1
Personality and financial behavior: What determines the
allocation of savings and the demand for debt?
Willem Boorsma1, supervised by Dr. A. Plantinga2 S2550113
University of Groningen, The Netherlands
Abstract
This paper studies whether personality characteristics affect the allocation of savings and the demand for debt. Using a sample of representative Dutch individuals I find that extravert individuals tend to hold less savings accounts, individuals that are more open to experience tend to hold more savings accounts and finally, more agreeable individuals hold less debt accounts.
Keywords: , big five personality traits, behavioral finance, household finance, household debt, household savings
JEL codes: C23, D14, D91
1 Correspondence to: w.boorsma.1@student.rug.nl
2
1. Introduction
Personality traits are “the individual’s styles of thought, feelings and behavior” (Costa and McCrae, 1986). This definition is still used in more recent literature (Borghans et al., 2008). Furthermore, personality explains the different motives one can have to behave in a certain way (Nyhus and Pons, 2005). Based on these sources it is very likely that personality characteristics also determine an individual’s financial behavior.
In this paper I focus on differences in personality traits between individuals and the effect of these differences on their financial behavior. To determine the personality of an individual I will use the Big Five personality measure, which is the most prominent personality measure that is available (Borghans et al., 2008). Originating from psychology research, the Big Five has been applied more and more in economic research over the last decades (Nyhus and Webley, 2001; Groves, 2005; Mueller and Plug, 2006; Donnelly et al., 2012; Pinjisakikool, 2017). This measure consists of 5 personality characteristics for which a score is based on multiple questions concerning personal behavior. In this study I will construct a score at the individual level for each of the big five personality traits. To do this I make use of survey questions concerning Dutch individuals3. The use of survey questions in economic research is a relatively young field of research (Alessie and Kapteyn, 2001). Rather than using data on actual behavior of subjects they illustrate that survey questions are very useful.
My paper takes a different approach compared to previous literature by investigating the number of savings or debt accounts an individual possesses rather than the amount of money stored on these accounts (Friedman, 1957; Brown and Taylor, 2014). I assume that opening or closing an account takes more effort than only transferring money from one account to another and therefore the number of accounts is an interesting measure to indicate saving and borrowing. Whereas transfers between financial categories describe financial needs in the short-run, opening a new savings account or getting involved with additional creditors is more affected by long-run planning. This strategy will also help to tackle the problem of determining what saving exactly is (Nyhus and Webley, 2001), because it ignores small changes within accounts. This new strategy to examine an individual’s financial behavior therefore is an interesting addition to the existing literature.
In this study I will take a broad view and combine the major financial products. The total group of financial products will be divided into four different categories, namely liquid savings, debts, investment savings and insurance savings. Apart from the personality traits I also incorporate information about income and personal characteristics in the analysis.
Apart from personality traits there are more factors affecting financial behavior. Firstly, it is generally known that adding more stocks to a portfolio decreases the total risk, however if someone owns more financial products , managing the portfolio will become more difficult. Although risk for holding savings accounts is relatively low, risk is involved when determining
3 investment savings. Secondly, another benefit of holding multiple accounts is that these can be used for specific expenses or savings purposes. Thaler (1985) refers to this phenomenon of labelling as mental accounting. Individuals that apply this are expected to hold more financial accounts. I will address this principle later in this thesis. Due to this combination between positive and negative aspects of holding a specific number of financial instruments there might exist an ideal number of financial accounts one should hold depending on personal preferences and other socio-economic variables.
As being outlined in the paragraph above, I expect a relationship between personality and the number of financial accounts and products an individual prefers to hold. Therefore it is interesting to see the effect of each personality trait across a large sample of individuals. Understanding drivers for financial behavior based on personality traits can be very useful for different parties. For the government this information is valuable to support individuals easily tempted to open new accounts or take on more debt. An interesting example is the campaign “Borrowing money is costly” of the Dutch authority for financial markets4. This campaign discourages Dutch consumers to take on debts with firms when buying goods. Also for savings, a government campaign could intervene to prevent consumer from losing insight in their financials and discourage opening more savings accounts. On the other hand commercial parties could exploit opportunities by designing advertisements that are suitable for the most vulnerable groups of customers.
This thesis will have the following structure, first I will provide more insight in the big five personality scale that I use extensively in this thesis. After that, I will provide an overview of the literature concerning the topic of personality and financial decision making. Then, data and methodology, results and conclusion will follow.
2. Literature
2.1 The big five personality scale
The big five personality scale historically consists of the factors Extraversion (or Surgency), Agreeableness, Conscientiousness (or Dependability), Emotional Stability (vs. Neuroticism) and Openness to experience, where the last factor previously has been described as culture or autonomy (Goldberg, 1990). For history in detail about the big five personality scale one should consult John and Srivastava (1999). In this chapter I shortly outline each of the personality factors and describe the expected effect of the specific trait on the individual’s financial behavior.
Extraversion
An Extravert individual is a person that is more open to other people and likes to be in the center of attention. They are more likely to be in contact with others which makes it likely that their
4 knowledge of consumption possibilities is larger than it is for more introvert people. Furthermore, an extravert cares more about his external image and is therefore more sensitive to consuming new products (Tsao and Chang, 2010). Following from this more-consuming behavior it is likely that extravert individuals save less which is in line with existing literature (Nyhus and Webley, 2001; Brown and Taylor, 2014; Hirsh, 2015). Apart from consumption goods, extravert individuals are also able to gain more knowledge about financial products compared to introverts. Therefore, the overall effect of extraversion on the number of savings accounts is ambiguous. On the other hand, I hypothesize that the effect of extraversion on holding debt is positive. Firstly, debt is needed to finance the extravert’s lifestyle and secondly, possibilities to obtain debt are more known to an extravert. This is in line with Brown and Taylor (2014) who find this result for both couples as singles by making using of data from the British Historical Statistics Project (BHSP). Furthermore, they find extraversion to be negatively related to the amount of money invested in financial assets. However, I follow the same reasoning as with the number of savings accounts and find the expected effect of extraversion on the number of investment savings to be ambiguous.
Agreeableness
People that are agreeable care a lot about the well-being and feelings of other people (Costa and McCrae 1986). Due to these characteristics agreeable people consider money to be less important which leads to economic hardship (Matz and Gladstone, 2018). However, Nyhus and Webley (2001) find that agreeableness is positively related to savings by using data from the DHS. Therefore the relationship between agreeableness and the number of liquid savings is ambiguous. Being indebted can create a feeling of guilt which has a larger impact for more agreeable individuals. Therefore, I hypothesize that more agreeable individuals hold fewer debt relative to less agreeable individuals. Bucciol & Zarri (2015) find that lower agreeableness predicts higher willingness to take on risk by investigating 7 years of the US health and Retirement study. I expect that investment savings are more risky whereas insurance savings are less risky. Therefore I hypothesize that more agreeableness predicts a lower number of investment savings and a higher number of insurance savings.
Conscientiousness
5 more conscientious individuals are less likely to have ever been in debt. Brown and Taylor (2014) study the relationship between conscientiousness and financial activity and find conscientiousness to be negatively related to both holding debt as financial assets. Donnely et al. (2012) find a positive relationship between conscientiousness and money management. To measure money management they study the responses of 936 visitors of yourmorals.org on six questions concerning financial responsibility. This suggests that more conscientious individuals behave more in line with their own preferences.
Emotional stability
Emotionally stable individuals are relaxed, more secure and independent relative to more emotionally unstable people (Costa and McCrae, 1986). There exists a relation between emotional stability and the number of savings accounts (Nyhus and Webley, 2001). Individuals that are emotional stable tend to save more than emotional unstable people. One explanation for this is that emotional stable individuals can easier control their emotions and purchases (Tsao and Chang, 2010). The relationship between emotional stability and debts, investment savings and insurance savings is more unclear (Brown and Taylor, 2015; Nyhus and Webley, 2001). Therefore I expect the effect of emotional stability on these variables to be ambiguous.
Openness
The variable openness, or often referred to as openness to experience, measures the degree to which an individual is interested in new ideas and thinks about taking new initiatives. Furthermore openness is related to someone who pursues new activities (Costa and McCrae, 1986; Nyhus and Webley. 2001; Barrick and Mount., 1991). Someone who is open to experience may be willing to try more financial products than someone who is less open to this. Also, openness is related to impulsive consumption (Shahjehan et al., 2012) which could have a stimulating effect on the number of all types of financial products. Therefore, I hypothesize that openness has a positive effect on the number of all financial products included in this research.
2.2 Control variables Age and personality
6 individual will influence the number of accounts one holds and therefore I include this in the analysis.
Gender and personality
Furthermore, literature suggests that gender does play a role in the determination of a person’s personality. Osborne (2000) argues that there exists a clear effect of earnings due to personality but that this is distinct between men and women. She finds that there exist differences between men and women for the effect each personality trait has on earnings. This also might be the case in the fields of saving, borrowing and investing. Weisberg et al. (2011) find that female respondents score significantly higher on the personality trait agreeableness in a Canadian study performed both online as in the laboratory. Semykina and Linz (2007) also find differences between male and female respondents in a Russian survey, but emphasize the difference of locus of control. Where men have more an internal locus of control, this is external for women. Because of the existence of differences between male and female respondents I include gender into the analysis.
Risk appetite and financial behavior
Due to the European deposit guarantee scheme for savings accounts5, I expect that risk appetite only has a small effect on the number of savings accounts for individuals with wealth levels lower than the maximum guaranteed amount of money of 100,000 euro. If the balance on a specific account exceeds this amount there exists an incentive to open an additional account, because the maximum guaranteed amount of money holds for every separate savings account. By applying this strategy one is able to safely secure his savings. People with a higher appetite for risk may possess less savings accounts due to the idea that they prefer more risky instruments and therefore have less savings accounts. On the other hand, the opposite is expected for holding debt. Brown et al. (2013) find that the appetite for risk has a positive effect on the amount of debt. Although they use risk aversion as an instrument instead of risk appetite, the implications are identical. Therefore, I hypothesize that an individual with high appetite towards risk has a larger number of debt accounts. Furthermore, I assume that the number of investment savings is positively related to appetite for risk. Alternatively, I expect the number of insurance savings accounts to be negatively related to risk appetite for obvious reasons.
Financial literacy and financial behavior
In the existing literature there is some attention on the relation between financial literacy and saving behavior of households. A higher level of financial literacy increases the likeliness of a household to plan for retirement (Van Rooij et al., 2011). Therefore, more financial literate
5
7 individuals are expected to save more. Perry (2008) finds that households that are not financially literate do not make optimal financial decisions. By applying this knowledge on saving, borrowing and investing I hypothesize that financially illiterate households are not able to manage their money in the most optimal way. Therefore I expect that financial illiteracy affects the relationship between personality traits and the number of savings products and debts.
3. Data and Methodology Data collection
Data for this research has been obtained from the Dutch Household Survey (DHS). Starting in 1993, the survey has collected data on more than 2,000 households covering many different topics such as work, pensions, economic and psychological concepts and personal characteristics.
The DHS collects data on the number of savings instruments and the amount allocated to each instrument. I collect this information for the survey waves of 2005, 2009, 2013-2015 and 2017 since these waves include a larger number of personality questions. These waves include for each personality trait 10 questions, whereas only 2 questions on each personality trait are available for other years. The DHS separates the collected data into different modules. For this research I use the modules containing data about the household, accommodation, income and psychological concepts. To combine the data I match all individual respondents for each year by creating the variable respondent. This variable been calculated as
𝑅 = 𝐻 + 𝑁/10 (1)
where R represents the individual respondent, H the number for the household the individual belongs to and N the number for the individual within the household. After merging the different modules of the DHS for each specific year, I append all different years leading to a dataset of 23,070 respondents of which 13,684 answered the 50 personality trait questions.
Construction variables explaining activity in financial products
The findings of Nyhus and Webley (2001) provide the insight that it is important to divide savings into saving categories rather than treating them as identical. They suggests that personality traits do affect financial behavior but that the effect differs over different categories of products. This approach has been copied by Brown and Taylor (2014) for financial assets and debts. For the construction of the dependent variables I follow the approach of Nyhus and Webley (2001) and group asset components that are similar. First of all I create a variable that represents the number of liquid savings accounts which I calculate as
𝐿𝑖𝑡 = 𝐶𝑖𝑡+ 𝑆𝑖𝑡 (2)
for each respondent i in year t. Liquid savings equals the number of checking accounts Cit
together with the number of saving accounts Sit.
8 no respondents with a significant deficit on their checking accounts. Therefore, I will treat all cases in which a respondent has a deficit on his checking account identical to the individuals that have a positive balance. I exclude saving certificates because less than 1% of the sample does own these and the number of saving certificates has decreased significantly in the survey period 2005-2018.The number of liquid savings accounts a respondent holds will be used as the first dependent variable in the analysis provided later in this paper. In case the number of checking accounts or the number of savings accounts is missing I create a missing value for Liquid savings. I create 12,871 values for Liquid savings in total with a mean value for the individual of 2.78.
The number of all debt contracts an individual holdsis the sum of the number of debts of each category outstanding. I calculate the number of debts as
𝐷𝑖𝑡 = 𝑃𝑖𝑡+ 𝐸𝑖𝑡+ 𝐻𝑖𝑡 + 𝐺𝑖𝑡+ 𝐹𝑖𝑡 + 𝐽𝑖𝑡+ 𝑂𝑖𝑡 (3) where Dit is the number of debts of individual i in year t. This includes 8 different types of debts
which can be found in table A2 of the appendix. In case one of the variables misses I create a missing value for the variable debts. In the literature exists some discussion whether mortgages should be included in the number of debts or not. According to Nyhus and Webley (2001), mortgages are relatively noisy and therefore should be excluded from debt. However, I include them since the number of mortgages an individual has on his personal accommodation is a determinant of his degree of financial activity. In total, I create 3,500 observations for Dit with
a mean number of debts of 3.22.
I calculate the variable for investment savings as
𝐼𝑖𝑡 = 𝐵𝑖𝑡+ 𝑆𝑖𝑡 + 𝑀𝑖𝑡 (4)
where Bit represents the number of bonds , Sit the the number of stocks and Mit the number of
mutual funds. I exclude options on financial instruments in this variable, because of lack of data availability. If there exists a missing value for one of the variables that are used for construction I make the assumption that it equals zero. In total, I create 23,070 observations for investment savings with a mean number of 0.32.
I calculate the variable insurance savings as
𝐾𝑖𝑡 = 𝐴𝑖𝑡+ 𝐸𝑖𝑡 (5)
where Ait equals the number of single-premium insurances or annuity insurances and Eit
represents number of endowment insurance policies. If information is missing on the ownership of these products, I assume that the respondent does not own this product. This is in line with the construction of the variable for investment savings.In total, I create a number of 23,070 values for insurance savings with a mean of 0.16.
9 skewed. Due to the inclusion of mortgages in the construction of the variable debts the mean and median are relatively large compared to the savings products.
Table 1: Summary statistics of the dependent variables
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. The table illustrates the descriptive statistics of the dependent variables in this research.
N Mean St.Dev min max p5 p25 Median p75 p99
Liquid savings 12,871 2.78 1.86 0 32 0 2 2 4 8
Debts 3,500 3.22 1.13 0 9 1 3 3 4 6
Investment savings 23,070 .32 1.59 0 56 0 0 0 0 7
Insurance savings 23,070 .16 .61 0 12 0 0 0 0 3
Construction of the big five personality traits
To construct a value for each individual concerning each personality trait, 50 questions about personality of the DHS will be used. The questions that I use to create a value can be found in table A1 in the appendix. The values range from 1, “not at all applicable”, to 5, “highly applicable”. Furthermore, the effect of each question on the personality trait can also be found in table A1 in the appendix. In this table a minus sign after the question represents that it has a negative effect on the personality trait and a positive effect otherwise. In each case where there is a minus sign attached to the statement, the scale of each score has been reversed. This means that for questions that have a negative effect on the specific personality trait, the answers “not at all applicable” are assigned a value of 5 whereas “highly applicable” is assigned a value of 1.To obtain a complete score I aggregate all scores which results in a 10 to 50 scale for each personality trait. Summary statistics can be found in table 2.
Table 2: Summary statistics of the personality traits
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. The table illustrates the descriptive statistics of the big five personality traits for each year of the DHS.
Total 2005 2009 2013 2014 2015 2017
mean mean mean mean mean mean mean
10 There are no large deviations between the mean value for a personality trait for a specific year and the overall sample mean score. One can observe that the level of agreeableness is relatively high compared to the other personality traits. On the other hand, the level of extraversion is low compared to the other personality traits. The order in which the personality questions are being asked has been adjusted in 2015, but it is unclear whether this has resulted in different responses. More editions of the DHS have to be performed to get a better idea about the impact of this change. Comparing the statistics with Thelken (2017), there are no large differences in means or standard deviations between both datasets when taking into account the different scales used in both studies.
To test for a potential risk due to multicollinearity I calculate the correlations between the personality traits. Results are presented in table 3. The correlation coefficients are not problematically high so I expect no trouble. Furthermore, there are only little missing values in the DHS psychology module. Therefore, the individuals that have failed to answer one of the personality questions are being dropped from the analysis.
Table 3: Correlation coefficients of the independent variables
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. The table illustrates the correlation coefficients between the big five personality traits and the independent variables Risky and Mental accounting. Variables (1) (2) (3) (4) (5) (6) (7) (1)Extraversion 1.00 (2)Agreeableness 0.34 1.00 (3)Conscientiousness 0.10 0.29 1.00 (4)Emotional stability 0.21 0.22 0.25 1.00 (5)Openness 0.33 0.30 0.15 0.16 1.00 (6)Risky 0.04 -0.17 -0.12 -0.04 0.03 1.00 (7)Mental accounting 0.06 0.02 0.01 -0.04 0.04 0.00 1.00
In order to test the consistency of each measurement constructed, I follow the standard approach used in previous research (Nyhus & Webley, 2001; Brown and Taylor, 2014) and calculate Cronbach’s Alpha for each of the five personality traits. As can be found in table 4 the Alpha for each personality trait in each specific year is well above the benchmark of 0.6. This implies that I can use the created values as a reliable measure for each personality trait.
Table 4: Overview of Crohnbach’s Alpha for the big five personality traits
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. The table illustrates the Crohnbach’s Alpha scores for the big five personality traits.
2005 2009 2013 2014 2015 2017
Variables Alpha Alpha Alpha Alpha Alpha Alpha
Extraversion 0.86 0.85 0.86 0.87 0.84 0.85
Agreeableness 0.82 0.84 0.84 0.85 0.85 0.85
Conscientiousness 0.78 0.80 0.79 0.79 0.76 0.76
Emotional stability 0.85 0.85 0.86 0.87 0.87 0.87
Openness 0.75 0.77 0.78 0.79 0.76 0.76
11 Bowles et al. (2001) address the potential problem of endogeneity when concerning personality traits in a labor-market setting. This study provides the insight that personality traits are both causes as determinants of success on the labor market. However, where an individual’s can only affect success on the labor market in a limited way, the number of accounts and products someone holds is more a decision based on personal preferences. Therefore I make the assumption that the problem of endogeneity is much smaller when our focus is on financial products rather than on earnings.
In Table 5 I outline the relationship between personality characteristics and the number of financial products owned by the respondent. It plots the average score on each personality trait for four different groups of respondents. The first group of respondents does not own any products in a particular class, and individuals in the other groups own an increasing number of products. It is interesting to observe that there are 966 individuals that do not own any liquid savings products. Furthermore, almost all respondents do possess at least 1 debt account where the majority of respondents owns more than 4 debt products. In most cases there is no clear pattern between the number of financial products and the average score for each personality trait. However, for liquid savings there seems to be a positive relationship between the value for openness to experience and the number of accounts. For debts there is a pattern between the number of debts and the personality traits emotional stability and openness. When addressing investment savings there is a negative pattern between the number of products and the personality trait extraversion. On the other hand, for emotional stability and openness there seems to exist a positive pattern whereas there is no clear pattern for conscientiousness and agreeableness. Lastly, a positive pattern between the personality trait openness can be observed for the number of insurance savings products.
Constructing control variables
12 indicates whether an individual uses a separate bank account for different spending purposes. I use this as a measure of self-perceived mental accounting.
Summary statistics of the control variables can be found in table 6. Table 7 compares summary statistics of individuals that do apply mental accounting and of individuals that do not apply this. From the table one can find that the number of all financial accounts is higher for individuals that apply mental accounting which is intuitive. To verify this observation I perform t-tests for all financial categories and find significant results at the 10% level for debts and at the 1% level for liquid savings, investment savings and insurance savings. Results of this test are presented in Table A4 of the appendix.
Table 5: Mean values personality traits depending on the number of products
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. The table illustrates the mean values for the big five personality traits for a specific number of liquid savings, debt, investment savings and insurance savings accounts. The number of individuals holding no debt is too small to provide meaningful statistics for.
Number of products
Liquid savings 0 1&2 3&4 >4
Extraversion 30.56 30.79 30.70 30.57 Agreeableness 38.60 38.92 39.05 38.95 Conscientiousness 36.44 36.24 36.62 36.09 Emotional stability 34.57 35.32 35.67 35.41 Openness 31.91 33.84 34.58 34.58 Observations 966 4,918 4,215 3,585 Debts Extraversion 30.62 30.57 30.72 Agreeableness 38.75 38.67 39.02 Conscientiousness 35.97 36.57 36.29 Emotional stability 36.67 36.66 35.05 Openness 34.71 34.66 33.98 Observations 499 2,433 10,752 Investment savings Extraversion 30.75 30.39 30.23 30.12 Agreeableness 39.04 38.38 38.65 38.21 Conscientiousness 36.30 36.46 36.59 36.51 Emotional stability 35.23 36.11 36.35 36.89 Openness 34.05 34.55 34.25 34.82 Observations 11,591 1334 353 406 Insurance savings Extraversion 30.69 30.56 30.80 33.24 Agreeableness 38.96 38.81 39.27 39.18 Conscientiousness 36.26 36.71 36.44 36.64 Emotional stability 35.27 36.07 35.99 37.55 Openness 34.03 34.57 34.98 35.75 Observations 11,571 1,861 185 67 Financial literacy
13 questions concerning the financial literacy of respondents of the DHS. Therefore it is very useful to add this data to our existing dataset. Limitation of this module is that it only contains data about 2005 and therefore cannot be used in a panel data study. To test for the effect of financial literacy a separate OLS regression will be performed on the data. I will make use of the self-reported financial literacy of the respondent which is scaled from 1, very little knowledge, to 7, a lot of financial knowledge. Using self-reported knowledge rather than financial literacy based on financial questions is in line with findings of van Rooij et al. (2011). Summary statistics for financial literacy are presented in table 6.
Table 6: Summary statistics independent variables
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. The table illustrates the descriptive statistics of the independent variables excluding the personality traits.
Variable Observations Mean Std.Dev. Min Max
Education 22,981 4.68 1.55 0 7 Gender 23,049 .49 .5 0 1 Age 23,038 49.75 17.07 18 97 Mental accounting 23,070 .19 .39 0 1 Children 23,070 .43 .50 0 1 Income 9,519 23,998 19,930 -4,800 623,000 Partner 23,049 .82 .38 0 1 Risky 12,853 16.53 6.09 6 41 Financial literacy 1,467 4.73 1.16 1 7
Table 7: Summary statistics of dependent variables depending on mental accounting
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. The table illustrates the descriptive statistics of the dependent variables for individuals that do apply mental accounting in panel 1 and individuals that do not apply this in table 2.
Variable Observations Mean Std.Dev. Min Max
Mental Accounting Liquid Savings 3518 3.13 1.99 0 22 Debts 1014 3.27 1.17 0 9 Investment Savings 4269 .38 1.54 0 27 Insurance savings 4269 .25 .74 0 10 No mental accounting Liquid Savings 9353 2.65 1.79 0 32 Debts 2486 3.19 1.11 0 9 Investment Savings 18,801 .30 1.60 0 56 Insurance savings 18,801 .14 .59 0 12
Construction of appetite for risk
14 address the risk appetite of an individual (Kapteyn and Teppa, 2011; Pinjisakikool, 2017). Summary statistics of the variable are provided in table 6. To test the reliability of this index I calculate Crohnbach’s reliability alpha that has a value of 0.66 which is just higher than the lowest acceptable value of 0.6. Later in this paper I will test whether individuals with a low appetite for risk hold more financial products. In table 8 I provide an overview of the number of financial products an individual holds for a specific risk appetite. From the table one can observe that the number of savings accounts is relatively stable if the appetite for risk is low or medium and starts to decrease after a certain level of risk appetite. By looking at the number of debt accounts there is no clear pattern. For investment savings the pattern is predictable and increasing if the risk appetite increases from low to medium. This can be explained by a large group of individuals with low risk appetite that do not hold any investment savings at all. The number of products decreases if the value for risk appetite increases from medium to high. An explanation for this phenomenon is the principle of diversification that assumes that having a larger number of products decreases the total amount of risk. Since individuals with a large appetite for risk do not value this lower risk, they may hold less investment savings. Lastly, the number of insurance savings accounts decreases if risk appetite increases from medium to high. Interestingly, the number of insurance savings accounts is higher for individuals with a medium appetite for risk than for low.
Table 8: Mean numbers of financial products relative to risk appetite
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. The table illustrates for each category the mean number of financial products depending on risk appetite.
Variable 0-15 16-25 26+ Liquid Savings 2.81 2.84 2.61 Debts 3.18 3.25 3.22 Investment Savings .21 .60 .21 Insurance savings .20 .29 .07 Standardization
Since I am working with survey data with a lot of different scales it is useful to standardize the variables. Therefore, all variables in this research are standardized values apart from the binary variables mental, children, gender and partner. As a result of standardization the mean of these variables equals 0 with a standard deviation of 1.
4. Results
Analyzing the four categories of financial products
15 By applying the Breusch-Pagan Lagrangian multiplier test I find evidence of significant differences between individuals. The implication of this test is that the random effects model is suitable in this case. Results of the Hausman test and the Lagrangian multiplier test can be found in table A6 and A7 of the appendix. Findings from the random effects model are presented in table 9. For completeness, the results of a pooled OLS model and a fixed effects model are included in the appendix and can be found in table A8 and A9. As a precautionary measure I include robust standard errors in the regression to deal with a potential problem due to heteroscedasticity.
First of all, I will address the question whether the number of liquid savings accounts is related to the personality traits and risk appetite by applying the model
𝐿𝑖𝑡 = 𝛼 + 𝛽1 ∗ 𝐸𝑥𝑖𝑡 + 𝛽2 ∗ 𝐴𝑔𝑖𝑡 + 𝛽3 ∗ 𝐶𝑜𝑖𝑡 + 𝛽4 ∗ 𝐸𝑚𝑖𝑡 + 𝛽5 ∗ 𝑂𝑖𝑡 + 𝛽6 ∗ 𝑅𝑖 + 𝛽7 ∗ 𝐶𝑉𝑖𝑡 (6) where Exit is extraversion, Agit is agreeableness, Coit is conscientiousness , Emit is emotional
stability Oit is the variable for openness, Riit indicates risk appetite and finally CV represents
the set of control variables. All variables have respect on individual i in year t. The results are presented in the first column of table 9. The coefficient for extraversion is negative and highly significant which is in line with our hypothesis. Furthermore the coefficient for openness is positive which indicates that individuals who are more open to experience have a higher number of savings accounts. This is in line with the hypothesis that openness to experience leads to a larger diversity of savings products and therefore to holding a larger quantity of products. From the control variables I find that the coefficients for education, gender, age, mental accounting, income and partner are all positive and highly significant. I also test separately for interaction effects between the personality trait agreeableness and the binary variables children and partner, however this does not lead to anything worth mentioning. Lastly, in line with the hypothesis there is a small negative but not significant relationship between risk appetite and the number of liquid savings accounts.
After the number of liquid savings accounts I focus on holding debt. To test the relationship between the personality traits and the number of debts I will use
𝐷𝑖𝑡 = 𝛼 + 𝛽1 ∗ 𝐸𝑥𝑖𝑡 + 𝛽2 ∗ 𝐴𝑔𝑖𝑡 + 𝛽3 ∗ 𝐶𝑜𝑖𝑡+ 𝛽4 ∗ 𝐸𝑚𝑖𝑡 + 𝛽5 ∗ 𝑂𝑖𝑡+ 𝛽6 ∗ 𝑅𝑖 + 𝛽7 ∗ 𝐶𝑉𝑖𝑡 (7) where the variable names are identical to the ones used to explain liquid savings. The results from this regression are presented in column two of table 9. The coefficient for agreeableness is negative and significant. This is in line with the hypothesis that more agreeable individuals have less creditors. An explanation for this phenomenon is that agreeable people dislike the fact of being indebted because they care about the feelings of others towards themselves. Also, the coefficient for risky is positive and highly significant. This implies that being more open to risk is positively related to holding a larger number of debt accounts. Furthermore, there is evidence that male respondents hold less debt compared to female respondents.
16
Table 9: Estimates of a Random effects model
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. This table shows the estimates of a random effects model on the number of financial products in each category with robust standard errors. The values for the variables Extraversion, Agreeableness, Conscientiousness, Emotional Stability, Openness, Education, Risky, Age, and Income are standardized.
(1) (2) (3) (4)
VARIABLES Liquid savings Debts Investment savings insurance savings
Extraversion -0.072*** 0.015 -0.035 0.009 (0.024) (0.031) (0.024) (0.011) Agreeableness 0.007 -0.079** 0.025 0.009 (0.025) (0.033) (0.029) (0.013) Conscientiousness 0.007 0.011 0.027 -0.003 (0.022) (0.030) (0.026) (0.010) Emotional stability 0.037 0.029 0.023 -0.013 (0.028) (0.031) (0.026) (0.012) Openness 0.046* -0.017 0.010 0.008 (0.025) (0.034) (0.031) (0.013) Education 0.340*** 0.010 0.157*** 0.030** (0.032) (0.029) (0.039) (0.013) Risky -0.011 0.081*** 0.272*** 0.017* (0.021) (0.028) (0.031) (0.010) Gender 0.235*** -0.194*** 0.323*** 0.178*** (0.062) (0.070) (0.058) (0.027) Age 0.137*** -0.039 0.215*** -0.047*** (0.031) (0.037) (0.038) (0.012) Mental accounting 0.293*** 0.024 -0.051 0.005 (0.041) (0.056) (0.035) (0.018) Children 0.057 0.287*** -0.156*** 0.060* (0.060) (0.074) (0.054) (0.032) Income 0.123*** 0.049** 0.132** 0.035*** (0.030) (0.020) (0.054) (0.013) Partner 0.169** 0.075 -0.003 0.000 (0.072) (0.079) (0.082) (0.026) Constant 2.430*** 3.205*** 0.421*** 0.176*** (0.064) (0.069) (0.063) (0.023) Observations 7,543 2,387 8,079 8,079 Number of respondent 3,430 1,234 3,710 3,710
Standard errors are provided in parentheses and significance levels are given by *** p<0.01, ** p<0.05, * p<0.1.
The number of investments savings account will be addressed in a similar fashion as the number of savings and debt accounts. Therefore I use the model
17 Finally, I look at the number of insurance savings products of the Dutch individuals and use a similar equation as before:
𝐾𝑖𝑡 = 𝛼 + 𝛽1 ∗ 𝐸𝑥𝑖𝑡 + 𝛽2 ∗ 𝐴𝑔𝑖𝑡 + 𝛽3 ∗ 𝐶𝑜𝑖𝑡+ 𝛽4 ∗ 𝐸𝑚𝑖𝑡 + 𝛽5 ∗ 𝑂𝑖𝑡+ 𝛽6 ∗ 𝑅𝑖 + 𝛽7 ∗ 𝐶𝑉𝑖𝑡 (9) The results are presented in column 4 of table 9. Just as with investment savings, none of the coefficients of the personality traits is statistically significant. However, the coefficients for education, gender, income, children and risky are all significant and positive. On the other hand, the coefficient for age is highly significant and negative.
Analyzing the Effect of ageing on financial behavior
From column 1 of Table 9 I already observed a highly significant positive coefficient for the variable age. This implies that the age of the respondent is positively related to the number of liquid savings accounts an individual possesses. However, it is interesting to examine the effect of ageing on the individual level. To study this, I make use of a fixed effect model which I will conduct separately on different age groups. This model is appropriate for this specific subject because it is able to explain the effect of a change in the age of the individual on the number of savings accounts at the personal level. I create separate groups of respondents where the spread of each category is 15 years. Results are presented in table 10. The table presents a changing sign for the number of liquid savings accounts when moving from the youngest group towards the older groups. Where the coefficient is positive and significant for the youngest group, it is highly significant and negative for the oldest group.
From column 2 of Table 9 I observed that the coefficient for age is negative and insignificant when testing equation 7 which has Dit as dependent variable. Table 10 provides more insight in
18
Table 10: Estimates of a fixed effects model respective of age
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. This table shows the results of a fixed effects model on the number of liquid savings products with robust standard errors. The value for Age, and Income are standardized. Each column represents the effect of ageing for each specific age group. For brevity I exclude the coefficients of the variables Extraversion Agreeableness Conscientiousness Emotional stability Openness Education Risky Gender Mental accounting Children Income and Partner in this table. The value for Age has been standardized.
VARIABLES 20-34 35-49 50-64 65+ Liquid Savings Age 0.69* 0.34 -0.15 -0.55*** (0.39) (0.38) (0.22) (0.18) Debt Age -3.74*** 1.19** 0.44 1.45*** (1.22) (0.52) (0.37) (0.28) Investment savings Age 0.45** -0.54 -0.55** -0.83*** (0.22) (0.34) (0.24) (0.22) Insurance savings Age 0.17 -0.27** -0.73*** -0.43*** (0.19) (0.13) (0.15) (0.09)
Standard errors are provided in parentheses and significance levels are given by *** p<0.01, ** p<0.05, * p<0.1.
Analyzing the Effect of financial literacy on financial behavior
19
Table 11: Estimates of an OLS model
Data is from the DHS from the year 2005. This table shows the estimates of an OLS model on the number of financial products in each category with robust standard errors. The values for the variables Extraversion, Agreeableness, Conscientiousness, Emotional Stability, Openness, Education, Risky, Age, Income and Financial Literacy are standardized.
(1) (2) (3) (4)
VARIABLES Liquid savings Debts Investment savings insurance savings
Extraversion -0.105* 0.073 -0.034 -0.003 (0.062) (0.049) (0.082) (0.040) Agreeableness -0.076 -0.050 -0.033 0.036 (0.069) (0.059) (0.112) (0.061) Conscientiousness -0.066 0.018 0.057 -0.042 (0.059) (0.064) (0.082) (0.041) Emotional stability 0.009 -0.027 -0.126 0.013 (0.056) (0.056) (0.089) (0.042) Openness 0.046 -0.020 0.193 -0.023 (0.067) (0.061) (0.126) (0.056) Education 0.251*** 0.024 0.205** 0.006 (0.064) (0.052) (0.104) (0.048) Risky -0.023 0.005 0.532*** 0.081* (0.058) (0.058) (0.113) (0.049) Gender 0.078 0.145 0.006 0.314*** (0.133) (0.119) (0.234) (0.095) Age 0.222*** 0.000 0.473*** -0.108** (0.068) (0.064) (0.099) (0.043) Mental accounting 0.442*** 0.170 -0.176 -0.025 (0.131) (0.124) (0.156) (0.079) Children 0.085 0.538*** -0.224 0.115 (0.152) (0.145) (0.211) (0.115) partner 0.390*** -0.207 0.061 0.205** (0.136) (0.135) (0.222) (0.082) Income 0.086 -0.027 0.430 0.061 (0.068) (0.017) (0.350) (0.075) Financial Literacy 0.245*** 0.011 0.468*** 0.066 (0.060) (0.057) (0.114) (0.042) Constant 2.636*** 3.435*** 0.962*** 0.230*** (0.120) (0.098) (0.179) (0.071) Observations 793 356 849 849 R-squared 0.100 0.080 0.156 0.062
Standard errors are provided in parentheses and significance levels are given by *** p<0.01, ** p<0.05, * p<0.1.
5. Conclusion
20 effects model on data from the DHS for the years 2005, 2009, 2013 to 2015 and 2017. In contrast to existing literature, the focus of this study is on the number of financial accounts an individual possesses instead of the amount of money allocated to these accounts. The summary statistics already provide a small indication that there might exist a relationship between the big five personality traits and the number of accounts held of the four categories in our research. In the data section I find evidence for the existence of mental accounting in the sample. By investigating the independent variables other than the big five personality traits I find that individuals mentioning self-perceived mental accounting have significantly more financial accounts than others. This is the case for each category of financial products that were addressed in this research. Furthermore, I find that risk appetite is positively related to the number of accounts for the categories debts, investment savings and insurance savings. There does not exist evidence for a relation between an individual’s the appetite for risk and the number of savings accounts he has. These findings are in line with the hypotheses for debts and investment savings, but contradicted the hypothesis that insurance savings should be negatively related to risk appetite. In addition, chapter 4.2 offers some more insight on the changing effect of ageing over the life-cycle. Where the effect of ageing on liquid savings first is positive and becomes negative over the life-cycle, the opposite is true for the number of creditors. For the categories investment savings and insurance savings there exist positive effects of ageing for the youngest group and negative effects for all other age groups. In chapter 4.3 I find that including financial literacy in the analysis has mixed effects on the significance coefficients for the personality traits extraversion and openness to experience. More research concerning these relations is needed in the future. A limitation of this research is that the data concerning financial literacy is limited to 1 year. Therefore, it is not possible to measure the effects of financial literacy in a longitudinal analysis.
This study adds to the literature by expanding the knowledge of financial behavior by addressing the number of accounts an individual holds. With this strategy it is possible to focus more on long-run determinants of saving. Furthermore, I find a new method to determine what saving and borrowing exactly is. Future research can apply this measure of saving and borrowing to enrich the existing knowledge about financial behavior.
6. References
Alessie, R., Kapteyn, A., 2001. New data for understanding saving. Oxford review of economic policy 17, 55-69.
Barrick, M. R., and Mount, M. K., 1991. The big five personality dimensions and job performance: a meta‐analysis. Personnel psychology, 44, 1-26.
21 Bowles, S., Gintis, H.,Osborne, M., 2001. Incentive-enhancing preferences: personality, behavior, and earnings American Economic Review 91,155-158.
Brown, S., Taylor, K., 2014. Household finances and the ‘Big Five’ personality traits. Journal of Economic Psychology 45, 197–212.
Brown, S., Garino, G., Taylor, K., 2013 Household debt and attitudes toward risk. Review of Income and Wealth 59, 283-304.
Bucciol, A., Zarri, L., 2015. Does Investors' Personality Influence their Portfolio. Unpublished working paper. University of Verona.
Costa Jr, P. T., & McCrae, R. R., 1986. Personality stability and its implications for clinical psychology. Clinical Psychology Review, 6, 407-423.
Donnelly, G. Iyer, R. and Howell, R., 2012. The Big Five personality traits, material values, and financial well-being of self-described money managers. Journal of Economic Psychology 33, 1129–1142.
Friedman, M., 1957. A Theory of the Consumption Function. Princetown University press. Goldberg, L., 1990. An Alternative description of Personality: The Big-Five Factor Structure. Journal of Personality and Social Psychology 59, 1216-1229.
Groves, M., 2005. How important is your personality? Labor market returns to personality for women in the US and UK Journal of Economic Psychology 26, 827-841.
Hirsh, J. B., 2015. Extraverted populations have lower savings rates. Personality and Individual Differences 81, 162-168.
Kapteyn, A., Teppa, F., 2011. Subjective measures of risk aversion, fixed costs, and portfolio choice Journal of Economic Psychology 32, 564–580.
Matz, S. C., & Gladstone, J. J. (2018). Nice guys finish last: When and why agreeableness is associated with economic hardship. Journal of personality and social psychology.
Mueller, G., Plug, E., 2006. Estimating the effect of personality on male and female earnings. Industrial and Labor Relations Review 60, 3-22.
Nyhus, E., Webley, P., 2001. The role of personality in household saving and borrowing behavior. European journal of personality 15, 85-103.
Nyhus E., Pons E., 2005. The effects of personality on earnings. Journal of Economic Psychology 26, 363–384.
Nyhus E., Pons E., 2012. Personality and the gender wage gap. Applied economics 44, 105-118.
22 Rooij, van M. Lusardi, A., Alessie, R. 2011. Financial literacy and stock market participation” Journal of Financial Economics 101, 449–472.
Semykina, A. Linz, S., 2007. Gender differences in personality and earnings: Evidence from Russia. Journal of Economic Psychology 28, 387–410.
Shahjehan, A., Zeb, F., & Saifullah, K., 2012. The effect of personality on impulsive and compulsive buying behaviors. African Journal of Business Management, 6, 2187-2194.
Perry, V. G., 2008. Is ignorance bliss? Consumer accuracy in judgments about credit ratings. Journal of Consumer Affairs, 42, 189-205.
Pinjisakikool, T., 2017. The Influence of Personality Traits on Households’ Financial Risk Tolerance and Financial Behavior. Journal of Interdisciplinary Economics 30, 32–54.
Thaler, R., 2017 Mental accounting and consumer choice. Marketing Science 4, 199-214. Thelken, H., 2017 A longitudinal analysis of the Big Five personality trait influence on individual portfolio choices. Unpublished master thesis. University of Groningen.
Tsao, W. C., & Chang, H. R., 2010. Exploring the impact of personality traits on online shopping behavior. African Journal of Business Management, 4(9), 1800-1812.
25
Table A3: Skewness/Kurtosis tests for Normality on the dependent variables.
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. This table shows the results of the Hausman test on the dependent variables Liquid savings, Debts, Insurance savings and investment savings.
Variable Observations Pr(Skewness) Pr(Kurtosis)
Liquid Savings 14,940 0.00 0.00
Debts 4,207 0.00 0.00
Investment Savings 28,061 0.00 0.00
Insurance savings 28,061 0.00 0.00
Table A4: Results of a comparison of means test for mental accounting
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017.This table presents the results of a t-test for mean comparison.
Variable Obs1 Obs2 Mean1 Mean2 Dif SE T value P value Liquid savings 9,353 3518 2.65 3.13 -.49 .04 -13.35 0.00
Debts 2,486 1014 3.19 3.27 -.08 .04 -1.9 .06
Investment savings 18,807 4269 .30 .38 -.08 .03 -2.95 .00 Insurance savings 18,807 4269 .142 .25 -.10 .01 -10 0.00
Table A5: Overview of the DHS questions used to construct the variable risky that describes the risk
appetite of the individual.
The following statements concern saving and taking risks. Please indicate for each statement to what extent you agree or disagree.
Please indicate on a scale from 1 to 7 to what extent you agree with the statement. 1 means ‘totally disagree’
7 means ‘totally agree’.
Question 1
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
Question 2
I do not invest in shares, because I find this too risky Question 3
If I think an investment will be profitable, I am prepared to borrow money to make this investment Question 4
I want to be certain that my investments are safe Question 5
If I want to improve my financial position, I should take financial risks Question 6
I am prepared to take the risk to lose money, when there is also a chance to gain money
Table A6: Results of the Hausman test
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. This table shows the results of the Hausman test on the dependent variables Liquid savings, Debts, Insurance savings and investment savings.
Liquid savings Debts Investment savings Insurance Savings Chi-square test
coefficient value
103.463 55.488 273,76 291,47
26
Table A7: Results of the Breusch and Pagan Lagrangian multiplier test
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. This table shows the results of the Breusch and Pagan Lagrangian multiplier test on the dependent variables Liquid savings, Debts, Insurance savings and investment savings.
Liquid savings Debts Investment savings Insurance Savings Chi-square test
coefficient value
5280.04 122,63 7432,55 4319,48
P-value 0.00 0.00 0.00 0.00
Table A8: Estimates of a Pooled OLS model
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. This table shows the estimates of a Pooled OLS model on the number of financial products in each category with robust standard errors. The values for the variables Extraversion, Agreeableness, Conscientiousness, Emotional Stability, Openness, Education, Risky, Age, and Income are standardized.
(1) (2) (3) (4)
VARIABLES Liquid savings debts investment saving insurance saving
Extraversion -0.161*** 0.007 -0.087*** -0.005 (0.023) (0.026) (0.023) (0.009) Agreeableness 0.008 -0.079*** 0.051* 0.033*** (0.025) (0.029) (0.028) (0.012) Conscientiousness 0.037* -0.002 0.039 0.004 (0.022) (0.026) (0.024) (0.010) Emotional Stability 0.081*** 0.044 0.065*** -0.003 (0.025) (0.027) (0.024) (0.010) Openness 0.045* 0.001 -0.002 0.001 (0.025) (0.029) (0.030) (0.012) Education 0.327*** 0.009 0.184*** 0.025** (0.027) (0.025) (0.032) (0.012) Risky 0.011 0.053** 0.521*** 0.047*** (0.023) (0.025) (0.034) (0.011) gender 0.176*** -0.140** 0.208*** 0.172*** (0.054) (0.062) (0.048) (0.021) Age 0.152*** -0.078** 0.314*** -0.019** (0.025) (0.031) (0.032) (0.009) Mental Accounting 0.517*** 0.049 -0.135*** -0.010 (0.049) (0.053) (0.045) (0.021) Children 0.041 0.240*** -0.196*** 0.069*** (0.053) (0.063) (0.048) (0.025) Income 0.298*** 0.033 0.216*** 0.066*** (0.060) (0.022) (0.057) (0.016) Partner 0.094** 0.069 -0.051 -0.019 (0.047) (0.064) (0.063) (0.021) Constant 2.503*** 3.177*** 0.575*** 0.213*** (0.046) (0.059) (0.050) (0.019) Observations 7,543 2,387 8,079 8,079 R-squared 0.108 0.032 0.109 0.030
27
Table A9: Estimates of a Fixed effects model
Data is from the DHS from the years 2005, 2009, 2013 to 2015 and 2017. This table shows the estimates of a Fixed effects model on the number of financial products in each category with robust standard errors. The values for the variables Extraversion, Agreeableness, Conscientiousness, Emotional Stability, Openness, Education, Risky, Age, and Income are standardized.
(1) (2) (3) (4)
VARIABLES Liquid savings debts investment saving insurance saving
Extraversion -0.018 0.012 0.059 0.015 (0.035) (0.075) (0.043) (0.021) Agreeableness 0.027 -0.050 0.059 -0.014 (0.035) (0.069) (0.047) (0.019) Conscientiousness -0.025 0.061 -0.006 -0.027 (0.032) (0.066) (0.043) (0.017) Emotional stability 0.022 -0.082 -0.011 -0.008 (0.045) (0.071) (0.047) (0.020) Openness 0.029 -0.105 -0.023 0.012 (0.035) (0.073) (0.046) (0.020) Education -0.026 -0.108 0.189** 0.092* (0.100) (0.241) (0.090) (0.051) Risky -0.035 0.133*** 0.039 0.008 (0.025) (0.051) (0.032) (0.013) Age 0.044 0.880*** -0.667*** -0.603*** (0.093) (0.135) (0.138) (0.064) Mental Accounting 0.172*** 0.048 -0.018 -0.022 (0.049) (0.085) (0.037) (0.021) children 0.087 0.286 -0.160* 0.116* (0.103) (0.201) (0.083) (0.064) Income 0.040*** 0.070*** 0.045 0.008 (0.015) (0.019) (0.065) (0.010) partner 0.224 -0.260 0.179 -0.040 (0.170) (0.219) (0.244) (0.048) Constant 2.665*** 3.042*** 0.733*** 0.468*** (0.134) (0.184) (0.191) (0.042) Observations 7,543 2,387 8,079 8,079 R-squared 0.008 0.056 0.013 0.061 Number of respondents 3,430 1,234 3,710 3,710
28
Table A10: Estimates of an OLS model
Data is from the DHS from the year 2005. This table shows the estimates of an OLS model on the number of financial products in each category with robust standard errors. The values for the variables Extraversion, Agreeableness, Conscientiousness, Emotional Stability, Openness, Education, Risky, Age, and Income are standardized.
(1) (2) (3) (4)
VARIABLES Liquid savings debts investment saving insurance saving
Extraversion -0.080 0.074 0.015 0.004 (0.063) (0.048) (0.084) (0.040) Agreeableness -0.096 -0.051 -0.068 0.031 (0.070) (0.058) (0.116) (0.060) Conscientiousness -0.038 0.019 0.104 -0.035 (0.059) (0.062) (0.083) (0.040) Emotional Stability 0.018 -0.026 -0.104 0.016 (0.057) (0.055) (0.088) (0.042) Openness 0.062 -0.018 0.215* -0.020 (0.067) (0.061) (0.129) (0.055) Education 0.246*** 0.023 0.195* 0.005 (0.065) (0.053) (0.106) (0.048) Risky -0.011 0.006 0.559*** 0.085* (0.058) (0.058) (0.113) (0.050) gender 0.141 0.148 0.129 0.332*** (0.134) (0.118) (0.241) (0.095) Age 0.182*** -0.002 0.405*** -0.117*** (0.068) (0.061) (0.094) (0.042) Mental Accounting 0.412*** 0.169 -0.235 -0.034 (0.132) (0.124) (0.160) (0.081) Children 0.052 0.538*** -0.281 0.107 (0.154) (0.145) (0.213) (0.115) Income 0.461*** -0.205 0.198 0.225*** (0.136) (0.135) (0.227) (0.081) Partner 0.096 -0.027 0.452 0.064 (0.082) (0.017) (0.376) (0.079) Constant 2.597*** 3.435*** 0.883*** 0.219*** (0.121) (0.098) (0.177) (0.070) Observations 793 356 849 849 R-squared 0.083 0.080 0.133 0.059
