock market participation besides financial literacy. Campbell (2006) argues that education and self-reported attitudes to risk could be important to predict the willingness to take financial risk, and thus to invest in stocks. To advocate why education is important, Campbell rephrases to Haliassos and Bertaut (1995) who put it as follows: ‘education and the free acquisition of information are important in overcoming the barrier to stockholding erected by ignorance and misperceptions’. Halliassos and Bertaut also state that cultural influences associated with race, gender or marital status, are factors that might influence portfolio choice. Finally, Guiso et al. (2002) show that the effect of age and wealth on the structure of household portfolios is very important.
2.3 Financial literacy measured over a period of time
There are hardly any papers that investigate financial literacy over time or during the financial crisis. A notable exception is the paper by Alessie et al. (2011). The authors investigate the financial literacy levels for respondents, who are in charge of the household finances, for the years 2005 and 2010 in the Netherlands. The advantage of the paper is that it captures an ideal time period, namely the years before and during the financial crisis. The disadvantage of the paper, however, is that the sample of respondents is not identical in both years. This means that unique individuals cannot be compared.
These authors measure financial literacy by making use of three basic financial literacy questions, originally designed for the Health and Retirement Study (HRS) and that are also used in several other financial literacy studies (Lusardi and Mitchell, 2011). The questions relate to the concepts of interest rates, inflation and risk diversification (See Appendix B). The outcomes indicate that the number of incorrect answers is approximately the same in both years, but the number of ‘do not know’ answers has increased in 2010. This might imply that there is less guessing and overconfidence than was present in 2005.
7 much improved in 2010 compared to 2005. This is noteworthy given the fact that the authors state that Dutch government and the financial sector have developed several initiatives to increase the financial awareness and skills of individuals during this period.
However, there might be an explanation why financial literacy has not much improved in the Netherlands. In general, policy makers over the world have embraced financial literacy education since individuals have become more responsible for financial choices and have to deal with more financial complex products (Mandell, 2008). Mandell states it as follows: ‘This education is widely believed to turn consumers into ‘responsible’ and ‘empowered ’ market players, who are motivated to make financial decisions that increase their welfare. However, as Mandell shows, education in personal finance and money management on high school is not effective in raising financial literacy. The financial literacy levels of the high school students did not significantly increase.
Another paper, which investigated financial literacy for longer periods, is from Ward and Lynch (2014). The authors examine the difference of financial literacy of couples with different relationship lengths in the United States. They suggest that initial distributions of financial responsibility could lead to differences in financial literacy over time between the members of the couple. It is shown that relationship length is associated with increased financial literacy for partners who take the responsibility in the financial domain and with decreased financial literacy for partners who offload this responsibility after 5.5 years. In other words, this research implies that the one responsible for the financial decisions improves his/her financial literacy, and for the one not responsible, financial literacy decreases.
2.4 Hypotheses
First, this thesis will study how financial literacy is related to stock market participation. According to the standard portfolio theory, all households should have some stocks in their portfolio. Based on the empirical research of several papers, it is shown that most households do not possess stocks. Campbell (2006) argues that limited participation in the equity market must be due to a failure in one of the assumptions of standard portfolio theory. Participation costs might explain why financially illiterate households do not possess any stocks. These costs can be psychological factors that refrain some households from equity ownership. Individuals, who are financially illiterate, have less knowledge about stock markets and therefore their participation costs will be higher. These ‘higher’ participation costs mean that they avoid buying stocks. Van Rooij et al. (2011), already find that there is a positive
8 Based on the argumentation and the empirical evidence, the following hypothesis is put forward:
H1: Individuals with a higher financial literacy are more likely to participate in stocks in the
United States in 2008 and 2011
This hypothesis adds to the existing literature examining the effect of financial literacy on stock market participation. This paper will be the first to test the relationship for two different years, namely 2008 and 2011, for the same unique respondents. It is especially interesting since both years encounter different economic circumstances. In 2008, the financial crisis started and equity prices collapsed heavily. In 2011, the Dow Jones rose again for the third consecutive year. Furthermore, new insights can be generated on the relationship by making use of different financial literacy indices. The indices will measure financial literacy by taking into account the number of correct answers, the confidence-level or the difficulty of
questions.
Second, this thesis will study the change in financial literacy of unique respondents for a three-year period during the financial crisis. To my knowledge, there are only two papers that measure financial literacy for a consecutive period. The hypothesis is based on the paper from Alessie et al. (2011). These authors find that the financial literacy of individuals, responsible for the financial decisions in the household, did not increase during the financial crisis. Based on this empirical evidence, the following hypothesis is put forward:
H2: Individuals, responsible for the financing decisions in the household, will hold the same
financial literacy over a three-year period during the financial crisis.
This hypothesis adds to the scarcely existing literature examining financial literacy for a longer time period. In the paper from Alessie et al. (2011), different respondents are
9
3.Data
This section is divided in two parts. In Section 3.1, a description of the database and the two datasets are given. In Section 3.2, the descriptive statistics of the two datasets are shown.
3.1 Database and Datasets
Most of the financial literacy studies use an experimental module from the 2004 HRS, to measure financial literacy. This module has been used very successfully for investigating the link between financial literacy and financial decision-making. However, this module only uses three basic financial literacy questions to assess numeracy and economic concepts like diversification and inflation (See Appendix B).
This thesis uses a more advanced module that measures financial literacy based on ten financial literacy questions (See Appendix B). These questions are divided into four different categories, which are: knowledge of markets, diversification, fees and numeracy. The
advanced module that is used is from the Cognitive Economics Survey1 (CogEcon), set up by the University of Michigan. CogEcon is a panel survey with observations from 2008, 2009, 2011 and 2013. In this way, it is not only possible to investigate the relationship between financial literacy and stock market participation for different years, but also to investigate the difference in financial literacy for the same respondents.
The CogEcon was designed by a team of economists from the University of Michigan to understand the cognitive bases on economic decision-making and has fielded four waves, in which the first and third wave are of interest for this research. These two waves are used because they contain a financial literacy questions module.
The first wave, CogEcon 2008, has collected information about income, work, assets and debts, financial sophistication, use of financial advice and other topics. The collection of information was done by mail and internet and made use of an American national sample of 1222 respondents, aged 44 years and older and their partners. The first wave achieved an overall response rate of 80.6%, with 985 responses from the 1222 persons invited.
The third wave, CogEcon 2011, has repeated many questions of the first wave, and also covers some new questions concerning health literacy and income taxes. Most
individuals who participated in the first wave study in 2008 or the second wave study in 2009 were invited for the third wave study in 2011. People who only partly filled in the 2008 study, and people who deceased or had a permanent condition were not invited. The CogEcon 2011
1 “The Cognitive Economic Study (CogEcon) is sponsored by the National Institute on Aging (grant number NIA
10 study achieved an overall response rate of 81.2%, with 772 responses from the 951 people invited. The CogEcon study aimed to be nationally representative, but due to the lengthy cognitive tests the final samples are slightly better educated and have higher cognitive abilities than the national average (Willis et al., 2014).
The databases of the two waves are merged and the final database serves as a starting point for the two datasets that are used. The first dataset consists of 214 individuals,
responsible for the financial decisions of the household in 2008 and 20112. It is assumed that these individuals are still responsible for the financial decisions in 2011. This dataset will be used to answer the first hypothesis. The second dataset consists of 647 individuals that answered all ten financial literacy questions in 2008 and 2011. The dataset also includes partners/spouses, next to the head of the household who makes the financial decisions. In this way, the differences in financial literacy for these individuals can also be shown. This dataset is used to answer the second hypothesis.
Financial literacy is measured using the financial literacy questions that were asked in the CogEcon study in 2008 and 2011. In the CogEcon study in 2008, 41 questions were asked to the participants to test their financial literacy. Of these 41 questions, 14 questions were asked again in the CogEcon study in 2011 to the same people. After subtracting questions, which asked for attitudes towards risk aversion and investing, ten questions remained that are used to measure financial literacy.
The respondents were asked to rate their certainty about each financial literacy question on a 12-point scale, from 100% false to 100% true3. The answers are recoded based on the correct responses. This means that 1 point on a question implies that the respondent was completely wrong and 12 points implies that he/she was completely right. When the respondent was very uncertain about an answer, he/she had to choose 50% false or 50% true, which resulted in 6 or 7 points. If the respondent obtained 1 till 5 points for a question, it means that the answer is incorrect. If the respondent obtained 6 or 7 points for a question, he/she did not know the question. And if the respondent obtained 8 till 12 points, it means the answer is correct.
Next to using a certainty scale the respondents also received different versions. In the different versions, questions were phrased differently, or verbalized in an alternative way. This is discussed in Section 5.
2 These individuals also answered all ten financial literacy questions in 2008 and 2011 and questions that are
relevant for the first hypothesis. These questions include their age, gender, marital status, risk tolerance and household income.
11
3.2 Descriptive statistics
In Table 1, the summary statistics for the first dataset are shown. The Table shows that there are relatively more woman head of the household and that the average respondent is 65 years old. It is also interesting to see that the average household income heavily decreased from 2008 to 2011. This could be explained by the fact that more respondents are retired. As the table shows, respondents who are in between 44 and 56 years old in 2008 earned even more household income in 2011. Most of these respondents are not retired yet.
Table 1- Descriptive statistics Dataset 1
This table presents the summary statistics(mean, median and standard deviation) of the individuals who are responsible for taking the financial decisions in the household. The data consist of 214 members of the CogEcon 2008 and 2011
Variable Variable Name Mean Median Std. Dev. Demographic characteristics Female (Female) .56 1 .50 Age in 2008 (Age) 64.89 63 9.18 Age 44-56 in 2008 (Age1) .20 0 .40 Age 57-61 in 2008 (Age2) .24 0 .43 Age 62-69 in 2008 (Age3) .26 0 .44 Age ≥70 in 2008 (Age4) .30 0 .46 Married in 2008 (Married2008) .39 0 0.49 Married in 2011 (Married2011) .39 0 0.49 Education characteristics
Education in years (Educ) 14.82 16 2.09
1-12 Years (Educ1) .19 0 .40
13-14 Years (Educ2) .25 0 .45
15-16 Years (Educ3) .27 0 .45
17 Years (Educ4) .29 0 .45
Physiological characteristics
Low Risk Tolerance (RiskAverse) .18 0 .38
Medium Risk Tolerance (RiskTolerant) .42 0 .49
High Risk Tolerance (RiskSeeking) .40 0 .49
Financial characteristics
Total household income in 2008 (HouseholdInc) 114,473 75,000 416,572
Total household income in 2008 DAge1 (HouseholdInc) 109835 75,000 124,803
Total household income in 2011 (HouseholdInc) 79,986 56,000 116,673
Total household income in 2011 DAge1 (HouseholdInc) 130,148 72,500 218,873
12 These stock market based assets are individual stocks or mutual funds, which hold at least some stocks.
It can be observed that most of the respondents are involved in stocks. This is a relative high percentage, and can be explained by the fact that the sample is probably better educated than the average American household (Willis et al. 2014). In 2008, more males were involved in stock market participation compared to females. However, the differences have become much smaller in three years and are not significant anymore. This can be explained by a significant increase (decrease) in stockholding by females (males) between 2008 and 2011. Finally, the table shows that earning a household income, lower than the median in Table 1, means a significant lower chance to be involved in the stock market in 2008 and 2011.
Table 2- Stock market participation in 2008 and 2011
This table reports the percentage of individuals, who are responsible for taking the financial decisions in the household, involved in stock market participation. The data consist of 214 members of the CogEcon Study 2008 and 2011. T-tests are performed for differences in stock market participation between gender and household income in the year 2008 and 2011. Furthermore, t-tests are performed for the difference in stock market participation for all variables between 2008 and 2011 The observations in brackets imply the amount of respondents who are married or not married in 2011. ***p<0.01, **p<0.05, *p<0.10
Variables 2008 2011 Diff. between Years Obs.
All respondents 71.5 73.4 1.9 214 Demographic characteristics Male 84.0*** 75.5 8.5** 94 Female 61.7 71.7 10.0*** 120 Age 44-56 in 2008 88.1 90.5 2.4 42 Age 57-61 in 2008 63.5 76.9 13.5** 52 Age 62-69 in 2008 71.4 64.3 7.1 56 Age ≥70 in 2008 67.2 67.2 0.0 64 Married 77.1 71.4 5.7 83(84) Not Married 67.9 74.6 6.7 131(130) Education characteristics 1-12 Years 53.7 53.7 0.0 41 13-14 Years 67.9 71.7 3.8 53 15-16 Years 75.9 75.9 0.0 58 17 Years 82.3 85.5 3.2 62 Physiological Characteristics
Low risk tolerance 51.3 56.4 5.1 39
Medium risk tolerance 78.7 82.0 3.4 89
High risk tolerance 73.3 72.1 1.2 86
Financial Characteristics
<$75,000 Household income in 2008 62.3*** 106
≥$75,000 Household income in 2008 80.6 108
<$56,000 Household income in 2011 60.4*** 106
13
The descriptive statistics for the second dataset are shown in Table D1 (See Appendix D). The summary statistics of this dataset do not differ much from the first dataset.
It is also important to show descriptive statistics of the financial literacy questions. First, the ten financial literacy questions will be put in several categories. The questions are grouped into four categories: knowledge of capital markets, risk diversification, knowledge of fees and savvy/numeracy. T-tests will be performed in other to show whether the answers significantly changed between 2008 and 2011.
Knowledge of capital markets
Table 3 shows that more than half of the respondents lack knowledge of capital markets. This can be seen from the low amount of correct answers and the relatively high amount of ‘do not know’ answers in both years for all three questions. Question 2, that asks respondents whether it is easy to pick individual stocks that will have better than average returns, is the only one that significantly changed compared with 2008. It is shown that the same amount of
respondents know that employees should have little or none of his or her retirement savings in the company’s stock (Question 3). Employees should have little or none of their savings in the company’s stock from a risk diversification perspective because if the company would go bankrupt, the employee would also lose all his savings.
Table 3 - Knowledge of capital markets
This table reports the percentage of respondents providing correct, incorrect and ‘do not know’ answers to each of the financial literacy questions. The data consists of 637 respondents from the CogEcon Survey. ***p<0.01, **p<0.05, *p<0.10 if percentage from 2011 is significantly different from the same answer choice in 2008.
Financial literacy Question Year Correct Incorrect Do not know
14
Risk diversification
The knowledge of risk diversification is very important if the respondent is considering investment options and it is therefore crucial in determining the financial literacy of the respondent. From Table 4 it can be seen that most respondents recognize risk diversification when they answered question 4 or 5. When a respondent is asked whether he/she should avoid holding foreign stocks the results are ambiguous (Question 6). Only the amount of incorrect answers significantly changed from 2008 and 2011 for this question. The reason that more people answered this question wrong, could be that they have less confidence in investing in other countries because of the economic crisis. However, if an individual holds foreign stocks, it means an increase in the diversification of the portfolio and consequently fewer risks. Overall, it seems that the financial literacy of respondents for risk diversification slightly decreased.
Table 4- Risk diversification
This table reports the percentage of respondents providing correct, incorrect and ‘do not know’ answers to each of the financial literacy questions. The data consists of 637 respondents from the CogEcon Survey. ***p<0.01, **p<0.05, *p<0.10 if percentage from 2011 is significantly different from the same answer choice in 2008
Financial Literacy Question Year Correct Incorrect Do not know
4 2008 73.94 5.49 20.57 2011 73.78 6.59 19.62 5 2008 78.02 10.99 10.99 2011 75.82 11.71 12.40 6 2008 56.67 15.23 28.10 2011 53.69 19.47*** 26.84 Knowledge of fees
The knowledge of fees is also essential for determining the financial literacy of a person. When a person wants to invest via a mutual fund, he/she will always pay an annual fee. Therefore, if a person invests via a mutual fund for a long time, annual fees are very
15
Table 5- Knowledge of fees
This table reports the percentage of respondents providing correct, incorrect and ‘ do not know’ answers to each of the financial literacy questions. The data consists of 637 respondents from the CogEcon Survey. ***p<0.01, **p<0.05, *p<0.10 if percentage from 2011 is significantly different from the same answer choice in 2008
Financial Literacy Question Year Correct Incorrect Do not know
7 2008 64.05 13.81 22.14
2011 67.03 12.56 20.41
Savvy/numeracy
Finally, the ability of the respondents to perform calculations is also relevant for determining financial literacy. Table 6 shows that only half of the respondents can calculate that an investment of $1000, -, with an annual return of 10%, will grow to more than $6000, - in 30 years (Question 8).
The last two questions also deal with interest (costs). Most people know that the interest costs will be lower with a 15-year old mortgage compared to a 30-years old mortgage (Question 9) and that using a savings account to pay off credit card debt is a good idea to decrease the total interest costs (Question 10). It can be observed, that more respondents answer Question 8 and 10 with ‘do not know’.
Table 6-Savvy/numeracy
This table reports the proportion of respondents providing correct, incorrect and ‘ do not know answers to each of the financial literacy questions. The data consists of 637 respondents from the CogEcon Survey. ***p<0.01, **p<0.05, *p<0.10 if percentage from 2011 is significantly different from the same answer choice in 2008.
Financial Literacy Question Year Correct Incorrect Do not know
8 2008 56.36 32.18 11.46 2011 53.22 29.51 17.27*** 9 2008 85.71 8.32 5.97 2011 87.28 7.69 5.02 10 2008 71.27 20.41 8.32 2011 65.93*** 21.04 13.03***
From the ten financial literacy questions, it can be observed that the results significantly change for only two questions4. The respondents are able to provide more correct answers for question 2 and less correct answers for question 10. For the other eight questions, the results do not significantly change. This is preliminarily evidence that the financial literacy for all respondents does not increase.
4Only these two questions show significant changes for two out of the three answer possibilities (correct,
16
4. Methodology
In this section, the methodology for the two hypotheses is given.
4.1 Financial literacy and stock market participation
As already shown in the literature review, several papers advocate that there is a relationship between financial literacy and stock market participation. To validate these findings, this thesis will also dive into the relationship.
Probit regressions are performed in order to assess the impact of financial literacy on the stock market participation decision of households in 2008 and 2011. The probit model is a type of regression where the dependent variable can take only two values, in this case whether the respondent is involved in stock market participation or not.
The independent variable of interest is a financial literacy index, which is measured in three different ways. Furthermore, several control variables are included in the model. These are the individual characteristics: gender, age, race, marital status, education, risk aversion and household income. Most of the control variables are categorical, therefore dummy variables are created. In order to prevent the dummy trap, one of the categories is excluded in the regression.
The first hypothesis is tested by interpreting the marginal effects of the financial literacy indices, that is, how much the probability of stock market participation changes when the value of the financial literacy index is increased by one unit, holding all the other
variables constant at their values.
The estimated probit equation for stock market participation for 2008 and 2011 is given here: πStocksit= Φ(β0 + β1 FinancialLiteracyIndexkit+ β2Femaleit
+ β3Age1it+ β4Age2it+ β5 Age3 it + β6 Marriedit
+ β7Educ2it + β7Educ3it +β8 Educ4it +β9DRiskaverseit (1) + β10Riskseekingit + β10LOG(HouseholdIncit) + ɛit
πStocksit is the probability that a respondent is involved in stock market participation. This is the case if the respondent owns stock market based assets either in retirement accounts or outside these accounts. These stock market based assets are individual stocks or mutual funds, which hold at least some stocks.
The variable FinancialLiteracyIndexkit is the financial literacy index k for household i at time t. The financial literacy indices are measured in three ways. The measurement of the indices is
17 First, the most straightforward index measures financial literacy on the amount of correct answers to the ten financial literacy questions. The index assumes that people, who are responsible for the financial decisions in 2008, are responsible as well in 2011. To make an easier assessment about the financial literacy level, it is assumed that respondents can only provide correct and incorrect answers. When a respondent filled in 60% true to 100% true and the answer was indeed correct, the respondent gets one point. Zero points for a question are obtained when the respondent filled in 60% false to 100% false, when the answer to the question was correct. When the respondent does not know the answer to the question, and fills in 50% true or 50% false, he/she also obtains 0 points.
The second way to create the financial literacy index is based on the confidence of the respondent in the answer. Besides the choice of the respondents to indicate whether a question is correct or incorrect, the answer scale also includes the amount of certainty that a participant has in the answer. Therefore, a participant who had a lot of confidence in an answer, could mention this, and the more confidence in the answer, the higher the score if he/she was right and the lower the score if the answer was incorrect. The scoring system for financial literacy index 1 and 2 is explained in Table D2 (See Appendix D).
The third way to create the financial literacy index is based on the research of Brocket et al. (2002). The financial index will now rely on a weighting scheme that also takes into consideration the difficulty of each question. This weighting scoring mechanism is called PRIDIT. The scoring mechanism is based on whether an answer is right or wrong and does not involve certainty. This index is built in two steps. In the first step, the response of the individuals on the questions is rescaled to how many people answered this question
incorrectly or correctly. This means that if one answers a question is incorrect, the respondent will get a negative penalty and if most respondents answer that particular question correct, the penalty for the respondent will be higher. Correspondingly, a right answer on a question that most respondents answered wrong gives more credit. The procedure transforms each answer in the interval [-1,1] and the results are RIDIT scores. For example, if 80% of the respondents give a wrong answer to a question, the scoring mechanism gives a value of -0.2 for a wrong answer and 0.8 for a correct answer. In the second step, principal component analysis is used to analyze rescaled responses, which permits to take into account the amount of correlation across questions. The less correlated the questions are with each other, the more ‘informative’ they are since they will measure different aspects of financial literacy. The principal
18 to the other eigenvectors. This vector is then used to calculate the so-called PRIDIT weights that are used to calculate the financial literacy index. The PRIDIT weights can be found in Table D3 (See Appendix D).
Table 7- Summary statistics financial literacy indices
This table reports the summary statistics of the financial literacy indices. The data consists of 214 respondents, responsible for financial decisions in the household from the CogEcon 2008 and 2011.
Financial Literacy Index Year Mean Median Std. Dev. Min. Max
1 2008 6.49 7 2.12 0 10 2011 6.27 6 2.30 0 10 2 2008 6.82 6.85 1.27 3.15 9.5 2011 6.70 6.575 1.26 3.425 9.75 3 2008 0 0.1968 1.49 -4.70 2.25 2011 0 0.2130 1.60 -4.42 2.37
The control variables are measured as follows. The first control variable Femaleit is a dummy variable which takes value 1, if the head of the household is female and 0 otherwise. The variables Age1it, Age2it and Ageit denote the dummy variables for the intervals 44 ≤age<57, 57≤age<62 and 62≤ age <70 respectively, representing the age of the respondents in 2008. The age of the respondents in 2008 is chosen to compare the same group of
respondents in 2011. The dummy variable Marriedit is equal to 1 for married and 0 for not-married respondents. The variables DEduc2it, DEduc3it and DEduc4it denote the dummy variables for the intervals 13≤years of education<15, 15≤years of education<17 and years of education=17 respectively. The dummy variables Riskaverseit and Riskseekingit denote whether the respondent is risk-averse or risk-seeking respectively. It is assumed that people have the same risk-taking behavior in 2008 and 2011. For a detailed description how the variables RiskAverseit and RiskSeekingit are created, see Appendix C. The variable HouseholdIncit is the household income for household i at time t.
4.2 Financial literacy over time
19
5.Results
This Section presents the results. First, the relationship between financial literacy and stock market participation will be tested. Second, the possible change in financial literacy over three years is presented. Lastly, a robustness test is done for hypothesis 1 to verify the results.
5.1 Financial literacy and stock market participation
In Table 8, the results of the probit regressions are shown. The coefficients show the marginal effects on stock market participation for the independent variables. The first hypothesis can also be answered based on Table 8. The results show that there is a strong positive
relationship between financial literacy and stock market participation, which is in accordance with the proposed hypothesis. The relationship is explained for financial literacy index 1,2 and 3.
Financial literacy index 1
The first and second regression show the results when the financial literacy index 1 is used. In both years, there is a highly significant positive relationship between financial literacy and stock market participation. In 2008 (2011), when a respondent answered one more question correct, the probability of stock market participation increases with 7.3% (4.0%).
Some control variables also have a significant influence on stock market participation. When the respondent was a woman in 2008, the chance of having stocks decreases with 11.2%. However, in 2011, the gender does not matter anymore. This can be explained by the fact that the proportion of males and females involved in stock market participation is approximately the same in 2011 (See Table 2).
Furthermore, an interesting result is that being a risk-seeking person decreases the chance of having stocks with 14.1% (13.2%) compared to risk tolerant respondents in 2008 (2011). Lastly, a higher household income increases the chance of having stocks. When the variable LogHouseholdInc increases with one unit, it means 8.4 % (7.8%) more chance to be involved in stocks in 2008 (2011).
Financial literacy index 2
20 increases with 11.0 % (9.2%). This ‘one unit’ increase is approximately the same as
answering one more question correct with full confidence. The second index shows some new results compared to the first financial literacy index with respect to the age of the respondents. In 2008, being 57 to 61 years old means that the chance of being involved in stocks decreases with 11.0 % compared to respondents who are 71 years or older.
Financial literacy index 3
The fifth regression, which is based on PRIDIT scores, gives comparable results to the first regression. Again, a higher financial literacy implies a higher chance of being involved in the stock market. An increase of one unit in the PRIDIT scores indicates that the probability of having stocks increases with 10.06%. It is very difficult to explain what this ‘one unit’ is for the PRIDIT scores because it is based on the difficulty of questions and the correlation between the questions. Since the respondents could score in a range of six units and the correlation between the financial literacy index 1 and this index is very high (namely 0.98), the difference with financial literacy index 1 is reasonable. The sixth regression shows no new insights. An increase with one unit for the PRIDIT score implies that the probability of
21
Table 8-Multivariate analysis of stock market participation in 2008 and 2011: Probit Results
This table reports Probit marginal effects of financial literacy and other control variables on stock market participation. The data are from the CogEcon 2008 and 2011. The data consist of 214 respondents, responsible for the financial decisions in the household. Probit 1,3 and 5 are the regressions from 2008 and Probit 2,4 and 6 are the regressions from 2011. Note: *** p<0.01, ** p<0.05, * p<0.1. This table reports estimates of the marginal effect of financial literacy on stock market participation.
Variables Probit1 Probit2 Probit3 Probit4 Probit5 Probit6
Fin. Lit. index 1 0.0725*** 0.0400***
Fin. Lit index 2 0.1098*** 0.0921***
Fin. Lit index 3 0.1006*** 0.0540***
Female -0.1117* 0.0292 -0.1247** 0.0180 -0.111** 0.0282 Age1 0.1539* 0.1464* 0.1330 0.1340* 0.1625 0.1429 Age2 -0.0946 0.0733 -0.1095 0.0705 -0.0913 0.0757 Age3 0.0125 -0.0515 -0.0054 -0.0562 0.0145 -0.0515 Married -0.0872 -0.0990 -0.0601 -0.1169 -0.0917 -0.1018 Educ2 0.0173 0.0598 0.0125 0.0353 -0.0269 0.0619 Educ3 0.0311 0.0667 0.0397 0.0615 0.0348 0.0707 Educ4 0.0421 0.1060 0.0514 0.0781 0.0458 0.1115 Riskaverse -0.1253* -0.1149 -0.1290* -0.1280* -0.1184* -0.1099 Riskseeking -0.1405** -0.1323** -0.1172** -0.1434** - 0.1385* -0.1333** Log(HouseholdInc) 0.0834*** 0.0783** 0.0668** 0.0775** 0.0830** 0.0812*** R-squared 0.2761 0.2623 0.2726 0.2272 0.2726 0.2065 F-statistic 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
5.2 Financial literacy over time
To answer the second hypothesis, whether respondents, who are responsible for the financial decisions of the household, increase their financial literacy over time, first the results of the ten literacy questions in Section 3.2 can be analyzed. These statistics suggest that the financial literacy of the respondents has not increased. However, these are the results for all
respondents and not for the one who is responsible for the financial decisions in the
household. Therefore Table 8 is constructed, which provides the financial literacy index based on the amount of correct answers for 2008 and 2011 for individuals, responsible for the financial decisions. In addition, the Table shows the results of different groups which share a same characteristic.
22 Alessie et al. (2011) who conclude that the financial literacy of respondents, who take the financial decisions, does not increase during a five year period in the Netherlands. In addition, the t-tests for all other groups show that financial literacy does not significantly change. This implies that none of the described groups have increased their financial literacy.
Besides looking at differences in financial literacy over time, other interesting conclusions can be made by inspecting the certain variables at one point of time. By performing a t-test it is shown that respondents, who are responsible for the financial decisions, have a significant higher financial literacy than respondents who are not responsible for the financial decisions. Households in which partners make the financial decisions together, seem to have a higher financial literacy than when the respondent or the partner takes the financial decisions. From intuition, this is obvious, since by sharing their knowledge, both can increase their financial literacy. However, by performing a t-test it is shown that the financial literacy is not significantly higher compared to when only one person is responsible for the financial decisions.
23
Table 9- Financial literacy in 2008 and 2011
This table reports the average amount of financial literacy questions that are answered correct out of the ten financial literacy questions for different groups. T-tests are performed to show whether financial literacy changed for the particular variable in 2011 with respect to 2008. For none of the variables these results are significant. T-tests are also performed to see whether financial literacy significantly differs for the first variable in a group compared to the last variable. *** p<0.01, ** p<0.05, * p<0.1.
Groups Correct answers 2008 Correct answers 2011 N
All respondents 6.22 6.19 637
Who makes financial decisions? 626
Me 6.30*** 6.16*** 256 Me and my partner 6.38 6.45 314 My Partner or spouse 5.18 5.04 56 Demographic variables 637 Gender Female 5.83*** 5.79*** 357 Male 6.72 6.70 280 Age 44-56 Years old in 2008 6.26 6.38 181 57-61 Years old in 2008 6.12 6.10 155 62-69 Years old in 2008 6.35 6.33 155 70+ Years or older in 2008 6.14 5.90 146 Education variables 637 College-Degree 6.74*** 6.76*** 364 No college-Degree 5.52 5.44 273 Years of Education 1-12 years 5.31*** 5.17*** 131 13-14 years 5.86 5.81 183 15-16 years 6.61 6.62 153 17 years 6.96 6.99 170 Physiological variables 637
Low Risk Tolerance 5.36*** 5.36*** 130
Medium Risk Tolerance 6.18 6.22 302
High Risk Tolerance 6.82 6.66 205
Note ***Significant at P=0.01
5.3 Robustness test
24 version uses a different wording. The respondents were randomly assigned to one of the two versions.
In Panel A of Table 10, the distribution of four financial literacy questions is given for the two versions. The wording of the questions causes large differences in the distribution of answering. For instance, Question 2 is formulated as follows in Version 1 and Version 2:
(2-1)If you are smart, it is easy to pick individual company stocks that will have better than average results.
(2-2) If you are smart, it is hard to pick individual company stocks that will have better than average results.
As can be seen from Table 10 , the pattern of answers is very different for both versions. The chi-square test confirms that the distribution of answering for this question is different for both versions.
Whereas the CogEcon 2008 distributed two versions to the respondents, the CogEcon 2011 used four versions. These versions are called 1A,1B,2A and 2B. The respondents, who received Version 1 (Version 2) in 2008, were randomly given Version 1A or 1B (Version 2A or 2B) in 2011. Version 1A and 1B (Version 2A and 2B) have the same wording of questions as in Version 1 (Version 2) in 2008, but are ordered differently. Relative to Version 1A and 2A, in Version 1B and 2B the second half of financial literacy questions are placed in front of the first half of the questions. Therefore, to test whether the distribution of the questions is the same, Version 1A (Version 1B) and Version 2A (Version 2B) are compared in Appendix D, in Table D4.
By comparing the chi-square tests for all ten questions in Table 10 and Table D4 (See Appendix D) it can be observed that there are only three financial literacy questions that have an equal distribution of answering in 2008 and 2011, when a significance level larger than 1% is used for the chi-square test5. Therefore it is assumed that these three questions, Question 5, 7 and 10, do measure financial literacy.
The three financial literacy questions are used for the construction of the financial literacy indices for the robustness test. The probit regressions are done again and can be found in Table D5 (See Appendix D). As can be seen from the table, there is still a significant positive relationship between financial literacy and stock market participation in 2008 and 2011 when the new indices are used. This means that the robustness test confirms that the results are significant.
5 Using a significance level larger than 5%, would imply that only one financial literacy question has the same
25 The wording of financial literacy questions is also tested by Van Rooij et al. (2011). These authors also find that the distribution of answering can change significantly when the wording of the questions is changed. The findings in this thesis provides further evidence that a lot of respondents do not know what stocks, bonds and fees are. In addition, it shows that most financial literacy questions are frequently noisy proxies of financial literacy.
Table 10- Financial Literacy Questions with inverted wording in 2008
This table reports the proportion of respondents providing correct, incorrect and ‘ do not know’ answers for four of the ten literacy questions that were asked in two versions with different wording in Panel A. Respondents were assigned randomly one of two versions. 256 respondents received Version 1 and 381 respondents received Version 2. The table also shows the chi-square statistic, which tests the null hypothesis that wording of the question is not relevant for the distribution of correct, incorrect and ‘do not know’ answers for all ten questions. Questions with a p >0.01 are statistically significant. This means that the distribution of answering is the same for both versions. Panel B shows the chi-square statistics for the other six financial literacy questions. The data is from the 2008 Cognitive Economics Study.
Panel A: Wording of four financial literacy questions
Question-Version Correct Incorrect Do not know
(2-1)If you are smart, it is easy to pick individual company 37.89 33.59 28.52
stocks that will have better than average returns.
(2-2)If you are smart, it is hard to pick individual company 56.69 26.25 17.06
stocks that will have better than average returns. (2)Pearson(chi(2)=23.12) (p =0.000)
(5-1) When an investor spreads money between 20 stocks 73.83 14.84 11.33
rather than 2, the risk of losing a lot of money decreases.
(5-2) When an investor spreads money between 20 stocks 80.84 8.40 10.76
rather than 2, the risk of losing a lot of money increases (5)Pearson chi(2)=6.70) (p =0.033)
(7-1) )If you invest for the long run, the annual fees 57.81 15.63 26.56
of mutual funds are unimportant (N=256)
(7-2) )If you invest for the long run, the annual fees 68.24 12.60 19.61
of mutual funds are unimportant . (N=381) (7)Pearson(chi(2)=7.41) (p =0.025)
(10-1)Using money in a bank savings account to pay off 68.36 22.27 9.38
credit card debt is usually a bad idea (N=256)
(10-2)Using money in a bank savings account to pay off 73.27 19.16 7.61
credit card debt is usually a good idea (N=381) (10)Pearson(chi(2)=1.81) (p =0.406)
Panel B: Chi-square test six other financial literacy questions
26
6.Conclusion
This thesis investigates the relationship of financial literacy on stock market participation for older American households for 2008 and 2011. Furthermore, this paper observes the financial literacy of unique respondents over these three years. It is especially interesting because the households are observed during the financial crisis. Data from the Cognitive Economics Study of the University of Michigan is used, which provides extensive information on several individual characteristics. In this way, it is possible to investigate other influences of stock market participation. In addition, financial literacy levels can be compared between specific groups.
Overall, this thesis finds that there is a strong positive relationship between financial literacy and stock market participation. This is confirmed by using three different financial literacy indices which measure financial literacy based on the amount of correct answers, confidence or difficulty for ten financial literacy questions. Furthermore, the financial literacy of individuals, who are responsible for the financial decisions in the household, does not increase over three years.
Apparently, the financial crisis does not change the relationship between financial literacy and stock market participation. People who have a higher financial literacy still have a higher chance of being involved in stock market participation after the financial crisis started. In addition, the financial crisis does not change the financial literacy of respondents. This could indicate that the financial awareness of respondents did not change during the financial crisis.
Finally, some remarks should be made. This paper was one of the first to investigate financial literacy for the same respondents for a longer time. Unfortunately, investigating the difference in financial literacy for three years is a short period, especially since a typical financial crisis lasts longer than three years. In addition, the sample was relatively old. Therefore, it is recommended to use a representative sample of the whole population in further research.
27
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.
Andersen, S. , Nielsen, K., 2010, Participation Constraints in the Stock Market: Evidence from Unexpected Inheritance due to Sudden Death, Working Paper University of Hong Kong Australia and New Zealand Banking Group, 2008. ANZ Survey of Adult Financial Literacy in Australia. Melbourne, Australia: Australia and New Zealand Banking Group.
Bernheim, D., 1998. Financial illiteracy, education and retirement saving, in: Mitchell O. and Schieber, S,. Living with Defined Contribution Pensions, University of Pennsylvania Press, Philadelphia, 38-68.
Brockett, P., Derrig, R. , Golden, L., Levine, A. , Alpert, M. (2002) Fraud classification using principal component analysis of RIDITs. Journal of Risk and Insurance, 69, 341–371.
Campbell, J., 2006. Household finance, Journal of Finance, 61, 1553-1604. Christelis, D., Jappelli, T., Padula, M. , 2010. Cognitive Abilities and Portfolio Choice, European Economic Review 54: 18-38.
Cognitive Economics Study 2008 and 2011, public use dataset. University of Michigan Documentation in English.
Fornero, E., Monticone, C, 2011. Financial literacy and pension plan participation inItaly. Journal of Pension Economics and Finance 10(4): 547–564.
Guiso, L., Haliassos,M., Jappelli,T.,2002. Household Portfolios.MIT Press, Cambridge,MA. Guiso, L., Jappelli, T., 2005. Awareness and stock market participation. Review of Finance, 9(4), 537-567.
Grinblatt, M., Keloharju, M., Linnainmaa, J. , 2011. IQ and Stock Market Participation. Journal of Finance 66(6): 2121-2164.
28 Huston, S, 2009. The Concept and Measurement of Financial Literacy: Preliminary Results from a New Survey on Financial Literacy Assessment. Conference Presentation, Academy of Financial Services Annual Conference, Anaheim, CA, October 9.
Huston, S, 2010. Measuring Financial Literacy. Journal of Consumer Affairs, 44: 296–316. Jickling, M. ,2009. Causes of the financial crisis.
Koenen, T., Ziegelmeyer, M. 2011. Who lost the most? Financial literacy, cognitive abilities and the financial crisis. MEA Discussion Paper 234–11.
Lusardi, A., Mitchell, O., 2005. Financial literacy and planning: Implications for retirement wellbeing. Michigan Retirement Research Center Research Paper No. WP, 108.
Lusardi, A.,Mitchell,O.,2007a. Baby boomers retirement security: the role of planning, financial literacy and housing wealth. Journal of Monetary Economics 54, 205–224. Lusardi, A. , Mitchell O. 2011d. Financial Literacy and Retirement Planning in the United States. Journal of Pension Economics and Finance 10(4): 509-525.
Lusardi, A., & Mitchell, O., 2011. Financial literacy around the world: an overview. Journal of Pension Economics and Finance, 10(04), 497-508.
Mandell, A. ,2008. Financial Literacy of High School Students. In Handbook of Consumer Finance
Marcolin, S., & Abraham, A., 2006. Financial literacy research: Current literature and future opportunities.
Merton, R., 1971. Optimum consumption and portfolio rules in a continuous-time model. Journal of Economic Theory 3, 373–413.
Moore, D.,2003. Survey of financial literacy in Washington State: knowledge, behavior, attitudes and experiences. Technical Report 03-39.Social and Economic Sciences Research Center, Washington State University.
29 Stango, V., Zinman J, 2007, Fuzzy math and red ink: When the opportunity cost of
consumption is not what it seems, Working Paper, Dartmouth College.
Van Rooij, M., Lusardi, A., Alessie, R. 2011. Financial literacy and stock market participation. Journal of Financial Economics, 101(2), 449-472.
Vissing-Jorgensen, A. (2004). Perspectives on Behavioral Finance: Does" Irrationality" Disappear with Wealth? Evidence from Expectations and Actions. In NBER Macroeconomics Annual 2003, Volume 18 (pp. 139-208). The MIT Press.
Ward, Lynch, G ,2014. “On a Need-to-Know Basis: Transactive Memory Processes, Financial Literacy, and Financial Outcomes.”
30
Appendix A -Figures
FIGURE 1
31
Appendix B - Financial Literacy Questions
Basic Financial Literacy Questions Lusardi and Mitchell (2006)
1.) Understanding of Interest Rate (Numeracy)
Suppose you had €100 in a savings account and the interest rate was 2% per year. After 5 years, how much do you think you would have in the account if you left the money to grow?
(i) More than €102; (ii) Exactly €102; (iii) Less than €102; (iv) Do not know; (v) Refusal.
2.) Understanding of Inflation
Imagine that the interest rate on your savings account was 1% per year and inflation was 2% per year. After 1 year, how much would you be able to buy with the money in this account?
(i) More than today; (ii) Exactly the same; (iii) Less than today; (iv) Do not know; (v) Refusal.
3.) Understanding of Risk and Diversification
Do you think that the following statement is true or false? Buying a company stock usually provides a safer return than a stock mutual fund.
32
Financial Literacy Questions CogEcon 2008 and 2011
The respondent indicates whether he/she thinks that the statement is true or false. If he/she is not sure, the best guess is given. The correct answer to the question is given in italic words.
Knowledge of capital markets
1.) If the interest rate falls, bond prices will rise.
Sophisticated investor: TRUE
2.) If you are smart, it is easy to pick individual company stocks that will have better than average returns.
Sophisticated investor: FALSE
3.) An employee of a company with publicly traded stock should have little or none of his or her retirement savings in the company’s stock.
Sophisticated investor: TRUE Risk diversification
4.) Buying a stock mutual fund usually provides a safer return than a single company stock. Sophisticated investor: TRUE
5.) When an investor spreads money between 20 stocks, rather than 2,the risk of losing a lot of money decreases.
Sophisticated investor: TRUE
6.) It is best to avoid owning stocks of foreign companies.
Sophisticated investor: FALSE Knowledge of fees
7.) If you invest for the long run, the annual fees of mutual funds are unimportant
Sophisticated investor: FALSE Savvy/numeracy
8.) If you start out with $1,000 and earn an average return of 10% per year for 30 years, the initial $1,000 will have grown to more than $6,000
Sophisticated investor: TRUE
9.) You could save money in interest costs by choosing a 15-year rather than a 30-year mortgage
Sophisticated investor: TRUE
10.) Using money in a bank savings account to pay off credit card debt is usually a bad idea
33
Appendix C – Physiological Variables Risk Tolerance
How would you describe yourself: Are you generally willing to take risks or do you try to avoid risks?
Not at all willing 0 1 2 3 4 5 6 7 8 9 10 Very Willing
34
Appendix D –Supplementary Tables
Table D1- Descriptive statistics Dataset 2
This table presents the summary statistics(mean, median and standard deviation) of the individuals who are responsible for taking the financial decisions in the household. The data consist of 637 members of the CogEcon Survey. Note: The dummies could not sum up to 1 due to rounding.
Variable N Mean Median Std. Deviation
Who makes financial decisions 632
Respondent .40 0 .49 Partner/Spouse .49 0 .50 Both partners .09 0 .28 Other .01 0 .10 Demographic characteristics 637 Female .56 1 .50 Age in 2008 62.65 61 9.19 Age 44-56 in 2008 .28 0 .45 Age 57-61 in 2008 .24 0 .43 Age 62-69 in 2008 .24 0 .43 Age ≥70 in 2008 .23 0 .42 Married in 2008/2011 .72 1 .45 Education characteristics 637 Education in years 14.64 15 2.16 1-12 Years 0.21 0 .40 13-14 Years 0.29 0 .45 15-16 Years 0.24 0 .43 17 Years 0.27 0 .44 College Degree .57 1 .50 Physiological characteristics 637
Low Risk Tolerance .20 0 .40
Medium Risk Tolerance .47 0 .50
35
Table D2 -Methodology Scoring of Financial Literacy Questions
This table provides the methodology for the scoring on each of the financial literacy questions. ‘Categorical Response’ shows the level of certainty that the respondent has for a question that in fact is true. The second column provides the points a respondent gets before the financial literacy indices are made. The third column provides the confidence interval of the respondent. The fourth and fifth column show the score that a respondent can get for each question.
Categorical response Points Belief statement is true(%) Fin. Literacy Index 1 Fin. Literacy Index 2
Surely false 100% 1 0% 5% 0 0.025 90% 2 5% 15% 0 0.10 80% 3 15% 25% 0 0.20 70% 4 25% 35% 0 0.30 60%| 5 35% 45% 0 0.40 Guess False 50% 6 45% 50% 0 0.475 Guess True 50% 7 50% 55% 0 0.525 60% 8 55% 65% 1 0.60 70% 9 65% 75% 1 0.70 80% 10 75% 85% 1 0.80 90% 11 85% 95% 1 0.9 Surely True 100% 12 95% 100% 1 0.975
Table D3 –PRIDIT weights
This table provides the PRIDIT weights, for each financial literacy question, that are used for obtaining financial literacy index 3. The data is from Cogecon2008 and 2011
Financial Literacy Question 2008 2011
36
Table D4: Financial literacy questions with inverted wording and ordering in 2011
This table reports the proportion of respondents providing correct, incorrect and ‘ do not know’ answers for three literacy questions that were asked in four versions with different wording and ordering of questions in Panel A. Respondents were assigned randomly to one of four versions. 137 respondents received Version 1A, 156 respondents received Version 1B,181 respondents received Version 2A and 163 respondents received Version 2B. The table shows the percentages and the Pearson chi-square statistic to test the null hypothesis that wording of the question is not relevant for the distribution of correct, incorrect and ‘do not know’ answers for all ten questions. Questions with a p >0.01 are statistically significant. This means that the distribution of answering is the same for both Versions. The data is from the 2011 Cognitive Economics Study. Panel B shows the chi-square statistics for the other seven financial literacy questions.
Panel A: Wording of three financial literacy questions
Financial Literacy Question Correct Incorrect Do not know
(5-1A)When an investor spreads money between 20 stocks, 75.18 16.06 8.76
rather than 2, the risk of losing a lot of money decreases.
(5-2A)When an investor spreads money between 20 stocks, 77.90 9.94 12.15
rather than 2,the risk of losing a lot of money increases. (5-1A/2A) Pearson chi(2) =3.23 (p=0.199)
(5-1B) 71.15 15.38 13.46
(5-2B) 78.53 6.75 14.72
(5-2A/2B )Pearson(chi(2)= 6.09 (p =0.048)
(7-1A)If you invest for the long run, the annual fees 65.69 10.22 24.09
of mutual funds are unimportant
(7-2A) If you invest for the long run, the annual fees 73.48 6.63 19.89
of mutual funds are important
(7-1A/2A) Pearson chi(2) =2.54 (p =0.281)
(7-1B) 56.41 21.15 22.44
(7-2B) 71.17 12.88 15.95
(7-1B/2B )Pearson(chi(2)=7.69 (p =0.021)
(10-1A)Using money in a bank savings account to pay off 69.34 14.60 16.06
credit card debt is usually a bad idea (N=256)
(10-2A)Using money in a bank savings account to pay off 70.72 20.44 8.84
credit card debt is usually a good idea (N=381) (10-1A/2A) Pearson chi(2) =4.91 (p =0.086)
(10-1B) 59.62 24.36 16.03
(10-2B) 63.80 23.93 12.27
(10-1B/2B )Pearson(chi(2)=1.03 (p =0.598)
Panel B: Chi-square test six other financial literacy questions
37
Table D5-Multivariate analysis of stock market participation in 2008 and 2011: Probit Results Robustness test
This table reports Probit marginal effects of financial literacy and other control variables on stock market participation. The data are from the CogEcon 2008 and 2011. The data consist of 214 respondents, responsible for the financial decisions in the household. Probit 1,3 and 5 are the regressions from 2008 and Probit 2,4 and 6 are the regressions from 2011. Note: *** p<0.01, ** p<0.05, * p<0.1. This table reports estimates of the marginal effect of independent variables on stock market participation.
Variables Probit1 Probit2 Probit3 Probit4 Probit5 Probit6
Fin. Lit. index 1 0.0684*** 0.0697**
Fin. Lit index 2 0.1221*** 0.1155**
Fin. Lit index 3 0.0461* 0.0552**
