1
Faculty of Economics and Business Thesis MSc Finance
The impact of financial advice on the financial asset allocation of individuals: evidence from the Netherlands
Author: Bahram Gerami (S2597128) Supervisor: Dr. V. Angelini
Date: 12-01-2017
Key words: Financial asset allocation, financial advice, Dutch individuals JEL classifications: D14, G11
A BSTRACT
This paper investigates and compares the impact of different sources of financial advice on the decisions of Dutch individuals to invest in risky financial assets and their financial asset allocation between safe and risky financial assets by analysing the data from the Dutch Household Survey for the period 2015. Logistic analyses show that those who receive informal financial advice are less likely to hold risky financial assets. Moreover, both OLS and tobit models provide sufficient evidence that those with informal financial advice tend to allocate less of their financial assets towards risky financial assets. Additionally, there is no significant difference between the impacts of those professionally advised and self-deciders on the decision to hold risky financial assets or on their allocation of financial assets towards risky assets. However, among risky investors, the only significant difference is between those with professional advice and self-deciders, with the former showing a higher propensity towards risky financial assets. Overall, it can be concluded that the source of financial advice affects individuals’ decisions to invest in risky financial assets as well as their financial asset allocation.
Acknowledgements: I would like to express my sincere gratitude to my supervisor Dr. V. Angelini for her support, patience, and extraordinary knowledge and my family, and friends for their unconditional love and support throughout writing this thesis and my life in general.
2
C ONTENTS
Abstract... 1
1. Introduction ... 3
2. Literature review ... 5
2.1 Financial advice and asset allocation ... 5
2.2 Socio-demographic and socio-economic determinants of asset allocation ... 9
3. Research framework and methodology ... 13
3.1 Research purpose and key questions ... 13
3.2 Hypotheses and methodology ... 14
3.2.1 Hypothesis one ... 14
3.2.2 Hypothesis two ... 17
4. Data and descriptive statistics ... 19
4.1 Data ... 19
4.2 Descriptive statistics ... 21
5. Empirical results... 26
6. Conclusion and discussion ... 33
6.1 conclusion ... 33
6.2 limitations and future research ... 34
References ... 35
Appendix ... 40
Appendix A: Financial assets ... 40
Appendix B: Variance inflation factor ... 41
Appendix C: Correlation matrix ... 42
3
1. I NTRODUCTION
Financial asset allocation places important investment decisions in the hands of individuals.
Simultaneously, these decisions often involve complex situations. Financial knowledge and sufficient information are necessary requirements to make optimal decisions. One important approach to improving the quality of individual investment decisions is to rely on financial advisors. Previous studies investigated the main reasons why individuals seek financial advice and there is ample evidence about the shortcomings of individuals in order to manage their own finances. According to Bluethgen et al. (2008) and Shum & Faigh (2006) individuals make sub-optimal choices in relative complex situations due to a lack of sufficient information. As individuals make their decisions based on the available information, more information may lead to a better outcome. However, acquiring more information also increases the costs of information acquisition. Therefore, seeking financial advice can be most valuable to those who may suffer from cognitive errors and face high information acquisition costs (Mccall, 1970).
Jansen et al. (2008) questioned the customers of a German bank about the main reasons why they used financial advice. According to the answers of the respondents, the two main reasons to use financial advice were time constraints and in order to avoid investment mistakes. Other reasons were inconvenience and safety. Other studies indicate lack of financial knowledge (Lusardi & Mitchell, 2007; Karabulut, 2013), lack of information (Guiso &
Jappelli, 2006) and behavioral biases (Huberman et al., 2007; Kahneman & Tversky, 1979) as the main reasons why individuals seek financial advice. However, overconfidence among individuals reduces investor’s propensity to seek financial advice (Guiso & Japelli, 2006). The results of Kramer (2014) confirm that a higher degree of overconfidence relates to lower demand for financial advice. Bhattacharya et al. (2012) state that those individuals who are in the most need of financial advice are less likely to seek out advising, due to a lack of financial sophistication, lack of trust and lack of familiarity.
Financial decisions of individuals are often based on advice from non-professionals,
such as relatives, friends, financial magazines, etc. According to the results of Hemantkumar
et al. (2015) the most preferred sources of financial information are newspapers and
magazines. However, in many countries it is common for individual investors to seek
professional financial advice when making their investment decisions; for example, more than
80% of German retail investors consult a financial advisor before making a financial
4
investment decision (Bluethgen et al., 2008). Moreover, 75% of U.S investors consult a financial adviser before making a stock market or mutual fund transaction (Hung et al, 2008).
In the Netherlands, 51% of households with an investment portfolio rely on financial advice (Brown, 2010).
Many studies have addressed the effect of professional financial advice on the performances of the portfolios of households and individuals (see e.g., Guiso et al., 2009;
Hoechle et al., 2013; Kramer, 2012; Kramer & Lensink, 2012; Montmarquette & Viennot-Briot, 2015; Von Gaudecker, 2015; Zhang, 2014; Zur & Veneiza, 2001). Numerous studies have explored the relationship between professional financial advice and asset allocation of individuals (see e.g., Bluethgen et al., 2008; Zhang, 2014; Foerster et al., 2014; Chalmers &
Reuter, 2012; Shum & Faigh, 2006; Georgarakos & Inderst, 2011; Kramer, 2012; Jansen et al., 2008). However, relatively little is known regarding the impact of different sources of advice on the financial asset allocation of Dutch individuals. In particular, empirical research with regard to the effect of informal financial advice on financial asset allocation is very scarce.
Therefore, this paper contributes to the existing literature in two ways: First by making a distinction and an impact comparison between different sources of financial advice:
professional, informal and self-deciders. Informal advice is defined here as advice received from friends, family and/or acquaintances. Self-deciders are those who make their financial decisions on the basis of other sources of advice, such as newspapers, books, etc. The distinction between these sources of financial advice will be further outlined in section 4.1.
Secondly, this paper contributes to the existing literature by investigating the effects of these sources of financial advice both on decisions of individuals to invest in risky financial assets and financial asset allocation between safe and risky financial assets. To the best of the authors knowledge, this is the first paper that empirically investigates and evaluates this research subject in such a framework. Therefore, this study provides new insights on the impact of different sources of financial advice on the decisions of individuals to invest in risky financial assets as well as financial asset allocation between safe and risky financial assets.
The remainder of this paper is structured as follows. In section 2, the relevant theoretical literature is presented. In section 3 and 4, the research framework and methodology, and data and descriptive statistics of this paper are addressed, respectively.
Section 5 provides the empirical results, and the conclusion and discussion are reported in
section 6.
5
2. L ITERATURE REVIEW
2.1 F INANCIAL ADVICE AND ASSET ALLOCATION
A number of studies have been done concerning the impact of financial advice on the decisions of individuals to invest in risky assets and on their financial asset allocation.
Bluethgen et al. (2008) analyzed the impact of professional financial advice on the composition of individual investors’ portfolios using a dataset provided by a large German retail bank from January 2003 to October 2005. The data obtained from the bank was aggregated by asset categories. The asset allocation of 4,363 professionally advised investors was compared to non-advised investors. The results indicated that those who received advice tend to be older, wealthier, more risk averse and more likely to be female. It was also found that clients who received advice held, on average, approximately 3% less equity (including stock equity investment funds). Moreover, it is reported that clients who received advice had significantly larger allocation to mutual funds in comparison to non-advised investors. Clients who received advice on average held 66% of their equity in mutual funds, whereas clients without advice on average held 38% of their equity in mutual funds. The general conclusion was that professional financial advice has a significant impact on household investment behavior.
Similar to Bluethgen et al. (2008), Zhang (2014) also found that professional financial advice affects the investment behavior of households in terms of portfolio composition. Zhang (2014) investigated the differences between investors who received professional financial advice and those who did not, in order to explore the differences in portfolio composition between these investors. A proprietary dataset was used from the period of 2007-2012, containing information of a large sample of 405,107 individuals, who represent approximately 10% of the total New Zealand population. This study addressed professional financial advisers who provided face-to-face advice to the investors rather than using brokerage firm data.
According to the results, professional financial advice did have an impact on the asset
allocation of investors. It was also found that there is a positive relationship between
professional financial advice and property and equity asset holdings and a negative
relationship between professional financial advice and cash and bond asset holdings. Those
6
who received professional financial advice held 5% less cash and 4% less bonds, while holding 2% more property and 6% more equity assets.
Similar research on the impact of financial advice on the investor’s portfolios was performed by Foerster et al. (2014). The impact of professional financial advice on the clients’
investment portfolios were investigated using the Canadian Financial Monitor (CFM), which is a household survey covering both advised and unadvised households, from the period 2001- 2010. The results indicated that advised households shift their portfolio allocation away from safe assets to riskier assets, such as equity and mutual funds. Moreover, it was found that advisors induce their clients to take more risk. However, they argue that it is unclear whether these differences are due to the preferences of the clients or due to the advice. The results of Chalmers & Reuter (2012) are in line with the findings of Foerster et al. (2014). They explored the differences between investors who chose to invest through brokers and self-directed investors in order to outline the effect of professional financial advisors on portfolio choices.
This was investigated by using data from the Oregon University System sent to Optional Retirement Plan (ORP) in 2012. They compared the responses of 791 participants consisting of 297 advised-investors and 494 self-directed investors. Significant differences between investors who chose to invest through brokers and self-directed investors were documented.
The results showed that broker clients’ portfolios were significantly riskier than portfolios of self-directed investors.
Shum & Faigh (2006) completed an empirical study of the determinants of stock
holdings. Information on financial characteristics of U.S households was obtained by analyzing
U.S Surveys of Consumer Finances (SCF) from 1992 to 2001. Households who did not have
sufficient funds (less than $1,000 financial net worth and/or zero income) to form a reasonable
portfolio and outliers in several independent variables were excluded. It was found that
households who sought professional financial advice had a higher stock market participation
rate and those who participated also had larger investment amounts. However, these overall
effects were relatively limited, implying that the distribution of households by their proportion
of equity in the financial portfolio was affected solely to a small extent by whether they sought
professional financial advice or not. Therefore, it was concluded that professional financial
advice did not easily mitigate the barriers to participation in the stock market. The results also
suggest that the decision to hold stocks is positively correlated with wealth, age and risk
attitude. Another article which studied the effect of financial advice on the decision of
7
investors to hold risky assets is the paper of Georgarakos & Inderst (2011). The empirical investigation was based on data from a 2003 Eurobarometer survey managed by the European Commission. The Eurobarometer survey interviews a representative sample of European households across fifteen EU countries and asks questions concerning attitudes of households to different products and services. The data also provides information on ownership of risky financial assets, such as stocks held through mutual funds and retirement accounts, as well as an array of socio-economic characteristics. The article provided evidence that trust in professional financial advice had significant effect on stock market participation. This is in accordance with results of Shum & Faigh (2006). However, Georgarakos & Inderst (2011) reported that professional financial advice does not have a significant effect on the stock market participation for households with high financial capability.
Kramer (2012) investigated the influence of financial advisers on the Dutch individual investor’s portfolio by comparing the portfolios of advised and self-directed investors. A data set from a medium-sized, full service retail and business bank which provides different financial products and services throughout the Netherlands through a network of bank branches was used. All clients of the bank were eligible for advice during the sample period of the study. The portfolios of 6,100 Dutch investors for a 52-month period from 2003 to 2007 were analyzed and compared to security investments using data from the Dutch Central bank (DNB, 2006) in order to determine the representativeness of the sample. According to average portfolio size and composition of the investors, the sample reasonably represented the average investor in the Netherlands. The findings showed significant differences in characteristics and portfolio composition between advised and self-directed investors. Both for advised and self-directed investors, equity and bonds were the main assets, whilst advised investors held less risky portfolios. The portfolio value of advised investors consisted of less than 50% equity, whereas self-directed investors allocated approximately 70% to equity.
However, for portfolios exceeding €100,000, there was less difference between advised and self-directed investors. Moreover, advised investors allocated 66% of their portfolio to mutual funds compared with 48% for self-directed investors. Furthermore, advised investors held more structured products (such as options and derivatives) in terms of number of portfolios than self-directed investors, especially for larger portfolios.
Jansen et al. (2008) studied the influence of professional financial advice on the
suitability of asset allocation based on the risk tolerances of individual investors. The analysis
8
was based on a combination of three data sets: a customer questionnaire, a corresponding advisor questionnaire and customer account records provided by a German retail bank (DAB bank, München). On the basis of the Ordinary Least Squares (OLS) regression analysis, it was concluded that there is strong evidence that financial advisors have an incentive to promote equity-concentrated asset allocations.
To summarize, there is sufficient theoretical support that professional financial advice affects the decisions of individuals to hold risky assets and to allocate their assets more towards risky assets. However, the articles discussed above make a distinction between professional advice and non-professional advice when comparing its impact on the decision to invest in risky assets and on asset allocation and generally do not make a further distinction among non-professionally advised individuals. As a result of this, comparatively little is known regarding self-deciders and even less is known about those who receive informal financial advice. In this paper, sources of financial advice are distinguished into three categories:
professional, informal and self-deciders. Therefore, this paper differs from the existing literature by comparing and outlining the impact of different sources of financial advice on the decisions of individuals to invest in risky financial assets as well as their financial asset allocation between safe and risky financial assets.
Other relevant articles include Von Gaudecker (2015), Kramer (2016), and Guiso et al.
(1996). Von Gaudecker (2015) and Kramer (2016) both used the same data as this paper (DHS) and both distinguished between different sources of financial advice and financial assets.
Moreover, both used broadly the same independent and control variables. Von Gaudecker
(2015) used data of DHS from the period 2005-2006. After excluding households with less than
1,000 euros in financial assets and excluding risky financial asset holders with incomplete
reports, 381 observations were left in the final sample. In this study, financial assets have been
categorized into risky and safe financial assets. Mutual and growth funds, bonds, options and
shares were defined as risky financial assets and checking and savings accounts and cash value
of insurances were defined as safe financial assets. As the data set not only provided
information about the number of stocks and mutual funds individuals held, but also provided
each item’s name and the quantity held, a connection between each item to its time series of
returns obtained from Datastream or Morningstar was made. The analysis showed that nearly
all households that rely on professional or private contacts for advice achieved reasonable
investment outcomes compared to those who trusted their own decision-making capabilities.
9
Kramer (2016) used data from DHS and a large Dutch retail bank from the year of 2005 containing more than 2,000 households. The final sample consisted of 1,276 households of which 354 were investors. Both financial literacy and professional advice-seeking behavior of these households were studied in order to explore the demand for and impact of professional financial advice. The descriptive statistics were categorized into three different samples: all households, investors only, and a retail bank sample. It was found that those with higher confidence in their own financial literacy were less likely to seek financial advice. This effect was greater among wealthy households. However, Kramer (2016) stated that there was no relationship between measured financial literacy and financial advice-seeking.
Guiso et al. (1996) extended previous studies of portfolio choice by outlining the effect of income risk and borrowing constraints in the financial asset allocation of households. An empirical analysis based on the data of a Bank of Italy Survey of Households Income and Wealth (SHIW) for the year 1989 was performed. SHIW is a random sample of the Italian resident population containing detailed information on financial and demographic characteristics of 8,274 households. In SHIW the financial assets of households is categorized into 13 categories. The article defined risky financial assets as the sum of long-run government bonds, corporate bonds, investment fund units and equities. Furthermore, they discussed that the demand for risky assets is a two-stage decision process: before deciding how to allocate total financial assets between safe and risky securities, households first choose whether or not to hold risky assets. It was reported, however, that not all households owned risky assets and that this selection bias thus leads to inconsistent estimates of a simple OLS regression of the proportion of risky assets. This problem was mitigated by using a tobit model. It was found that investors who are confronted with uninsurable income risk tend to allocate their assets towards relatively safer assets.
2.2 S OCIO - DEMOGRAPHIC AND SOCIO - ECONOMIC DETERMINANTS OF ASSET ALLOCATION
Empirical literature shows that there are also other factors which have an impact on the
financial decision-making of individuals. Hemantkumar et al. (2015) did an exploratory study
regarding the factors which have an influence on financial investment decision-making of
investors. On the basis of the results it was concluded that socio-demographic and socio-
economic factors affect the financial investment decision-making of investors. Therefore,
10
these factors should be included in the regression analyses as control variables (see section 3) and will be argued separately in this subsection.
One of these factors is gender. Empirical literature provides ample evidence with regard to gender differences and shows that gender has an influence on the financial decision- making and risk behavior of individuals. For example, Jacobsen et al. (2014) investigated the gender differences in optimism and asset allocation. It is found that males tend to hold more stocks than females due to risk aversion behavior and also because of differences in either optimism or in perceived risk of financial markets. The results showed that men tend to be significantly more optimistic than women and therefore hold more riskier assets. Sunden et al. (1998) indicated that gender also has a significant effect on how individuals choose to allocate their assets. The results of numerous studies (see, e.g., AL-Ajmi, 2008; Barber &
Odean, 2001; Deo & Sundar, 2015; Frino et al., 2015; Hlako et al., 2012; Niessen & Reunzi, 2007) are similar to the findings of Jacobsen et al. (2014) and Sunden et al. (1998).
According to the literature, age has an impact on risk behavior and asset allocation of individuals. Charles & Kasilingman (2015) focused on the influence of age on investment behavior. The findings suggest that investor’s age has a significant effect on the investment behavior. Many results of other articles (see, e.g., AL-Ajmi, 2008; Ameriks et al., 2001;
Campbell, 2006; Canner et al., 1997; Hemantkumar et al., 2015; Mehta & Sharma, 2015;
Sultana, 2010; Summers et al., 2016) are in line with the results of Charles & Kasilingman (2015). Additionally, Zhang (2014) found that age is positively associated with cash and bond asset allocation and negatively associated with property and equity asset allocation. This implies that as investors get older, they tend to hold relatively safer assets. Moreover, it was reported that this finding is in line with life-cycle theory, indicating that relatively younger investors are more flexible in their future savings rate and hence can afford to take more risks.
Marital status is another factor which has an influence on the asset allocation of individuals (see, e.g., Bernasek & Bajtelsmit, 2002; Dobbelsteen & Kooreman, 1997;
Jianakoplos et al., 2003; Woolley, 2003; Zagorsky, 2003). According to Riley & Chow (1992)
married people are more risk averse than those who are not married. Moreover, married
individuals generally do not make investment decisions on their own. Their investment
choices tend to be affected by their spouse. This may be due the fact that their spouse is the
household decision-maker in financial matters or because the couple makes the financial
decisions together. In addition, Love (2010) argues that marital-status transitions can have
11
significant effects on financial household decisions. These effects are even greater in the cases of widowhood and divorce. However, the results of Bertocchi et al. (2011) and Hemantkumar et al. (2015) are contradictory. Bertocchi et al. (2011) reported that married individuals tend to invest more in risky assets than single individuals. In contrast, according to the exploratory study of Hemantkumar et al. (2015) marital status does not have a significant impact on financial investment decision of investors.
Having children plays an important role in asset allocation of individuals. (Love, 2010) argues that households with children hold riskier assets compared to households without children. Bogan (2013) stated that having children affects the financial decisions of households. According to Bogan (2013) single male and female respondents with no children have the highest proportion of stockholding, mutual fund holding, fixed income security holding and US savings bond holding. Furthermore, Xia et al. (2014) indicated that there is a positive correlation between the number of children in each household and stock market participation.
Education level affects the financial behavior of individuals as well. Al-Ajmi (2008) researched the risk tolerance of individual investors in an emerging market of Bahrain and concluded that the investor’s education level has a significant influence on the risk tolerance of the investor. Hibbert et al. (2012) investigated the role of financial education in the management of retirement savings. The retirement savings allocations of finance were compared to English professors. The results indicated that finance professors allocate a larger share of their retirement savings to equities and they also managed their retirement portfolios more actively. Moreover, many other studies indicate that higher educated investors tend to hold more stocks (see, e.g., Campbell, 2006; Guiso et al., 2003; Haliassos & Bertaut, 1995;
Hong et al., 2004).
Both Al-Ajmi (2008) and Hemantkumar et al. (2015) suggest that the level of income affects the financial decision-making and behavior of the investors. These results are similar to those of Campbell (2006), Anbar & Eker (2010) and Jansen et al. (2008). According to Riley
& Chow (1992) there is a negative relationship between risk aversion and income. Very low
income families appear to be the most risk averse. The results of Cohn et al. (1975) confirm
the results of Riley & Chow (1992). Cohn et al. (1975) found that the proportion of risky assets
held increases as the income of the individuals increases.
12
Another factor which influences the financial decision-making of investors is wealth.
According to Shum & Faigh (2006) there is a positive correlation between the decision to hold stocks and wealth. Campbell (2006) found that besides income, wealth of investors also affects financial behavior of investors and indicated that wealthy investors tend to take more risk in their portfolios compared to less wealthy investors. This is due to greater stock market participation and higher portfolio shares conditional on participation. Anbar & Eker (2010) also found that total net assets of investors have significant influence on the financial risk behavior of investors. Additionally, Zhang (2014) stated that wealth of investors is related to their asset allocation and reported that as the balances of accounts increase, investors tend to hold more equity and property assets and less cash and fixed interest assets.
According to the empirical literature, home-ownership also affects asset allocation of individuals. Heaton & Lucas (2000) showed that home-owners hold fewer risky financial assets. The results of Yamishita (2003) are similar to those of Heaton & Lucas (2000). Cardak
& Wilkins (2009) conducted a study on the portfolio allocation decisions of Australian households, while focusing on household allocations to risky financial assets. The results were contrary to those obtained by Heaton & Lucas (2000) and Yamishita (2003). Cardak & Wilkins (2009) concluded that home-ownership is associated with greater risky asset holdings.
Financial knowledge, risk tolerance and occupation also have an influence on financial decision-making and behavior of investors. Asaad (2015) explored how financial literacy and financial knowledge affect financial decisions and found that both financial literacy and knowledge influence the financial decisions of investors. The results of Disney & Gathergood (2013) are in accordance with Asaad (2015). However, according to Von Gaudecker (2015) financial knowledge does not have an effect for those who seek professional financial advice.
Furthermore, Campbell (2006) found that there is a relationship between self-reported risk-
attitude of investors and their financial risk behavior. Nguyen et al. (2016) examined the
influence of financial risk tolerance on investment decision-making. Their results revealed that
more risk tolerant investors tend to invest more in risky assets. Das et al. (2008) indicated that
the occupation of investors has a great influence on their investment decisions. For example,
government employees prefer life insurance, while private sector employees prefer mutual
funds. Furthermore, Christelis et al. (2013) argue that work status, such as retired, working or
unemployed, could affect the income risk of households, which in turn can affect asset
allocation of households.
13
3. R ESEARCH FRAMEWORK AND METHODOLOGY 3.1 R ESEARCH PURPOSE AND KEY QUESTIONS
In general, empirical literature indicates that professional financial advice does affect the decisions of individuals to hold risky assets and to allocate their assets more towards risky assets. However, the effect of informal financial advice cannot be readily inferred from these studies since these studies solely distinguish between those receiving professional financial advice and those not receiving professional financial advice. This paper further divides those not receiving professional financial advice into two categories: those receiving informal financial advice and self-deciders.
According to Guiso et al. (1996) the demand for risky assets is a two-stage decision process: first individuals make the decision whether or not to hold risky assets and then they decide how to allocate their total assets between safe and risky financial assets. Therefore, the objective of this paper is to first outline the impact of different sources of financial advice on the decision of individuals to invest in risky financial assets and then to investigate its impact on the financial asset allocation between safe and risky financial assets. This leads to the following two key questions:
1. Does the source of financial advice affect the decision of Dutch individuals whether or not to invest in risky financial assets?
2. Does the source of financial advice affect the financial asset allocation of Dutch individuals between safe and risky financial assets?
The first key question concerns whether individual’s decision to invest in risky financial assets is affected by the primary source of their financial advice. In other words, do individuals with different primary source of financial advice have the same propensity to invest in risky financial assets? The second key question involves whether individual’s financial asset allocation between safe and risky financial assets is related to the individual’s primary source of financial advice.
There are different sources of financial advice. In order to categorize these sources the
same approach as Von Gaudecker (2015) will be applied, and hence the sources of financial
advice will be distinguished into three different categories, namely ‘professional advice’,
14
‘informal advice’ and ‘self-deciders’. Individuals who belong to the category of professional advice are those who have professional advice as their primary source of financial advice. The informal-advised individuals are those who rely on the financial advice of their friends, family and/or acquaintances when making important financial decisions. ‘Self-deciders’ refers to those who receive financial advice via other sources, such as newspapers, financial magazines, guides, books, brochures from banks, etc., and make their financial decisions based on their own interpretation.
3.2 H YPOTHESES AND METHODOLOGY
3.2.1 H YPOTHESIS ONE
Hypothesis one refers to the first key question of this paper. In section 2.1, several relevant studies have been discussed. The overall conclusion of these studies is that those who receive professional financial advice are more likely to invest in risky assets compared to those without professional financial advice. However, this paper attempts to outline the effects of other sources of advice as well. These other sources are subdivided into self-deciders and informal financial advice. Therefore, the following hypothesis will be tested:
H0: The decision of Dutch individuals whether or not to invest in risky financial assets is not significantly affected by the source of financial advice.
H1: The decision of Dutch individuals whether or not to invest in risky financial assets is significantly affected by the source of financial advice.
According to the empirical literature, it is expected that individuals who receive professional financial advice are more likely to invest in risky assets compared to self-deciders and those who receive informal financial advice.
The linear probability model (LPM) is a linear regression model which can be used to
test the first hypothesis since this model deals with a binary dependent variable; one holds
risky financial assets or does not (Brooks, 2014). LPM is the simplest way of dealing with binary
dependent variables. Therefore, three regressions will be conducted to test the first
hypothesis. Equation 1 below is one of the three equations which are used for the regression
analyses.
15
𝑃
𝑖= 𝑃(𝑦
𝑖= 1) = 𝛽
0+ 𝛽
1𝐷𝑃𝑟𝑜𝑓𝑎𝑑𝑣
𝑖+ 𝛽
2𝐷𝐼𝑛𝑓𝑜𝑟𝑎𝑑𝑣
𝑖+ 𝛽
3𝐷𝑀𝑎𝑙𝑒
𝑖+ 𝛽
4𝐷𝐴𝑔𝑒1
𝑖+ 𝛽
5𝐷𝐴𝑔𝑒2
𝑖+ 𝛽
6𝐷𝐴𝑔𝑒3
𝑖+ 𝛽
7𝐷𝐴𝑔𝑒4
𝑖+ 𝛽
8𝐷𝐸𝑑𝑢𝑐𝐻
𝑖+ 𝛽
9𝐷𝐹𝑖𝑛𝑘𝑛𝑜𝑤𝐻
𝑖+
𝛽
10𝐷𝑅𝑖𝑠𝑘𝑡𝑜𝑙𝑒𝐻
𝑖+ 𝛽
11𝐷𝑅𝑖𝑠𝑘𝑡𝑜𝑙𝑒𝐿
𝑖+ 𝛽
12𝐷𝐻𝑜𝑚𝑒
𝑖+ 𝛽
13𝐷𝐶ℎ𝑖𝑙𝑑𝑟𝑒𝑛
𝑖+ 𝛽
14𝐷𝑀𝑎𝑟𝑟𝑖𝑒𝑑
𝑖+ 𝛽
15𝐷𝐸𝑚𝑝𝑙𝑜𝑦
𝑖+ 𝛽
16𝑇𝑜𝑡𝑖𝑛𝑐𝑜𝑚𝑒
𝑖+ 𝛽
17𝑇𝑜𝑡𝑤𝑒𝑎𝑙𝑡ℎ
𝑖+ µ
𝑖(1)
Where the fitted values from this regression are the estimated probabilities for 𝑌
𝑖= 1 for each observation i. In equation 1 a dummy variable 𝑌
𝑖for whether an individual has invested in risky financial asset is defined as follows:
𝑦
𝑖= 0 𝑖𝑓 𝑡𝑜𝑡𝑎𝑙 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑟𝑖𝑠𝑘𝑦 𝑎𝑠𝑠𝑒𝑡𝑠 = 0 (2)
𝑦
𝑖= 1 𝑖𝑓 𝑡𝑜𝑡𝑎𝑙 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑟𝑖𝑠𝑘𝑦 𝑎𝑠𝑠𝑒𝑡𝑠 > 0 (3)
β
0refers to the constant. The independent variables are the dummy variables 𝐷𝑃𝑟𝑜𝑓𝑎𝑑𝑣
𝑖, 𝐷𝐼𝑛𝑓𝑜𝑟𝑎𝑑𝑣
𝑖, and 𝐷𝑆𝑒𝑙𝑓𝑑𝑒𝑐
𝑖, which refer to professional financial advice, informal financial advice and self-deciders, respectively. The latter one is the reference category, which is excluded in this regression.
Control variables are gender, age, education level, financial knowledge, risk tolerance, home-ownership, children, marital status, occupation, total income and total wealth. 𝐷𝑀𝑎𝑙𝑒
𝑖is dummy variable which has a value of 1 when the respondent is a male and a value of 0 when the respondent is a female. The ages of the respondents are categorized into the following dummies: 𝐷𝐴𝑔𝑒1
𝑖, 𝐷𝐴𝑔𝑒2
𝑖, 𝐷𝐴𝑔𝑒3
𝑖, 𝐷𝐴𝑔𝑒4
𝑖, 𝐷𝐴𝑔𝑒5
𝑖, which represent the following age intervals: ‘’younger than 35’’, ‘’35-44’’, ‘’45-54’’, ‘’55-64’’ and ‘’65 and older’’, respectively.
The latter age interval is the reference category, which is excluded. Dummy variable 𝐷𝐸𝑑𝑢𝑐𝐻
𝑖takes the value of 1 if the respondent has a high education level and 0 otherwise. The dummy variable 𝐷𝐹𝑖𝑛𝑘𝑛𝑜𝑤𝐻
𝑖takes the value of 1 if the respondent has high financial knowledge and 0 otherwise. The dummy variable 𝐷𝑅𝑖𝑠𝑘𝑡𝑜𝑙𝑒𝐻
𝑖stands for high risk tolerance of the respondent. It has a value of 1 when the respondent has a high risk tolerance and 0 otherwise.
The dummy variable 𝐷𝑅𝑖𝑠𝑘𝑡𝑜𝑙𝑒𝐿
𝑖refers to low risk tolerance and has a value of 1 if the
respondent has a low risk tolerance and 0 otherwise. 𝐷𝑅𝑖𝑠𝑘𝑡𝑜𝑙𝑒𝑁
𝑖is a dummy variable as
well, which refers to neutral risk tolerance. It takes the value of 1 if the respondent is
indifferent between risky and safe assets and 0 otherwise. 𝐷𝑅𝑖𝑠𝑘𝑡𝑜𝑙𝑒𝑁
𝑖is the reference
category, which is excluded. 𝐷𝐻𝑜𝑚𝑒
𝑖is a dummy variable, which refers to the home-
16
ownership of the respondent and takes the value of 1 if the respondent is the owner of house and 0 otherwise. Dummy 𝐷𝐶ℎ𝑖𝑙𝑑𝑟𝑒𝑛
𝑖has a value of 1 if the respondent has one or more children and a value of 0 otherwise. Dummy variable 𝐷𝑀𝑎𝑟𝑟𝑖𝑒𝑑
𝑖refers to the marital status.
It takes the value of 1 when the marital status of the respondent is married or registered partnership and 0 otherwise. Furthermore, the dummy variable 𝐷𝐸𝑚𝑝𝑙𝑜𝑦
𝑖refers to the occupation and takes the value of 1 if the respondent is employed or self-employed or works in own business and 0 otherwise. The last two control variables are 𝑇𝑜𝑡𝑖𝑛𝑐𝑜𝑚𝑒
𝑖and 𝑇𝑜𝑡𝑤𝑒𝑎𝑙𝑡ℎ
𝑖, which represent the total gross income and total wealth in 2015, respectively.
The wealth of individuals is calculated according to equation 4. Finally, µ
𝑖presents the error term.
𝑇𝑜𝑡𝑎𝑙 𝑤𝑒𝑎𝑙𝑡ℎ = 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠 − 𝑡𝑜𝑡𝑎𝑙 𝑑𝑒𝑏𝑡𝑠 (4)
The slope estimates for this linear probability model imply the change in the probability that the dependent variable will equal 1 for a one-unit change in a given independent variable, whilst holding the effect of all other independent variables fixed. The estimated probabilities should lie within the range (0,1). A drawback of this model, however, is that the estimated probabilities may be outside this range. In addition, the estimated probabilities of observations of this model may result in exactly zero or one. This is simply not realistic; one cannot be really certain that professional financial advice always results in investment in risky financial assets. Therefore, a logistic specification can be used in order to overcome the limitations of LPM. The logistic specification uses logistic function F, which is a function of any random variable z:
F(𝑧 𝑖 ) = 𝑒
𝑧𝑖1+𝑒
𝑧𝑖= 1+𝑒 1
−𝑧𝑖(5)
Where e is the exponential under the logistic approach. The function F is the cumulative logistic distribution. Therefore, the logistic model estimated is given in equation 6.
𝑃 𝑖 = 1
1+𝑒
−(𝛽0+𝛽1𝐷𝑃𝑟𝑜𝑓𝑎𝑑𝑣𝑖+𝛽2𝐷𝐼𝑛𝑓𝑜𝑟𝑎𝑑𝑣𝑖+𝛽3𝐷𝑀𝑎𝑙𝑒𝑖+𝛽4𝐷𝐴𝑔𝑒1𝑖+𝛽5𝐷𝐴𝑔𝑒2𝑖+⋯+µ𝑖)(6)
17
Where 𝑃
𝑖is the probability that 𝑦
𝑖= 1. By applying equations 5 and 6, 0 and 1 are asymptotes to the function and therefore the probabilities will never be exactly zero or one; they will always have a positive value smaller than one. Since this is a non-linear model, maximum likelihood should be used. The constant, independent and control variables of equation 6 remain the same as in equation 1.
3.2.2 H YPOTHESIS TWO
Hypothesis two refers to the second key question of this paper. After testing the first hypothesis, the logical next step is to check whether the source of financial advice affects the financial asset allocation of individuals between safe and risky financial assets. Empirical literature shows that individuals who receive professional financial advice tend to allocate more of their assets towards risky assets, compared to non-advised individuals. This study attempts to outline the effect of other sources of financial advice as well. Therefore, the following hypothesis is proposed:
Hypothesis 2:
H0: The source of financial advice does not significantly affect the financial asset allocation of Dutch individuals between safe and risky financial assets.
H1: The source of financial advice significantly affects the financial asset allocation of Dutch individuals between safe and risky financial assets
According to the empirical literature, it is expected that those individuals who receive professional financial advice tend to shift their assets away from safe assets towards riskier assets compared to those who do not receive professional financial advice. In order to test hypothesis two, 6 OLS regressions will be performed to the data. One of these regressions is shown in equation 7.
𝑌
𝑖= 𝛽
0+ 𝛽
1𝐷𝑃𝑟𝑜𝑓𝑎𝑑𝑣
𝑖+ 𝛽
2𝐷𝐼𝑛𝑓𝑜𝑟𝑎𝑑𝑣
𝑖+ 𝛽
3𝐷𝑀𝑎𝑙𝑒
𝑖+ 𝛽
4𝐷𝐴𝑔𝑒1
𝑖+ 𝛽
5𝐷𝐴𝑔𝑒2
𝑖+ 𝛽
6𝐷𝐴𝑔𝑒3
𝑖+ 𝛽
7𝐷𝐴𝑔𝑒4
𝑖+ 𝛽
8𝐷𝐸𝑑𝑢𝑐𝐻
𝑖+ 𝛽
9𝐷𝐹𝑖𝑛𝑘𝑛𝑜𝑤𝐻
𝑖+ 𝛽
10𝐷𝑅𝑖𝑠𝑘𝑡𝑜𝑙𝑒𝐻
𝑖+
𝛽
11𝐷𝑅𝑖𝑠𝑘𝑡𝑜𝑙𝑒𝐿
𝑖+ 𝛽
12𝐷𝐻𝑜𝑚𝑒
𝑖+ 𝛽
13𝐷𝐶ℎ𝑖𝑙𝑑𝑟𝑒𝑛
𝑖+ 𝛽
14𝐷𝑀𝑎𝑟𝑟𝑖𝑒𝑑
𝑖+ 𝛽
15𝐷𝐸𝑚𝑝𝑙𝑜𝑦
𝑖+
𝛽
16𝑇𝑜𝑡𝑖𝑛𝑐𝑜𝑚𝑒
𝑖+ 𝛽
17𝑇𝑜𝑡𝑤𝑒𝑎𝑙𝑡ℎ
𝑖+ µ
𝑖(7)
18
Where 𝑌
𝑖refers to the proportion of risky financial assets, which is calculated according to equation 8 (see also appendix A). The constant, independent and control variables remain the same as in equation 1.
𝑃𝑟𝑜𝑝𝑜𝑟𝑡𝑖𝑜𝑛 𝑜𝑓 𝑟𝑖𝑠𝑘𝑦 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠 =
total risky financial assetstotal financial assets
(8)
The dependent variable for testing hypothesis 2 is the proportion of risky financial assets (see equation 8). The descriptive statistics (table 1) show, however, that many respondents do not hold any risky financial asset and hence many observations take the value of 0 on the dependent variable. These dependent variables are referred to as censored dependent variables. According to Guiso et al. (1996) the parameter estimates obtained by a OLS regression might be biased and inconsistent as a result. They overcame this problem by using a tobit model. The tobit model can be used to estimate models with censored dependent variables (Brooks, 2014). In this paper, the tobit model estimates the relationship between variables when there is censoring from below at zero. Therefore, it is called a censored regression model as well. According to the tobit model, the dependent variable is defined according to equation 9.
𝑌
𝑖= { 𝑌
𝑖∗if 𝑌
𝑖∗> 0
0 if 𝑌
𝑖∗≤ 0 (9)
Where
𝑌
𝑖∗= 𝛽
0+ 𝛽
1𝐷𝑃𝑟𝑜𝑓𝑎𝑑𝑣
𝑖+ 𝛽
2𝐷𝐼𝑛𝑓𝑜𝑟𝑎𝑑𝑣
𝑖+ 𝛽
3𝐷𝑀𝑎𝑙𝑒
𝑖+ 𝛽
4𝐷𝐴𝑔𝑒1
𝑖+ 𝛽
5𝐷𝐴𝑔𝑒2
𝑖+ 𝛽
6𝐷𝐴𝑔𝑒3
𝑖+ 𝛽
7𝐷𝐴𝑔𝑒4
𝑖+ 𝛽
8𝐸𝑑𝑢𝑐𝐻
𝑖+ 𝛽
9𝐷𝐹𝑖𝑛𝑘𝑛𝑜𝑤𝐻
𝑖+ 𝛽
10𝐷𝑅𝑖𝑠𝑘𝑡𝑜𝑙𝑒𝐻
𝑖+
𝛽
11𝐷𝑅𝑖𝑠𝑘𝑡𝑜𝑙𝑒𝐿
𝑖+ 𝛽
12𝐷𝐻𝑜𝑚𝑒
𝑖+ 𝛽
13𝐷𝐶ℎ𝑖𝑙𝑑𝑟𝑒𝑛
𝑖+ 𝛽
14𝐷𝑀𝑎𝑟𝑟𝑖𝑒𝑑
𝑖+ 𝛽
15𝐷𝐸𝑚𝑝𝑙𝑜𝑦
𝑖+
𝛽
16𝑇𝑜𝑡𝑖𝑛𝑐𝑜𝑚𝑒
𝑖+ 𝛽
17𝑇𝑜𝑡𝑤𝑒𝑎𝑙𝑡ℎ
𝑖+ µ
𝑖(10)
Where 𝑌
𝑖∗presents the true latent variable value of the proportion of risky financial assets,
which is calculated according to equation 8, and 0 refers to the lower limit. The constant,
independent and control variables remain the same as in equation 1.
19
4. D ATA AND DESCRIPTIVE STATISTICS 4.1 D ATA
The Dutch Household Survey (DHS) of the year 2015 can provide the required data for this investigation. DHS is a representative sample of the Dutch population and is a panel survey that started in 1993. The data are collected each year with a panel of more than 2,000 households. The questionnaires are presented to each of the household members who are older than 16. DHS data are unique since they contain ample information about the Dutch households, such as employment, pensions, accommodations, mortgages, income, assets, debts, health, psychological and economic aspects of financial behavior, personal characteristics, and more. Their data are collected through CentERpanel.
According to Campbell (2006), there are five key criteria that an ideal data set investigating household finance behavior should have: It should cover a representative sample of the entire population, measure total wealth and have a breakdown of wealth categories, distinction between asset classes, high level of accuracy in reporting of data, and presence of panel data. DHS meets all five key criteria of Campbell (2006).
DHS provides all required information for the objective of this study. First of all, the primary source of financial advice is the key variable of interest of this study. In DHS, the respondents were asked the following question: ‘What is your most important source of advice when you have to make important financial decisions for the household?’ The respondents had the following nine options to choose:
1. Parents, friends or acquaintances;
2. Information from the newspapers;
3. Financial magazines, guides, books;
4. Brochures from my bank or mortgage adviser;
5. Advertisements on TV, in the papers, or in other media;
6. Professional financial advisers;
7. Financial computer programs;
8. Financial information on the Internet;
9. Other.
20
In this paper, these sources of financial advice are categorized into three groups, which is in accordance with the approach of Von Gaudecker (2015). The first group solely consists of those who chose for ‘professional financial advisers’ (option 6). The second group contains those who chose for ‘parents, friends or acquaintances’ (option 1), which is considered as informal advice in this paper. The remaining respondents are categorized into group 3, which are classified as self-deciders. Thus, self-deciders are those who receive financial advice via information from the newspapers, financial magazines, guides, books, brochures from their bank or mortgage adviser, advertisements on TV, or in the papers, or in other media, financial computer programs, financial information on the Internet, or other sources of financial advice (options 2, 3, 4, 5, 7, 8, and 9).
DHS provides detailed information regarding the assets of the respondents. DHS distinguishes the assets into different asset classes, and provides the (market) value of all these assets. However, since the sample size does not allow looking at the whole assets, the approach of Von Gaudecker (2015) is also applied in order to aggregate and classify the assets as well. Von Gaudecker (2015) categorizes all assets and debts of the respondents of DHS into 5 levels starting at level 0, which is the lowest level of aggregation. Since this paper is strictly concerned with financial assets, all assets will be aggregated and categorized into 4 levels. By using this approach eventually all financial assets will be aggregated and distinguished into safe or risky financial assets. The financial asset aggregation and categorization is denoted in appendix A. First, mutual funds and growth funds are aggregated into mutual and growth funds. Bonds and options are aggregated into bonds and options. Checking accounts, saving
& deposit accounts, bank certificates & deposits, and saving certificates are all aggregated into checking and savings accounts. Saving or endowment insurance policy, mortgage-related life insurance, life-cycle savings plan and single premium annuity insurance policy are aggregated into cash value of insurance. Secondly, these financial assets are categorized into risky and safe financial assets. Risky financial assets consist of mutual and growth funds, shares, and bonds and options, whilst safe financial assets contain checking and savings accounts, and cash value of insurances. Finally, risky and safe financial assets together comprise the total financial assets of an individual.
The financial knowledge of the respondents can be determined by using the following
question of DHS to its respondents: ’How knowledgeable do you consider yourself with
respect to financial matters?’. The possible answers are: not knowledgeable, more or less
21
knowledgeable, knowledgeable and very knowledgeable. In this paper, the first two answers will be considered as ‘low financial knowledge’ and the last two as ‘high financial knowledge’.
Furthermore, the risk tolerance of the respondents is determined by using the following statement: ‘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.’ The respondents are asked to indicate to what extent they agree with this statement on a scale from 1 to 7, where 1 refers to ‘totally disagree’ and 7 refers to ‘totally agree’. In this paper, respondents who chose for 1,2 or 3, are considered to have a high level of risk tolerance and respondents who chose for 5, 6 or 7 are considered to have a low level of risk tolerance. The respondents who chose for 4 can be considered as neutral regarding this matter. The level of education regards the highest level of education completed, in which ‘HBO’ or ‘WO’ refers to a high level of education, and the remaining education levels refer to a low level of education. In addition, all other required information, such as gender, age, marital status, home-ownership, number of children, occupation and gross income are provided by DHS as well.
4.2 D ESCRIPTIVE STATISTICS
DHS provides the data of 5,137 individuals in the year 2015. However, not all the respondents are adults. Therefore, after excluding respondents younger than 18 years old 4,075 observations are left. A part of these respondents did not fill out the question regarding their most important source of financial advice, in spite of questioning them. Since the most important source of financial advice is the key variable of interest, those respondents from whom information is missing were excluded, which led to 2,655 respondents. However, 536 of these 2,655 respondents had no financial assets and as a result were excluded as well.
Finally, after selecting those who answered all relevant questions concerning the objective of
this paper, the total sample consists of 1,651 observations. In this paper, the approach of
Kramer (2016) will be applied concerning the sample selection. To this end, one subsample is
composed out of the total sample. The subsample solely contains respondents who hold risky
financial assets (309 observations); thus, those who have a positive value of risky financial
assets (see equation 3). In this paper, this group of investors will be referred to as ‘risky
investors’. The descriptive statistics of total sample and subsample of risky investors are
22
Table 1: descriptive statistics of the total sample
This table presents the descriptive statistics of the total sample obtaining total observations, means, standard deviations, minima, and maxima. The descriptive statistics are organized in general information, financial information and characteristics.
The total sample consists of 1,651 observations.
Variables Observations Mean Std. Dev. Min. Max.
General information
Male 1,651 0.58 0.49 0.00 1.00
Female 1,651 0.42 0.49 0.00 1.00
Younger than 35 years 1,651 0.10 0.30 0.00 1.00
35-44 years 1,651 0.17 0.37 0.00 1.00
45-54 years 1,651 0.16 0.37 0.00 1.00
55-64 years 1,651 0.20 0.40 0.00 1.00
65 years and older 1,651 0.37 0.48 0.00 1.00
One child or more 1,651 0.34 0.47 0.00 1.00
No children 1,651 0.66 0.47 0.00 1.00
Married or registered partner 1,651 0.66 0.47 0.00 1.00
Not married or registered partner 1,651 0.34 0.47 0.00 1.00
Employed 1,651 0.50 0.50 0.00 1.00
Unemployed 1,651 0.50 0.50 0.00 1.00
Low education level 1,651 0.62 0.49 0.00 1.00
High education level 1,651 0.38 0.49 0.00 1.00
Home-owner 1,651 0.78 0.41 0.00 1.00
No home-owner 1,651 0.22 0.41 0.00 1.00
Financial information
Total gross income 1,651 32,533.06 23,097.56 0.00 304,550.00
Total assets 1,651 237,619.40 323,653.00 -97,560.84 4,387,611.00
Total non-financial assets 1,651 193,701.80 282,053.40 0.00 4,272,091.00
Total financial assets 1,651 43,917.59 111,270.60 -97,660.84 2,646,672.00
Total risky financial assets 1,651 8,849.76 45,306.81 0.00 834,079.40
Total safe financial assets 1,651 35,067.84 91,435.99 -97,660.84 2,622,183.00
Total debts 1,651 72,613.55 142,805.80 0.00 2,175,000.00
Total wealth 1,651 165,005.90 302,079.80 -1,824,708.00 4,387,611.00
Total risky financial assets>0 1,651 0.19 0.39 0.00 1.00
Total risky financial assets=0 1,651 0.81 0.39 0.00 1.00
Proportion of risky financial assets 1,651 0.07 0.20 -0.64 1.10
Proportion of safe financial assets 1,651 0.93 0.20 -0.10 1.64
Characteristics
Professional advice 1,651 0.23 0.42 0.00 1.00
Informal advice 1,651 0.24 0.43 0.00 1.00
Self-deciders 1,651 0.53 0.50 0.00 1.00
Low financial knowledge 1,651 0.70 0.46 0.00 1.00
High financial knowledge 1,651 0.30 0.46 0.00 1.00
Low risk tolerance 1,651 0.64 0.48 0.00 1.00
High risk tolerance 1,651 0.21 0.41 0.00 1.00
Neutral risk tolerance 1,651 0.15 0.36 0.00 1.00
23
denoted in table 1 and 2, respectively. More than half of the total sample is male (58%), is older than 55 years (57%), has no children (66%), is married or has a registered partner (66%), has low education level (62%) and owns a home (78%). 50% of the total sample is employed.
The average gross income of these 1,651 respondents is €32,533 and the average wealth is
€165,006. In the regressions both gross income and wealth are rescaled; gross income is divided by 100,000 and wealth is divided by 1,000,000. 23% of the total sample receives professional financial advice when making important financial decisions, whereas 24%
receives informal financial advice, and the remaining 53% are self-deciders. Furthermore, 19%
of the total sample holds risky financial assets, whilst 81% of the respondents hold no risky financial assets. The average proportion of safe financial assets of total financial assets is 93%, and the average proportion of risky financial assets is 7%. In addition, 30% of the respondents consider themselves highly knowledgeable with respect to financial matters. Furthermore, 64% of the total sample has a low risk tolerance, while 21% and 15% have high and neutral risk tolerance, respectively.
Descriptive statistics of the subsample of risky investors, however, show several different outcomes (table 2) in comparison with the descriptive statistics of the total sample (table 1). The subsample of risky investors consists for the most part of males (76%), approximately half of the subsample (47%) is 65 or older, 77% of the subsample has no children, more individuals have higher education level (52%), and are home-owners (88%).
Additionally, total gross income and wealth are higher, namely €42.616 and €310.790, respectively. The average proportion of risky financial assets of total financial assets is 38%.
The proportion of the subsample who receives professional advice remains the same (23%), whereas less people receive informal financial advice (12%), and the proportion of self- deciders is greater (65%). Furthermore, it is noteworthy that the proportion of individuals who have low risk and high risk tolerance is approximately the same as the total sample, namely 63% and 21%, respectively.
In order to test the multicollinearity of the data set, a Variance Inflation Factor (VIF)
can be applied. The VIF values of regressions for both total sample and subsample are shown
in appendix B. The VIF results indicate that multicollinearity does not seem to be a problem,
since all VIF values are lower than 4. Additionally, the correlation matrix between the variables
is presented in appendix C and it shows no multicollinearity either. The correlation matrix
shows several interesting results. Home-owners, males, highly educated individuals,
24
Table 2: descriptive statistics of the subsample risky investors
This table presents the descriptive statistics of the subsample of risky investors obtaining total observations, means, standard deviations, minima, and maxima. Risky investors refer to those who have a positive value of risky financial assets. The descriptive statistics are organized in general information, financial information and characteristics. The total subsample consists of 309 observations.
