Ambiguity aversion and household‘s investment decisions: The role of ﬁnancial advice

(1)

Ambiguity aversion and household‘s investment

decisions:

The role of financial advice

Ivaylo Momchilov

University of Groningen, Faculty of Economics and Business

MSc Finance

Supervisor: Dr. Annika Mueller

Abstract

The aim of this paper is to explore whether professional financial advice can mitigate the negative effect of ambiguity aversion on stock market participation and foreign stock ownership in the Netherlands. According to the literature, ambiguity aversion is associated with less participation in the capital mar-kets and less ownership of stocks from foreign companies. On the other hand, financial advice can improve individual‘s decision-making when considering in-vestment decisions. It is also associated with more stock market participation and more diversified portfolios. Using a large panel data from DNB Household Survey (DHS) for the period of time between 2004 and 2016, ambiguity aver-sion is measured with similar questions to the Ellsberg paradox experiment. Thus, it is possible to test whether financial advice can mitigate ambiguity and hence lead to more participation in the capital markets and more investments in foreign stocks by Dutch individuals. The findings of this thesis suggest that professional financial advice mitigates the effect of ambiguity aversion on stock market participation and foreign stock ownership. This means that financial advice can improve individual‘s decision-making and thus it can potentially lead to more investments in the capital markets which has large practical im-plications for policy makers.

Keywords: Ambiguity aversion, financial advice, stock market participation, foreign stock ownership, DHS

(2)

1. Introduction

In economic theory, ambiguity aversion is the preference for known over the un-known risks. Ellsebrg (1961) found evidence in his experiment that the majority of people are ambiguity-averse because in a situation where they have to choose between a lottery of known probabilities to a similar lottery with unknown probabilities, they choose the one with the known outcomes.1

In the field of Finance, previous studies like the one from Dimmock et. al (2016) found the importance of ambiguity attitudes toward investment decisions, showing a negative relation between ambiguity aversion and stock market participation for people who consider stock returns as ambiguous. The same authors argue that even in case an individual already own shares of companies, very often those shares are only of companies from the individual‘s home country because stocks from foreign firms are also argued to be ambiguous. This is because ambiguity plays a role in the so-called home country bias and Guidolin and Rinaldi (2013) reported that investors, who do not invest in foreign stocks, experience under-diversification, often leading to lower return and higher risk for their investment portfolios.

According to Posner (1998), people are not rational and suffer from different be-havioral biases. A typical one is ambiguity aversion, which decreases individual‘s willingness to invest and it is one of the reasons why investors hold under-diversified portfolios, argued in Dimmock et al. (2016). While the literature already found that this cognitive bias is associated with less stock market participation and less diversified portfolios, there are no studies that test empirically how ambiguity could be mitigated in the context of investment decisions. This thesis aims to test one possible way how ambiguity can be alleviated when individuals are considering in-vestments in the stock market and inin-vestments in foreign companies. Von Gaudecker (2015) proposed financial education and financial advice as tools which can possibly improve individual‘s investment choices and mitigate behavioral biases. His results

(4)

show that an increase of the availability of financial advice is more promising and more convenient for policy makers to mitigate behavioral biases of the society than financial education in terms of financial decision-making.

In this thesis, I investigate whether professional financial advice has an impact on both the relation between ambiguity aversion and stock market participation and the relation between ambiguity aversion and foreign stock ownership. Using data from DNB Household Survey for the period of time between 2004 and 2016, I measure ambiguity aversion with a similar to Ellsberg (1961) experiment questions. I argue that professional financial advice can reduce the effect of ambiguity-aversion when an individual is considering investment in the stock market and investment in for-eign stocks. Hence, I claim that financial advisers can help people to overcome this behavioral bias and become more rational in their financial choices.

The results in this thesis are straightforward and show that professional financial ad-vice has a positive and significant impact on the relation between ambiguity aversion, owning shares and foreign shares. This means that professional financial advisor can reduce the negative effect of ambiguity aversion on stock market participation and foreign stock ownership which can potentially lead to more investments in the capital markets and more diversified portfolios. Moreover, the results are robust when the most ambiguity-averse people are observed. The robustness test shows that the more ambiguity-averse is the individual, the more professional financial advisor can miti-gate ambiguity for this individual. The results also remain robust when the outliers in the sample are excluded.

(5)

de-cision to invest in ambiguous assets and they only suggest this link and did not test it. Hence, the results from this study contribute to Behavioral Finance field. They also further support the findings of Von Gaudecker (2015) that financial advice can be a strong tool for mitigating behavioral biases (de-biasing) of individuals when they make investment decisions. However, he studied the impact of financial advice on household portfolio diversification and not the impact of financial advice on the relation between ambiguity aversion, stock market participation and foreign stock ownership. The findings of this thesis have large practical implications for policy makers because they have to pay more attention to the availability of financial ad-vice for households in order for them to make better investment decisions.

This thesis is structured as follows: the second part will explain the ambiguity aver-sion phenomena and summarize the empirical findings of previous studies for the relation between ambiguity preferences and investing in the stock market and own-ing foreign shares. Then financial advice is goown-ing to be discussed as a de-biasown-ing mechanism. Part three will provide more information about the data collection process and the methodology used for the empirical analysis. Part four shows the empirical results of this study and a robustness test of these results. Part five is going to discuss the findings from this thesis and part six provides a conclusion.

2. Literature Review

2.1. Ambiguity aversion and how does it affect the investment

deci-sions:

(6)

2 urns – the first one contains 50 black balls and 50 white balls, while the second one contains 100 balls with unknown proportions of black and white balls. Most of the people in this experiment chose urn 1 - the one with known probabilities. According to Fox and Weber (2002), Ellsberg‘s experiment has a large contribution to the lit-erature because it shows that the experiment violates the subjective expected utility theory because people prefer the 50/50 lottery rather than the one with unknown probabilities, although the sum of the probabilities of the final outcome are exactly the same. In addition, the experiment suggests that people, in general, choose based on preference for the known, and do not prefer the unknown. Different studies stated some possible explanations for this human behavior, however, only the first explana-tion is relevant to this thesis:

Heath and Tversky (1991) argued that the willingness to play in a game with uncer-tain outcome depends not only on the estimated probability of winning this game but also on the individual‘s competence and knowledge about the game. They also show that if the estimations of the probabilities remain constant, people prefer to play a game for which they consider themselves knowledgeable. This applies also to the Ellsberg example and also to this study. People avoid the urn with unknown probabilities because they feel less competent because there is less information about the proportion of black and white balls in the urn. The same holds true when people are considering investment in the stock market – if they don‘t feel competent enough, they most probably won‘t invest.

(7)

Another interesting findings for people‘s ambiguity attitudes in the context of the Ellsberg experiment were made by Perner and Kuhberger, (2003). They distinguish two situations with respect to this experiment: if people find the experiment like a competition with other people (the competitor is responsible for the composition of the boxes), then people tend to avoid ambiguity and choose the risky box. On the other hand, if the experiment is represented as team game, then people are indifferent or sometimes they are even ambiguity-seeking.

2.2. Ambiguity aversion, stock market participation and foreign stock

ownership:

(8)

There is another important role of ambiguity aversion in the context of investing in the stock market. The literature states that the so-called home bias puzzle is mainly due to the agent‘s level of ambiguity aversion. According to Dimmock et al. (2016), home bias puzzle is related to a situation, where investors do not invest in foreign stocks because they are ambiguous and there is lack of information about them. Barberis and Thaler (2003) proposes that ambiguity aversion is a reasonable explanation about why individuals do not hold sufficiently diversified portfolios in-cluding international stocks. According to Huberman, G. (2001), familiarity is the most important determinant of the home bias. People tend to invest in stocks which are in their home country or region because they are more familiar with them and lack of information about international companies usually discourage people in their decision to obtain shares from a foreign company.2 Lewis et al. (1999) further claim that U.S. investors should hold about 50 percent in foreign stocks rather than 10 percent as they do when we consider mean-variance analysis and the value-weighted world equity indices.

Clearly, theoretical models and previous studies like the one from Dimmock et al. (2016) already explored that the ambiguity aversion is negatively related with stock market participation and foreign stock ownership, therefore, this will not be the main focus of this thesis. The emphasis of this research will be on the impact of financial advice on the relation between ambiguity aversion, owning shares and foreign shares.

2.3. The impact of financial advice:

The role of financial advice is important for the household‘s investment decision-making. Households rely on financial advice for different choices like mortgages, retirement and investing in the stock market. According to Von Gaudecker (2015), the sources of advice are different but the most common ones are usually from par-ents or relatives, internet, financial magazines (books) or from a professional financial

2_{There are also rational explanations about the home bias which are not due to behavioral}

(9)

advisor. This study focuses on the latter source of advice. Professional financial ad-vice is needed, and in some cases mandatory for households who consider making a new investment because the financial product that they choose has to meet their risk tolerance and their available resources. A later study written by Bergstresser, Chalmers and Tufano (2009) found that most of the US households invest in mutual funds and equities after consulting with a professional adviser. While investing in mutual fund is not considered as risky investment, investing in the stock market is, and as more risky and complex the financial decision become, the more neces-sary is the advice from professionals. Lusardi and Mitchell, (2014) mention that for simple financial decisions like opening a deposit account, advisor is not needed but for retirement plans or investment decisions which typically require more advanced understanding in Finance, professional advisor is necessary. Assuming that financial advisor is required for more complex decisions, Georgarakos and Inderst (2011) per-formed a research on the impact of financial advice on stock market participation with a sample of European households and they suggested that financial advice is a key determinant of household‘s willingness to invest in risky assets, however this advice matters most for households with relatively low available wealth. In this sit-uation for households it is crucial to have trust in the adviser.3

The impact of the financial advice could be both positive and negative. Kramer (2012) found evidence that it has a small but significant positive impact on in-vestor‘s portfolio performance. Shapira and Venezia (2001) further argues that port-folios which are managed by professionals are better diversified and those portport-folios experience weaker disposition effect.4 Collins (2010) contributed to the literature by showing evidence that professional financial advisers can help the households to reduce their debt. Furthermore, Von Gaudecker (2015) finds that households which rely on both professional or non-professional perform financial advice perform well in terms of investment outcomes. On the other hand, households which rely on their

3_{This is one of the reasons why trust is included as a control variable in the empirical model of}

this paper.

4_{The disposition effect is the propensity of investors to sell stocks whose prices are increasing}

(10)

own decisions have worse investment outcomes. Despite the fact that studies like the one from Von Gaudecker (2015) focus on the positive impact of financial advisers for the society, there are also academic papers which put emphasis on the negative aspects of financial advice.5

2.4. Ambiguity aversion and professional financial advice:

The impact of financial advice on ambiguity could be observed as a de-biasing mechanism, which can have a positive outcome for people who are ambiguity-averse toward participating in the stock market, as well as for those who already partici-pate in the stock market but do not diversify their portfolio with international assets. This can be found later in Figure 1. Currently, there is lack of literature about the role of financial advice in the relation between ambiguity aversion and stock market participation. The only paper which suggests a link between ambiguity aversion and financial advice is from Engle-Warnick,Pulido and Montaignac (2016) but they study how trust affects investment decisions and not how financial advice impact the rela-tion between ambiguity aversion and investment decisions. Thus, in order to further support the model of this thesis, a study which has a similar idea will be discussed because there is no available literature for the link between ambiguity aversion and professional financial advice. The closest to ambiguity aversion behavior which has a relation with stock market participation and is positively influenced by professional financial advice is the risk aversion. There are studies where, risk and ambiguity are considered as correlated to each other. The paper of Butler and Guiso (2011), claims that risk aversion and ambiguity aversion are not independent, showing that those who do not like risk also do not like ambiguity. Dimmock et al. (2013) also argue that there might be a positive correlation between risk aversion and ambiguity aversion. Lee et al. (2015) found a relation between risk-aversion and stock mar-ket expectations (and participation), while including financial advice in their model. They found that the source of financial advice and the financial advice itself has

5_{Bergstresser, Chalmers and Tufano (2009) found out that advisers are associated with higher}

(11)

influence on the relation between risk-aversion and stock market expectations (and participation). Since in their research financial advice has a positive impact on the relation between risk-aversion and stock market expectations (and participation), it might also have positive influence on the relation between ambiguity aversion and stock market participation. Introducing an interaction term between financial ad-vice and ambiguity aversion in the relation between ambiguity aversion, stock market participation and foreign stock ownership is the main contribution of this thesis to the literature. The findings of the literature suggest that financial advice should be able to mitigate the effect of ambiguity aversion on stock market participation and owning foreign company shares. Based on the past literature this paper propose the following hypothesis:

H1: Financial advice mitigates the negative effect of ambiguity aversion on stock market participation and foreign stock ownership.

(12)

Fig. 1. Theoretical framework

3. Data and methodology

3.1. Data collection

The data for this research is acquired from the Dutch Central Bank (DNB) House-hold Survey - DHS. It is a panel survey supported by CentER, which is part of Tilburg University, the Netherlands. Started in 1993 on average about 2000 households fill out the online surveys which make it a valuable source for academic research. Each year new households are introduced in the panel in order to keep the panel repre-sentative for the Dutch population. The reason why DHS is used for this research is that this panel contains a large amount of information about the sources of income, wealth, the demographic composition of the household, health and psychological con-cepts of the Dutch households.6

Considering the empirical part of this thesis, the required variables can be found into the following sections of data in the DHS panel: General information on the household, Household and Work, Health and Income, Assets and Liabilities,

Eco-6_{The questions in DHS are split into six categories including: General information on the}

(13)

nomic and Psychological Concepts and Aggregate data on Wealth. Furthermore, this research uses data from the questions which were added by Guiso, Sapienza and Zingales (2008) in the 2003 wave for measuring risk and ambiguity aversion.7 Waves from 2004 to 2016 are used, resulting into a large sample of 2563 households and a total of 11,869 observations for the whole period, after exclusion of those individuals who did not answer the question to choose between risky and ambiguity lottery.

Choosing this time period is motivated by the paper of Guiso, Sapienza and Zingales (2008) who introduced their question measuring ambiguity aversion in the 2003 wave. In this case, it is possible to explore how financial advice impact ambiguity prefer-ences and investment decisions of Dutch citizens after 2003. The usage of a large period of time, however, assumes that ambiguity preferences do not experience a dramatic change over time. The assumption is supported by the findings of Roberts and DelVecchio (2000) who made a research about whether people experience time-invariant preferences and attitudes over their life. They show evidence that people change more often their preferences when they are young and those preferences be-come more stable in college years. After the age of 30, people have minimum change in their preferences and after 50, they do not change them at all. In the sample which is used, only about 3 percent of the population is below 30 years old which assumes that most people in the sample do not experience a large difference in their preferences.8 _{This thesis will provide results with both including and excluding those}

3 percent of the population which are under the age of 30. First of all, the outcomes of the regressions including all observations will be shown because it can provide us with more accurate results and then the results excluding the individuals who are under 30 years old will be represented in Appendix A. Those people have minimum change in their ambiguity preferences and that is why we are interested to see the re-sults including only them. Furthermore, this study will use a large panel data which offers a possibility for larger diversity of methods which can provide more accurate

7_{I would like to thank CentER for the professional attitude and for providing me with the}

questions from Guiso, Sapienza and Zingales (2008), which are essential part of this study.

(14)

results.

3.2. Summary statistics

This section provides the summary statistics of the variables included in this study. Table 1 represents the summary statistics of the dependent variables and the control variables in the model, while Table 2 shows the same information for one of the key independent variables in this study – ambiguity aversion.

Table 1: Summary statistics of the dependent variables and the control variables.

This table represents the summary statistics for the dependent and control variables in this study. The last column shows the observations for each variable available in the data.

Mean Std.Dev. Minimum Maximum N

Dependent variables:

Stock Ownership* 0.117 0.321 0 1 11,869

Foreign Stock Ownership* 0.189 0.392 0 1 467

Control variables: Age 55.616 14.245 19 93 9,719 Male* 0.562 0.496 0 1 10,801 Married* 0.627 0.484 0 1 8,717 Number of children 0.685 1.069 0 6 10,801 University education 0.105 0.307 0 1 10,789 Employed* 0.266 0.442 0 1 5,401 Wealth 54,399.12 145,180.4 -97,560.84 4,665,001 10,974 Income 32,623.92 28,016 0 1,165,420 6,052 Risk Aversion 0.030 1.006 -3.07109 1.241861 8,390 Financial literacy 2.074 0.705 1 4 8,390 Prof. advice* 0.246 0.431 0 1 8,914

Note: *Variables are measured in percentages.

(15)

observations for this variable is not in line with the literature. The small amount of observations for foreign stock ownership might also have an impact on the results later in the empirical part of this thesis because if there were more observations, the results would be more accurate.

Table 2: Ambiguity aversion in the Netherlands.

The table below illustrates the ambiguity attitudes in the Netherlands, using data from DNB Household Survey (DHS). Section 1 provides information about the answers of the participants from the survey question, where participants were asked whether they prefer the risky lottery or ambiguous lottery or they are indifferent between those two. Section 2 shows summary statistics of the answers of the respondents about how much they value the risky lottery relative to the ambiguous lottery. Ambiguity is measured by taking the difference between what price the respondent assigned to the risky lottery and what price she assigned to the ambiguity lottery. Extensive explanation about this method can be found where the variable measurement is discussed.

Section 1 Percentage Ambiguity-averse 0.343 Ambiguity-neutral 0.609 Ambiguity-seeking 0.047 Section 2

Mean Std.Dev. Minimum Maximum N

Ambiguity 42.318 42.318 -2000 9990 11877

(16)

lottery while the maximum value is 9990 which is associated with the strongest preference for the risky lottery. Clearly, most of the individuals priced the risky lottery much more, indicating that they prefer it more than the alternative ambiguous lottery.

3.3. Measuring the dependent variables:

Although the two dependent variables look quite similar, they have large differ-ence in economic interpretations with regards to the field of behavior finance. Stock ownership is associated with the relation between ambiguity aversion and stock mar-ket participation, while foreign stock ownership is related with the so-called home country bias. The questions from DNB Household Survey and their answers which are used for measuring the stock market participation and foreign stock ownership are included in Appendix B1. The associated answer from those questions is binary (yes or no), therefore two dummy variables are constructed - one for each dependent variable. The value of the dependent variables is equal to 1 if the respondent owns shares and foreign shares, respectively and 0, otherwise.

3.4. Measuring ambiguity aversion:

(17)

which it can be estimated whether a person is ambiguity-averse or not, depending on his answer of the question which is also stated in Trautmann, Vieider and Wakker (2011). If a participant in the survey is willing to give less money for an ambiguous lottery than for a risky one, it means that he is ambiguity-averse, disregarding the initial wealth of the participant and his utility, because both lotteries give the same amount of money in case the participant wins. In this thesis ambiguity aversion is measured with a dummy variable, which is equal to 1 if the respondent chose the risky lottery instead of the ambiguous lottery (the respondent was willing to pay more for the risky lottery relative to the ambiguous lottery) and 0 if the respondent choose the ambiguous lottery or she is indifferent.

3.5. Measuring financial advice:

People, in general, do not only rely only on professional financial advice when making financial decisions. DHS includes a question which investigates the sources of advice of Dutch people. The exact question and a table with the proportion of different sources of financial advice can be found in Table 9, Appendix B3. Using this question, there in an opportunity for exploring which is they key source of financial advice for Dutch people. According to Table 9, 24.59 of the people use professional financial advice as their main source of advice in the context of investment decisions. This study uses a dummy variable which equals to 1 if the respondent answered that she uses professional financial advisor and 0, otherwise.

3.6. Measuring control variables:

(18)

(19)

3.7. Methodology

This part of the thesis aims to discuss the empirical methods that will be used in this study. Referring to Figure 1, the focus of this thesis is to estimate the indirect effect of professional financial advice on the relation between ambiguity aversion with stock market participation and foreign stock ownership. Therefore, two equations will be formulated for each dependent variable. In the first one, the variable in the left-hand side of the equation is the dependent variable, while on the right-hand side is the ambiguity aversion measure, financial advice and the interaction term. The second one adds the control variables, which were explained in details previously. To interpret the findings in a more concise way, equations (1) and (2) will test the suggested hypothesis that financial advice mitigates the effect of ambiguity aversion on stock ownership, while (3) and (4) will test whether financial advice mitigates the effect of ambiguity aversion on foreign stock ownership. The equations are rep-resented as follows:

Equation (1): The impact of financial advice on the relation between ambiguity aver-sion and stock market participation:

STOCKi= α + β1AM BIGU IT Y + β2ADV ICE + β3AM BIGU IT Y ∗ ADV ICE + i (1)

Where STOCK is a dummy variable equal to 1 if the respondent participates in the stock market and 0, otherwise, AMBIGUITY is the measure of ambiguity aversion, ADVICE is equal to 1 if the respondent use professional financial advice and 0, otherwise and AMBI-GUITY*ADVICE is the interaction term between ambiguity aversion and financial advice.

Equation (2): The impact of financial advice on the relation between ambiguity aversion and stock market participation including control variables:

STOCKi= α + β1AM BIGU IT Y + β2ADV ICE + β3AM BIGU IT Y ∗ ADV ICE + δCON T ROL + i (2)

(20)

Equation (3): The impact of financial advice on the relation between ambiguity aversion and foreign stock ownership:

FSTOCKi= α + β1AM BIGU IT Y + β2ADV ICE + β3AM BIGU IT Y ∗ ADV ICE + i (3)

Where FSTOCK is a dummy variable equal to 1 if the respondent owns foreign stocks and 0, otherwise. The rest of the variables are the same as equation (1) and (2).

Equation (4): The impact of financial advice on the relation between ambiguity aversion and foreign stock ownership including the control variables:

STOCKi= α + β1AM BIGU IT Y + β2ADV ICE + β3AM BIGU IT Y ∗ ADV ICE + δCON T ROL + i (4)

The above equations are tested with OLS fixed effects regression. In this case vari-ables which do not vary over time (gender, education, trust and the ambiguity aver-sion measure) will drop out from the equation when estimating the regresaver-sion results.9

Also, when the dependent variable is binary (dummy variable) in a linear model such as the OLS, there is heteroskedastisity by construction because the error term is con-stant. Therefore, robust standard errors are used instead of the regular ones.

While this thesis uses OLS fixed effects for its empirical part, previous studies on this topic like the one from Dimmock et al. (2016) uses a probit model. The results from the probit regression can be found in the Appendix E. Although by using a probit regression variables which do not change over time will not drop out, there are two main reasons why the empirical results of this thesis are based on OLS fixed effects and not on a probit model. Firstly, the panel data used in this study is large, including many individuals over a large period of time, while the previous literature relies mostly on cross-sectional data. The second important reason is that fixed ef-fects model can solve potential problems with omitted variables bias.

9_{OLS fixed effects relies on mathematical approach called within transformation. The within}

(21)

Clearly, using OLS fixed effects might seem appropriate for this study, however, without testing whether it is indeed relevant, the results cannot be trusted. First of all, it is important to verify whether OLS fixed effects is necessary at all. This can be done by checking whether the probability value associated to the F-test in the regression output is less than 5 percent (0.05 <p). This F-test estimates whether all the coefficients in the model are statistically different than zero. In all regression outputs of this study the probability value is less than 5 percent which indicates that OLS fixed effects model is relevant for this study. OLS fixed effects results cannot be biased because of omitted characteristics which do not vary over time (culture, gender). In this case the pure effect of the time-varying variables on the dependent variable can be calculated using this model. The interaction term also changes over time and using fixed effects can lead to more precise results. However, fixed effects also have one drawback – variables that do not vary over time drop out from the model.

On the other hand, in order to keep those variables in the model and see their impact on stock market participation and foreign stock ownership, it is possible to choose OLS random effects because it allows the estimation of the impact of both variables which change and do not change over time. However, in order to use OLS random effects we need to assume that the error term is uncorrelated with the explanatory variables. This is the main disadvantage of the random effects model because in-dividual characteristics, which are not observed, are rarely uncorrelated with the independent variables.10 This is the reason why it is important to test whether OLS fixed effects or random effects should be used in this thesis. Hausman test

10_{The assumption for OLS random effects that the error term is uncorrelated with the}

explana-tory variables is represented by the following equation: γit = β0+ β1∗ xi1+ ... + βk∗ xitk+ αi+ µit, (5)

where Cov(xitj, αi) = 0, t = 1, 2...T ; j = 1, 2..., k

(22)

provides information about which estimator is more appropriate. By rejecting the null hypothesis that the error term is uncorrelated with the independent variables at 5 percent level (0.05 <p), we conclude that fixed effect is more appropriate than random effects model.

After following the F-test in the OLS fixed effects regression and the Hausman test it is clear that fixed effects is a relevant model for this thesis. We lose the ability to see the impact of ambiguity aversion on stock market participation because it is assumed to be constant over time. The same holds for gender, education and trust. However, the focus of this study is to show whether financial advice has a positive effect on the relation between ambiguity aversion, stock market participation and foreign stock ownership. It is not our main interest to show the relation itself. The reason behind this is that the impact of financial advice on the both the relation between ambiguity aversion with stock ownership and foreign stock ownership is not studied previously in the literature, while those relations are extensively investigated in past academic work. In the literature review of this thesis it is clearly shown that according to previous studies, there is a negative relation between ambiguity aversion with stock market participation and foreign stock ownership. Thus, by using OLS fixed effects, we do not lose important information which is not studied yet.

4. Results

(23)

(24)

Table 4 represents the Pearson‘s correlation matrix which aims to show the rela-tionship between the independent variables. The Pearson correlation coefficients is presented with the associated significance level. According to the table, there are no variable which are extremely high correlated with each other. The only variables with large correlation coefficients between each other are retired and age (69 percent), children and age (50 percent) and having a job and age (41 percent). Intuitively, the interaction term has also a large correlation coefficient of 42 percent with ambiguity aversion and 53 percent with financial advice, however, those values are still in a reasonable range. Meanwhile, in the correlation matrix where the dependent vari-able is foreign stock ownership we can observe that the same correlation coefficients which we discussed are larger. The correlation between retired individuals and age is 74 percent, while for job and age it is 63 percent. The interaction term and fi-nancial advice are also highly correlated (65 percent) but not extremely high. The correlation matrix is included just before the regression results because it provides a clear idea how the independent variables are related to each other. Based on the correlation matrix, there is no need to drop a variable because of an extremely high correlation with another one.

4.1. The impact of financial advice on ambiguity aversion, stock

mar-ket participation and foreign stock ownership.

(25)

Table 5: The impact of financial advice on the relation between ambiguity aver-sion,stock market participation and foreign stock ownership.

This table provide the results of an OLS fixed effects regression. Column (1) illustrates the results when the dependent variable is stock ownership, while the independent variables are financial advice and the interaction term between financial advice and ambiguity aversion. Column (2) show the results of the regression when all variables are included. The same applies for Column (3) and Column (4) but the only difference is that the dependent variable is foreign stock ownership. Variables are named, accordingly: “Advice” is indicator whether the respondent use professional financial advisor‘s services. “Ambiguity*Advice” is the interaction term between those two variables. “Knowledge” is the self-assessed financial knowledge of the participant. “Risk” is the measure of risk aversion of the respondent. Robust standard errors are clustered by households and reported in parentheses. All non-binary variables are standardized.

Variables (1) (2) (3) (4) Advice 0.006 -0.012 -0.032 -0.007 (0.01) (0.02) (0.04) (0.05) Ambiguity*Advice 0.007* 0.022** 0.032* 0.067** (0.01) (0.01) (0.02) (0.03) Age -0.096** -0.782* (0.05) (0.5) Married -0.002 0.026 (0.01) (0.11) Children -0.006 -0.247** (0.01) (0.11) Employed 0.047** -0.103* (0.02) (0.06) Self Employed 0.055 0.153 (0.04) (0.15) Retired -0.022 0.097 (0.04) (0.11) Knowledge 0.006 0.047 (0.02) (0.07) Risk -0.001 0.002 (0.01) (0.02) Wealth 0.013 0.022 (0.01) (0.08) Income -0.003 -0.048 (0.01) (0.05) R-squared 0.0011 0.0314 0.0051 0.1409 N 11057 1445 446 126

(26)

Table 5 represents the outcome of an OLS fixed effects regression, where the vari-ables of interest are only those which vary over time because the time-invariant ones drop out from the model. The reported results are straightforward and support the proposed hypothesis that professional financial advice has a significant positive impact on the relation between ambiguity aversion with stock market participation and foreign stock ownership. In column (2) and (4), where the control variables are included in the model, the significance of the interaction term is at 5 percent level (0.05 <p). It shows how much impact has financial advice on the effect of ambiguity aversion on stock market participation. For example, in column (1) the coefficient of the interaction term is 0.007 which implies that one standard deviation increase in ambiguity aversion is associated with 0.7 percentage point increase in stock market participation for those who are using financial advice, assuming that the relation between ambiguity aversion and stock market participation is negative. This finding suggests that financial advice can reduce ambiguity and de-bias individuals when they consider investment in the stock market.

The impact of financial advice in column (1) is statistically significant but not eco-nomically significant. However, in the rest of the columns, the coefficients of the interaction term are statistically and economically significant. In column (2), the significance of the moderator is at 5 percent level and it shows that one standard deviation increase in ambiguity aversion leads to an increase of 2.2 percentage point in owning stocks for those who use professional financial advice. This result has large economic meaning since it represents about 19 percent relative to a basis of 11.7 per-cent participation of Dutch individuals in the stock market represented in Table 1. The within R-squared is very low (3.1 percent) indicating that the regression model does not fit the data very well, despite the fact that most of the control variables in previous studies are included.11 _{In terms of owning foreign stocks, the economic}

meaning of the results is even stronger. The output of the regression represented in column (3) shows that one standard deviation increase in ambiguity aversion leads

11_{The within R-squared is used because the OLS fixed effects regression is calculated using within}

(27)

to 3.2 percentage point increase in owning foreign shares for those who consult with financial adviser, if we assume that the relation between ambiguity aversion and owning foreign stocks is negative. In column (4) the coefficient indicates that the increase of foreign stock ownership for those who use financial advice will be 6.7 percentage point which is about 35 percent increase relative to a basis of 18.9 per-cent represented in Table 1. This means that individuals who ask for advice from professional adviser will hold more diversified portfolios. The within R-squared in this case is much larger (14.1 percent) compared to column (2) but still it is not large enough to say that the regression model well fits the data.

According to the output in column (2), there are control variables which have sig-nificant impact on stock market participation. Age is sigsig-nificant at (0.05 <p) and it has the expected sign of the coefficient. According to the life-cycle theory of invest-ing, as age increases, individuals are supposed to change their allocation from risky to riskless assets, because they no longer can rely on labor income.12 Employment has positive influence on stock market participation as theory predicts and it is also significant at 5 percent level. The rest of the control variables are insignificant and not all of them are in line with the findings in previous studies. For example, income, self-employed and individuals who are married have the opposite sign of the coeffi-cients from what the literature predicts. In column (4), age and employment status are significant at 10 percent level, however, employment has a negative sign of the coefficient, which is in contrast with what previous studies suggest. Rodrigo Garc´ıa (2013) found that employment is associated with more stock market participation because those who are working have higher probability to participate indirectly in the stock market. The reason behind this is due to the fact that a lot of the worker‘s pension plans are allocated to their company‘s stock or in mutual funds. This means

12_{In empirical studies which explore the relation between age and stock market participation and}

(28)

that employment is positively related with stock market participation but the same is not necessary true for holding foreign stocks. In column (4) number of children is the only control variable significant at 5 percent level, which is in line with the findings of Dimmock et al. (2015) about the negative relation between the size of the household and stock market participation.

4.2. Robustness check

The results from the previous regressions in Table 5 suggest that financial advice mitigates the effect of ambiguity aversion on stock market participation and foreign stock ownership. This part of the study aims to test the robustness of the results and investigate whether professional financial advice has more impact on people who are more ambiguity-averse. For this purpose, we turn back to the questions measuring ambiguity aversion. Our ambiguity aversion measure is equal to 1 if the respondent put a higher price for risky lottery relative to ambiguous lottery and 0, otherwise. However, individuals put different prices for those lotteries. It is logical to ask to what extend financial advice impacts those individuals who put the highest price for risky lottery compared to ambiguous lottery because those people are the most ambiguity-averse. Following the methodology for measuring ambiguity aversion in this study, after we take the difference between the price of risky and ambiguous lottery, it can be observed in Table 2 that the price for risky lottery ranges from -2000 (the least ambiguity-averse) to 9990 (the most ambiguity-averse). In order to test the robustness of the results, a ranking of the answers of those questions is created, where 10 percent of individuals who put the highest price for risky lottery compared to ambiguous lottery are considered as the most ambiguity-averse.13 For this test, the ambiguity aversion measure now has a value of 1 if the respondent belongs to the top 10 percent ambiguity-averse people and 0, otherwise. Thus, it is possible to see whether financial advice has larger impact on more ambiguity-averse individuals. Results from OLS fixed effects regression are shown in Table 6.

(29)

Table 6: The impact of financial advice on the subsample of top 10 percent ambiguity-averse people in The Netherlands.

This table provide the results of an OLS fixed effects regression. Column (1) illustrates the results when the dependent variable is stock ownership, the independent variables are financial advice and the interaction term between financial advice and ambiguity aversion. Column (2) show the results of the regression when all variables are included. The same applies for Column (3) and Column (4) but the only difference is that the dependent vari-able is foreign stock ownership. Varivari-ables are named, accordingly: “Advice” is indicator whether the respondent uses professional financial advisor‘s services. “Ambiguity*Advice” is the interaction term between those two variables. “Knowledge” is the self-assessed finan-cial knowledge of the participant. “Risk” is the measure of risk aversion of the respondent. Robust standard errors are clustered by households and reported in parentheses. All non-binary variables are standardized.

Variables (1) (2) (3) (4) Advice 0.006 -0.013 -0.032 -0.006 (0.01) (0.02) (0.04) (0.05) Ambiguity*Advice 0.007 0.038** 0.053* 0.078** (0.01) (0.02) (0.03) (0.03) Age -0.041 -0.796 (0.05) (0.56) Married 0.002 0.053 (0.02) (0.13) Children -0.013 -0.21** (0.01) (0.10) Employed 0.072*** -0.129 (0.03) (0.06) Self Employed 0.021 0.026 (0.04) (0.12) Retired -0.019 0.081 (0.04) (0.13) Knowledge -0.005 -0.015 (0.02) (0.05) Risk -0.005 -0.004 (0.01) (0.03) Wealth 0.02*** 0.033 (0.01) (0.08) Income -0.004 -0.049 (0.01) (0.06) R-squared 0.0005 0.0419 0.0089 0.1427 N 8419 1076 359 114

(30)

Table 6 reports the results from the OLS fixed regression. The sample size is reduced because from 34 percent of the respondents who are ambiguity-averse this robustness test considers only the top 10 percent, however, we have sufficient observations in order to make this robustness check.14 The outcome of the regressions are compared with the outcome from Table 5. While in Table 5 the interaction term is significant in all 4 columns, in Table 6 it is significant in all columns except the first one. On the other hand, the coefficients corresponding to the interaction term are also larger in all columns except the first one, where they are equal. In Column (2), where the coefficient is statistically significant at 5 percent level, the increase in the value of the coefficient is almost doubled – from 0.022 to 0.038. In Column (4) the coefficient is again significant at 5 percent and it is also larger compared to the same one in Table 5 – from 0.067 to 0.078 percentage points. It implies that one standard deviation increase in ambiguity aversion is associated with 7,8 percentage point increase in foreign stock ownership for those who consult with financial adviser. This is a very large increase because it represents 41 percent of those 18,9 percent individuals who own foreign shares.

Assuming that the relation between ambiguity aversion, stock market participation and foreign stock ownership is negative, as past literature predicts, the results from Table 6 clearly show that financial advice mitigates ambiguity and can potentially increase stock market participation.The estimates are both statistically and econom-ically significant. They show that for those 10 percent who are the most ambiguity-averse, financial advice does really matter a lot. Those individuals theoretically are expected to participate less in the capital market than others and financial advis-ers can mitigate ambiguity the most for them. In Column (2) employment remains significant, while wealth now has a positive and significant coefficient, as theory predicts. The results of the second robustness test are shown in Table 7.

14_{The results remain robust if we investigate different subsample of ambiguity-averse people.}

(31)

Table 7: The impact of financial advice on the relation between ambiguity aversion and stock market participation and foreign stock ownership excluding the outliers.

This table provides the results of an OLS fixed effects regression. Column (1) illustrates the results when the dependent variable is stock ownership, while the independent vari-ables are financial advice and the interaction term between financial advice and ambiguity aversion. Column (2) show the results of the regression when all variables are included. The same applies for Column (3) and Column (4) but the only difference is that the de-pendent variable is foreign stock ownership. Variables are named, accordingly: “Advice” is indicator whether the respondent uses professional financial advisor‘s services. “Ambi-guity*Advice” is the interaction term between those two variables. “Knowledge” is the self-assessed financial knowledge of the participant. “Risk” is the measure of risk aversion of the respondent. Robust standard errors are clustered by households and reported in parentheses. All non-binary variables are standardized.

Variables (1) (2) (3) (4) Advice 0.006 -0.013 -0.032 -0.002 (0.01) (0.02) (0.04) (0.05) Ambiguity*Advice 0.007 0.024** 0.032* 0.066** (0.01) (0.02) (0.02) (0.03) Age -0.097** -0.866 (0.05) (0.54) Married -0.002 0.03 (0.02) (0.11) Children -0.009 -0.238* (0.01) (0.12) Employed 0.047** -0.074 (0.02) (0.07) Self Employed 0.094 0.172 (0.06) (0.14) Retired -0.028 0.111 (0.04) (0.11) Knowledge 0.006 0.042 (0.01) (0.06) Risk -0.001 0.007 (0.01) (0.03) Wealth 0.042 0.003 (0.04) (0.07) Income -0.004 -0.022 (0.01) (0.04) R-squared 0.0002 0.0448 0.0041 0.1347 N 11057 1445 446 126

(32)

In regression analysis, extreme outliers can affect the results. In order to control for extreme outliers in the results from Table 5, we will follow the approach of Lee et al. (2015) and winsorise the sample with 1 percent from each tail. In the sample of this study, extreme outliers can be found in almost all variables. This can substantially influence the regression results. An example for such extreme cases can be found in the ambiguity aversion measurement where there are extreme observations which are associated to both ambiguity-averse and ambiguity-seeking individuals. Wealth is also an example because the wealthiest person owns 4,665,001 assets (measured in euros) while the poorest individual is in a situation of indebtedness because she owns -97,560.84. Similar extreme values can be found in the risk aversion and etc. There are doubts that those responses might be biased or not representative for the Dutch population and that is why a regression excluding them is conducted. Table 7 illustrates the results from an OLS fixed effects regression when outliers are excluded from the sample. Similar to the results in the previous robustness test, the interac-tion term is significant in all columns except the first one. When control variables are included, the significance is at 5 percent level. The output of Table 7 imply that even when the outliears are excluded from the sample, the results from Table 5 remain robust which highlights their importance. There is not much difference between the coefficients in Table 5 and Table 7. According to the output of Column (2) in Table 7, the coefficient associated to the interaction term is positive and significant at 5 percent level. The value of the coefficient is 0.024 which means that one standard deviation increase in ambiguity aversion will lead to 2.4 percentage point increase in stock market participation for individuals who use professional financial advice compared to 2.2 percent in Table 5. In column (2) age and employment are with the same significance compared to Table 5, while in column (4) the only variable which remains signficicant is the number of children.

(33)

sample. Moreover, the robust test which was estimated on the subsample of most ambiguity-averse individuals shows that professional financial advice has the largest impact exactly on them. This is an important implication for policy makers because professional financial advice can potentially lead to more stock market participation for this group of people, resulting in more gains for individuals and the society.

5. Discussion and Limitations

5.1. Results:

The outcomes of the empirical part of this research supports the proposed hy-pothesis and also supports the findings of Von Gaudecker (2013) that financial advice can be used as a tool for mitigating behavioral biases of individuals, which can lead to more rational investment decisions. On the other hand, by using OLS fixed ef-fects we lose the ability to see the impact of the time-invariant variables like gender, education and ambiguity aversion on stock market participation and foreign stock ownership for our sample, which is a potential limitation of this research. In partic-ular, this study is mostly interested in ambiguity aversion, however, previous studies clearly show that ambiguity aversion is negatively related with investments in the capital market and in foreign shares. Since this is already explored in the literature, the primary focus on this thesis is to provide an accurate results for the impact of financial advice on the relation between ambiguity aversion with owning stocks and foreign stocks.

(34)

bias. Therefore, we can assume that the time-varying variables are more robust with OLS fixed-effects. On the other hand, we found out that the interaction term is not significant in any of the columns in the Table 11 in Appendix E. Other variables like age are also not significant in the probit regression and are significant in the OLS fixed effect. This suggests that we cannot trust the results from the probit regression where the relation between ambiguity aversion with stock market participation and ambiguity aversion with foreign stock ownership is positive and insignificant, which is clearly in contrast with the theory.

The outcomes of the OLS fixed effects regression in Table 5 differ also when they are compared to the results including separately wealth and income in the regression.15

This rise an important point of discussion with regards to the regression outcomes due to doubts for multicollinearity issue in the model, since including those control variables leads to a change in the sign of the coefficients of some of the other control variables. Moreover, the interaction term is no longer significant when only wealth is included in both the regression with stock ownership and foreign stock owner-ship. Although, no high correlation is observed between the regressors, there is still possibility of collinearity between them. In particular, wealth and income seems to be very similar in economic sense which might be problematic.16 _{In an attempt to}

prevent this problem, all non-binary variables in the regressions are standardized. Dimmock et al. (2016) also uses standardized measures of the coefficients in their paper. The reason behind standardizing the variables in this thesis is mainly because including an interaction term in the model also introduces multicollinearity and by standardizing the interaction term as well as the non-binary variables the problem is mitigated.

Foreign Stock Ownership:

One possible limitation of this study is the small amount of observations for one

15_{These results can be found in Table 13 in Appendix G.}

16_{VIF test is used for determining multicollinearity in the model. Variance inflation factor (VIF)}

(35)

of the dependent variables. In contrast to the question for stock ownership, which have enough responses, the question whether individuals hold foreign stocks was answered by only 467 individuals. Furthermore, this is not in line with the findings of Bekaert and Wang (2009), who argue that The Netherlands is the least affected from home bias. Considering the large amount of control variables in the model which substantially reduce the sample size, the regression outputs are based on 126 observations which is still sufficient but no always enough for expecting accurate results.

5.2. Endogeneity and reverse causality:

In this part of the study, the problem with endogeneity and reverse causality will be addressed. In general, endogeneity is often a problem in empirical studies which is difficult to be solved. First of all, Dimmock et. al (2016) claim that the problem with reverse causality can be avoided by measuring ambiguity aversion with the Ellsberg experiment. According to the same author, it is unlikely that there is a reverse causality problem between the dependent variables and ambiguity aversion because ambiguity aversion is measured by respondent choice of risky and ambiguous urns which does not involve anything financial (stocks or bonds, for example). Secondly, in order this thesis to overcome the problem with omitted variable bias, a sufficient amount of control variables are used, which appear in most of the previous studies. Furthermore, OLS fixed effects model is used which avoids potential problem with omitted variable bias.

6. Conclusion

(36)

over the period of time between 2004 and 2016, I measure ambiguity aversion with a similar to Ellsberg experiment questions. I test whether professional financial advice has a positive impact on the relation between ambiguity aversion with stock market participation and foreign stock ownership. I found that the professional financial advice can mitigate the negative effect of ambiguity aversion when individuals are considering investments in the capital market and in foreign companies, which is in line with the proposed hypothesis. The results are statistically and economically sig-nificant. They are also in line with previous studies which argue that financial advice can be used as a de-biasing mechanism in the context of investment decisions. The first robustness test of this thesis provides a very strong finding that financial advice has more impact on the subsample of individuals who are more ambiguity-averse than for those who are less ambiguity-averse. This robustness test clearly show the practical implications of this thesis that policy makers should focus on increasing of the availability of financial advisers to households and individuals in order for them to start investing in the capital market and to diversify their portfolios by investing in foreign stocks. This could potentially have positive gains both for individuals and the society. The second robustness test shows that the results are not driven by outliers. In addition to the important implications for policy makers, this study is a good starting point for further research on this topic. A possible extension of the paper would be the usage of different data, especially because the data from DNB Household Survey do not have many observations for foreign stock ownership which is one of the dependent variables in this study. In addition, it would be also inter-esting to see whether professional financial advice can mitigate ambiguity, and thus increase the stock market participation and owning foreign shares in other countries.

7. References

Antoniou, C., Harris, R. D., Zhang, R. (2015). Ambiguity aversion and stock market participation: An empirical analysis. Journal of Banking Finance, 58, 57-70.

(37)

Economics of Finance, 1, 1053-1128.

Bekaert, G., Wang, X. S. (2009). Home bias revisited.

Bergstresser, D., Chalmers, J. M., Tufano, P. (2009). Assessing the costs and bene-fits of brokers in the mutual fund industry. The Review of Financial Studies, 22(10), 4129-4156.

Brooks, C. (2014). Introductory econometrics for finance. Cambridge university press.

Cao, H. H., Wang, T., Zhang, H. H. (2005). Model uncertainty, limited market participation, and asset prices. The Review of Financial Studies, 18(4), 1219-1251.

Christiansen, C., Rangvid, J., Joensen, J. S. (2010). The Effects of Marriage and Divorce on Financial Investments: Learning to Love or Hate Risk?. CREATES Re-search Papers, 57.

Collins, J. M. (2010). A review of financial advice models and the take-up of financial advice. Center for Financial Security WP, 10(5).

Collins, J. M., O’ROURKE, C. M. (2010). Financial education and counseling—Still holding promise. Journal of Consumer Affairs, 44(3), 483-498.

Coval, J. D., Moskowitz, T. J. (1999). Home bias at home: Local equity preference in domestic portfolios. The Journal of Finance, 54(6), 2045-2073.

Dimmock, S. G., Kouwenberg, R., Wakker, P. P. (2015). Ambiguity attitudes in a large representative sample. Management Science, 62(5), 1363-1380.

(38)

Ambi-guity aversion and household portfolio choice puzzles: Empirical evidence. Journal of Financial Economics, 119(3), 559-577.

DNB Household Survey, Available at www.centerdata.nl

Ellsberg, D. (1961). Risk, ambiguity, and the Savage axioms. The quarterly journal of economics, 643-669.

Engle-Warnick, J., Pulido, D., de Montaignac, M. (2016). Trust, ambiguity, and financial decision-making (No. 2016s-44). CIRANO.

Fagereng, A., Gottlieb, C., Guiso, L. (2017). Asset Market Participation and Port-folio Choice over the Life-Cycle. The Journal of Finance, 72(2), 705-750.

Fox, C. R., Tversky, A. (1995). Ambiguity aversion and comparative ignorance. The quarterly journal of economics, 110(3), 585-603.

Fox, C. R., Weber, M. (2002). Ambiguity aversion, comparative ignorance, and decision context. Organizational behavior and human decision processes, 88(1), 476-498.

Garc´ıa, R., Tessada, J. (2013). The effect of education on financial market partici-pation: evidence from Chile. Santiago, Chile: Working Paper.

GAUDECKER, H., VON, M. (2015). How does household portfolio diversification vary with financial literacy and financial advice?. The Journal of Finance, 70(2), 489-507.

(39)

Guidolin, M., Rinaldi, F. (2013). Ambiguity in asset pricing and portfolio choice: A review of the literature. Theory and Decision, 74(2), 183-217.

Guiso, L., Sapienza, P., Zingales, L. (2008). Trusting the stock market. the Journal of Finance, 63(6), 2557-2600.

Heath, C., Tversky, A. (1991). Preference and belief: Ambiguity and competence in choice under uncertainty. Journal of risk and uncertainty, 4(1), 5-28.

Huberman, G. (2001). Familiarity breeds investment. The Review of Financial Stud-ies, 14(3), 659-680.

Kapteyn, A., Teppa, F. (2011). Subjective measures of risk aversion, fixed costs, and portfolio choice. Journal of Economic Psychology, 32(4), 564-580.

Kramer, M. M. (2012). Financial advice and individual investor portfolio perfor-mance. Financial Management, 41(2), 395-428.

K¨uhberger, A., Perner, J. (2003). The role of competition and knowledge in the Ellsberg task. Journal of Behavioral Decision Making, 16(3), 181-191.

Malkiel, B. G., Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work. The journal of Finance, 25(2), 383-417.

Mason, C. H., Perreault Jr, W. D. (1991). Collinearity, power, and interpretation of multiple regression analysis. Journal of marketing research, 268-280.

(40)

Lee, B., Rosenthal, L., Veld, C., Veld-Merkoulova, Y. (2015). Stock market expec-tations and risk aversion of individual investors. International Review of Financial Analysis, 40, 122-131.

Lewis, K. K. (1999). Trying to explain home bias in equities and consumption. Jour-nal of economic literature, 37(2), 571-608.

Lusardi, A., Mitchell, O. S. (2011). Financial literacy around the world: an overview. Journal of pension economics finance, 10(4), 497-508.

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

Odean, T. (1998). Are investors reluctant to realize their losses?. The Journal of finance, 53(5), 1775-1798.

Posner, R. A. (1998). Rational choice, behavioral economics, and the law. Stanford Law Review, 1551-1575.

Roberts, B. W., DelVecchio, W. F. (2000). The rank-order consistency of person-ality traits from childhood to old age: a quantitative review of longitudinal studies. Psychological bulletin, 126(1), 3.

Shapira, Z., Venezia, I. (2001). Patterns of behavior of professionally managed and independent investors. Journal of Banking Finance, 25(8), 1573-1587.

Trautmann, S. T., Vieider, F. M., Wakker, P. P. (2011). Preference reversals for ambiguity aversion. Management Science, 57(7), 1320-1333.

(41)

8. Appendix

Appendix A. Results when only individuals above 30 years old are in-cluded.

Table 8: The impact of financial advice on the relation between ambiguity aversion and stock market participation and foreign stock ownership (subsample of individuals who are above 30 years old.)

This table provide the results of an OLS fixed effects regression. Column (1) illustrates the results when the dependent variable is stock ownership, while the independent variables are financial advice and the interaction term between financial advice and ambiguity aversion. Column (2) test the same relation but this time all variables are included. The same applies for Column (3) and Column (4) but the only difference is that the dependent variable is foreign stock ownership. Control variables are named, accordingly: “Advice” is indicator whether the respondent use professional financial advisor‘s services. “Ambiguity*Advice” is the interaction term between those two variables. “Knowledge” is the self-assessed financial knowledge of the participant. “Risk” is the measure of risk aversion of the respondent. Robust standard errors are clustered by households and reported in parentheses. All non-binary variables are standardized.

Variables (1) (2) (3) (4) Advice 0.007 -0.012 -0.032 -0.007 (0.01) (0.02) (0.04) (0.05) Ambiguity*Advice 0.007 0.022** 0.032* 0.067** (0.01) (0.01) (0.02) (0.03) Age -0.096** -0.782 (0.05) (0.5) Married -0.002 0.026 (0.01) (0.11) Children -0.006 -0.247** (0.01) (0.10) Employed 0.047** -0.103 (0.02) (0.06) Self Employed 0.055 0.172 (0.04) (0.15) Retired -0.022 0.097 (0.01) (0.11) Knowledge 0.006 0.047 (0.01) (0.07) Risk -0.001 0.002 (0.01) (0.02) Wealth 0.013 0.022 (0.01) (0.08) Income -0.003 -0.048 (0.01) (0.04) R-squared 0.0012 0.0314 0.0051 0.1409 N 10832 1445 445 126

(42)

Appendix B. DHS questions used in this study.

B1. Questions which are used for measuring the two dependent variables.

“Did you, on 31 December 2015, own any SHARES? Do not include shares of your own private limited company here, nor bonds through MUTUAL FUNDS?”

“Did this include shares of foreign companies?”

B2. Questions which are used for measuring ambiguity aversion.

Risky lottery:

“Consider the following hypothetical lottery. Imagine a large urn containing 100 balls. In this urn, there are exactly 50 red balls and the remaining 50 balls are black. One ball is randomly drawn from the urn.If the ball is red, you win 5,000 euros; otherwise, you win nothing. What is the maximum price you are willing to pay for a ticket that allows you to participate in this lottery?”

Ambiguous lottery:

“Consider now a case where there are two urns, A and B. As before, each one has 100 balls, but urn A contains 20 red balls and 80 blacks, while urn B contains 80 reds and 20 blacks. One ball is drawn either from urn A or from urn B (the two events are equally likely). As before, if the ball is red you win 5,000 euros; otherwise, you win nothing. What is the maximum price you are willing to pay for a ticket that allows you to participate in this lottery?”

(43)

“What is your most important source of financial advice?”

Table 9: Sources of financial advice for Dutch peoples.

This table represents the proportion of people using different sources of financial advice, including professional financial advice. This statistic is based on survey question from DHS including 8,914 observations over the period of time between 2004 and 2016.

Frequency Percentage parents, friends or acquaintances 1,925 21.6 information from the newspapers 916 10.28 financial magazines, guides, books 916 8.64 brochures from my bank or mortgage advisor 675 7.57 advertisements on TV or in the papers 268 3.01 professional financial advisers 2,192 24.59 financial computer programs 85 0.95 financial information on the Internet 1,420 15.93

other 663 7.44

Total 8,914 100

B4. Questions which are used for measuring risk attitudes.

The following statements concern saving and taking risks. Please indicate for each statement to what extent you agree or disagree.

To what extent do you agree with the following statements?

Please indicate on a scale from 1 to 7 to what extent you agree with the statement. 1 means ‘totally disagree’

1 means ‘totally disagree’

Totally disagree Totally agree

1 2 3 4 5 6 7

(44)

“I do not invest in shares, because I find this too risky.”

“I want to be certain that my investments are safe.”

B5. Question measuring Trust.

“Generally speaking, would you say that most people can be trusted or that you have to be very careful in dealing with people?”

Possible answers:

1. One has to be very careful with other people. 2. Most people can be trusted.

3. I don’t know.

B6. Question measuring financial knowledge.

“How knowledgeable do you consider yourself with respect to financial matters?”

Possible answers:

1. not knowledgeable

2. more or less knowledgeable 3. knowledgeable

4. very knowledgeable

Appendix C. Principal component analysis

(45)

that is shared with the others, while factor loadings measure the correlation coefficients between each risk preference question and the specific factor.

Table 10: Principal component analysis

This table represents the outcome of the principal component analysis performed on 3 survey questions, measuring the risky attitudes of Dutch citizens. Communalities and Factor loadings are reported below.

Factor loadings Communalities ”I think it is more important to have safe investments and

guaranteed returns, than to take a risk to

have a chance to get the highest possible returns.” 0.6478 0.4197 ”I do not invest in shares, because I find this too risky.” 0.4197 0.1654 “I want to be certain that my investments are safe.” 0.677 0.4583

Appendix D. Histogram of the distribution of age of Dutch people

Fig. 2. Age

Appendix E. Results of the regression with OLS panel probit model.

(46)

Table 11: Ambiguity aversion, stock market participation and foreign stock owner-ship (probit regression)

This table provide the results of a panel probit regression. Column (1) illustrates the results when the dependent variable is stock ownership, while the independent variables are the ambiguity aversion measure, financial advice and the interaction term between financial advice and ambiguity aversion. Column (2) test the same relation but this time all variables are included. The same applies for Column (3) and Column (4) but the only difference is that the dependent variable is foreign stock ownership. Variables are named, accordingly: “Ambiguity” is dummy variable equals to 1 if the respondent is ambiguity averse and 0 otherwise. “Advice” is indicator whether the respondent use professional financial advisor‘s services. “Ambiguity*Advice” is the interaction term between those two variables. “Knowledge” is the self-assessed financial knowledge of the participant. “Risk” is the measure of risk aversion of the respondent. Robust standard errors are clustered by households and reported in parentheses. All non-binary variables are standardized.

Variables (1) (2) (3) (4) Ambiguity 0.131 -0.002 0.012 0.009 (0.19) (0.01) (0.03) (0.10) Advice -0.111 -0.002 0.003 0.294** (0.11) (0.01) (0.02) (0.16) Ambiguity*Advice 0.081 0.004 0.003 -0.030 (0.05) (0.00) (0.01) (0.06) Age -0.004 -0.086 (0.01) (0.10) Married 0.002 0.091 (0.01) (0.10) University 0.024 0.153 (0.02) (0.12) Male 0.016 -0.010 (0.01) (0.14) Children -0.003 0.018 (0.00) (0.07) Employed 0.008 -0.216** (0.01) (0.12) Self Employed -0.019 -1.096 (0.02) (0.36) Retired 0.000 0.151 (0.01) (0.19) Knowledge 0.009 -0.065 (0.01) (0.06) Risk -0.005 -0.020 (0.00) (0.04) Wealth 0.004 -0.029 (0.00) (0.04) Income 0.000 -0.038 (0.00) (0.03) Trust 0.014 -0.119 (0.01) (0.12) N 11057 1445 446 126

(47)

Ambiguity aversion and household‘s investment decisions: The role of ﬁnancial advice