MSc. Finance Thesis

(1)

University of Groningen. Faculty of Economics and Business.

MSc. Finance Thesis

Financial risk tolerance of individuals during and after the global financial crisis in the Netherlands.

Author: A.E. Zeilmaker - S2976137 A.E.Zeilmaker@student.rug.nl Supervisor: Drs. S. Pool

Abstract

In this thesis I test whether the financial risk tolerance of individuals in the Netherlands changed during the global financial crisis in comparison to the post financial crisis years. This thesis distinguishes from other studies because it is the first study of crisis’s consequences on risk taking on the general population in the Netherlands. I perform a regression on a sample of 693 individuals in the Netherlands between 2007 and 2012. The data used comes from the DNB Household survey. The main finding of this thesis is that the global financial crisis has positive significant impact on the financial risk tolerance level of individuals in the Netherlands. The results show a decreasing trend from 2007 to 2012. Although, the global financial crisis has positive impact, the economic impact is minor.

JEL classification: D14, D81, G02

Field keywords: financial risk tolerance, global financial crisis, the Netherlands, behavioural finance

(2)

2 1. Introduction

In a time span of one year, August 2007 - August 2008, the mortgage crisis in the United States has become a deep global financial crisis. On September 14, 2008 Lehman Brothers, then the fifth largest investment bank in the world, announced its bankruptcy. Consequently, countries all over the world encountered an economic downturn (Ershov, 2010). This economic downturn results in individuals who lose trust in the financial markets and the economy. Nowadays economists are extremely interested in the consequences of this global financial crisis on individuals. Several studies have conducted a research regarding the effects of a crisis on investors (Gerrans, Faff and Hartnet, 2015; Weber, Weber and Nosic, 2012; Necker and Ziegelmeyer, 2016). The aim of this study is to examine whether the financial risk tolerance of individuals in the Netherlands has changed during the financial crisis in comparison to the post crisis years.

The study of Gerrans, Faff and Hartnett (2015) investigated the individual financial risk tolerance associated with the global financial crisis. They concluded that financial risk tolerance is a stable trait in the short-term, but can be influenced and changed by events more gradually over time. In addition to their result, Weber, Weber and Nosic (2012) show similar findings. Besides, Necker and Ziegelmeyer (2016) investigated the effects of the financial crisis on the household willingness to take risks in Germany. Their study concluded that households’ wealth change during the global financial crisis has no effect on the financial risk tolerance. However, households which encounter losses to the global financial crisis decreased their financial risk tolerance and planned their long run risk taking. Necker and Ziegelmeyer (2016) interpret their results as an emotional reaction to the crisis.

Former researchers came to the conclusion that financial risk tolerance is a rather stable trait over the lifetime of an individual. The aim of this research is to investigate whether this is the case for the Dutch population. This research adds value, because, to my knowledge, this is the first study of the crisis’s consequences on risk taking of the general population in the Netherlands.

(3)

3 The empirical analysis is based on data from the DNB Household Survey from 2007, the start of the financial crisis, to 2012, the start of the recovery of the economy. The information from these years provides the opportunity to unravel the effects of the financial crisis on the financial risk tolerance of individuals. The DNB Household survey consists of data of information on work, pensions, housing, mortgages, income, possessions, loans, health, economic and psychological concepts and personal characteristics. There are six risk statements which measure the risk tolerance of individuals. The average of these six risk statements is used as dependent variable and is named average level of risk tolerance. This survey is held under 2,000 households, which are supposed to reflect the Dutch population. Thus, this thesis is about the population rather than just investors. Besides, the DNB Household Survey contains personal information on income, wealth, age, gender and education which are used as control variables.

In order to answer the research question a linear regression is created. The results show that the global financial crisis is positive statistically significance. This means that we can reject the null hypothesis at every significance level that the financial risk tolerance of individuals in the Netherlands does not change during the financial crisis. There is a decreasing risk tolerance level from 2007 until 2012. Yet, the change of the risk tolerance level is minor. Besides this, the six risk tolerance variables show also a significant result for the global financial crisis. This thesis contributes to the behavioural finance literature by showing that the global financial crisis influences risk tolerance on the complete population of the Netherlands. This study is the first to show that risk taking in the general population in the Netherlands is positively affected by the crisis in comparison to the post crisis years.

This paper is organized as follows. The second Section contains a review of the related literature and will end with the hypothesis. In Section 3 the approach of the empirical analysis and the research method are described. Section 4 contains the results and Section 5 concludes. In Section 6 the shortcomings and improvements that can be made in the future are discussed.

2. Literature review

2.1. What is financial risk tolerance?

(4)

4 comfort level that an individual is willing to accept while risking their current wealth for future growth”. Gerrans, Faff and Hartnett (2015) state that financial risk tolerance can be best defined as the extent to which the investor “... is willing to risk experiencing a less favourable financial outcome in the pursuit of a more favourable financial outcome.” Various studies tried to define financial risk tolerance. This shows that financial risk tolerance is a rather difficult understanding.

Determinants of financial risk tolerance are whether an individual is risk neutral, risk loving or risk averse. Individuals who only care about expected values and not about the risk level are called risk neutral. This corresponds to a linear utility function. Then, people are said to be risk loving when taking an additional risk for an investment that has a relatively low return (Ackert and Deveas, 2010). The utility function of a risk loving individual is convex. Finally, risk averse can be defined as “a person who dislikes risk”. According to Modern Portfolio Theory assets with a high variance corresponds to a higher expected return (Markowitz, 1952). This means that the utility function of the investor is concave. The more concave the utility function is, the less willing the investor is to accept variation in consumption over time. Thus, the investor is risk averse. Therefore, these individuals need to be compensated, to accept the greater consumption variation, with a higher expected return (Guillemete and Nanigian, 2014).

2.2. What factors influence financial risk tolerance?

Many studies investigated the factors regarding the financial risk tolerance. The income or wealth level of individuals is one of these factors. Gollier (2001) states that the microeconomic theory assumes that the willingness to take risk decreases as wealth decreases. Carducci and Wong (1998) investigated personality factors as determinants of financial risk taking in personal investments and household affairs. They suggest that income explains the financial risk tolerance of individuals. Individuals who possess a higher income level are more willingly to take greater financial risks. Yet, several studies (Brunnermeier and Nagel, 2008; Sahm, 2012) find no significant relationship between changes in wealth or income and changes in the risk level of individuals.

(5)

5 Research conducted by Hoffmann, Post and Pennings (2013) makes a distinction between risk tolerance and risk perception. Their results show a decrease in risk tolerance and an increase in risk perception during the financial crisis. Overall, they state that investors do not take less risk during the crisis. A follow-up study conducted by Hoffmann, Post and Pennings (2015) find that the perceptions of investors are important influencers of risk-taking behaviour. Their results show that investors with higher level of upward revisions in risk tolerance have a higher risk tolerance level.

Finally, an increasing number of studies investigated the risk tolerance level by examining demographic, socioeconomic, and attitudinal characteristics of individuals (Dohmen et al., 2011; Gibson, Michayluk and Van de Venter, 2013). They find that such characteristics have a significant effect on the financial risk tolerance of individuals; women and older individuals tend to have a lower risk tolerance. Besides, Barber and Odean (2001) and Bailey et al. (2011) show that financial risk tolerance is influenced by investor psychology. Barber and Odean (2001) find that individuals are overconfident regarding their knowledge leading to high levels of trading. According to Sivaramakrishnan, Srivastava and Rastogi (2017) and Rooij, Lusardi and Alessie (2011) financial literacy has a positive significant relation with the financial risk tolerance of individuals. Figner and Weber (2011) state that risk taking is not a single trait but a behaviour influenced by characteristics and characteristics of the situation and the decision. Controversy, Van de Venter, Michayluk and Davey (2012) argue that the financial risk tolerance is a stable personality trait which does not change over the lifetime of an individual.

Additionally, Nicolosi, Peng and Zhu (2009) state that individuals learn from their investment experiences and consequently adjust their trading and achieve higher investment performance. Gervais and Odean (2001), Koestner et al., (2017) and Seru et al., (2011) agree on the statement that individuals learn slowly from their failures if they even learn from it due to overconfidence. Controversy, Glaser and Weber (2017) find that individuals fail in updating their behaviour to match their experiences and are unwitting of their return performance.

Overall, researchers seem to agree that financial risk tolerance is statistically related and influenced by several factors, such as age, education, gender, income and wealth

2.3. The Global Financial Crisis

(6)

6 the Ice Save Bank in Iceland (Financieel.InfoNu.nl, 2018 and Centre for Economic Policy Research, 2017).

Many Dutch citizens lost their money with the fall of this bank. Besides this, the nationalization from parts of ABN AMRO and Fortis was also noticeable in the Netherlands. Consequently, Dutch households lost 27% of their equity, which equals 66 milliard euro. Pension funds and insurers lost 33% percent of their equity, equal to 160 milliard euro and banks lost 21% of their shares, this caused investors to lose trust in the financial markets (Cbs.nl, 2018).

In the first quarter of 2008 the financial crisis reached its peak in the Netherlands (Centre for Economic Policy Research, 2017). Until 2012 households do not move, companies do not invest and the consumer confidence is low (Cbs.nl, 2018). Yet, since 2012 the Dutch economy settles and from the second half of 2012 the recovery of the economic growth started (Cpb.nl, 2018). Since 2013 the GDP has surpassed the peak it had reached in 2008 (Centre for Economic Policy Research, 2017).

2.4. What happened to the financial risk tolerance level of individuals during and after the financial crisis?

Several studies conducted research to the effect of the global financial crisis on the financial risk tolerance of individuals. The studies of Hoffmann, Post and Pennings (2013) and Weber, Weber and Nosic (2012) find that during the global financial crisis the risk attitudes and risk taking behaviour of individuals vary.

Research conducted by Hoffmann, Post and Pennings (2013) states that individuals continue to trade actively and do not lower their risk level during the global financial crisis. Moreover, Weber, Weber and Nosic (2013) investigated in a period of September 2008 – June 2009 the willingness to take risk by customers in the United Kingdom. In their study they concluded that there was hardly a change in the risk tolerance during the financial crisis. Yet, this study does not show a contrast between pre or post global financial crisis and the global financial crisis.

(7)

7 risk and therefore change their personal financial risk tolerance. This is in line with the research conducted by Thaler and Johnson (1990). They state that making generalizations about risk tolerance is difficult, but experiencing losses leads to individuals taking less risk and vice versa.

2.5. Hypothesis development

A large body of research examines the effects of the global financial crisis on the risk tolerance level of individuals, with outcomes varying from positive to negative to no change, and offering different explanations for why such outcomes occur. Despite extensive research, there is no consensus about the effect of the global financial crisis on the risk tolerance level of individuals, but previous literature makes it clear that a reaction could follow from the global financial crisis. The following research question has been developed.

Research question: What is the financial risk tolerance of individuals during and after the global financial crisis in the Netherlands?

In order to test this research question, the following hypothesis is tested.

: The financial risk tolerance of individuals in the Netherlands does not change during the financial crisis.

3. Research Design 3.1. Data collection

In this paper use is made of data of the DNB Household Survey. The DNB Household Survey is launched in 1993 and is an active panel data set held under 2,000 Dutch households every year. These Dutch households show a good reflection of the Dutch population. In order to give every household the opportunity to participate to the survey, a so-called set-top box with built-in internet is given to households without computer and internet connection. Even a television set was provided, if needed, to the households to give them the opportunity to participate in the survey.

The panel consists of data of information on work, pensions, housing, mortgages, income, possessions, loans, health, economic and psychological concepts and personal characteristics. The analysis in this thesis is based on information from 2007, the start of the global financial crisis, to 2012, the start of the recovery of the economy.

(8)

8 those observations and after matching the ID’s of the individuals, only 693 individuals filled in all the questions for the period 2007-2012. This could mean that the outcomes are not representative for the Netherlands. It is difficult to assess whether the representative is affected, because all the individuals are anonymous and there is no reason found why the individuals did not answer these questions.

The main goal of this thesis is to investigate whether there is a difference in the financial risk tolerance level of individuals during the years of the financial crisis in comparison to the years after the financial crisis. Therefore, the global financial crisis is also a variable in this model. In this thesis the years 2007 and 2008 correspond to the global financial crisis and the years 2009-2012 correspond to the post global financial crisis. Although the crisis is since 2008 noticeable in the Netherlands, in this thesis 2007 corresponds also to the financial crisis. The reason for this is to make sure that we take all individual’s reactions to this event into account. The years during the global financial crisis correspond to a dummy variable “1” and the years after the financial crisis correspond to “0”.

The psychological part of the DNB Households Survey contains six hypothetical statements which measure the financial risk tolerance of the Dutch Households. These six risk statements are shown in table 1. The average of the six statements is used as one dependent variable in the model, which is called the average level of risk tolerance.

Table 1 Survey questions

Name Survey variable Answer categories

Spaar1a 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

1 (totally disagree) – 7 (totally agree)

Spaar2a I would never consider investments in shares because I find this too risky

1 (totally disagree) – 7 (totally agree) Spaar3 If I think an investment will be profitable, I am

prepared to borrow money to make this investment

Spaar4a I want to be certain that my investments are safe 1 (totally disagree) – 7 (totally agree) Spaar5 I get more and more convinced that I should take

greater financial risks to improve my financial position

Spaar6 I am prepared to take the risk to lose money, when there is also a chance to gain money

1 (totally disagree) – 7 (totally agree) Notes: This table presents the questions as used in the DNB Household Survey. A 7-point Likert scale is used to record individuals’ response to each question. Answer option “9 don’t know” is deleted from the observations.

a

Denotes a reverse-scored question.

(9)

9 (1)

Furthermore, the DNB Household Survey contains information on age, gender and education of the respondents. These data are used to create control variables. Grable (2000) states that age, gender and education influences the financial risk tolerance of individuals. First, older people are seemed to be less risk tolerant. Secondly, a higher education level leads to a more risk tolerant level. Finally, Grable (2000) states that men are more risk tolerant than women.

Each respondent is asked for the year of birth (gebjaar), which is transformed to the age of each respondent, and gender (geslacht). Gender indicates the sex of the respondents, with a “0” indicating for a male and “1” for a female. Besides, in the DNB Household Survey the respondents indicate their highest level of education they have completed (oplmet). Education is measured with the highest completed and received diploma or certificate. The following answer options are possible:

1. Special education 2. Primary education

3. Pre-vocational education (VMBO) 4. Pre-university education (HAVO/VWO) 5. Senior vocational or apprentice system 6. Vocational colleges (HBO)

7. University education (WO) 8. No education (yet)

9. Other education

“Other education” is an answer that does not describe the education level. Therefore, this answer is deleted from the observations. Due to missing variables, there are no respondents who filled in the answer “no education (yet)”.

Besides this, income and wealth are also control variables used in this thesis. The DNB Household Survey contains annual information on household income. Several studies (Gollier, 2001; Carducci and Wong, 1998) conclude that the income level has a positive influence on the financial risk tolerance of individuals.

(10)

10 Inkhh. Into which of the categories mentioned below did the total net income of your

household go in the past 12 months? 1. Less than €10,000 2. Between €10,000 and €14,000 3. Between €14,000 and €22,000 4. Between €22,000 and €40,000 5. Between €40,000 and €75,000 6. €75,000 or more 9. Don’t know

Inknorm. Is this income unusually high or low compared to the income you would expect in a “regular” year, or is it regular?

1. Unusually low 2. Regular

3. Unusually high 9. Don’t know

Another variable is the level of wealth of households, which is measured by the question “finsitu”. The higher the wealth level of the households the higher the risk tolerance level is (Gollier, 2001). Controversy, Guiso et al. (2013) find that an increase in risk aversion is unrelated to changes in wealth.

Finsitu. How is the financial situation of your household at the moment? 1. There are debts

2. Need to draw upon savings 3. It is just about manageable 4. Some money is saved 5. A lot of money can be saved

In this thesis the independent variable crisis is used as an interaction term with the two income variables and the wealth variable. All three variables are multiplied with the crisis and used as an interaction variable in the regression. An interaction effect occurs when the effect of one variable depends on the value of another variable. Interaction effects indicate that a third variable influences the relationship between an independent and dependent variable.

(11)

11 financial literacy of individuals. Yet, this survey is not held every year. Therefore it is not possible to check whether the respondents improved their financial literacy over years or not.

Furthermore, there is no information on current and former jobs of households. Consequently, a control variable for financial literacy and job-related changes is not created in this thesis.

3.2. The empirical model

To check whether the financial risk tolerance of individuals is different during the years of the crisis in comparison to the post crisis years the following regression can be tested.

(2)

On the left hand side of the equation is the dependent variable which is indicated by financial risk tolerance for person i at time t. The right hand side of the equation starts with a constant term, followed by the independent variable, financial crisis, denoted , where “1” corresponds to the years during the global financial crisis, 2007 and 2008, and “0” to the years after the global financial crisis, 2009 until 2012. A positive coefficient means that the risk tolerance level increases. The independent variable is then followed by the control variables, denoted . The regression includes control variables for socio-demographic characteristics, such as income, wealth, age and education. After this, a dummy variable is included for gender, which is denoted by . Finally, an error term is added to the equation, which is indicated as .

3.3. Research method

The data of the DNB Household Survey is considered to be unbalanced panel data. Time series data is collected over a period of time on one or more variables and cross-sectional data is collected at a single point in time. Panel data contains observations on multiple entities, where each entity is observed at two or more points in time. Basically, panel data, also known as longitudinal data, is a combination of time series data and cross-sectional data.

The simplest way to deal with unbalanced panel data would be to estimate a pooled regression, which would involve estimating a single equation on all the data together. This equation would be estimated using Ordinary Least Squares (OLS). Yet, it has some severe limitations. Namely, pooling the data assumes that the average values of the variables and the relationships between them are constant over time and across all of the cross-sectional units in the sample.

(12)

12 predictor variables. Within the fixed effects model there are two different models; the entity-fixed effects and time-entity-fixed effects model. The entity-entity-fixed effects model allows the intercept in the regression model to differ cross-sectionally, but not over time. The time-fixed effects model allows the intercept in the regression model to differ over time but not cross-sectionally. A combined entity and time-fixed effects regression model eliminates omitted variables bias arising both from unobserved variables that are constant over time and from variables that are constant across states.

Unlike the fixed effects model, the random effects model assumes the variation across entities to be random and uncorrelated with predictor or independent variables included in the model. An advantage of random effects is that time-invariant variables (i.e. gender) can be included. The random effects model assumes that the entity’s error term is not correlated with predictor which allows for time-invariant variables to play a role as explanatory variables.

To determine which model is most appropriate to use in this thesis a standard pooled

OLS and a model including entity fixed effects are compared with each other. There are hardly any differences between the results of these two models, which could mean that the fixed effects

model is not necessary. After this, a redundant fixed effects test checks whether you prefer the

pooled OLS or the fixed effects estimates. The redundant fixed effects test shows a probability of 0.0000. Therefore, the null hypothesis that the effects are redundant is rejected and the fixed

effects model is more appropriate to use.

After this the Hausman-test is conducted to determine whether the fixed or random effects model is more appropriate to use. The corresponding hypothesis for this test is: is uncorrelated with the explanatory variables. If the null hypothesis is rejected, the random effects method is not appropriate and fixed effects model should be preferred. The Hausman-test results in a probability of 0.0000, which means that the null hypothesis is rejected at each significance level and the fixed effects models is the most appropriate model to use.

Although the fixed effects model is the most appropriate model to use, gender is a crucial control variable which does not change over time and therefore an entity-fixed effects model is not able to use. To maintain comparability of the framework across specification, we focus on the pooled OLS approach. Thereby we should take into account that a pooled OLS neglects the panel aspect of the data and interprets everything as one big cross-section. It should be noted that fixed-effects regressions yield very similar effects, which can be found in Appendix A.

(13)

13 4. Results

4.1. Descriptive statistics

(14)

14 Table 2

Descriptive statistics of the dependent variable “average level of risk tolerance” and the risk tolerance variables for the years 2007-2012

a

Denotes a reverse-scored question.

Table 2 report the results of the mean, the standard deviation, the minimum, the median, the maximum and the number of observations. Looking at the differences in years in Table 2 it can be concluded that the average level of risk tolerance is slightly higher for the years 2007 and 2008, which are indicated as the global financial crisis. This means that Variables Average level of risk

tolerance

Spaar1a Spaar2a Spaar3 Spaar4a Spaar5 Spaar6

(15)

15 individuals have a higher risk tolerance during the financial crisis. A higher risk tolerance during the crisis is against the earlier called intuition that we expect a lower risk tolerance level during the financial crisis. In the years after the peak of the global financial crisis, 2009 until 2012, the average risk tolerance level is lower, which means that the individuals are more risk averse. This indicates that there is a decreasing line from 2007 until 2012.

Besides, from Table 2 we can conclude that the average level of risk tolerance has a low value for each year, between 2.50 and 2.73. This means that the individuals are on average risk averse. In line with this result are the medians, which also show a relatively low value. Besides, the value of the mean of the six risk statements is also quite low. Looking at these six risk statements, noticeable is that the second statement, “I would never consider investments in shares because I find this too risky”, has a higher result than the other five risk statements for each year, between 3.00 and 3.60. This means that the individuals are slightly more risk loving in this statement.

A possible explanation for this result is the negative framing of a statement. According to Ackert and Deaves (2010) individuals are risk seeking in a negative frame and risk averse in a positive frame. The statement contains the word “never”, which could lead to risk seeking answer.

The six risk statements are also shown in Figure 1 to distinguish the value differences between them. In this Figure it is clear that the second statement results in a higher risk tolerance level. Interesting is that the trend of “Spaar2” is in line with the fluctuations of the AEX for 2007 until 2012 which could imply that this statement shows sentiment with shares on the market (Beleggen.nl, 2018).

Figure 1

The average level of risk tolerance per variable per year.

0 0,5 1 1,5 2 2,5 3 3,5 4 2007 2008 2009 2010 2011 2012

(16)

16 Figure 2 shows the average level of risk tolerance over the years 2007-2012. In the Figure the decrease in 2008 is noticeable. The period 2007 until 2009 has fluctuations within the area of 0.25

Figure 2

The average level of risk tolerance per year

Year

In Table 3 are the characteristics of the restricted sample, which are used as control variables, presented.

Table 3

Descriptive statistics of the independent and control variables

Variables Age Gender Education Income Inc. norm Financial situation Mean 57.6429 0.4113 4.6941 3.9052 1.9711 3.0772 Std. Dev 13.3749 0.4924 1.5461 1.3753 0.2262 1.2792 Min 24 0 1 2 1 1 Median 59 0 5 4 2 3 Max 87 1 7 6 3 5 N 693 693 693 693 693 693

(17)

17 Furthermore, Table 3 shows that 41% of the respondents are women and the average level of education is 4.69. Interesting is that the average income of the respondents correspond to the range €22,000 and €40,000, which is the average income in the Netherlands (Opendata.cbs.nl, 2018).

Figure 3

Normal distribution age

Figure 4

(18)

18 Table 4

Mean risk tolerance per risk variable and the average level of risk tolerance for control variables Variables Average level

of risk tolerance

Age 16-45 2.7136 2.6291 3.4225 2.0493 2.4883 2.9624 2.7300 46-65 2.7049 2.8753 3.4237 1.9650 2.7748 2.7290 2.4618 66 and older 2.4859 2.8491 3.1662 1.7931 2.5315 2.3704 2.2051 Gender Men 2.7612 2.9134 3.5915 2.0580 2.6638 2.7218 2.6189 Women 2.3498 2.7263 2.8579 1.6357 2.5550 2.3444 1.9795 Education Special education 3.0556 3.0556 3.8333 2.3889 2.6111 3.3889 3.0556 Primary education 2.7389 3.6115 3.2994 1.8662 3.4395 2.0828 2.1338 Pre-vocational edu. (VMBO) 2.4659 3.0522 2.8632 1.8085 2.7430 2.3284 2.0000 Pre-university edu. (HAVO/VWO) 2.5775 2.6974 3.3687 1.8196 2.4449 2.6673 2.4669 Senior vocational/ apprentice system 2.5493 2.7866 3.2317 1.8628 2.58470 2.6098 2.2302 Vocational colleges (HBO) 2.5465 2.6507 3.3098 1.8475 2.5057 2.5819 2.3832 University edu. (WO) 2.9571 2.6821 4.1482 2.1893 2.5446 3.0196 3.1589

No education (yet) - - - -

a

Denotes a reverse-scored question

Table 4 reports the means of the answers for the control variables of the average risk tolerance variable and the risk statements. The results observed in Table 4 are comparable to those of earlier studies (Grable, 2000 and Gollier, 2001). The age distribution shows that the younger the individuals are the more risk they take. This is in line with the research conducted by Grable (2000). Yet, the average level of risk tolerance shows a low value, around 2.50. Figure 4 shows the number of individuals per age distribution for the years 2007-2012. The younger individuals are underrepresented in this sample, which could lead to biased results. Furthermore, women tend to have a lower level of risk tolerance than men, which means that they are less risk tolerant, which is in line with Grable (2000). Looking at education there is no obvious trend shown. However, the only noticeable aspect regarding education is that the individuals who completed a special education are the most risk tolerant. According to Grable (2000) the respondents with a high educational level should be the most risk tolerant.

(19)

19 Table 5

Mean average level of risk tolerance for Income, Income norm and Financial situation Variables Average

level of risk tolerance

Income Less than €10,000 2.3690 3.3214 2.5000 1.6964 2.7876 2.0893 1.8214 Between €10,000 and €14,000 2.4342 3.1338 2.8567 1.6561 2.8217 2.2516 1.8854 Between €14,000 and €22,000 2.4673 2.8585 3.0183 1.8402 2.5967 2.3359 2.1540 Between €22,000 and €40,000 2.5872 2.8234 3.2409 1.8514 2.6548 2.6120 2.3407 Between €40,000 and €75,000 2.6843 2.7281 3.5210 1.9863 2.5529 2.7783 2.5392 €75,000 or more 2.9258 2.8410 4.1908 2.0989 2.5442 2.8410 3.3039 Income during crisis

Less than €10,000 - - - - Between €10,000 and €14,000 - - - - Between €14,000 and €22,000 2.1462 2.8421 2.3158 1.7018 2.2807 2.1053 1.6316 Between €22,000 and €40,000 2.6516 3.0000 3.1375 1.9688 2.8094 2.7219 2.2719 Between €40,000 and €75,000 2.6575 2.7483 3.3953 2.0848 2.5404 2.7168 2.4596 €75,000 or more 2.8966 2.7931 4.1877 2.1111 2.5019 2.8008 2.9847 Income after crisis

Less than €10,000 2.3690 3.3214 2.5000 1.6964 2.7876 2.0893 1.8214 Between €10,000 and €14,000 2.4342 3.1338 2.8567 1.6561 2.8217 2.2516 1.8854 Between €14,000 and €22,000 2.4859 2.8595 3.0591 1.8483 2.6151 2.3493 2.1843 Between €22,000 and €40,000 2.5673 2.7686 3.2730 1.8151 2.6070 2.5779 2.3621 Between €40,000 and €75,000 2.7379 2.6877 3.7726 1.7890 2.5781 2.9014 2.6986 €75,000 or more 3.2727 3.4091 4.2273 1.9545 3.0455 3.3182 3.6818 Income norm Unusually low 2.7934 3.2281 3.2807 2.1228 2.6374 2.9649 2.5263 Regular 2.5803 2.8152 3.2868 1.8735 2.6174 2.5416 2.3473 Unusually high 2.6446 2.6324 3.5588 1.9118 2.2941 3.0000 2.4706 Income norm during crisis

Unusually low 2.7825 3.0000 3.3898 2.2881 2.6949 2.9322 2.3898 Regular 2.6734 2.8079 3.4425 2.0358 2.5731 2.6874 2.4938 Unusually high 2.8269 2.4231 4.1154 2.0769 2.4231 3.1538 2.7692 Income norm after crisis

Unusually low 2.7991 3.34821 3.2232 2.0357 2.6071 2.9821 2.5982 Regular 2.5342 2.8189 3.2096 1.7931 2.6393 2.4694 2.2748 Unusually high 2.5317 2.7619 3.2143 1.8095 2.2143 2.9048 2.2857 Financial situation

There are debts 2.6547 2.9272 3.3087 1.9812 2.6826 2.6110 2.4178 Need to draw upon

savings

(20)

20 Table 5 (continued) It is just about manageable 2.5610 2.9526 3.1024 1.8976 2.7222 2.5736 2.1178 Some money is saved 2.5666 2.7695 3.2992 1.8690 2.5739 2.5140 2.3737 A lot of money can

be saved

2.7989 2.6884 4.1161 1.8499 2.5297 2.6686 2.9405 Financial situation during crisis

savings 2.5409 2.6140 3.1667 2.0965 2.4211 2.7281 2.2193 It is just about manageable 2.6620 2.9444 3.4167 1.8611 2.7500 2.6806 2.3194 Some money is saved 2.7700 2.8145 3.6145 2.1768 2.4986 2.8841 2.6319 A lot of money can

be saved

2.9460 2.6901 4.2394 2.2958 2.5070 2.7465 3.1972 Financial situation after crisis

savings 2.4675 2.9114 2.8785 1.6987 2.6203 2.5544 2.1418 It is just about manageable 2.5382 2.9545 3.0314 1.9058 2.7159 2.5495 2.0722 Some money is saved 2.5154 2.7582 3.2200 1.7917 2.5929 2.4210 2.3088 A lot of money can

be saved

2.7618 2.6879 4.0851 1.7376 2.5355 2.6489 2.8759

a

Denotes a reverse-scored question

The results of Table 5 show that on average the higher the income the more risk tolerant individuals are, which is in line with Grable (2000). Noticeable is that during the financial crisis none of the individuals said to have an income below €10,000 or between €10,000 and €14,000. This could lead to biased results. The second statement, “I would never consider investments in shares because I find this too risky”, shows a higher risk tolerant level than the other risk statements.

Looking at the income norm of individuals it is hard to find a trend. Even the years during the crisis and after the crisis do not show a trend. A noticeable affect regarding the financial situation is that during the global financial crisis individuals are slightly more risk tolerant than the years after the global financial crisis. Another noticeable affect is that for the second statement, “I would never consider investments in shares because I find this too risky”, individuals are most risk tolerant, especially if individuals say “a lot of money can be saved”. This means that the higher the wealth level the more risk tolerant a household is, which is in line with Gollier (2001).

(21)

21 Figure 5

The income level of the individuals per year

Figure 6

(22)

22 Figure 7

Financial situation of the individuals per year

The financial situation of individuals per year is shown in Figure 7. This Figure shows that in 2008 there is a peak in which individuals experience their financial situation as “there are debts”. While in the year previous or the years following the answers on the question about the financial situation are more diversified. Most of the individuals say that “some money can be saved”. This means that during the global financial crisis individuals experience difficulties in their financial situation.

(23)

23 4.2. Regression results

In this thesis the risk tolerance level is the dependent variable and the global financial crisis is the independent variable in the regression. The socio-demographic characteristics such as age, education, income and wealth are used as control variables. The dummy variable in this regression is gender, of which male is removed.

There are seven different variants of the model. For this purpose, seven different sets of control variables X are defined which are included on after another. The basic regression is the financial risk aversion and the financial crisis. With these variables I intend to capture how the risk tolerance level changes during and after the global financial crisis.

(24)

24 Table 6

Determinants of average level of risk tolerance

Hypothesis tests are based on robust standard errors. Significance levels: *:10% **: 5% ***:1% Average level of risk

tolerance Model 1 (robust st. error) Model 2 (robust st. error) Model 3 (robust st. error) Model 4 (robust st. error) Model 5 (robust st. error) Model 6 (robust st. error) Model 7 (robust st. error) Alfa 2.5474*** (0.0142) 3.3110*** (0.0106) 3.1789*** (0.1265) 2.1148*** (0.0437) 2.3326*** (0.0638) 2.1186*** (0.0624) 3.1214*** (0.1293) Crisis 0.1338*** (0.0376) 0.1340*** (0.0376) 0.1226*** (0.0279) 0.1150*** (0.0337) 0.1239** (0.0528) 0.0518 (0.484) Age -0.0103*** (0.0013) -0.0103*** (0.0012) -0.0101*** (0.0012) Female -0.4399*** (0.0222) -0.4341*** (0.0235) -0.4311*** (0.0239) Education 0.0189** (0.0077) 0.0139** (0.0235) 0.0504*** (0.0058) 0.04773*** (0.0057) 0.0127* (0.0067) Income 0.0353*** (0.0091) 0.0447*** (0.0131) 0.0664*** (0.0138) 0.0851*** (0.0126) 0.0639*** (0.0164) Income norm -0.0325** (0.0137) -0.0312*** (0.0116) -0.0461*** (0.0133) -0.0433** (0.0181) Financial situation 0.0253** (0.0102) 0.0279** (0.0114) -0.0019 (0.0090) -0.0025 (0.0100) Crisis * Income -0.0597*** (0.0128) -0.0393** (0.0165)

Crisis * Income norm 0.0392*

(0.0212)

0.0298 (0.0274)

Crisis * Financial situation 0.0505***

(25)

25 Table 7

Determinants of “Spaar1”

Hypothesis tests are based on robust standard errors. Significance levels: *:10% **: 5% ***:1%

Spaar1 Model 1 (robust st. error) Model 2 (robust st. error) Model 3 (robust st. error) Model 4 (robust st. error) Model 5 (robust st. error) Model 6 (robust st. error) Model 7 (robust st. error) Alfa 2.8478*** (0.0139) 3.4637*** (0.1639) 3.8024*** (0.2513) 3.7258*** (0.1665) 2.9857*** (0.0889) 4.0093*** (0.0897) 4.0993*** (0.2250) Crisis -0.0339** (0.0156) -0.0302 (0.0189) -0.0875 (0.0600) -0.0913 (0.0586) -0.6846*** (0.0853) -0.7109*** (0.0919) Age 0.0018 (0.0019) 0.0017 (0.0019) -0.0018 (0.0019) Female -0.2481*** (0.0457) -0.2556*** (0.0448) -0.2590*** (0.0457) Education -0.1313*** (0.0141) -0.1217*** (0.0112) -0.1118*** (0.0096) -0.1061*** (0.0100) -0.1155*** (0.0109) Income -0.0136 (0.0143) -0.0102 (0.0126) -0.0382 (0.0235) -0.0397* (0.0217) -0.0483** (0.0242) Income norm -0.0701*** (0.0169) -0.0699** (0.0159) -0.0814*** (0.0261) -0.0805*** (0.0281) Financial situation -0.0573** (0.0230) -0.0530** (0.0233) -0.1061*** (0.0141) -0.107*** (0.0142) Crisis * Income 0.0540** (0.0223) -0.0621** (0.0249)

Crisis * Income norm 0.0370

(26)

26 Table 8

Spaar2 Model 1 (robust st. error) Model 2 (robust st. error) Model 3 (robust st. error) Model 4 (robust st. error) Model 5 (robust st. error) Model 6 (robust st. error) Model 7 (robust st. error) Alfa 3.2078*** (0.0779) 3.5126*** (0.1358) 2.895* (0.435) 1.5775*** (0.1456) 2.6992*** (0.1888) 1.2391*** (0.1196) 2.5258*** (0.1715) Crisis 0.2460** (0.1047) 0.2132** (0.0999) 0.343* (0.075) 0.3347*** (0.1138) 1.0556*** (0.1239) 0.9523*** (0.1192) Age -0.0115*** (0.0013) -0.011** (0.005) -0.0114*** (0.0012) Female -0.7041*** (0.0387) -0.681* (0.122) -0.6762*** (0.0410) Education 0.1403*** (0.0090) 0.119* (0.043) 0.1694*** (0.0130) 0.1592*** (0.0144) 0.1108*** (0.0139) Income 0.076** (0.037) 0.0898** (0.0352) 0.1512*** (0.0422) 0.1724*** (0.0468) 0.1416*** (0.0422) Income norm -0.014 (0.032) -0.0124 (0.0150) -0.0202 (0.0262) -0.0164 (0.0253) Financial situation 0.149* (0.034) 0.1553*** (0.0205) 0.1872*** (0.0195) 0.1839*** (0.0164) Crisis * Income -0.1349*** (0.0507) -0.1053** (0.0451)

(27)

27 Table 9

Spaar3 Model 1 (robust st. error) Model 2 (robust st. error) Model 3 (robust st. error) Model 4 (robust st. error) Model 5 (robust st. error) Model 6 (robust st. error) Model 7 (robust st. error) Alfa 1.8041*** (0.0343) 2.6207*** (0.1410) 2.5845*** (0.1520) 1.4954*** (0.154) 1.6586*** (0.0627) 1.6707*** (0.0598) 2.7795*** (0.1288) Crisis 0.2406*** (0.0518) 0.2066*** (0.0508) 0.2063*** (0.0574) 0.2368*** (0.0630) -0.1234 (0.0849) -0.2015** (0.0861) Age -0.0114*** (0.0012) -0.0114*** (0.0012) -0.0113*** (0.0012) Female -0.4601*** (0.0391) -0.4584*** (0.0378) -0.4605*** (0.0384) Education 0.0083 (0.0098) 0.0070 (0.0093) 0.0464*** (0.0074) 0.0492*** (0.0075) -0.0110 (0.0090) Income 0.0096 (0.0128) 0.0180 (0.019) 0.0578*** (0.0084) 0.0123 (0.0226) -0.0106 (0.0260) Income norm -0.0128 (0.0137) -0.015 (0.0107) -0.0222 (0.0149) -0.0193 (0.0209) Financial situation 0.0092 (0.0173) 0.0140 (0.0191) -0.0283*** (0.0085) -0.0287*** (0.0085) Crisis * Income 0.0149 (0.0270) 0.0370 (0.0287)

(28)

28 Table 10

Spaar4 Model 1 (robust st. error) Model 2 (robust st. error) Model 3 (robust st. error) Model 4 (robust st. error) Model 5 (robust st. error) Model 6 (robust st. error) Model 7 (robust st. error) Alfa 2.6374*** (0.0271) 3.4307*** (0.0622) 3.6547*** (0.0956) 3.2218*** (0.0874) 2.7125*** (0.0498) 3.2713*** (0.1044) 3.7117*** (0.1049) Crisis -0.0552* (0.0295) -0.0700** (0.0281) -0.1186*** (0.0367) -0.1065*** (0.0412) -0.1970** (0.0804) -0.2278*** (0.0782) Age -0.0045*** (0.0014) -0.0045*** (0.0013) -0.0045*** (0.0014) Female -0.1754*** (0.0240) -0.1800*** (0.0224) -0.1806*** (0.0219) Education -0.0967*** (0.0083) -0.0907*** (0.0091) -0.0751*** (0.0102) -0.0745*** (0.0104) -0.0897*** (0.0092) Income -0.0022 (0.0066) 0.0011 (0.0057) -0.0239*** (0.0088) 0.0038 (0.0129) -0.0052 (0.0129) Income norm -0.0551*** (0.0110) -0.0546*** (0.0093) -0.0531*** (0.0104) -0.0519*** (0.0130) Financial situation -0.0398*** (0.0123) -0.0380*** (0.0126) -0.0563** (0.0229) -0.0564** (0.0231) Crisis * Income -0.0019 (0.0136) 0.0068 (0.0140)

Crisis * Income norm -0.0019

(29)

29 Table 11

Spaar5 Model 1 (robust st. error) Model 2 (robust st. error) Model 3 (robust st. error) Model 4 (robust st. error) Model 5 (robust st. error) Model 6 (robust st. error) Model 7 (robust st. error) Alfa 2.4986*** (0.01172) 3.4406*** (0.1604) 3.2350*** (0.1986) 1.6978*** (0.0668) 2.0807*** (0.01398) 1.7319*** (0.1632) 3.2560*** (0.2904) Crisis 0.2042 (0.1389) 0.1481 (0.1402) 0.0403 (0.1628) 0.0899 (0.1628) 0.0861 (0.1488) 0.0000 (0.1424) Age -0.0190*** (0.0025) -0.0188*** (0.0025) -0.0185*** (0.0025) Female -0.4104*** (0.0439) -0.4018*** (0.0432) -0.3938*** (0.0413) Education 0.0726*** (0.0221) 0.0633*** (0.0197) 0.1112*** (0.0180) 0.1048*** (0.0153) 0.0591*** (0.0170) Income 0.0840** (0.0340) 0.0929** (0.0362) 0.1244*** (0.0318) 0.1958*** (0.0607) 0.1716** (0.0678) Income norm -0.0208 (0.0216) -0.0192 (0.0198) -0.0439** (0.0171) -0.0406* (0.0239) Financial situation -0.0055 (0.0316) -0.0031 (0.0335) -0.0943*** (0.0212) -0.0899*** (0.0238) Crisis * Income -0.1511** (0.0610) -0.1275* (0.0683)

Crisis * Income norm 0.0675*

(30)

30 Table 12

Spaar6 Model 1 (robust st. error) Model 2 (robust st. error) Model 3 (robust st. error) Model 4 (robust st. error) Model 5 (robust st. error) Model 6 (robust st. error) Model 7 (robust st. error) Alfa 2.2890*** (0.0207) 2.9584*** (0.0896) 2.5281*** (0.1028) 0.9705*** (0.0649) 1.8586*** (0.1626) 0.7891*** (0.0778) 2.3562*** (0.1069) Crisis 0.2009*** (0.0366) 0.1524*** (0.0346) 0.1819*** (0.0560) 0.2263*** (0.0543) 0.6063*** (0.1003) 0.4989*** (0.0977) Age -0.0167*** (0.0014) -0.0166*** (0.0014) -0.0164*** (0.0014) Female -0.6386*** (0.0439) -0.6241*** (0.0477) -0.6163*** (0.0478) Education 0.1215*** (0.0127) 0.1071*** (0.0131) 0.1625*** (0.0135) 0.1538*** (0.0122) 0.1007*** (0.0119) Income 0.0650*** (0.0249) 0.0767*** (0.0294) 0.1273*** (0.0392) 0.1657*** (0.0208) 0.1345*** (0.0241) Income norm -0.0212 (0.0276) -0.0194 (0.0255) -0.0559*** (0.0201) -0.0519* (0.0279) Financial situation 0.0861*** (0.0182) 0.0924*** (0.0177) 0.0863*** (0.0250) 0.0865*** (0.0279) Crisis * Income -0.1394*** (0.0222) -0.1092*** (0.0268)

Crisis * Income norm 0.0886**

(31)

31 Table 6 reports the results of estimating equation 2. The first column contains the results from the basic regression including the financial crisis as independent variable. The different sets of controls are added step by step. A positive coefficient implies an increase in financial risk tolerance from 2007 to 2012. Table 7 until 12 reports the results of the risk tolerance variables.

From Table 6 we observe that the crisis is positive significance related to the financial risk tolerance for all the models. Our result is, at first sight, not in line with the intuitive that investors take less risk during the financial crisis. The coefficient value of 0.1338, in the first model, suggests that on average and holding everything else equal, the financial risk tolerance is around 0.1338 higher during the years of the financial crisis than the average for the years after the financial crisis.

A study conducted by Gerrans, Faff and Hartnett (2015) compares the years prior to the financial crisis to the years during the financial crisis. Their study concludes that there is a decrease in the financial risk tolerance during the financial crisis. However, they do not investigate whether this line continues post financial crisis. Therefore, in our research it could be that the financial risk tolerance prior to the financial crisis is even higher than during the financial, which indicates a decreasing trend. Besides this, research conducted by Hoffman, Post and Pennings (2013) concludes that during the financial crisis the level of risk does not change, but is rather stable over years. Although our result is positive the result is minor and therefore we can say that the financial risk tolerance is a stable trait.

The second column of Table 6 shows the results of the regression in which the control variables age, gender and education are taken into account. The model shows a significant result at each significance level. Interesting results on the control variables are the following. In each model the control variable age has a negative significant coefficient, which means that the average risk tolerance decreases with age. This is in line with Grable (2000), who states that an older individual is less risk tolerant. In model 2 and 3 the effect is -0.0103, this can be interpreted as when the individual is aging the risk tolerance level decreases with 0.0103 scale points. Furthermore, the control variable female shows a negative significant coefficient for each model. This implies that a woman is less risk tolerant than a man, which is in line with Grable (2000). The effect of gender is -0.4399 which indicates that women tend to decrease their risk tolerance level with 0.44 scale points. Finally, the control variable education shows a positive significant result, which implies that the higher educated an individual is the higher the risk tolerance level is. This is in line with Grable (2000).

(32)

32 financial situation is, the higher the risk tolerance level is. These results are in line with Gollier (2001) and Carducci and Wong (1998). Yet, the income norm shows a significant negative coefficient, which implies that the higher the income level is valued by individuals the lower the risk tolerance level is.

Interesting is that in model 5, where only income is included in the regression, income has a larger positive impact on the financial risk tolerance than in the other models. This is in line with the studies of Gollier (2001) and Carducci and Wong (1998).

Column 6 and 7 includes the interaction variables. Overall, the three interaction variables in the regression are significant for the two models. The interaction variable “crisis x income” shows a negative significant result. This means that during the financial crisis the higher the income level is the less risk tolerant the individuals are. This tells us that for an increase in the income level we will see a decrease of 0.0597 scale point in the risk tolerance level during the crisis. In this case, every increase in income level leads to 2.1186-0.0597= 2.0589. In model 3 the results show that a higher income corresponds to a higher risk tolerance level. The results from the interaction variable could mean that during the crisis the individuals with the highest income become more risk averse. Besides, the interaction variable “crisis x income norm” shows a positive significant result in model 6 and a positive result in model 7. This implies that during the financial crisis the higher the income level is valued the higher the risk tolerance level of individuals is, which is not in line with the previous interaction variable “crisis x income”. Finally, the interaction variable “crisis x financial situation” is positive significant related to the risk tolerance level. This means that during the crisis the better the financial situation is the higher the risk tolerance level is.

Comparing the models, adding control variables increases the adjusted R-squared, which means that these control variables explains more of the dependent variable. When focusing on model 6 and 7, when the interaction variables are included, the R-squared and the adjusted R-squared are slightly higher than without the three interaction variables.

(33)

33 Looking at Table 7 until 12 the independent variable crisis shows similar results as in Table 6. Only Table 7 and 10, which correspond to the first and the fourth statement, the independent variable crisis shows a negative coefficient. This implies that individuals take less risk during the financial crisis. The other four statements show a positive coefficient which could have led to the positive coefficient for the average level of risk tolerance.

The control variables education, income and financial situation show in Table 7 and 10 a negative coefficient while in Table 6 a positive coefficient is presented. These results lead to the opposite interpretation regarding these control variables. A negative coefficient for education, income and the financial situation suggests that the higher the educational level, the income level and the financial situation of the individual, the lower the risk tolerance level is. A clear explanation is not found for these results.

5. Further analysis

5.1. Serial autocorrelation

In this thesis the sample is tested for autocorrelation, which is the relationship between a variable and its lag value. The lagged value is simply the value that the variable took during a previous period. It is assumed that the errors are uncorrelated with one another. If the errors are not uncorrelated with one another, it would be stated that they are autocorrelated or that they are serially correlated.

The Durbin-Watson (DW) is a test for first order autocorrelation; it tests for a relationship between an error and its immediately previous value, the lagged value. The corresponding null hypothesis is that the errors at time t-1 and t are independent of one another. If the null hypothesis is rejected it would be concluded that there was evidence of a relationship between successive residuals.

In this case the DW test statistic value is 0.6438. The relevant critical values for the test are =1.44 and =1.65. So and . The test statistic is clearly lower than the lower critical value and hence the null hypothesis of no autocorrelation is rejected and it would be concluded that that the residuals from the model appear to be positively autocorrelated. This means that the standard errors might be smaller than the true standard errors.

5.2. Sample selection bias

(34)

34 (57 years vs. 49 years), have an average income (between €22,000 and €40,000 vs. €27,000), and is quite high educated (senior vocation or apprentice system vs. 43% senior vocation). There are differences in the background of the respondents concerning gender and age. This may lead to biased results.

5.3. Gender effects

Earlier research shows that women tend to be less risk tolerant than men (Grable, 2000). To determine whether the women in this sample are also less risk tolerant than men a dummy variable is put into the regression. Also an interaction variable “crisis x female” is added to the regression. This will tell us whether women respond in the same manner to crises as men. The results are presented in Table 13.

Table 13

Determinants of average level of risk tolerance variable and risk variables for gender and crisis Average

level of risk tolerance

Spaar1 Spaar2 Spaar3 Spaar4 Spaar5 Spaar6

Alfa 2.7003*** (0.0173) 2.9001*** (0.0116) 3.4902*** (0.0947) 1.9565*** (0.0489) 2.6746*** (0.0326) 2.6575*** (0.1487) 2.5227*** (0.0240) Crisis 0.1829*** (0.0411) 0.0398*** (0.0131) 0.3039*** (0.1107) 0.3045*** (0.0637) -0.0325 (0.0464) 0.1930 (0.1856) 0.2886*** (0.0451) Female -0.3716*** (0.0099) -0.1273*** (0.0465) -0.6867*** (0.0409) -0.3705*** (0.0290) -0.0904*** (0.0317) -0.3864*** (0.0779) -0.5683*** (0.0252) Crisis * Female -0.1195*** (0.0116) -0.1793*** (0.0466) -0.1408*** (0.0513) -0.1554*** (0.0319) -0.0552 (0.0605) 0.0272 (0.1180) -0.2132*** (0.0319) F 62.9103*** 4.3321*** 47.5446*** 42.6792*** 2.2146* 23.5839*** 72.6488*** R2 0.0435 0.0031 0.0332 0.0299 0.0016 0.0167 0.0499 Adjusted R2 0.0428 0.0024 0.0325 0.0292 0.0009 0.0160 0.0492 N 693 693 693 693 693 693 693

(35)

35 6. Discussion and conclusion

Economists are increasing interested in the impact of the global financial crisis on the economy. The crisis in 2007 and 2008 is unprecedented and individuals lost their trust in the financial markets and economy. The main goal of this thesis is to answer the research question: What is the financial risk tolerance of individuals during and after the global financial crisis in the Netherlands? In order to answer this question the DNB Household Survey of 2007-2012 is analysed. The dependent variable is the average level of risk tolerance, the global financial crisis is used as the independent variable and various socio- demographic variables are used as control variables.

This thesis shows a positive significant result for the global financial crisis, which indicates that the global financial crisis has economic impact on the risk tolerance level of individuals. Therefore, we can reject the null hypothesis and conclude that the global financial crisis has a positive impact on the financial risk tolerance of individuals in the Netherlands. We conclude that the financial risk tolerance level is higher during the crisis in comparison to the post crisis years. This suggest that the impact of the crisis on the financial risk tolerance decreases from 2007 until 2012. Research conducted by Gerrans, Faff and Hartnet (2015) shows a decreasing risk tolerance level from the years prior to the crisis until the crisis. Besides, the impact of the financial crisis on the financial risk tolerance is negligibly small. This suggests that the level of financial risk tolerance of individuals is a rather stable trait over a lifetime, which is in line with the study performed by Hoffmann, Post and Pennings (2013).

Besides, the control variables income, wealth, gender, age and education show significant results, which is in line with earlier studies (Grable, 2000 and Gollier, 2001). The control variable age has a negative coefficient, implying that the older an individual the less risk tolerant. Furthermore, the control variable gender shows that women are more risk averse than men are. Education shows a positive coefficient, meaning that the higher educated the higher the risk tolerance level is. Then, the income and financial situation coefficient are positive saying that the higher the income level or the financial situation the higher the risk tolerance level is.

(36)

36 7. Shortcomings and future research

In this thesis the years during the financial crisis are compared to the post crisis years. In our result the financial risk tolerance of individuals is higher during the financial crisis in comparison to the post crisis. This suggest a decreasing trend from 2007 until 2012. As the years prior to the financial crisis are not taken into account this could be a biased result. Additional research should be conducted to check how high the financial risk tolerance is in the years prior to the financial crisis and to check whether there is a decreasing line from the year prior to the crisis to the post crisis years.

The initial sample consists out of 2,000 respondents. Yet, only 639 respondents answered all the questions needed for this thesis. Therefore, it is possible that the sample does not represent the complete Dutch population. Besides this, the sample in this thesis could show biased results, because the average age of the individuals is 57 years, which is nearly retirement age.

Then, according to Ackert and Deaves (2010) the way in which a question or statement is asked or formulated to a decision maker has a strong impact on the answer given or the decision made. Positive or negative framing of the same problem can have a substantial influence. Looking at our descriptive statistics it is clear that individuals respond different to the second statement, “I would never consider investments in shares because I find this too risky”. Besides this, when looking at the regression results, statement 1, “I think it is more important to have safe investments and guaranteed returns, than to take risk to have a chance to get the highest possible returns”, and statement 4 “I want to be certain that my investments are safe”, show odd results for the independent variable crisis and the control variables education, income and financial situation, which could be caused by the framing. Hereby, it is not clear from the data how the DNB Household Survey reached a consensus to formulate these statements.

Previous studies showed that financial literacy and job-related changes could influence the financial risk tolerance of individuals. Therefore it is possible that the results of this thesis are biased due to these missing control variables. Therefore, additional research is needed to control for financial literacy and job-related changes.

(37)

37 References

Ackert, L.F., Deaves, R., 2010. Behavioural finance, psychology, decision-making, and markets. South Western Cengage Learning, Mason, OH, USA.

Bailey, W., Kumar, A., Ng, D., 2011. Behavioral biases of mutual fund investors. Journal of Financial Economics, v102(1), 1-27.

Barber, B.M., Odean, T., 2001. Boys will be boys: gender, overconfidence and common stock investment. The Quarterly Journal of Economics, v19(3), 262-292.

Beleggen.nl. (2018). AEX index grafieken - Koersgrafieken AEX index. [online] Available at: https://www.beleggen.nl/aex_index/grafieken [Accessed 16 May 2018].

Brooks, C. 2014. Introductory econometrics for finance. Cambridge University Press, Cambridge.

Brunnermeier, M.K., Nagel, S., 2008. Do wealth fluctuations generate time-varying risk aversion? Micro-evidence on individuals’ asset allocation. American Economic Review, v98(3), 713-736.

Bucher-Koenen, T., Ziegelmeyer, M., 2014. Once burned, twice shy? Financial literacy and wealth losses during the financial crisis. Review of Finance, v18(6), 2215-2246.

Cameron, I., Shah, M., 2012. Risk-taking behavior in the wake of natural disasters. National bureau of economic research. Cambridge, MA, USA.

Caplin, A., Leahy, J., 2001. Psychological expected utility theory and anticipatory feelings. The Quarterly Journal of Economics, v116(1), 55-79.

Carducci, B.J., Wong, A.S., 1998. Type a and risk taking in everyday money matters. Journal of Business and Psychology, v12(3), 355-359.

Cbs.nl. (2018). De Nederlandse economie 2008. [online] Available at: https://www.cbs.nl/nl-nl/nieuws/2009/37/de-nederlandse-economie-2008 [Accessed 31 Jan. 2018].

(38)

38 Centre for Economic Policy Research, (2017). Euro area business cycle dating committee: A slow but steady euro area recovery. Asian Social Science, v5(8).

Cpb.nl. (2018). Nederland in recessie, vooruitzichten 2012 afhankelijk van Europese crisis | CPB.nl. [online] Available at: https://www.cpb.nl/persbericht/3211182/nederland-recessie-vooruitzichten-2012-afhankelijk-van-europese-crisis [Accessed 4 Apr. 2018].

Dohmen, T., Huffman, D., Schupp, J., Falk, A., Sunde, U., Wagner, G.G., 2011. Individuals risk attitudes: measurement, determinants, and behavioral consequences. Journal of the European Economic Association, v9(3), 522-550.

Ershov, E., 2010. The global financial crisis, one year later. Problems of Economic Transition, v53(7), 37-60.

Fernando, C.S., May, A.D., Megginson, W.L., 2012. The value of investment banking relationships: Evidence from the collapse of Lehman Brothers. The Journal of Finance, v67(1), 235-270.

Financieel.InfoNu.nl (2018). Waardoor is de economische crisis eigenlijk ontstaan? [online] Available at: https://financieel.infonu.nl/geld/97560-waardoor-is-de-economische-crisis-eigenlijk-ontstaan.html [Accessed 2 Feb. 2018].

Financieel.InfoNu.nl (2018). Zwarte maandag: dieptepunt kredietcrisis 29 september 2008. [online] Available at: https://financieel.infonu.nl/diversen/25316-zwarte-maandag-dieptepunt-kredietcrisis-29-september-2008.html [Accessed 30 Mar. 2018].

Gerrans, P., Faff, R., Hartnett, N., 2015. Individual financial risk tolerance and the global financial crisis. Accounting & Finance, v55(1), 165-185.

Gervais, S., Odean, T., 2001. Learning to be overconfident. The Review of Financial Studies, v14(1) 1-27.

Gibson, R., Michayluk, D., Van de Venter, G., 2013. Financial risk tolerance: An analysis of unexplored factors. Financial Services Review, v22, 23-50.

(39)

39 Grable, J.E., 2000. Financial risk tolerance and additional factors that affect risk taking in everyday money matters. Journal of Business and Psychology, v14(4), 625-626.

Grable, J.E., Roszkowski, M.J., 2008. The influence of mood on the willingness to take financial risks. Journal of Risk Research, v11(7), 905-923

Guillemette, M., Nanigian, D., 2014. What determines risk tolerance? Financial Services Review, v23(3), 207-218.

Hoffmann, A.O.I., Post, T., Pennings, J.M.E., 2013. Individual investor perceptions and behavior during the financial crisis. Journal of Banking & Finance, v37, 60-74.

Hoffmann, A.O.I., Post, T., Pennings, J.M.E., 2015. How investor perceptions drive actual trading and risk-taking behavior. Journal of Behavioral Finance, v16(1), 94-103.

Kahneman, D., Tversky, A., 1972. Subjective probability: a judgment of representativeness. Cognitive Psychology, v3(3), 430-454.

Koestner, M., Loos, B., Meyer, S., Hackethal, A., 2017. Do individual investors learn from their mistakes? Journal of Business Economics, v87(5), 669-703.

Kuhnen, C.M., Knutson, B., 2011. The influence of affect on beliefs, preferences and financial decisions. Journal of financial and quantitative analysis, v46(3), 606-626.

Lin, H.W., 2011. Does the disposition effect exhibit during financial crisis? International Conference on Economics and Finance Research, v4, 6-10.

Loewenstein, G.F., Weber, E.U., Hsee, C.K., Welch, N., 2001. Risk as feelings. Psychological bulletin, v127(2), 267-286.

Malmendier, U., Nagel, S., 2011. Depression babies: Do macroeconomic experiences affect risk-taking? Quarterly Journal of Economics, v126, 373-416.

Markowitz, H., 1952. Portfolio selection. Journal of Finance, v7, 77-91.

(40)

40 Necker, S., Ziegelmeyer, M., 2016. Household risk taking after the financial crisis. The Quarterly Review of Economics and Finance, v59, 141-160.

Nicolosi, G., Peng, L., Zhu, N., 2009. Do individual investors learn from their trading experience? Journal of Financial Markets, v12(2), 317-336.

Opendata.cbs.nl. (2018). CBS Statline. [online] Available at: https://opendata.cbs.nl/statline/#/CBS/nl/dataset/83686NED/table [Accessed 15 May 2018]. Rooij, van, M., Lusardi, A., Alessie, R., 2011. Financial literacy and stock market participation. Journal of Financial Economics, v101(2), 449-472.

Seru, A., Shumway, T., Stoffman, N., 2010. Learning by trading. The Review of Financial Studies, v23(2), 705-739.

Sivaramakrishnan, S., Srivastava, M., Rastogi, A., 2017. Attitudinal factors, financial literacy, and stock market participation. International Journal of Bank Marketing, v35(5), 818-841. Thaler, R.H., Johnson, E.J., 1990. Gambling with the house money and trying to break even: the effects of prior outcomes on risky choice. Management Science, v36(6), 643-660.

Venter, van de, G., Michayluk, D., Davey, G., 2012. A longitudinal study of financial risk tolerance. Journal of Economic Psychology, v33, 794-800.

Weber, E.U., Milliman, R., 1997. Perceived risk attitudes: Relating risk perception to risky choice. Management Science, v43(2), 123-144.