The Big Five, Quality of Government, and Stock Market Participation:

Master Thesis

The Big Five, Quality of Government, and Stock Market Participation:

A cross-country analysis in the European Union

For the joint degree 2018-2020

MSc International Financial Management at University of Groningen (RUG) MSc Business and Economics at University of Uppsala (UU)

Supervisor (RUG) Dr. C. Laureti

Second Assessor (RUG) Dr. R.O.S. Zaal

Submitted by Sophie Jasper s3721698 (RUG)

soja0960 (UU)

10.01.2020

Abstract

Until now, the majority of individuals does not hold stocks. In our sample, only 20.5%

of individuals directly or indirectly participate in the stock market. Using data from the SHARE Study (N=54,636), we investigate at the individual level the effect of personality traits and at the country level the effect of Quality of Government while controlling for several socio- demographic factors. We prove that Openness to experience and Agreeableness significantly influence stock market participation. Agreeable people who are less open to new experiences are more likely to hold stocks. We also report mixed effects for Conscientiousness, Extraversion and Neuroticism. Extraversion seems to influence stock market participation through its effect on education and Neuroticism through its effect on both education and Quality of Government.

Additionally, countries with a higher level of Quality of Government have a higher participation rate. Lastly, we find that the strength of the effects of Openness and Extraversion depends on the level of Quality of Government. We contribute to the explanation of the non-participation puzzle and give implications for policy makers.

Field Keywords – stock market participation, personality traits, Big Five, Quality of

Government.

Table of Contents

1 Introduction ... 1

2 Literature review ... 5

3 Hypotheses development ... 17

3.1 The Big Five and stock market participation ... 17

3.2 The interaction between the Big Five and Quality of Government ... 22

4 Data and methodology ... 23

4.1 Data sources ... 23

4.2 Sample construction ... 25

4.3 Variable construction ... 26

4.4 Descriptive statistics ... 33

4.5 Methodology ... 38

5 Results ... 39

5.1 The effect of the Big Five ... 39

5.2 The interaction effect of Quality of Government and the Big Five ... 47

5.3 Robustness tests ... 56

5.4 Limitations ... 58

6 Conclusion ... 59

References ... 63

Appendix ... 68

1 Introduction

Participating in the stock market should improve the financial well-being of individuals and households in the long term (Xia et al., 2014). In theory, all individuals will benefit from participating in the stock market, independent of their wealth and income (Beshears et al., 2018;

Guiso and Sodini, 2013; Haliassos and Bertaut, 1995). Indeed, individuals and households have become a little more active in the stock market in the last three decades (Shum and Faig, 2006).

Lowered participation costs, wider available information about financial markets, and the development of stock markets are some factors which support the growth of the participation rate (Guiso et al., 2003). Still, the stock market participation rates are considerably low.

Previous research has contributed much to the understanding of this non-participation puzzle. At the individual level, scholars examined economic, demographic and cognitive factors. For instance, the level of participation costs influences stock market participation since higher costs are a hurdle for most individuals (e.g., see Guiso et al., 2003; Guiso and Sodini, 2013). The probability to participate in the stock market is higher for individuals with a higher income and education (e.g., see Christiansen et al., 2008; Guiso et al., 2003; Haliassos and Bertaut, 1995). Next to that, individuals who are willing to take higher risks participate more in the stock market (Donkers and van Soest, 1999). Also, intelligent individuals are associated with a higher probability to hold stocks (Guiso and Japelli, 2005).

Some scholars (e.g., see Beshears et al., 2018; Breuer et al., 2014; Shum and Faig, 2006) conclude that these factors are not sufficient enough to solve the non-participation puzzle.

Researchers recently have started to investigate the influence of personality traits on stock

market participation, because personality appears to have an independent predictive power in

explaining stock market participation rates (Bucciol and Zarri, 2017).

A common framework used to measure personality traits is the five-factor model, also referred to as the Big Five. The model consists of five traits, namely Openness to experience, Conscientiousness, Extraversion, Agreeableness and Neuroticism. The advantage of this model is that it categorizes personality traits comprehensively and universally across countries. Most studies that examine the effect of personality traits on stock market participation have been conducted only in one country (e.g., see Brown and Taylor, 2014; Bucciol and Zarri, 2017;

Conlin et al., 2015). Thus, this study aims to contribute to literature by investigating the influence of the Big Five on stock market participation at an international level, in particular, across countries in the European Union (EU). Thus, the first research question of this work is:

Do the Big Five influence stock market participation across countries in the European Union?

Besides the individual factors influencing stock market participation, some prior studies were conducted at a country level. Stock market participation differs across European countries, with Denmark and Sweden having twice as high participation rates as Austria, Spain and Italy (Georgarakos and Pasini, 2011). Scholars investigated economic shocks, the institutional and legal systems as well as the level of trust and confidence in those systems across countries in regard to stock market participation (e.g., see Guiso and Sodini, 2013; Tabellini, 2010). To elaborate more upon those cross-country differences, we introduce a measure that describes to what extent people believe in institutions, laws, equalities and corruption, namely Quality of Government (Charron et al., 2019). Hence, the second research question of this work is: Does Quality of Government influence stock market participation across countries in the European Union?

Previous research has studied the isolated effects of personality traits on stock market

participation. However, the interaction between Quality of Government and the Big Five should

not be neglected, since both the personality and the outside environment of individuals affect

the willingness to participate in the stock market (Xiao et al., 2009). Interestingly, the Big Five and Quality of Government are correlated to trust and trust is positively correlated to stock market participation (Georgarakos and Pasini, 2011; Hiraishi et al., 2008; Tabellini, 2010).

Depending on a country’s Quality of Government, the Big Five may have distinct effects on the decision of individuals to participate in the stock market. Therefore, this study seeks to explore whether differences in the Quality of Government of countries in the EU influence the relation between the Big Five and stock market participation. This leads to the third research question of this study: Does Quality of Government influence the relation between the Big Five and stock market participation?

Our research is conducted in three steps. First, the effect of the Big Five on stock market participation is analyzed. Second, the effect of the country factor Quality of Government on stock market participation is investigated. Third, Quality of Government is interacted with each of the Big Five to investigate whether Quality of Government affects the relation between the Big Five and stock market participation. The empirical analysis is conducted using as sample of 54,636 observations from 28,709 individuals residing in 14 different countries of the EU in the time span 2011 to 2017.

The results of this study show that while controlling for a host of socio-demographic

factors, both the Big Five and Quality of Government do help solving the non-participation

puzzle. We find that Openness to experience and Agreeableness have statistically significant

effects on SMP. Those who score high on Openness are less likely to hold stocks, whereas those

who score high on Agreeableness are more likely to participate in the stock market. Our results

also suggest that, depending on our regression, the effects of Conscientiousness, Extraversion

and Neuroticism are mixed. Extraversion and Neuroticism are likely to influence stock market

participation through their effect on education. In addition, neuroticism may affect the stock market participation of individuals through its effect on QOG.

Furthermore, countries with a higher level of Quality of Government have a higher stock market participation rate. Referring to our interactions, only the interactions Quality of Government and Openness as well as Quality of Government and Extraversion are significant.

The effects of Openness and Extraversion become more influential in countries with higher levels of Quality of Government. The effect of people scoring low on Openness and Extraversion on stock market participation compared to people scoring high on those traits is larger in high Quality of Government countries than in low Quality of Government countries.

This indicates that the effect of Openness and Extraversion is strengthened by a country’s Quality of Government.

This study contributes to solving the non-participation puzzle and giving implications for policy makers to increase stock market participation. The research is one of the first that investigates the effect of the Big Five on stock market participation across countries in the European Union. In addition, this study deepens the understanding of the interplay between personality and the country-specific outside environment, measured by Quality of Government.

The paper is organized as follows. Section 2 presents a review of the literature on the

drivers of stock market participation. It stresses the importance of personality traits as well as

Quality of Government with respect to stock market participation. In section 3, hypotheses for

the relation between the Big Five and stock market participation as well as the interaction

Quality of Government are built. Section 4 describes the data and methodology used in this

study. Section 5 elaborates on the results and limitations of the empirical study. Section 6 draws

the conclusion of our research.

2 Literature review

In theory, all individuals should participate in the stock market to take advantage of the high equity premium. The equity premium puzzle defines the phenomenon of too large differences of the higher expected rate of return on the stock market over the riskless rate of interest (e.g. government bonds) (Mankiw and Zeldes, 1991). Individuals will profit from the equity premium by holding at least some risky assets, as long as their non-stock income is uncorrelated with stock returns (Guiso and Sodini, 2013). This is supported by Beshears et al.

(2018), who state that holding zero stocks cannot be optimal. If an individual does not hold stocks, the returns have no covariance with the marginal utility. Thus, the individual should be risk-neutral when holding a small stock position (Beshears et al., 2018). Individuals that want to maximize their expected utility should always be willing to invest a small amount in assets with a higher expected return (Haliassos and Bertaut, 1995). In addition, the Merton model implies that all investors, independent of their wealth and risk preferences, should participate in all risky asset markets and should invest in the market portfolio (Guiso and Sodini, 2013).

Hence, participating in the stock market is strongly recommended by theoretical predictions.

Empirical studies show a significant difference between theory and reality. Indeed, the

stock market participation rate has slightly increased over time. Higher stock returns in the

1990s, increased financial market liquidity, and wider availability of financial information

supported stock market participation in the last decades (Guiso et al., 2003). In addition, the

privatization of public utilities, increasing competition, and cost reduction in the managed fund

sector reinforced the growth of the stock market (Guiso et al., 2003). Still, most people do not

invest in the stock market. In 1998, only 24% of households participated in the stock market in

Europe either directly or indirectly through mutual funds (Guiso et al., 2003). In addition, the

small amount of households who invest do not hold the market portfolio (Guiso and Sodini,

2013). The low participation rate and the holding of a portfolio deviating from the market

portfolio reveal a puzzling discrepancy between the theoretical prediction and the real observation of stock market participation. This is often known as the non-participation puzzle.

It is of great value to understand individuals’ motives to decide to not participate in the stock market. Individuals’ choices have far-reaching implications concerning their individual wealth and consumption behavior on the one side and the development of financial markets on the other side (Christiansen et al., 2008; Guiso et al., 2003). Previous literature investigated various factors explaining the low stock market participation rate. At the individual level, scholars examined economic, demographic and cognitive factors. At the country level, scholars investigated economic shocks, the institutional and legal environment as well as the level of trust and confidence. We firstly elaborate upon the individual factors in the next parts.

The effect of risk. Prior literature has tried to explain why individuals fail to participate

in the stock market and fail to invest in the market portfolio. The standard theories in finance predict that high-risk investments are rewarded by higher expected rates of returns. Individuals are driven by their cognitive assessment of the optimal trade-off between risk and expected return, before making decisions for their individual investment strategy (Breuer et al., 2014).

Empirical evidence of the effect of risk on stock market participation is mixed. On the one hand, some scholars, for example Donkers and van Soest (1999) and Shum and Faig (2006), show that risk aversion has a negative effect on the decision to participate in the stock market.

Hence, the larger the degree of risk aversion, the smaller the likelihood of participating in the stock market. These studies use qualitative risk measures conducted through simple questions.

Donkers and van Soest (1999) consider a question about the individual’s opinion about their

risk-return trade-off to derive a measure of risk aversion. Also, Shum and Faig (2006) apply a

variable that measures the self-reported risk-attitude with a question about the risk-return trade-

off of an hypothetical investment. However, these questions do not distinguish between risk

aversion and risk perception (Guiso and Sodini, 2013). As a result, some respondents may be more risk-averse in the data collection.

On the other hand, Guiso et al. (2008) use a quantitative measure for risk aversion, in particular a questionnaire about the willingness to pay for lotteries. They discover that risk aversion has little predictive power. This is supported by Breuer et al. (2014) who also employ a set of lottery questions as an actual measure of risk aversion. Their study disprove a significant relation between risk aversion and risk-taking in financial matters. This supports the increasing evidence that risk aversion does not have the sole explanatory power for the individual’s decision to participate in the stock market.

The effect of income and wealth. Low-income individuals participate less in stock

markets than high-income individuals. One explanation has been given by Haliassos and Bertaut (1995). The scholars provide an explanation based on inertia (from cultural influences) and on departures from expected-utility maximization. Departures from the expected utility are especially relevant for low-income households. Intuitively, people with a lower income have less resources available to invest in the stock market and more important, less resources to overcome the departures from their expected utility. Thus, the departures from expected utility may have more, far-reaching effects for those people (Haliassos and Bertaut, 1995).

Additionally, for low-income households, only one single factor, as for example education or risk aversion, can already majorly influence the choice of participating in the stock market, whereas several factors are needed for high-income households (Haliassos and Bertaut, 1995).

Empirical evidence for the effect of income and wealth is mixed. Guiso et al. (2003)

find out that stock market participation in Europe increases with the resources of an investor,

measured by income and wealth. Also, Shum and Faig (2006), who investigate stock holdings

in the United States, prove that several measures of wealth, such as financial net wealth and

labor income, are positively correlated with the decision to hold stocks. However, the scholars exclude households with very low holdings of financial assets or wealth, in particular households with financial net worth lower than $1,000. A substantial fraction of households with a net worth greater than $1,000 still do not participate in the stock market. In 2001, between 72% and 77% of those households did not invest in stocks (Shum and Faig, 2006). In addition, more than 65% of Austrian, Spanish and Greek households within the top 5% of the wealth distribution do not hold stocks (Beshears et al., 2018). Thus, the low participation rate cannot be explained solely with a lack of funds to invest.

The effect of participation costs. The cost of participation is another driver of stock

market participation. Vissing-Jorgensen (2003), among others, support the explanation that wealthier people can more easily access the stock market since they are more likely to be able to afford the participation costs. Her study shows that a very small fixed amount of participation costs would rationalize 75% of non-participation (Vissing-Jorgensen, 2003).

By this time, the costs of participating in the stock market have decreased and have encouraged individuals to participate in it more actively. The availability of information on the internet and reforms in the mutual funds industry allowed investors in the EU to hold diversified stock positions at lower costs than through direct holdings (Guiso et al., 2003; Guiso and Sodini, 2013). Still, despite the decrease in participation costs, the participation rate is below a quarter in Europe (see Figure 1 in section 4.4). This contradicts the observation that participation costs are a main explanation for limited stock market participation. Grinblatt et al. (2011) stress that participation costs do not have an independent power. The scholars state that participation costs are by now so low that they are unlikely to influence the low participation rate.

The effect of cognitive factors. Costs from participating in the stock market include

also the time spent on understanding risk-return trade-offs and general information about stock

markets (Christiansen et al., 2008). Individuals who are better at collecting information about stock markets have lower effective costs which will increase their likelihood to participate (Christiansen et al., 2008). This cost is related to individual cognitive factors such as intelligence, education, financial literacy, and financial awareness.

This theory is supported by several scholars. First, individuals with a high intelligence quotient are more likely to invest in the stock market because these investors are able to make high quality financial decisions (Grinblatt et al., 2011). Second, Haliassos and Bertaut (1995) find that people with at least a college degree are significantly more likely to hold stocks than those who have not attended college. Moreover, the type of education is important for the likelihood of participation. Economists have a higher probability to participate in the stock market (Christiansen et al., 2008). Third, Van Rooij et al. (2011) investigate financial literacy and prove that most people do not have the basic knowledge to enter the stock market. Lastly, a lot of people are not even aware of several financial assets due to the growing complexity of financial markets (Guiso and Jappelli, 2005). Therefore, intelligent and well-educated people with financial knowledge who are aware of financial markets have a higher probability to hold stocks.

This theory is partly disproved by Shum and Faig (2006). The academics examine

whether providing assistance to decrease the costs of gathering information lowers those costs

and the burden to participate. They show that households did not invest sufficiently more in the

stock market when they sought professional investment advice. Even if households with very

low holdings of financial assets are excluded, only 12% to 18% of households invested between

20% and 80% of their financial assets in stock holdings in the United States (Shum and Faig,

2006). Offering financial advice to individuals with a lower intelligence quotient, education,

and financial literacy and awareness does not help mitigating the participation problem.

To draw up an interim balance, prior literature has investigated different factors, such as risk, income, wealth, participation costs, and cognitive function to solve the non-participation puzzle. However, empirical evidence is mixed. Clear is that none of these factors have the exclusive power to explain the non-participation puzzle. Consequently, other barriers to enter the stock market must exist.

The effect of personality traits. Standard finance theory describes the logical and

rational choices that maximize an individual’s welfare (Breuer et al., 2014). In contrast, behavioral finance theory portrays the actual choices of individuals by focusing on exploring the motives behind their heterogeneous behavior (Conlin et al., 2015). This research area has recently started to explain why so many individuals do not participate in the stock market. In particular, light is shed on the role of personality traits in the context of stock market participation. Roberts (2009) reviewed several definitions of personality traits and concluded that these can be defined as “the relatively enduring patterns of thoughts, feelings, and behaviors that reflect the tendency to respond in certain ways under certain circumstances” (p.

140). Thus, personality trait theory tries to measure repeated behaviors and thoughts in similar situations (Roberts, 2009). In the context of economic behavior, Rustichini et al. (2012) declare that personality traits have a comparable or even stronger predictive power than economic preferences. In fact, the scholars suggest that personality trait theory could be integrated in standard portfolio models. Therefore, this research field seems to be important and promising in explaining the low stock market participation.

Prior literature investigated the effect of personality traits on risk aversion and financial

decision-making. Several scholars apply one of the most commonly used taxonomies, the Big

Five, which will be elaborated upon in section 3.1. For instance, Ameriks et al. (2009) prove

that the Big Five are related to investment decisions by retirement account holders. Also, Brown

and Taylor (2014) examine the relation between the Big Five and holding unsecured debt and financial assets in Great Britain. The scholars find, for example, that extraversion and conscientiousness influence the probability of holding credit card debt.

Other scholars further supplement the Big Five with other traits or use several personality trait models. Bucciol and Zarri (2017) investigate the effect of the Big Five and other personality traits on the propensity to take financial risk in the United States. Financial risk is measured by the holding and the amount of risky assets. Cynical hostility, agreeableness and anxiety have a negative effect on financial risk-taking. The connections between heterogeneous personality traits and financial risk-taking contribute to the understanding of the heterogeneity in portfolio allocation (Bucciol and Zarri, 2017). Durand et al. (2008) connect investment behavior and performance in Australia with several personality trait models, including the Big Five. In their empirical paper, the authors find a relation between personality traits and both investment decisions and performance. For instance, extroverted investors have a lower propensity to trade (Durand et al., 2008).

Another personality model is used by Conlin et al. (2015) who examine the relation between personality traits and stock market participation in Finland. The scholars use Cloninger et al.’s (1993) Temperament and Character Inventory model (TCI) instead of the Big Five, that measures four traits that again consist of subscales. One of the four traits, namely harm avoidance, negatively shapes stock market participation. Several sub-traits also correlate with stock market participation. When Conlin et al. (2015) control for risk aversion and wealth, results remain robust. This underlines the independent power of personality traits.

To sum up, this literature review by now shows various factors to solve the non-

participation puzzle. We highlight that one single factor cannot solely explain why individuals

do not participate in the stock market. Special attention is drawn to the relevant role of

personality traits in the context of stock market participation. There is mounting evidence that personality traits have an independent power in predicting stock market participation (Bucciol and Zarri, 2017). However, most scholars study the effect of personality traits on financial decision-making in general. Conlin et al. (2015) are one of the few scholars concentrating on the solely effect of personality traits on stock market participation. Moreover, most studies investigate stock market participation in one specific country, rather than including several countries. However, substantial differences exist in the stock market participation rates across countries (Georgarakos and Pasini, 2011). This research will contribute to the existing literature by exploring the effect of personality traits on stock market participation across several countries in the EU. Therefore, this study is close to prior literature but distinguished in meaningful ways. The next parts will elaborate more upon cross-country differences in the EU.

Cross-country differences. Besides the low stock market participation rate in general,

the non-participation puzzle also relates to large cross-country differences. The average wealth

holdings do not differ substantially between European countries (Georgarakos and Pasini,

2011). Nevertheless, Denmark, Sweden and Switzerland are countries with widespread stock

market participation whereas Austria, Spain and Italy are classified as low-participation

countries. Moreover, below-median net wealth households in Denmark, Sweden and

Switzerland have participation rates twice as high as above-median net wealth households in

Austria, Spain and Italy (Georgarakos and Pasini, 2011). Even though the disparate stock

market participation rates across countries exist, prior research has based explanations for the

discrepancies mainly on individual factors. However, country-level institutional variables also

contribute to the explanation of the non-participation puzzle (Guiso and Sodini, 2013). Thus, it

is of great value to shed special attention on the factors responsible for cross-country

differences. The next part sheds light on economic shocks, the institutional and legal

environment as well as on trust and confidence.

We first focus on the economic rationale behind cross-country differences in stock market participation. First, economic shocks seem to have an effect on stock market participation. One example is the financial crisis from 2007-2009. Guiso et al. (2018) compare risk aversion of households in Italy just before the eruption of the crisis in 2007 with the end of the crisis in summer 2009. Before the financial crisis started, 16% of the respondents were not willing to take any financial risk. A few months after the economy stopped falling, the fraction of individuals who were not willing to take any financial risk increased to 43% (Guiso et al., 2018).

Next to economic shocks, households’ decisions and their outcomes are often influenced by the institutional environment in which they operate (Guiso and Sodini, 2013).

Institutional- and policy-related changes, such as the privatization of public utilities and wider

availability of financial information appear to have an influence on stock market participation

(Guiso and Sodini, 2013). The quality of both formal and informal institutions and governments

are a key factor for different economic developments (Charron et al., 2019). This is further

supported by Farole et al. (2011) who declare that regional performance in general is a result

of formal and informal institutions prevailing in a region. Guiso and Sodini (2013) suggest that

the role of both institutional and regulatory differences should be exploited. Moreover, La Porta

et al. (2006) prove that financial markets do not prosper when left to market forces alone, but

that they must be supported by legal reforms ensuring transparency. Larger stock markets

exhibit more disclosure requirements and liability standards facilitating investor recovery of

losses (La Porta et al., 2006). An appropriate set of securities laws can guarantee several

conditions supporting financial markets, such as keeping payment obligations, issuing

enforceable contracts, and fair and fast judicial processes (Georgarakos and Pasini, 2011; La

Porta et al., 2006). Since these laws differ across countries, they can be a factor influencing

stock markets.

Empirical evidence supports these explanations. Balloch et al. (2015) show that economic shocks are a key factor in explaining stock market participation. Indeed, all countries in Europe were influenced by the financial crisis. Nevertheless, countries were differently affected. The divergence between the European countries’ economic growth has in general increased over time, even though the convergence in the well-being of the European citizens was an initial goal of the EU (Farole et al., 2011). Quality of governance is another factor explaining distinct stock market development (Boadi and Amegbe, 2017). Governance systems which vary across countries affect financial and capital markets differently. Quality of governance is captured by some of the above-mentioned factors, such as law systems, political stability, government effectiveness, regulatory quality, and control of corruption (Boadi and Amegbe, 2017).

Next to these “hard factors”, some scholars investigate “soft factors” such as trust and confidence. For instance, Tabellini (2010) explores variation in the quality of European political institutions in the past to identify the causal role of culture for economic development. Culture in this research is measured by an individual’s values and beliefs, trust in and respect for others, and his or her confidence in the link between individual effort and economic success (Tabellini, 2010). Furthermore, Guiso et al. (2008) and Georgarakos and Pasini (2011) contribute to the literature concerning discrepancies in stockholding across European households by showing that the level of trust is positively associated with stock market participation. Georgarakos and Pasini (2011) prove, just as Tabellini (2010), that historical disparities in political institutions affect household stockholding through prevailing trust. Further, Guiso et al. (2008) argue that individuals must believe that available data is reliable and that the overall system is fair. This is, in turn, connected to the above-mentioned findings of La Porta et al. (2006). While Guiso et al. (2008) point out to the level of trust in institutional and judicial systems, La Porta et al.

(2006) shed light on the general role of laws to ensure transparency and fairness.

Based on the influence of institutional and legal factors as well as the level of trust on stock markets, the relation between stock market participation and Quality of Government will be investigated to explain participation differences across countries. Quality of Government has recently been defined by Charron et al. (2019) as “capturing average citizens’ perceptions and experiences with corruption, and the extent to which they rate their public services as impartial and of good quality in their region of residence” (p. 4). This country factor measures to what extent people believe in institutions, laws, equalities and corruption (Charron et al., 2019).

Quality of Government is the strongest determinant of trust, which in turn fosters economic development (Charron and Rothstein, 2014; Georgarakos and Pasini, 2011; Guiso et al., 2008).

Furthermore, it is highly correlated with variables that capture the well-being of the people living in the different countries (Charron et al., 2019). For instance, high quality governments increase economic growth and foster returns on investments (Rodríguez-Pose and Garcilazo, 2015). Thus, Quality of Government is viewed as a broad, latent multi-dimensional concept that is built upon three pillars: corruption, impartiality and quality (Charron et al., 2019).

Quality of Government is high when a country has low corruption, high impartiality and high quality of public service delivery (Charron et al., 2019). On the contrary, countries suffering from high corruption, weak rule of law and low impartiality display inter alia lower levels of economic development and a lower overall subjective well-being of society (Helliwell and Huang, 2008; Mauro, 2004).

The interaction effect of Quality of Government and the Big Five. Until now, we

assume that the Big Five and Quality of Government affect stock market participation

independently. However, there are reasons to believe that Quality of Government influences

the relation between the Big Five and stock market participation. First, personality traits seem

to be relatively stable across Europe. Schmitt et al. (2007) divide Europe into three different

areas, and results show only small differences in the personality traits between these areas.

Thus, they cannot fully explain the different participation rates in the EU.

Second, investigating Quality of Government as a sole factor, without personality, could not be sufficient enough. Personality should always be taken into account when attitudes and behavioral intentions are measured under the assumption that people will behave according to their attitudes (Brandstätter, 1993). Xiao et al. (2009) emphasize that both the environment and personality affect the willingness to invest. According to Brandstätter (1993), predicting economic behavior from individual personality measures is based on the following assumption:

Individuals who share a personality structure respond in the same way to events or circumstances in their environment. These responses differ from the ones of individuals with other personality structures. In this case, individuals’ behavior regarding the decision to participate in the stock market is dependent on their personality.

Furthermore, individuals develop certain relations with external “objects”. In order to explain behavior, in this case stock market participation, it is relevant to understand the interaction between the characteristics of the individual and the object (Brandstätter, 1993).

Quality of Government will be seen as the external object, which interacts with an individual’s personality characteristics differently across countries. Brandstätter (1993) states that the interaction of the personality characteristics and the object takes place through prior experience.

As already mentioned, Georgarakos and Pasini (2011), Guiso et al. (2008) and Tabellini (2010)

all state that experiences of trust in institutional systems appear to be relevant in explaining

stock market participation. These experiences of trust are, besides other factors, measured with

the country factor Quality of Government. However, trust is not only controlled by those

environmental factors, but also by personality (Hiraishi et al., 2008). Thus, depending on the

level of Quality of Government, the effect of the Big Five on stock market participation could

either be reinforced or mitigated. In this study, it is explored for the first time whether the country factor Quality of Government influences the relation between the Big Five and stock market participation.

3 Hypotheses development

3.1 The Big Five and stock market participation

Most scholars investigate the effect of personality traits on stock market participation with the help of the Big Five (e.g., see Brown and Taylor, 2014; Bucciol and Zarri, 2017;

Durand et al., 2008). This frequently used framework was developed in the late 1980s. It is described as the best omnipresent trait structure, especially because its five dimensions represent personality at the broadest level of abstraction (John and Srivastava, 1999). The model allows to classify individuals’ personalities without losing its descriptive power of behavior (Borghans et al., 2008). In the following part, we develop a hypothesis for each of the five personality traits of the Big Five. The five traits defined hereafter are as summarized by Borghans et al. (2008).

Openness to experience (O). Openness to experience describes the degree to which an

individual requires intellectual stimulation, change or variety. Higher levels of Openness

indicate that they are open-minded, insightful, imaginative, and are willing to learn and to enjoy

new experiences (Borghans et al., 2008). People score low on this trait when they need low

intellectual stimulation and change. On the one hand, Matz et al. (2016) prove that people

scoring higher on Openness are associated with higher levels of spending on entertainment,

eating out, pubs, and tourism. Those people are more likely to consume and to buy and thus,

they may not use their money to participate in the stock market. On the other hand, the trait

Openness to experience is associated with self-employment, and self-employed people are

typically seen as being risk tolerant (Parker, 2004). Therefore, higher levels of Openness may lead to a higher participation probability.

In line with the second argument, prior literature proves that people with a higher Openness to experience score in Great Britain are positively associated with the likelihood of holding financial assets (Brown and Taylor, 2014). Thus, we assume that people who score high on Openness to experience have a higher probability to participate in the stock market.

Therefore, we hypothesize:

Hypothesis 1. Openness to experience will have a positive effect on the likelihood of individuals participating in the stock market in countries of the EU.

Conscientiousness (C). Conscientiousness regards the degree to which an individual is

willing to comply with conventional rules, norms and standards. The trait refers to people who are reliable, organized, and self-disciplined (Borghans et al., 2008). This trait further describes to which extent people favor spontaneous or planned behavior. Conscientious people may be more likely to overcome financial or informational costs associated with participating in the stock market because they are organized and self-disciplined. Those people may remain undeterred when they face challenges, such as the decision to invest in the stock market.

Existing evidence is mixed. On the one hand, both Brown and Taylor (2014) and

Bucciol and Zarri (2017) report in their studies in Great Britain and the United States that

Conscientiousness appears to be less important in regard to stock market participation. On the

other hand, Conlin et al. (2015) prove that the trait persistence correlates positively with stock

market participation. This attitude describes how hard someone is willing to push her or himself

to achieve her or his goals (Conlin et al., 2015). De Fruyt et al. (2000) find a high positive

correlation between persistence and Conscientiousness. Both traits describe people who are

self-disciplined. Therefore, one can expect that Conscientiousness is also positively associated with stock market participation. Hence, we argue that individuals who score high on Conscientiousness are more likely to be stock market participants. Thus, we hypothesize:

Hypothesis 2. Conscientiousness will have a positive effect on the likelihood of individuals participating in the stock market in countries of the EU.

Extraversion (E). Extraversion is the degree to which a person needs attention and

social interaction. An individual that is talkative, social, and outgoing is known to be extroverted (Borghans et al., 2008). Extroverts are further seen by Brown and Taylor (2014) as assertive, ambitious and energetic. These people enjoy interacting with the outside world.

Introverts, on the other hand, spend more time alone and interact less with the outside world (Borghans et al., 2008). The number of social connections allow for more information sharing and the acquisition of advice about stock markets which will decrease the cost of participation (Conlin et al., 2015). Thus, extroverts may be more likely to be stock market participants.

In the literature, contrasting views with regard to the influence of Extraversion on stock market participation can be found. On the one side, Brown and Taylor (2014) and Durand et al.

(2008) support the view that extroverted people have a lower propensity to hold financial assets.

Instead, extroverts have a higher probability to hold unsecured debt (Brown and Taylor, 2014).

On the other side, some scholars support the assumption that Extraversion is positively

correlated with stock market participation. Hong et al. (2004) and Christelis et al. (2010)

support this view by showing that the probability of holding stocks is higher for socially active

people, e.g. who interact with neighbors or attend church. Also, De Fruyt et al. (2000) find a

high negative correlation between Extraversion and harm avoidance. Harm avoidance is

defined to range from neurotic introversion to stable Extraversion and is negatively correlated

with stock market participation (Conlin et al., 2015; De Fruyt et al., 2000). Thus, Extraversion

may positively influence the likelihood of holding stocks. Lastly, extroverts show more overconfident behavior which positively influences stock market participation (Grinblatt and Keloharju, 2009). As a conclusion, extroverts are expected to more likely be stock market participants. Hence, we hypothesize:

Hypothesis 3. Extraversion will have a positive effect on the likelihood of individuals participating in the stock market in countries of the EU.

Agreeableness (A). Agreeableness regards the degree to which a person needs pleasant

and harmonious relations with others. Individuals who score high on this trait are likely to be kind, tactful, cooperative and group-oriented. Individuals scoring low on this trait are seen to be more distant, hostile and self-interested (Borghans et al., 2008). People scoring low on Agreeableness are more prone to take financial risk because they try to earn more for themselves (Bucciol and Zarri, 2017). People scoring high on Agreeableness may spend more on others and thus, may have less money to invest in the stock market (Choi and Laschever, 2018).

In line with this, Brown and Taylor (2014) and Bucciol and Zarri (2017) report a

negative correlation between Agreeableness and stock market participation. These findings are

in accordance with the empirical paper of Gambetti and Giusberti (2012). The scholars report

that anger is positively associated with the willingness to participate in the stock market. Anger

is a feeling that can range from mild irritation or annoyance to fury and rage (Gambetti and

Giusberti, 2012). Thus, people who score high on anger are not friendly, tactful and

harmonious. This description fits to people who score low on Agreeableness. Thus, we

hypothesize:

Hypothesis 4. Agreeableness will have a negative effect on the likelihood of individuals participating in the stock market in countries of the EU.

Neuroticism (N). Neuroticism is the degree to which a person experiences the world as

threatening and beyond their control. This trait is also often called emotional stability. Neurotic people are often moody and experience stress. People who score low on this trait are stable and balanced (Borghans et al., 2008). For neurotic people, stock markets may be viewed as dangerous. Participating in the stock market comes along with taking some amount of risk and accepting some amount of uncertainty. Neurotic people may focus on these risks and uncertainties instead of seeing the potential reward offered by the market. Neurotic people are thus less likely to be stock market participants.

Brown and Taylor (2014) report no significant influence of neuroticism on the propensity to hold financial assets. Several scholars, however, report a negative effect of neuroticism on stock market participation. Rustichini et al. (2012) find a strong negative correlation between Neuroticism and the willingness to take risk. Also, Conlin et al. (2015) investigate that the personality trait harm avoidance negatively influences stock market participation. This trait is strongly and positively correlated to Neuroticism (De Fruyt et al., 2000). Attitudes defining harm avoidance are pessimism, fear of uncertainty, shyness and fatigability (Conlin et al., 2015). These attitudes are in common with those that define Neuroticism. Additionally, several scholars (e.g., see Bucciol and Zarri, 2017; Gambetti and Giusberti, 2012) find that anxiety encourages non-participation in the stock market. An anxious person is defined to be apprehensive, worried and avoidant. Anxiety and Neuroticism are therefore very similar in their description. Hence, we hypothesize:

Hypothesis 5. Neuroticism will have a negative effect on the likelihood of individuals

participating in the stock market in countries of the EU.

To sum up, the Big Five have different effects on stock market participation. Whereas Openness to experience, Conscientiousness and Extraversion are hypothesized to positively influence stock market participation, Agreeableness and Neuroticism are hypothesized to have a negative effect on stock market participation.

3.2 The interaction between the Big Five and Quality of Government

With the help of the country factor, we investigate whether Quality of Government has an influence on the relation between the Big Five and stock market participation. As mentioned in section 2, Quality of Government reflects to what extent people believe in laws, equalities and corruption and how much people trust formal and informal institutional systems (Charron et al., 2019). A lower overall well-being of a country’s society will reduce its inhabitants’ stock market participation (Guiso and Sodini, 2013). Thus, a higher well-being of people is expected to increase stock market participation. Hence, we hypothesize:

Hypothesis 6. Quality of Government will have a positive effect on the likelihood of individuals participating in the stock market in countries of the EU.

Furthermore, it is of high interest to interact Quality of Government with the Big Five.

Recall that the relation between individuals’ personality and their behavior, in this case stock market participation, is influenced by the relation between individuals’ personality and the external object, in this case Quality of Government (Brandstätter, 1993). In order to explain stock market participation, it is relevant to understand the interactions between the characteristics of the individuals and the object. Individuals develop relations with the object through prior experiences (Brandstätter, 1993). Quality of Government captures citizens’

perceptions and experiences with institutions, laws, equalities and corruption and is a strong

determinant of trust. Trust is further affected by personality (Hiraishi et al., 2008). Openness to

experience, Conscientiousness, Extraversion and Agreeableness are positively associated with

trust (Hiraishi et al., 2008). Agreeableness and Extraversion have the strongest correlations to trust. Neuroticism is negatively associated with trust, indicating that neurotic people trust normally less (Hiraishi et al., 2008). Thus, someone scoring high on Openness to experience, Conscientiousness, Extraversion and Agreeableness and low on Neuroticism trusts more.

Since a high Quality of Government is also associated with a high level of trust, Quality of Government may positively or negatively reinforce the effect of the Big Five on stock market participation. The outcome of this interaction effect depends on the following: the level of Quality of Government in the respective country; and the positive or negative influence of each personality trait on stock market participation. As a conclusion, we assume that Quality of Government influences the relation between the Big Five and stock market participation.

Hence, we hypothesize:

Hypothesis 7. The effects of the Big Five on stock market participation are influenced by the level of Quality of Government.

4 Data and methodology 4.1 Data sources

The database SHARE. For this research, data from two different databases is required.

To firstly investigate the effect of the Big Five on stock market participation, the Survey of Health, Ageing and Retirement in Europe (SHARE) has been used. SHARE is a multidisciplinary and cross-national database that offers micro-level data about individuals’ life histories, housing and financial history, and employment and health history (Bergmann et al., 2019). Households consisting of a main respondent aged 50 years or older and having their residence in Europe are interviewed. The minimum age can be criticized as a constraint.

However, Georgarakos and Pasini (2011) argue that “older households control a large fraction

of society’s resources and therefore their investment decisions can have broader

macroeconomic implications” (p. 695). Thus, our research is not affected by the limitation in respondents’ age.

The first wave of data was collected by SHARE in 2004. Since then, SHARE has been continuously collecting and expanding the number of households and countries participating in the questionnaire. In addition, SHARE is adding new questions and modules for every new wave. In 2017, SHARE collected the latest data for wave 7. For this work, individuals who participated in a SHARE interview between 2011 to 2017 will be considered. Using this time period ensures that the research will not be affected by the financial crisis in 2007/2008. It should be mentioned that SHARE collected data on personality traits for the first time in wave 7, meaning that the other waves used in this thesis do not offer this particular information.

However, the prevailing view in literature is that personality traits become stable among working age people, with rank-order stability for many traits peaking after age 50 (e.g., see Borghans et al., 2008; Cobb-Clark and Schurer, 2012; McCrae and Costa, 1994). Since respondents of SHARE are 50 years or older, our research will not be affected by the problem of potentially unstable personality. Bucciol and Zarri (2017) also used an unbalanced panel of data with personality traits only measured in the last year of the time span. Hence, we follow Bucciol and Zarri (2017) and use the data of SHARE wave 4 to wave 7, with time-invariant personality traits measured in wave 7. Wave 4, 5, and 6 will be complemented with the data about personality traits in wave 7. Data for wave 4, 5 and 6 were collected in 2011, 2013, and 2015. Wave 7 was collected in 2017 and published in the beginning of 2019.

The EQI. Second, the influence of the country factor Quality of Government on the

relation between the Big Five and stock market participation is investigated. For this, the

SHARE data is supplemented with the European Quality of Government Index (EQI) funded

by the EU Commission. The EQI is based on a large citizen survey in which respondents are

asked about perceptions and experiences with corruption and with public services in their region of residence. The goal of this survey is to measure the extent to which respondents believe public sector services are impartially allocated and of good quality. This index is the first providing valid data for comparing Quality of Government across regions within and across countries (Charron et al., 2019). The EQI dataset was collected in three rounds, in particular in 2010, 2013 and 2017. The last data collection from 2017 was released in the beginning of 2019.

Similar to the data released by SHARE, all rounds from 2010 to 2017 will be used.

4.2 Sample construction

As mentioned above, we first use data from SHARE wave 4 to wave 7, which was collected in a two-year interval from 2011 to 2017. In 2011, the database consisted of 16 participating countries. In 2017, SHARE expanded its database to 27 countries. Nevertheless, not all countries are regarded for this analysis. SHARE recommends using only countries in which the personality traits have a congruence coefficient above 0.85 to ensure a fair similarity of the theoretical Big Five (Bergmann et al., 2019). In order to construct the dataset, data from 2011, 2013, 2015 and 2017 are appended using a unique identification number provided by SHARE for each individual. Observations with missing values for the dependent variable, independent variables and control variables were not considered by the statistical software STATA that we use for our research. Thereby, complete data and reliable regression outcomes are ensured. Also, SHARE recommends to not rely on observations with insufficient answers.

These answers can be identified with several codes provided by SHARE. We follow SHARE’s suggestion to change insufficient answers to missing values.

To arrive at the final dataset, SHARE data is further combined with data from the EQI.

Thereby, the sample is restricted to countries that are EU member states, since data for the

country factor Quality of Government is only collected within the EU. SHARE data was

combined with the mean of the EQI score for each country with the scores from 2010, 2013 and 2017. The datasets for 2010, 2013 and 2017 have shown that the EQI score did not change fundamentally. Therefore, a mean is suitable for the analysis of the influence of Quality of Government on the relation between personality traits and stock market participation. Further, the focus was rather set on differences between countries than on changes within countries. The final sample consists of the following 14 countries: Austria, Germany, Sweden, Spain, Italy, France, Denmark, Belgium, Czech Republic, Poland, Luxembourg, Portugal, Estonia, and Croatia.

4.3 Variable construction

Dependent variable. The dependent variable is stock market participation (SMP). A

respondent is a stock market participant when he or she directly or indirectly participates in the stock market. If a respondent engages in both, direct and indirect SMP, he or she is also considered to be a stock market participant. All other respondents are non-stock market participants. Hence, the dependent variable SMP is a binary variable taking the value of 1 for stock market participants and 0 otherwise.

SHARE defines SMP as a form of investment that allows a person to own a part of a

corporation and gives this person the right to receive dividends from it. According to Conlin et

al. (2015), Georgarakos and Pasini (2011) and Guiso et al. (2008), stock asset shares can be

either held directly or indirectly, e.g. through mutual funds. The SHARE database asks

households multiple questions about their holdings of financial assets held in various forms. In

accordance with the scholars, SHARE differentiates between direct and indirect SMP. SHARE

asks respondents whether they currently have any money in stocks or shares that are listed or

unlisted on the stock market to determine whether he or she participates directly in the stock

market. The answer can either be “yes” or “no”. If the respondent answers yes, he or she is a

direct stock market participant. SHARE also offers information of respondents who have indirect holdings in mutual funds or managed investment accounts. Indirect SMP is derived by asking respondents whether they currently have any money in mutual funds or managed investment accounts. Again, the answer can either be “yes” or “no”. If the respondent answers no, he or she is not considered to be an indirect stock market participant. Further, in the questionnaire, the database uses a (rather simple) distinction of the type of mutual funds by differentiating between stocks and bonds. Since both stocks and bonds are risky investments, here it is not distinguished between those two investments.

Independent variables. The traits of the Big Five, namely Openness to experience,

Conscientiousness, Extraversion, Agreeableness and Neuroticism, are our independent,

explanatory variables. In wave 7, SHARE introduced a personality inventory to measure the

five personality traits of the Big Five. Waves 4 to 6 will be complemented with this personality

inventory for respondents who participated in SHARE wave 7 and in one or more of the waves

before. SHARE follows Rammstedt and John (2007) for measuring the traits. Rammstedt and

John (2007) developed an ultra-short Big Five questionnaire (BFI-10) consisting of only two

items for each personality trait to provide specific multi-theme surveys with an efficient

questionnaire in regard to time and space. Table A1 in the appendix presents all questions from

the BFI-10 used to measure the personality traits of the respondents. Respondents were then

asked to rate themselves on a five-point scale. The scale ranges from 1 (“fully disagree”) to 5

(“fully agree”). SHARE further generated five personality trait variables that are derived from

the ten questions in order to facilitate the measurement of each personality trait. Each

personality trait variable is aggregated from the two respective items. The variable consists of

the mean of the pair of respective items. Thus, a low score indicates that a trait is not strongly

pronounced in a respondent’s personality whereas a high score indicates that a trait is strongly

pronounced.

This abbreviated questionnaire of the Big Five experiences sufficient reliability and has been used several times in prior research for measuring personality (e.g., see Brown and Taylor, 2014). However, the possibility of biased results exists due to acquiescent responding. Rating scales are often biased, because respondents generally tend to agree with items by selecting

“yes” or “true”, regardless of the content. To circumvent this problem, SHARE in advance formulated one item in the positive direction and one in the negative direction within each pair of items reflecting the same personality trait. With this procedure, SHARE avoids acquiescent responding. The database further analyzed respondents’ answers. After correcting for acquiescence, the Big Five structure emerged with greater clarity.

Quality of Government. As mentioned in section 4.1, we use data from the EQI and

create a variable (QOG) by the construction of a mean of the EQI score for each country with the data from 2010, 2013 and 2017. The EQI score is constructed in the following way:

Respondents of the survey were asked 29 questions which had to be either answered with “yes”

or “no”, or with a scale from “strongly disagree” to “strongly agree”, or “very poor” to

“excellent quality”. With these questions, a score for each country is calculated. This score is added up with another score calculated by the country’s average from the World Bank’s WGI (Worldwide Governance Indicators) data from four indicators: control of corruption, government effectiveness, rule of law, and voice and accountability. The final score is standardized within its sample of 30 European countries. Furthermore, the EU Commission provides a normed score, ranging from 0 – 100. We use the normed score to create the variable.

Control variables. To measure the effect of personality traits on SMP as well as the

influence of the country-specific QOG precisely, controlling for additional variables affecting

SMP is required. First, in line with several scholars in the research field of SMP, there will be

controlled for age (Age) (e.g., see Brown and Taylor, 2014; Conlin et al., 2015; Georgarakos and Pasini, 2011). Guiso et al. (2003) report different SMP rates for young and old people.

Second, it is of relevance to control for gender. All scholars investigating factors that influence SMP control for gender (e.g., see Conlin et al., 2015; Georgarakos and Pasini, 2011).

In addition, recent evidence shows that gender still influences SMP (Almenberg and Dreber, 2015). Females have a lower propensity to participate in the stock market (Conlin et al., 2015).

Gender (G) is a binary variable, taking the value of 1 for female respondents and 0 for male respondents.

Third, Bucciol and Zarri (2017) and Grinblatt and Keloharju (2009), besides others, control for the marital status (MS). Married people are less likely to participate in the stock market (Conlin et al., 2015). Respondents will be classified into two groups: being married or in a registered partnership, and being unmarried (never married, divorced, widowed). Thus, this variable is a binary variable, equaling 1 if respondents are married and 0 otherwise.

Fourth, in line with several scholars, we control for the number of children (NCH) (e.g., see Conlin et al., 2015; Georgarakos and Pasini, 2011). People having more children are less likely to participate in the stock market (Conlin et al., 2015). This variable ranges from 1 to 17.

Fifth, Rosen and Wu (2004) show that the health status (HS) is a significant predictor for SMP. Households in poor health conditions will hold a larger proportion in safe assets instead of risky assets in the stock market (Rosen and Wu, 2004). SHARE offers several variables to measure health, but we use a respondent’s self-reported health status, following both Rosen and Wu (2004) and Brown and Taylor (2014). Respondents of SHARE were asked to rate their health with five options: excellent (1), very good (2), good (3), fair (4), poor (5).

Again, a binary variable will be used. This variable takes the value of 1 if respondents report

an excellent (1), very good (2) or good (3) health status, and 0 if respondents have a fair (4) or poor (5) health status.

Sixth, income is considered to be another factor that must be controlled for (e.g., see Gambetti and Giusberti, 2012). Conlin et al. (2015) prove, as expected, that income increases the likelihood to participate in the stock market. Our conduction of research is based in several countries. Thus, differences in a country’s income average as well as its purchasing power make it difficult to use an absolute, objective measure. To enhance the explanatory power of the variable, a relative measure will be created. SHARE derives the total household income per month by asking how much the overall income, after taxes and contributions, was that the entire household had in an average month in the last year. Since SHARE asks this question only to one respondent of a household, the household income is added to observations of other household members. The monthly income will be multiplied by 12 months to receive the annual total household income (ICY). A respondent’s annual total income will then be classified into four percentage quartiles (ICQ) for each country. For example, quartile 3 indicates that the respondent has an annual total income higher than the median earner but lower than the top 25% earner in his or her country of residence. To ensure the explanatory power of the quartiles, income is winsorized at the 0.1% and 99.9% level. Some respondents specified very high income which skewed the country average. Therefore, income is relativized.

Seventh, several scholars control for the labor status because stock market participants

are less likely to be unemployed (e.g., see Grinblatt et al., 2011). In particular, there are

distinctions between working, unemployed and retired respondents (Georgarakos and Pasini,

2011). SHARE differentiates between the following groups: employed or self-employed,

unemployed, permanently sick or disabled, homemaker, retired, other. In this research, a binary

variable (UE) controls for the employment status by taking the value of 1 if respondents are

unemployed and 0 otherwise. Furthermore, we do not control for the retirement status, even though a large fraction of respondents was already retired. The variable age is strongly correlated with the retirement status. Thus, these two variables try to explain the same variance in the regression model. For this reason, the retirement status can be excluded.

Eighth, there will be controlled for education (e.g., see Gambetti and Giusberti, 2012;

Grinblatt et al., 2011; Guiso and Jappelli, 2005). We follow Grinblatt et al. (2011) who break down education into several categories in their research in Finland. SHARE provides data of respondents’ educational level by using the International Standard Classification of Education (ISCED) 1997. This instrument makes an international comparison of education possible and consists of six stages. ISCED 1997 begins with stage 1 being the primary education starting between the age of 5 to 7. The final stage of secondary education belongs to stage 3. The highest is stage 6 which is the second stage of tertiary education. This stage leads to the award of an advanced research qualification, e.g. a Ph.D. For this work, two binary variables were created:

The first binary variable (SE) equals 1 if respondents reached the final stage of secondary education (stage 3) and 0 otherwise. The second binary variable (TE) equals 1 if respondents have graduated from university, thus having reached at least stage 5, and 0 otherwise.

Ninth, scholars agree on the influence of cognitive function (CF) on SMP (e.g., see Grinblatt et al., 2011). SHARE provides different variables as proxies for cognitive function.

In line with Christelis et al. (2010) who also use SHARE for their study about cognitive function and stock market investments, their cognitive indicator has been employed as a control variable.

The scholars use the respondents’ ability to recall ten words correctly out of a list that is read

to them by the interviewer. A 0 indicates that the respondent did not remember one word,

whereas a 10 indicates that the respondent remembered all words.