Future Orientation and Financial Market
Participation
Natascha de Boer Supervisor: S.S.H. Eriksen
University of Groningen Faculty of Economics and Business
January 2020
Abstract
The purpose of this study is to investigate if future orientation influences the financial behaviour of individuals. Exploiting Dutch household survey data, we test whether future orientation influences the probability of participating in financial markets. In addition, using a Heckman two-stage model, we test if future orientation influences the total market value of investments of individuals. No significant relation is found between future orientation and financial market participation, and future orientation and total market value of investments. These results are robust for gender, household size, living in large cities, and a variable for future orientation from an earlier wave of data.
1. Introduction
Traditional economic theories and models are based on the assumption that people are rational and consistent. However, since the rise of behavioural finance over the past three decades, evidence has shown that the rationality assumption is often not valid, and that people show inconsistent behaviour. Despite low interest rates inducing individuals to increase their appetite for risk, still many choose not to invest in financial markets (Lian, & Ma, 2018). In this paper, we try to find a behavioural determinant, namely future orientation, to explain the non-participation of individuals in financial markets.
Many individuals feel a constant struggle between optimizing their consumption pattern for the future and wanting to spend now. This trade-off is often referred to as ‘time preference.’ The decisions that individuals make regarding this matter can have a significant impact on their lives. For instance, saving enough to maintain living standards after retirement, or abide by a long-term investment strategy, considering investing is often a long-term process that pays off when staying patient and committed during boom-and-bust cycles. Inconsistent or irrational behaviour can help explain certain anomalies and puzzles found in the past few decades. One of these puzzles is the equity premium puzzle first documented by Mehra & Prescott (1985). The equity premium puzzle refers to the fact that historical real returns of stocks are anomalously higher than government bonds. The premium reflects the relative risk of stocks compared to bonds that are thought of as “risk-free”. The high premium implies an level of risk aversion that is assumed to be unreasonable.
Not one individual is completely similar to another. Dissimilarities in upbringing and character traits explain why two individuals in the exact same situation may not necessarily act identical. These differences in character and behaviour are an interesting base for investigating why some individuals optimize their expenditures over time, invest in financial markets, or prepare for pension, while others have difficulty or no intention to plan ahead. Given that some individuals are more willing to sacrifice well-being in the present to achieve goals in the future, whereas others believe that things will work themselves out in the future, it raises the question whether future-orientation affects financial decision-making and the success of its outcome.
how they can positively change the future, and are, therefore, more aware of possibilities that lie within financial markets. Furthermore, they are better capable of sacrificing current benefits in return for future benefits. Secondly, we look at whether future orientation positively affects the total market value of investments because we expect that future-oriented individuals are more likely to anticipate future difficulties of their investments.
Prior research found several behavioural and non-behavioural aspects that influence the probability of investing in financial markets [Guiso and Jappelli (2005); Almenberg & Dreber (2015); Hong, Kubik & Stein (2004)]. This paper adds to existing research by introducing a new behavioural facet to the probability of financial market participation and to total market value of investments, as well as finding direct predictors of the total value of investments. Limitations of this research lie in the measure used to determine future orientation.
To answer the hypotheses we make use of the 2018 wave of the DNB Household Survey. This has a unique collection of data that contains both economic and psychological aspects of financial behaviour. From survey questions regarding the future we construct a future orientation variable, which is regressed on market participation and total market value of investments using several estimation models. We find no relation between future orientation and financial market participation, and future orientation and total market value of investments. These results are robust for gender, household size, living in large cities and a variable for future orientation from the 2017 wave of data. We do observe low rates of participation in financial markets, as can be seen in many countries over the world. Furthermore, we affirm prior findings by Almenberg & Dreber (2015) that gender is a factor that plays a role in the probability that one participates in financial markets.
The plan in this paper is as follows. We discuss current literature on the topic in section 2. We describe the data in section 3, and the methodology in section 4. This is followed by the results in section 5, and discussion, conclusion and recommendations in section 6.
2. Literature Review
saving accounts is low, individual are expected to be “reaching for yield” in financial markets (Lian, & Ma, 2018). This means that the appetite for risk of individuals increases, leading them to participate on financial markets. Cocco, Gomes, & Maenhout (2005) concluded that non-participation has important implications on individual welfare. According to their calibrated life cycle model, not participating in the stock market can lead to 1.5 to 2% of welfare loss. Limited stock market participation, among other market frictions, also helps to explain the equity premium puzzle. Because of limited stock market participation shareholders require a liquidity premium (Guo, 2004).
Thus, financial market participation is a decision that can have impact on someone’s life. What might explain the limited financial market participation? According to Guiso & Jappelli (2005) traditional research has primarily looked at transaction costs and information costs as an explanation. Guiso & Jappelli (2005), explore determinants of awareness. Using a survey of Italian households they found that 35% of potential investors lack awareness of stocks and 50% of mutual funds. In addition, they found a positive correlation between socioeconomic variables and awareness of stocks and mutual funds. Furthermore, they find that awareness is positively associated with proxies for social interaction.
However, over the last few decades increasing attention is paid to characteristics and behaviour to determine financial market participation. Almenberg & Dreber (2015) found that women participate less in stock markets compared to men. In addition, they found that this gender gap diminishes when controlled for financial literacy. Van Rooij, Lusardi, & Alessie, (2011) found that the majority of the respondents of the DNB Household Survey, data which is used in this paper, have basic financial knowledge, but few go beyond basic concepts. In addition, they found that financial literacy has a positive effect on stock market participation. Grinblatt, Keloharju, & Linnainmaa (2011) looked at a different form of intelligence. They found that market participation is positively related to IQ and that this high correlation even exists among wealthy individuals. Furthermore, they found that high-IQ investors are more likely to earn high Sharpe ratios and experience lower risk. Hong, Kubik, & Stein (2004), using a different approach than Guiso & Jappelli (2005), found a positive relation between social interaction and stock market participation.
According to Frederick, Loewenstein, & O'donoghue (2002) many (conflicting) psychological factors impact one’s time preference, despite time preference often being assumed exogenous. These psychological factors are, therefore, indirect determinants of if and how much one invests. In addition, Becker & Mulligan (1997) found that income, addictions, mortality, uncertainty, education, and wealth are predictors of the degree of time preference. Furthermore, they found that time preference varies not only across individuals, but also across countries, and that wealth causes patience. Daly, Harmon, & Delaney (2009) elaborate more on specific psychological as well as biological determinants of time preference. They found significant results for self-control, conscientiousness, extraversion, experimental avoidance, and consideration of future consequences.
Angelini, & Cavapozzi (2017) found that not only considering future consequences is important in financial decision-making, how one looks at the future is important as well. They found that for a sample of European investors aged 50+ dispositional optimism as well as personality matters for financial decisions. Dispositional optimism refers to generally expecting that good outcomes will happen rather than bad outcomes. Angelini, & Cavapozzi (2017) found that optimism is related to both stock market participation and the share of gross financial wealth invested in stocks. Future orientation relates both to considering future consequences and dispositional optimism. Future orientation can be broadly defined as the extent to which a person is concerned with the future. Strathman, Gleicher, Boninger, & Edwards (1994) nicely formulate future-oriented individuals as ones that ‘believe certain behaviours are worthwhile because of future benefits even if immediate outcomes are relatively undesirable, or even if there are immediate costs’. On the other hand, individuals who have a more present time perspective ‘maximize immediate benefits at the expense of cost or benefits that will not occur for some time’. Strathman et al., (1994) found that future orientation is relatively stable over time. However, individuals could experience events that can affect their future orientation. An example they give is a dramatic change in one’s socioeconomic status.
orientation has on economic decision making (Lusardi, 1999; Anong, & Fisher, 2013). A few studies have been done regarding future orientation and saving. For example, Howlett, Kees, & Kemp (2008) find that individuals with higher levels of future orientation are more likely to save through a 401(k) plan. However, the amount of influence of future orientation is depending on the financial knowledge of the consumers. Anong & Fisher (2013) studied patterns and behaviours among future-oriented individuals. They found that among future-oriented individuals, men and women exhibit similar saving behaviours. However, a model testing saving implementation was rejected. According to the researchers this can be explained by norms, attitudes and behaviour control.
3. Data
In this study, we make use of data drawn from the DNB Household Survey. This is an annual survey among more than 1,500 Dutch households about both psychological and economic aspects of their financial behaviour. The panel is active since 1993 and the data is collected through CentERpanel, which collects data every other week on various topics. If the questionnaires are not (completely) answered in the first week, they will be repeated in the second week. The participants are selected through a random sample based on addresses. The residents of those addresses are asked to become a panel member. It is, therefore, impossible, to register yourself as a panel member. The sample is representative of the Dutch society in terms of age of the head of the household, degree of urbanisation, housing, income, political preference, and region. All members of the household aged 16 and older participate in the questionnaires. The data is collected via internet surveys. Households without a computer and/or internet access will be given a simple computer and internet access. The DNB Household Survey is a relatively long questionnaire and is therefore divided into six parts administered over several weeks. According to Teppa & Vis (2012) panel members of CentERpanel receive some compensation for their expenses (mainly use of internet). For each completed survey they receive, on average, 25 CentERpoints, which is paid out once every three months. 1 CentERpoint is worth about 1 eurocent. Respondents can also choose to donate the compensation to a charity.
Outcome Variables
shares or options as a person who participates in the financial market. However, pension investments are disregarded. The reason for this is that in the Netherlands there exists an pay-as-you-go financed state pension (Van Els, Van Rooij and Schuit, 2007). In addition, more than nine out of ten employees compulsory save for pension through their employer. Of the respondents about 28% indicated that they own at least one of these financial assets on 31 December 2017. Among the various financial assets, mutual funds and shares are the most popular to invest in.
To determine the if future-oriented individuals own higher total market value of investments compared to their less future-oriented counterparts, we need information on the total market value of investments. The DNB Household survey distinguishes between several types of assets. As few respondents invest in anything other than bonds, mutual funds, shares, and derivatives, we only consider these investments. Again, any pension investments are disregarded since retirement plans in the Netherlands often provide little freedom of choice (Van Rooij, Kool & Prast, 2007). To derive the total market value on 31 December 2017,
TMVAL, of investments, we compile MUTUAL, SHARE, BOND, and DERIV. The variable MUTUAL is the total market value of investments in one or multiple mutual funds, where no
distinction is made for type of mutual funds. The variable SHARE consists of the total amount of value invested in shares excluding shares through mutual funds. This can be any type of share, foreign or not, small or large, etc. The variable BOND consists of the total market value of bonds invested by an individual excluding bonds through mutual funds. This can be standard bonds, issued by governments, companies, or other institutions, or mortgage bonds issued by a mortgage bank. The variable DERIV consists of the total market value of put-options, call-options, falcons, warrants, sprinters, and trackers.
Independent Variables
of the DNB Household Survey. The results of the mean comparison tests can be found in Table
A2 in the Appendix.
To control for other observable factors’ that otherwise would bias the results, we implant some control variables in the regressions. We intend to use the following variables: age (AGE), gender (GEN), income (INC), education (EDU), degree of urbanization (URB), savings (SAV), partner (PART), financial skills (SKL), usage of internet banking (INT), and twelve dummy variables for the twelve provinces the Netherlands has (PROV1 – PROV12). All these variables can be found in the DNB Household Survey (see Appendix Table A1). Since only the birth year is given, AGE is constructed by subtracting birth year from the year of the survey, 2018. The variable INC is constructed by compiling the various sources of income, such as income through work and social security benefits. The education variable, EDU, is composed of a scale from 1 to 8, where each number stands for a level education, ranging from no education (1) to university level education (8). The variable for urbanization, URB, is also constructed by a scale. The scale goes from 1 to 5, where 1 means a very low degree of urbanization and 5 a very high degree of urbanization. In addition, the variables for financial skills and usage of internet banking are also ordinal variables ranging from not knowledgeable (1) and not using internet banking (1) to very knowledgeable (4) and very often using internet banking (5). The DNB Household Survey distinguishes between many types of assets, among which are several types of bank accounts, investment in saving certificates, mutual funds, bonds, shares, real estate and options, as well as several types of valuable items. Since few respondents owned saving assets other than bank saving, mutual funds, and shares, we focus on these assets to derive savings (SAV). Furthermore, a few dummy variables are used to check for robustness. These variables are NUMHH and CITY, where NUMHH indicates whether the respondent lives in a one person or multiple person household, and CITY indicates whether the respondent lives in one of the three largest cities in the Netherlands, namely Amsterdam, Rotterdam, and The Hague. In addition, a variable, PROV, is used to account for intraclass correlation. This variable indicates in which of the twelve provinces of the Netherlands the respondent lives. The survey questions that are related to the variables can be found in Table A1 in the Appendix.
not know this, they were allowed to answer a follow-up question, where they could indicate on a scale how much is on their savings account. Since most respondents gave exact amounts and the scales are uneven, we only consider respondents who gave the exact amounts for all questions used in this research. Furthermore, not all household members are involved with financial decision making. Therefore, only members who are most involved with the financial administration are taken into account. After all these adjustments, we consider 746 and 731 observations. The sample characteristics are summarized in Table 1 and Table 2.
Table 1: Summary Statistics
Market Participation No Market Participation Difference
Age 61.14 55.94 -5.20*** (14.57) (16.83) (1.45) Gender (1 = Female) 0.19 0.44 0.25*** (0.39) (0.50) (0.04) Income 45,814 34,700 -11,114*** (31,179) (48,819) (2,219) Education (1-8) 6.27 5.89 -0.38*** (1.40) (1.44) (0.13) Urbanization (1-5) 3.03 3.16 0.13 (1.33) (1.27) (0.11) Partner (1 = Partner) 0.67 0.58 -0.09** (0.47) (0.49) (0.04) Future Orientation 52.06 50.51 -1.55* (9.95) (9.47) (0.85) Savings 94,347 23,482 -70,865*** (132.548) (61,949) (7,321) Future Orientation 2017 (N=682) 52.91 50.80 -2.11** (9.17) (9.31) (0.86) Financial Skills (1-4) 2.55 2.35 -0.21*** (0.05) (0.03) (0.06)
Usage of Internet Banking (1-5) 4.04 3.93 -0.12
(0.07) (0.04) (0.09)
Size Household (1 = multiple) 0.68 0.65 -0.03
(0.47) (0.48) (0.04)
City (1 = living in one of largest 3 cities) 0.15 0.16 0.01
(0.36) (0.36) (0.03)
N 163 583 746
Notes: A description of the variables can be found in section 3, Data. The twelve dummy variables that represent the provinces are shown in the Appendix Table A3. ***p<0.01, **p<0.05, *p<0.
most respondents are male. Since we took only the member of the household most involved with financial decision-making into consideration in this thesis, a possible explanation for the lower percentage of females in the sample is that males are more likely to make financial decisions in the household. On average, the highest education finished is MBO, which is a non-academic higher education that provides workforce education.
Table 2: Sample Characteristics of Assets
Low Future Orientation High Future orientation Difference
Mutual Fund 3,584 8,894 -5,310 (18,255) (44,877) (4,215) Shares 2,571 3,053 -483 (21,181) (20,207) (2,038) Bonds 990 781 209 (8,637) (10,598) (1,031) Derivatives 0 .16 -0.16 (0) (3.67) (0.37)
Total Market Value 7,206 12,884 -5,678
(30,386) (53,446) (5,093)
N 117 614 731
Notes: A description of the variables can be found in section 3, Data. The boundary for high/low future orientation is chosen at 42 since this is the middle of feasible future orientation scores. ***p<0.01, **p<0.05, *p<0.1
Table 2 describes the financial assets owned by respondents with a low and high future
orientation score. Though the group with a high future orientation score has on average the highest total market value of investments, none of the asset types show a significant difference between the two groups.
4. Methodology
This section presents and discusses the methodology in this thesis. First, we discuss potential issues in the form of survey biases. Secondly, the problem of measurement error in asset data is discussed. Then, we discuss the methods for estimating the effect of future orientation on financial market participation. Next, the methods for estimating the effect of future orientation on the total market value of investments are discussed and presented. And finally, some robustness checks are described.
bias the results. According to Teppa & Vis (2012), the willingness to participate after being contacted in 2009 was 55.3%. By using multiple waves of data it is possible to correct for non-response bias by using non-non-response adjusted waves. The reason for this is that there is information available on non-respondents from their responses on earlier waves. However, we use only one wave of data. Therefore, it is infeasible to directly test for a non-response bias. It is, however, possible to test whether there is a difference in answers between early and late respondents via a simple difference in means test. Table A4 in the Appendix shows the results from this test. We found no difference between early and late respondents for the future orientation variable. For both savings and income we found a significant difference, and thus, there may exists a non-response bias for these variables.A similar problem can arise when non-random panel members decide not to answer this particular survey, the DNB Household Survey, or parts of this survey. According to Teppa & Vis (2012), the response rate of already recruited individuals in 2009 was 81.6%. Unfortunately, no system to track these participants is in place. This bias is, therefore, unobservable and uncontrollable.
Another survey bias is social desirability bias, where individuals answer sensitive questions with socially desirable answers, rather than truthful answers. In our study the main variables of interest, future orientation and financial market participation, do not seem to be very sensitive questions. The variable for total market value of investments is perhaps a little bit more receptive for social desirability bias. The reason for this is that the questions used in constructing this variable ask for a precise monetary value of different assets. Although impossible to test for, the social desirability bias may affect results. Two other biases that can have the same effect are extreme response bias and neutral response bias. Both are most common in surveys that offer a scale for responses, such as the Likert scale. Extreme response bias is characterized by participants choosing extreme answers to questions. In contrast, with the neutral response bias participants provide a neutral answer every time, usually as a result of disinterest in the survey. These factors are often not observable and are, therefore, impossible to control for. The reason for these unobservable factors to bias the results is that it will affect the measures of impact.
high end of the savings variable is winsorized1 at 1.2%. The same procedure is done for the income variable, INC. In this case, the high end of the income variable is winsorized at 0.5%.
To estimate the effect of future orientation on financial market participation we apply the following specification:
𝑀𝑃𝐴𝑅𝑖 = {
1 𝑖𝑓 𝑡ℎ𝑒 𝑖𝑡ℎ 𝑖𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙 𝑑𝑜𝑒𝑠 𝑝𝑎𝑟𝑡𝑖𝑐𝑖𝑝𝑎𝑡𝑒 𝑖𝑛 𝑡ℎ𝑒 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑚𝑎𝑟𝑘𝑒𝑡 0 𝑖𝑓 𝑡ℎ𝑒 𝑖𝑡ℎ 𝑖𝑛𝑑𝑖𝑣𝑖𝑑𝑢𝑎𝑙 𝑑𝑜𝑒𝑠 𝑛𝑜𝑡 𝑝𝑎𝑟𝑡𝑖𝑐𝑖𝑝𝑎𝑡𝑒 𝑖𝑛 𝑡ℎ𝑒 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑚𝑎𝑟𝑘𝑒𝑡
𝑀𝑃𝐴𝑅𝑖 = 𝛽0+ 𝛽1𝐹𝑈𝑇𝑖+ 𝛿1𝑋′ + 𝜀𝑖 (1)
To estimate the model, we use two regressions, where market participation, MPAR, is a limited dependent variable and future orientation, FUT, is an ordinal variable. 𝛿1𝑋′ is a vector of control variables2, which are described in Table 1 in the data section. 𝜀 is the error term.
First of all, we estimate the equation above using a linear probability model (LPM). A drawback of this model is that the error term has a highly non-normal distribution, and thus, suffers from heteroscedasticity. Robust standard errors are used to correct for this phenomenon. However, a far more serious drawback from a LPM is that it is possible to predict probabilities outside the unit interval [0,1]. Caution is necessary if predicted values are near or exceed these boundaries. For this reason often a logit or probit model is used. A benefit is using a LPM model is that interpretation is straightforward. Since the LPM predicted probabilities of (1) does indeed exceed boundaries, it is useful to use a second model.
To overcome the issue mentioned above, a probit specification is applied. This model avoids the unboundedness problem that arises with the LPM. In a probit regression a cumulative standard normal distribution function is used. Since the dependent variable is a nonlinear function of the regressors, it is estimated by the maximum likelihood procedure. Again, robust standard errors are used to control for possible heteroskedasticity.
1 Winsorization is a way to minimize outliers by censoring data. Extreme values are replaced by the
next highest value in the sample.
To estimate the effect of future orientation on total market value of investment, where the investments3 can be in mutual funds, shares, bonds, and/or derivatives, we apply the following specification:
𝑇𝑀𝑉𝐴𝐿𝑖 = 𝛽0+ 𝛽1𝐹𝑈𝑇𝑖+ 𝛿1𝑋′ + 𝜀𝑖 (2)
To estimate the model, we use two regressions, where total market value of investments,
TMVAL, is the dependent variable and future orientation, FUT, the independent ordinal
variable. 𝛿1𝑋′ is a vector of control variables4, which are described in Table 1 in section 3 and
Table A3 in the Appendix. 𝜀 is the error term.
The first method of estimation makes use of an ordinary least squares (OLS) model. A major drawback is that this model potentially suffers from endogeneity in the form of selection. Participants self-select whether they invest in the financial market. If this occurs the sample is not random, and, therefore, probably not representative of the population. The decision whether or not to invest in financial markets influences the total market value of investments, TMVAL, since not investing automatically indicates that TMVAL is zero. A situation where TMVAL is not observed for those in the sample that choose to not invest, MPAR=0, is an example of a phenomenon called truncation. Not correcting for truncation leads to biased estimates of true population parameters, as we do not know what the total value of investment would be of those individuals who chose not to invest. To correct for truncation a different method of estimation is needed. A common solution for this incidental truncation is a Heckman correction.
Heckman first developed a sample selection model in 1974. But after criticism on the sensitivity of the parameter estimates to the normality assumption Heckman developed a more robust model, which is known as Heckman’s two-stage model, in 1979 (Marchenko & Genton, 2012). As the name indicates, Heckman’s two-stage model, has a two-step statistical approach. In the first stage, the likelihood of participating in the financial market is predicted for all observations using a probit model and an exclusion restriction. An exclusion restriction is necessary because the first stage and the second stage share the same vector of independent variables. Not including such a restriction will lead to inconsistent estimates due to collinearity issues. In our model the variables for gender and usage of internet banking are excluded in the first stage equation, but appear in the selection equation. The variable for usage of internet banking is
3 Owned on December 31 2017.
chosen because without proper online banking skills the step to join financial markets is larger since more effort has to be put in to open an investment account, to change asset allocations, and to quickly respond to new information. On the other hand, the usage of internet banking would not affect the total market value of investments since the procedure for buying or selling financial assets hardly differs for small or large amounts. The variable for gender is chosen since gender affects the probability of financial market participation (Almenberg & Dreber, 2015). However, there is no evidence nor reason to believe that gender effects the total market value of investments. From the probit model in the first stage, the inverse Mills ratio is generated. This is a selection parameter that is used to account for the potential biases. More formally, the ratio is the probability density function divided by the cumulative distribution function. If the inverse Mill’s ratio is statistically equal to zero, then one can conclude that there is no evidence of sample selection. After the first stage, the second stage uses OLS on the selected sample to predict the dependent variable. The inverse Mill’s ratio, generated in the first stage, is included in the second stage as a predictor, where it is referred to as Lambda.
The first stage equation is the following:
𝑇𝑀𝑉𝐴𝐿∗= 𝛽
0+ 𝛽1𝐹𝑈𝑇 + 𝛿1𝑋′ + 𝜀 (3)
Where 𝑇𝑀𝑉𝐴𝐿∗ is the total market value for all respondents, and FUT is ordinal independent
variable for future orientation. 𝛿1𝑋′ is a vector of control variables5 described in Table 1 in
section 3 and Table A3 in the Appendix, excluding the variables for gender and usage of internet banking, as these variables are used as an exclusion restriction. And finally, 𝜀 is the error term.
The second stage equation is the following:
𝑇𝑀𝑉𝐴𝐿 = 𝛽0+ 𝛽1𝐹𝑈𝑇 + 𝛿1𝑋′ + 𝜌1λ + 𝑢 (4)
Where 𝑇𝑀𝑉𝐴L is the total market value of individuals the selected sample, and FUT is ordinal independent variable for future orientation. 𝛿1𝑋′ is a vector of control variables described in
Table 1 in section 3. To prevent overspecification of the model, the control variables for
provinces are left out of this second stage. λ is the inverse Mill’s ratio, and u is the error term. A more detailed and mathematical explanation of the Heckman correction is described by Heckman (1976); Heckman (1979); Puhani (2000).
Even though the Netherlands might seem a small and homogenous country, several studies have indicated that there are regional differences. For example, Groot, de Groot, and Smit (2014) found that there exist regional wage differences, which can in part be attributed to individual characteristics and in part to variations in employment density. In addition, Arnold, and Vrugt (2002) found that the transmission of monetary policy varies greatly between regions and sectors. Finally, Groenewegen, Westert, and Boshuizen (2003) studied the healthy life expectancy at regional level. They found substantial differences between regions and discovered a link with lifestyle and social conditions differences. Taken these findings into consideration, we clustered the sample for the twelve provinces the Netherlands has (see Table
A3 in the Appendix for summary statistics). Thereafter, we performed the regressions for
market participation with clustered standard errors to control for intraclass correlation effects in the model. To control for possible heteroskedasticity the errors are also bootstrapped6. We
performed 1,000 replications to get accurate estimation (Cameron et al, 2008).
5. Results
This section presents the results of estimations and tests described in the previous section. First, we display and interpret the estimation results for the effect of future orientation on financial market participation. Thereafter, the results for the effect of future orientation on total market value of investments are presented and discussed.
Table 3 below presents the results7 from the analysis on financial market participation and
future orientation. The LPM and probit estimation show similar results for all variables involved, despite the LPM exceeding unit boundaries when producing in-sample forecasts. From the results we can see that future orientation does not have a significant or economic impact on financial market participation. Furthermore, we found that the probability of participating in the financial market decreases if one is a woman. According to the linear probability model, the probability of participating in financial markets decreases by 10.5 percent. The probit output is more complicated to interpret. If gender increases by one unit (in other words, going from male to female) we expect the log-odds of financial market participation to decrease by 0.481, holding all else equal. This result is in accordance with a study by Almenberg & Dreber (2015). The income variable is also statistically significant.
6 Bootstrapped errors allow for more assigned accuracy. Bootstrapping is a type of resampling
technique where from a known sample, one samples with replacement.
According to the LPM model, a one euro increase in income increases the probability of participating in financial markets by 0.00001%. This means that an increase in income of 10.000 euros increases the probability by 0.01%. The probit model shows an increase in the log-odds of the dependent variable of 5.22e-6. For an average household this result has little economic significance. The same appears to be the case for savings. According to the LPM model, a one euro increase in savings increases the probability of participating in financial markets by 0.00002%. And according to the probit model a one euro increase in savings increases the log-odds of financial market participation by 6.41e-6. How individuals perceive their own financial knowledge affects the probability of them participating in financial markets. According to both the LPM and the probit model an increase in the perception of one’s financial skills increases the probability of market participation. An increase in one unit, increases the probability by 3.3 percent (LPM) and increases the log-odds of market participation by 0.151 (probit).
Table 3: Results Market Participation
(1) (2) VARIABLES LPM Probit Future Orientation 0.000 0.001 (0.002) (0.009) Age 0.001 0.004 (0.001) (0.005) Gender (1 = Female) -0.105*** -0.481*** (0.032) (0.147) Income (€) 0.000*1 0.000**2 (0.000) (0.000) Education (1-8) 0.012 0.050 (0.009) (0.039) Urbanization (1-5) -0.008 -0.040 (0.013) (0.056) Savings (€) 0.000***3 0.000***4 (0.000) (0.000) Partner (1 = Partner) -0.019 -0.108 (0.023) (0.098) Financial Skills (1-4) 0.033* 0.147* (0.020) (0.082)
Usage of Internet Banking (1-5) 0.0005 0.008
(0.014) (0.066)
Constant -0.033 -1.873***
(0.106) (0.442)
Observations 741 741
Adjusted R-squared/Pseudo R-squared 0.175 0.168
Table 4 below shows the results8 from the analysis on total market value of investments and future orientation. Not unexpectedly, we do see differences between the two methods of analysis. OLS shows a positive and significant value for education and savings, whereas the Heckman method shows a significant value for urbanization. The inverse Mill’s ratio is significantly different from zero. This means that we can conclude that there is evidence of sample selection and it is, therefore, useful to apply the twostep Heckman model. The OLS model is likely to suffer from endogeneity in the form of selection.
Despite the OLS likely being biased, we look at the output of both estimation models. It can be seen that future orientation does not have a significant relation with the total market value of investments according to both estimations. OLS concludes that having a higher education increases the value of investments, as does savings. A one level of education higher, for example from HBO (higher professional education) to university, increases on average the total market value of investments by 1,759 euros. And for every extra euro saved the total market value of investments increases on average by 0.52 cent. Furthermore, the OLS model finds that living in the provinces of Noord-Holland or Limburg instead of the reference province, Flevoland, decreases on average the total market value of investments by 10,296 and 8,385 euros respectively.
The Heckman model finds that urbanization significantly affects the total market value of investments. By moving to a more urbanized area while keeping everything else equal increases the total market value of assets on average by 0.14 euro. This results has no economic significance.
To see if the results are consistent across samples, several robustness tests are performed. One of these tests is investigating whether results will be the same if the future orientation score of participants in the data wave of 2017 is used. This variable for the future orientation score of 2017 is constructed in the same way as our ‘normal’ 2018 variable and has almost the same sample as the 2018 variable. The only difference is that for FUT17 there are 64 missing values. The estimation results of market participation across the future orientation score of 2017 are displayed in Table 5 below. The results compared to the 2018 wave of future orientation are quite similar. The main difference can be found in the significance of perceived financial skills and income. In Table 3 we can see that in the probit regression of the 2018 wave the variable
Table 4: Results Total Market Value
(1) (2)
VARIABLES OLS Heckman
Future Orientation 47.063 0.018 (123.362) (0.014) Age 86.622 0.007 (59.260) (0.008) Gender (1 = Female) -1,973.874 0.005 (2,983.900) (0.624) Income (€) -0.074 0.0001 (0.059) (0.000) Education (1-8) 1,758.824* -0.084 (963.722) (0.138) Urbanization (1-5) 1,122.425 0.142* (1,286.050) (0.080) Savings (€) 0.521*** 0.0002 (0.082) (0.000) Partner (1 = Partner) -4,707.292 -0.170 (3,010.245) (0.686) Financial Skills (1-4) 1,640.713 0.094 (1,700.774) (0.125)
Usage of Internet Banking (1-5) -770.359 0.023
(1,182.215) (0.130) Province 1 (Friesland) -306.905 (3,222.140) Province 2 (Noord-Brabant) -4,852.328 (3,951.965) Province 3 (Noord-Holland) -10,296.416** (4,132.012) Province 4 (Utrecht) 6,855.878 (8,299.967) Province 5 (Limburg) -8,385.064* (5,056.131) Province 6 (Gelderland) -1,856.407 (4,479.206) Province 7 (Zuid-Holland) -4,056.024 (3,284.563) Province 8 (Overijssel) -5,063.395 (3,279.570) Province 9 (Zeeland) -3,691.277 (4,088.623) Province 10 (Drenthe) 4,752.981 (4,857.884) Province 11 (Groningen) -2,486.942 (4,787.765) Constant -19,388.763 0.625 (11,967.833) (1.193)
Inverse Mill’s Ratio 302,773.303**
(132,236.352)
Observations 728 741
Adjusted R-squared/Pseudo R-squared 0.521 0.511
for income is found significant at 5 percent and in both regressions perceived financial skills are significant at a 10 percent level. In the robustness test for the wave of 2017 these results cannot be found. The estimation results of total market value of investment across the future orientation score of 2017 are displayed in Table 6 below. The biggest difference is that future orientation is significant in the Heckman model if estimated with the 2017 wave of future orientation. Though this result may not be completely reliable as the inverse Mill’s ratio is not significant, which indicates that there is no evidence for sample selection. Therefore, this Heckman model might not be an appropriate model. Other than that, the results are consistent with the 2018 wave.
Table 5: Results Robustness Market Participation for Future orientation 2017
(1) (2) VARIABLES LPM Probit Future Orientation (2017) 0.001 0.005 (0.001) (0.006) Age 0.001 0.004 (0.001) (0.004) Gender (1 = Female) -0.114*** -0.521*** (0.032) (0.150) Income (€) 0.0001 0.0002 (0.000) (0.000) Education (1-8) 0.010 0.041 (0.009) (0.037) Urbanization (1-5) -0.006 -0.033 (0.013) (0.052) Savings (€) 0.000***3 0.000***4 (0.000) (0.000) Partner (1 = Partner) -0.026 -0.135 (0.025) (0.104) Financial Skills (1-4) 0.022 0.102 (0.021) (0.085)
Usage of Internet Banking (1-5) 0.001 0.018
(0.012) (0.052)
Constant -0.051 -1.896***
(0.109) (0.441)
Observations 678 678
Adjusted R-squared/Pseudo R-squared 0.173 0.166
Table 6: Results Robustness Total Market Value for Future Orientation 2017
(1) (2)
VARIABLES OLS Heckman
Future Orientation (2017) -171.346 -0.025** (138.596) (0.010) Age 77.550 0.006 (66.401) (0.010) Gender (1 = Female) -1,934.531 0.077 (3,259.775) (0.878) Income (€) -0.051 0.0001 (0.061) (0.000) Education (1-8) 1,811.758* -0.106 (1,059.704) (0.118) Urbanization (1-5) 1,070.473 0.145* (1,379.123) (0.084) Savings (€) 0.539*** 0.0002 (0.085) (0.000) Partner (1 = Partner) -4,452.596 -0.060 (3,166.870) (0.924) Financial Skills (1-4) 1,932.912 0.060 (1,806.593) (0.127)
Usage of Internet Banking (1-5) -833.382 0.052
(1,242.442) (0.132) Province 1 (Friesland) -1,470.077 (3,339.817) Province 2 (Noord-Brabant) -4,615.976 (4,124.115) Province 3 (Noord-Holland) -11,697.959*** (4,511.962) Province 4 (Utrecht) 3,728.293 (8,973.672) Province 5 (Limburg) -8,897.951* (5,322.918) Province 6 (Gelderland) -2,922.462 (5,369.115) Province 7 (Zuid-Holland) -4,834.741 (3,475.517) Province 8 (Overijssel) -6,002.831 (3,706.296) Province 9 (Zeeland) -3,783.289 (4,162.451) Province 10 (Drenthe) -4,281.272 (5,421.343) Province 11 (Groningen) -1,934.531 (3,259.775) Constant -9,013.699 0.375 (12,410.093) (1.369)
Inverse Mill’s Ratio 65,045.319
(97,645.362)
Observations 666 678
Adjusted R-squared/Pseudo R-squared 0.532 0.545
Other robustness tests were performed too. One of these tests is whether there exists a difference between one person households and multiple person households for market participation and total market value of investments. The results are shown in Tables A8 and A9 in the Appendix. The biggest difference that can be seen for market participation is that gender is only significant in multiple person households. This could indicate that when an individual is living alone and is more used to making important financial decisions alone, the gender gap disappears. On the other hand, education is only significant for multiple person households. For those households a 1 level increase in highest education level finished increases the probability of participating in financial markets by 2.1 percent. Furthermore, we find that income is only significant in the LPM model for a one-person household, and that perceived financial knowledge is not significant for any of the regressions. In addition we find that age is significant and increases probability of market participation for one-person household according to the probit model. For the estimations where total market value of investments is the dependent variable, we find that the Heckman model shows no significant value for any of the variables. A reason for this can be that the inverse Mill’s ratio is not significant, and thus, the model might not be reliable. If we look at the OLS regressions we find again that gender is only significant in multiple person households, as is age.
Another robustness test performed is the difference between gender. The results are displayed in Tables A10 and A11 in the Appendix. The results regarding market participation and future orientation are quite robust. We do see differences in some control variables such as age. Age is only significant for women. In addition, income in the probit model and financial skills in the LPM are also only significant for women. As is the case for the robustness test for household size, we find that the Heckman model for total market value of investments might not be reliable as the inverse Mill’s ratio is not significant. The OLS regressions robust regarding most variables including future orientation. We only find deviations in the variables for age and education.
Furthermore, we find that financial skills are not affecting market participation for all regressions and income is only significant for those individuals living in one of the largest three cities. In addition, we see that for individuals living in one of the largest three cities, that women instead of men are more likely to participate on financial markets. When looking at the results for total market value of investments, displayed in Table A13, we see again that the inverse Mill’s ratio is not significant indicating that there is no evidence of sample selection. The OLS results are robust with the exception of education being insignificant for individuals living in one of the largest three cities in the Netherlands and age being significant for this group.
Overall, we can see that the results for the variable of future orientation are quite robust for both market participation and market value of assets. On the other hand, we can see slight differences for certain groups of individuals, such as one person households, in the control variables. Furthermore, we find that the significance of the inverse Mill’s ratio is not robust.
6. Conclusion
found that some control variables, for example financial skills and age, were not robust. We do confirm the results by Almenberg & Dreber (2015) that women are less likely to participate in financial markets, but also find that this result is not consistent for individuals living in the three largest cities in the Netherlands, namely Amsterdam, Rotterdam, and The Hague. In fact, for individuals living in these cities, the opposite is the case. In addition, there is no significant relationship between gender and financial market participation when households consists of one person.
Despite these results showing no relation between future orientation and financial market participation, a different measure could perhaps show a different result. The measure used in this paper is based on a survey, where participants indicated on a scale questions with regards to how future-oriented they are. This measure thus incorporates a bias since each respondent might have a difference reference frame and is not the best judge of character (Alicke, Klotz, Breitenbecher, Yurak, & Vredenburg, 1995). In addition, several other survey biases could have affected the results. A more experimental setting will improve the value of the results.
Secondly, we studied whether there exists a relationship between future orientation and total market value of investments. We estimated this using both OLS and Heckman models, since not participating in financial markets may lead to truncation of the sample. A Heckman correction can correct for bias from non-randomly selected samples, such as truncated samples. For both models we found no relationship between future orientation and total market value of investment. Several robustness tests, for example for gender, household size, and living in large cities like Amsterdam, Rotterdam and The Hague, showed confirmed this result.
7. References
Alicke, M. D., Klotz, M. L., Breitenbecher, D. L., Yurak, T. J., & Vredenburg, D. S. (1995). Personal contact, individuation, and the better-than-average effect. Journal of personality and social
psychology, 68(5), 804.
Almenberg, J., & Dreber, A. (2015). Gender, stock market participation and financial literacy. Economics Letters, 137, 140-142.
Angelini, V., & Cavapozzi, D. (2017). Dispositional optimism and stock investments. Journal of economic psychology, 59, 113-128.
Anong, S. T., & Fisher, P. J. (2013). Future orientation and saving for medium‐term expenses. Family and Consumer Sciences Research Journal, 41(4), 393-412.
Arnold, I. J., & Vrugt, E. B. (2002). Regional effects of monetary policy in the Netherlands. International Journal of Business and Economics, 1(2), 123.
Becker, G. S., & Mulligan, C. B. (1997). The endogenous determination of time preference. The Quarterly Journal of Economics, 112(3), 729-758.
Cameron, A. C., Gelbach, J. B., & Miller, D. L. (2008). Bootstrap-based improvements for inference with clustered errors. The Review of Economics and Statistics, 90(3), 414-427.
Cocco, J. F., Gomes, F. J., & Maenhout, P. J. (2005). Consumption and portfolio choice over the life cycle. The Review of Financial Studies, 18(2), 491-533.
Daly, M., Harmon, C. P., & Delaney, L. (2009). Psychological and biological foundations of time preference. Journal of the European Economic Association, 7(2-3), 659-669.
Frederick, S., Loewenstein, G., & O'donoghue, T. (2002). Time discounting and time preference: A critical review. Journal of economic literature, 40(2), 351-401.
Grinblatt, M., Keloharju, M., & Linnainmaa, J. (2011). IQ and stock market participation. The Journal of Finance, 66(6), 2121-2164.
Groenewegen, P. P., Westert, G. P., & Boshuizen, H. C. (2003). Regional differences in healthy life expectancy in the Netherlands. Public Health, 117(6), 424-429.
Groot, S. P., de Groot, H. L., & Smit, M. J. (2014). Regional wage differences in the Netherlands: Micro evidence on agglomeration externalities. Journal of Regional Science, 54(3), 503-523.
Guiso, L., & Jappelli, T. (2005). Awareness and stock market participation. Review of Finance, 9(4), 537-567.
Guo, H. (2004). Limited stock market participation and asset prices in a dynamic economy. Journal of Financial and Quantitative Analysis, 39(3), 495-516.
Heckman, J. J. (1976). The common structure of statistical models of truncation, sample selection and limited dependent variables and a simple estimator for such models. In Annals of Economic and Social Measurement, Volume 5, number 4 (pp. 475-492). NBER.
Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica: Journal of the econometric society, 153-161.
Howlett, E., Kees, J., & Kemp, E. (2008). The role of self‐regulation, future orientation, and financial knowledge in long‐term financial decisions. Journal of Consumer Affairs, 42(2), 223-242.
Kahana, E., Kahana, B., & Zhang, J. (2005). Motivational antecedents of preventive proactivity in late life: Linking future orientation and exercise. Motivation and emotion, 29(4), 438-459.
Keough, K. A., Zimbardo, P. G., & Boyd, J. N. (1999). Who's smoking, drinking, and using drugs? Time perspective as a predictor of substance use. Basic and applied social psychology, 21(2), 149-164. Lea, S. E., Webley, P., & Walker, C. M. (1995). Psychological factors in consumer debt: Money management, economic socialization, and credit use. Journal of economic psychology, 16(4), 681-701. Lian, C., & Ma, Y. (2018). Low Interest Rates and Investor Behavior: A Behavioral Perspective. Lusardi, A. (1999). Information, expectations, and savings for retirement. Behavioral dimensions of retirement economics, 81, 115.
Marchenko, Y. V., & Genton, M. G. (2012). A Heckman selection-t model. Journal of the American Statistical Association, 107(497), 304-317.
Mehra, R., & Prescott, E. C. (1985). The equity premium: A puzzle. Journal of monetary Economics, 15(2), 145-161.
Puhani, P. (2000). The Heckman correction for sample selection and its critique. Journal of economic surveys, 14(1), 53-68.
Strathman, A., Gleicher, F., Boninger, D. S., & Edwards, C. S. (1994). The consideration of future consequences: Weighing immediate and distant outcomes of behavior. Journal of personality and social psychology, 66(4), 742.
Teppa, F., & Vis, C. (2012). The CentERpanel and the DNB household survey: Methodological aspects (No. 1004). Netherlands Central Bank, Research Department.
Van Els, P. J. A., Van Rooij, M. C. J., & Schuit, M. E. J. (2007). Why mandatory retirement saving?. In Costs and benefits of collective pension systems (pp. 159-186). Springer, Berlin, Heidelberg. Van Rooij, M. C., Kool, C. J., & Prast, H. M. (2007). Risk-return preferences in the pension domain: are people able to choose?. Journal of public economics, 91(3-4), 701-722.
Van Rooij, M., Lusardi, A., & Alessie, R. (2011). Financial literacy and stock market participation. Journal of Financial Economics, 101(2), 449-472.
8. Appendix
Table A1: Survey Questions related to Variables
Variable Question Number of
answer options Future
Orientation
1. I think about how things can change in the future, and try to influence those things in my everyday life.
7 Future
Orientation
2. I often work on things that will only pay off in a couple of years. 7 Future
Orientation
3. I am only concerned about the present, because I trust that things will work themselves out in the future.
7 Future
Orientation
4. With everything I do, I am only concerned about the immediate consequences (say a period of a couple of days or weeks).
7 Future
Orientation
5. Whether something is convenient for me or not, to a large extent determines the decisions that I take or the actions that I undertake.
7 Future
Orientation
6. I am willing to sacrifice my well-being in the present to achieve certain goals in the future.
7 Future
Orientation
7. I think it is important to take warnings about negative consequences of my acts seriously, even if these negative consequences would only occur in the distant future.
7
Future Orientation
8. I think it is more important to work on things that have important consequences in the future, than to work on things that have immediate but less important
consequences.
7
Future Orientation
9. In general, I ignore warnings about future problems because I think these problems will be solved before they get critical.
7 Future
Orientation
10. I think there is no need to sacrifice something now for problems that lie in the future, because it will always be possible to solve these future problems later.
7 Future
Orientation
11. I only respond to urgent problems, trusting that problems that come up later can be solved in a later stage.
7 Future
Orientation
12. I find it more important to do work that gives short-term results, than work where the consequences are not apparent until later
7
Gender Sex of the respondent 2
Education Highest level of education completed 9
Size Household
Number of household members 9
Urbanization Degree of urbanization of the town/city of residence 5
Province Province 12
Partner Is there a partner present in the household? 2
Age Year of birth of the respondent Open
Living in largest 3 Cities Region 5 Usage of Internet Banking
Banks also offer the possibility to arrange banking affairs through Internet (Internet banking). Do you use Internet banking?
5
Financial Skills
Table A2: Stability Future Orientation Variable 2018 2017 Difference Question 1 4.37 4.49 -0.13 (1.49) (1.42) (0.08) Question 2 3.78 3.78 0.00 (1.57) (1.54) (0.08) Question 3 4.44 4.35 0.09 (1.62) (1.55) (0.08) Question 4 4.24 4.33 -0.09 (1.62) (1.61) (0.09) Question 5 3.52 3.54 -0.01 (1.34) (1.34) (0.07) Question 6 3.35 3.39 -0.04 (1.52) (1.51) (0.04) Question 7 4.83 4.87 -0.03 (1.47) (1.53) (0.08) Question 8 4.15 4.23 -0.07 (1.37) (1.37) (0.07) Question 9 4.96 5.03 -0.07 (1.43) (1.41) (0.08) Question 10 4.36 4.30 0.06 (1.52) (1.48) (0.08) Question 11 4.41 4.44 -0.03 (1.53) (1.51) (0.08) Question 12 4.42 4.50 -0.07 (1.44) (1.36) (0.07)
Total Future Orientation Score 50.85 51.26 -0.41
(9.59) (9.31) (0.50)
N 746 682
Table A3: Summary Statistics Provinces
Market Participation No Market Participation
N Mean St. Dev. N Mean St. Dev.
Province 1 (Friesland) 163 0.04 0.19 583 0.04 0.20 Province 2 (Noord-Braband) 163 0.15 0.36 583 0.17 0.37 Province 3 (Noord-Holland) 163 0.19 0.39 583 0.14 0.35 Province 4 (Utrecht) 163 0.09 0.29 583 0.05 0.23 Province 5 (Limburg) 163 0.05 0.23 583 0.08 0.27 Province 6 (Gelderland 163 0.12 0.33 583 0.15 0.36 Province 7 (Zuid-Holland) 163 0.18 0.39 583 0.19 0.39 Province 8 (Overijssel) 163 0.06 0.24 583 0.06 0.24 Province 9 (Zeeland) 163 0.03 0.17 583 0.02 0.15 Province 10 (Drenthe) 163 0.02 0.13 583 0.03 0.16 Province 11 (Groningen) 163 0.05 0.23 583 0.05 0.21 Province 12 (Flevoland) 163 0.01 0.08 583 0.02 0.14 Notes: These are all dummy variables where 1 means that the individual is living in that province.
Table A4: Results t-test comparing means between early and late responders
In this table the p-value is given of a t-test comparing the means for several variables of early and late responders. Early respondents are respondents that answered the questionnaire within three weeks. Late respondents are respondents who answered the questionnaire after three weeks. H0: difference = 0, Ha: difference ≠ 0
WEEK(FUT) WEEK(INC) WEEK(SAV) Future Orientation 0.1165
Income (€) 0.0013***
Savings (€) 0.0001***
Market Participation (1 = Yes) 0.0423**
Table A5: Results Total Market Value Heckman Stage 1 (1) VARIABLES Heckman Future Orientation 529.937** (234.805) Age 296.357** (150.197) Gender (1 = Female) Income (€) -0.029 (0.073) Education (1-8) -486.402 (1,350.161) Urbanization (1-5) 5,017.290** (2,077.645) Savings (€) 0.538*** (0.084) Partner (1 = Partner) -8,985.294* (4,698.396) Financial Skills (1-4) 4,400.815* (4,698.396)
Usage of Internet Banking (1-5)
Province 1 (Friesland) -1,997.024 (3,355.857) Province 2 (Noord-Brabant) -4,722.181 (4,037.962) Province 3 (Noord-Holland) -10,189.790** (4,406.069) Province 4 (Utrecht) 7,418.864 (8,354.962) Province 5 (Limburg) -8,066.952 (5,144.094) Province 6 (Gelderland) -1,502.641 (4,649.551) Province 7 (Zuid-Holland) -4,758.482 (3,569.145) Province 8 (Overijssel) -4,635.075 (3,429.170) Province 9 (Zeeland) -4,454.831 (4,426.673) Province 10 (Drenthe) 2,332.752 (5,270.875) Province 11 (Groningen) -3,843.961 (5,012.995) Constant -77,571.530*** (27,979.014) Observations 741
Table A6: Results Market Participation without Winsorizing (1) (2) VARIABLES LPM Probit Future Orientation 0.001 0.003 (0.002) (0.009) Age 0.002 0.007 (0.001) (0.005) Gender (1 = Female) -0.125*** -0.534*** (0.032) (0.141) Income (€) 0.0001 0.0002 (0.000) (0.000) Education (1-8) 0.020** 0.078** (0.009) (0.034) Urbanization (1-5) -0.005 -0.029 (0.014) (0.056) Savings (€) 0.000***3 0.0004 (0.000) (0.000) Partner (1 = Partner) -0.012 -0.077 (0.021) (0.083) Financial Skills (1-4) 0.038* 0.163* (0.022) (0.085)
Usage of Internet Banking (1-5) 0.001 0.010
(0.014) (0.062)
Constant -0.118 -2.157***
(0.113) (0.468)
Observations 741 741
Adjusted R-squared/Pseudo R-squared 0.140 0.136
Notes: A market participation dummy variable is regressed on a future orientation variable, and control variables of which gender, and partner are dummy variables, and education, urbanization, financial skills, usage of internet banking are ordinal variables described in Table 1 in section 3. Bootstrapped and clustered (for province) standard errors with 1,000 replications in parentheses. ***p<0.01, **p<0.05, *p<0.1. 1) 3.48e-7 2) 1.34e-6 3) 1.32e-6 4)
Table A7: Results Total Market Value without Winsorizing
(1) (2)
VARIABLES OLS Heckman
Future Orientation 47.063 0.018 (125.710) (0.014) Age 86.622 0.007 (61.576) (0.008) Gender (1 = Female) -1,973.874 0.005 (2,949.502) (0.624) Income (€) -0.074 0.0001 (0.060) (0.000) Education (1-8) 1,758.824* -0.084 (995.459) (0.138) Urbanization (1-5) 1,122.425 0.142* (1,295.288) (0.080) Savings (€) 0.521*** 0.0002 (0.080) (0.000) Partner (1 = Partner) -4,707.292 -0.170 (2,986.917) (0.686) Financial Skills (1-4) 1,640.713 0.094 (1,675.482) (0.125)
Usage of Internet Banking (1-5) -770.359 0.023
(1,205.368) (0.130) Province 1 (Friesland) -306.905 (3,535.456) Province 2 (Noord-Brabant) -4,852.328 (4,852.328) Province 3 (Noord-Holland) -10,296.416** (4,445.288) Province 4 (Utrecht) 6,855.878 (8,061.791) Province 5 (Limburg) -8,385.064* (5,073.963) Province 6 (Gelderland) -1,856.407 (4,678.979) Province 7 (Zuid-Holland) -4,056.024 (3,467.612) Province 8 (Overijssel) -5,063.395 (3,422.000) Province 9 (Zeeland) -3,691.277 (4,220.383) Province 10 (Drenthe) 4,752.981 (5,132.707) Province 11 (Groningen) -2,486.942 (4,886.721) Constant -19,388.763 0.625 (12,209.610) (1.193)
Inverse Mill’s Ratio 302,773.303**
(132,236.352)
Observations 728 741
Adjusted R-squared/Pseudo R-squared 0.521 0.480
Table A8: Results Robustness Market Participation for Household Number
(1) (2) (3) (4)
LPM LPM Probit Probit
VARIABLES NUMHH=0 NUMHH=1 NUMHH=0 NUMHH=1
Future Orientation 0.003 -0.001 0.012 -0.004 (0.002) (0.003) (0.009) (0.011) Age 0.002 -0.001 0.009* -0.003 (0.001) (0.002) (0.005) (0.007) Gender (1 = Female) -0.063 -0.144*** -0.280 -0.716*** (0.042) (0.032) (0.194) (0.148) Income (€) 0.000*1 0.0002 0.0003 0.0004 (0.000) (0.000) (0.000) (0.000) Education (1-8) -0.006 0.021* -0.025 0.089* (0.012) (0.012) (0.056) (0.049) Urbanization (1-5) -0.007 -0.006 -0.033 -0.029 (0.025) (0.015) (0.106) (0.064) Savings (€) 0.000***5 0.000***6 0.000*7 0.000***8 (0.000) (0.000) (0.000) (0.000) Financial Skills (1-4) 0.028 0.037 0.138 0.151 (0.029) (0.035) (0.157) (0.144)
Usage of Internet Banking (1-5) -0.002 0.003 0.001 0.023
(0.222) (0.014) (0.106) (0.063)
Constant -0.146 -0.024 -2.419** -2.111***
(0.222) (0.126) (0.959) (0.589)
Observations 258 483 258 483
Adjusted R2/Pseudo R2 0.150 0.189 0.162 0.196
Table A9: Results Robustness Total Market Value for Household Number
(1) (2) (3) (4)
OLS OLS Heckman Heckman
VARIABLES NUMHH=0 NUMHH=1 NUMHH=0 NUMHH=1
Future Orientation 29.747 61.336 0.029 0.018 (179.885) (177.367) (0.700) (0.423) Age -70.795 144.769* -0.012 0.016 (99.969) (83.768) (0.447) (1.871) Gender (1 = Female) 688.756 -4,734.243* 0.042 0.055 (4,065.272) (2,736.194) (21.505) (15.260) Income (€) -0.127 -0.047 0.0001 0.0002 (0.144) (0.071) (0.001) (0.001) Education (1-8) 1,232.353 1,745.747 -0.127 -0.092 (1,364.523) (1,288.793) (9.014) (5.295) Urbanization (1-5) -526.831 1,848.677 0.252 0.113 (2,341.357) (1,410.725) (6.452) (9.550) Savings (€) 0.675*** 0.457*** 0.0003 0.0004 (0.181) (0.085) (0.000) (0.003) Financial Skills (1-4) 2,428.066 1,003.567 0.330 0.053 (2,427.620) (2,173.102) (6.391) (18.696) Usage of Internet Banking (1-5) -3,651.012* 307.286 0.119 -0.018 (2,025.625) 1,521.485 (7.256) (11.657) Province 1 (Friesland) -7,867.730 53.576 (7,879.015) (4,590.463) Province 2 (Noord-Brabant) -11,588.939 -1,930.311 (7,375.613) (5,180.297) Province 3 (Noord-Holland) -6,306.086 -12,541.496** (5,904.913) (5,716.761) Province 4 (Utrecht) 15,454.442 141.285 (16,306.131) (6,258.405) Province 5 (Limburg) -7,293.040 -10,044.154 (7,501.251) (6,568.292) Province 6 (Gelderland) 1,079.219 -4,518.198 (6,217.399) (6,097.654) Province 7 (Zuid-Holland) -4,389.156 -5,165.413 (5,710.706) (4,474.067) Province 8 (Overijssel) -6,647.023 -4,092.264 (5,119.610) (4,402.192) Province 9 (Zeeland) -1,093.805 -5,523.723 (5,079.944) (6,161.834) Province 10 (Drenthe) 5,948.780 3,940.203 (9,551.877) (6,471.581) Province 11 (Groningen) -6,051.969 -35.698 (6,222.421) (7,112.684) Constant 4,484.291 -29,488.490* -0.189 5.024 (18,972.374) (16,776.016) (105.508) (9.177)
Inverse Mill’s Ratio -179,296.545 156,540.691
(1.503e+11) (5.875e08)
Observations 254 474 258 483
Adjusted R2/Pseudo R2 0.623 0.474 0.631 0.484
Table A10: Results Robustness Market Participation for Gender
(1) (2) (3) (4)
LPM LPM Probit Probit
VARIABLES GEN=0 GEN=1 GEN=0 GEN=1
Future Orientation -0.0001 0.001 -0.001 0.007 (0.003) (0.002) (0.010) (0.012) Age 0.0002 0.002** 0.001 (0.013)* (0.001) (0.001) (0.005) (0.007) Income (€) 0.0003 0.0004 0.0005 0.000**6 (0.000) (0.000) (0.000) (0.000) Education (1-8) 0.019 0.005 0.060 0.032 (0.015) (0.007) (0.055) (0.066) Urbanization (1-5) -0.016 0.005 -0.057 0.018 (0.021) (0.012) (0.076) (0.084) Savings (€) 0.000***7 0.000***8 0.000***9 0.000***10 (0.000) (0.000) (0.000) (0.000) Partner (1= Partner) -0.001 -0.034 -0.010 -0.198 (0.027) (0.033) (0.080) (0.257) Financial Skills (1-4) 0.025 0.045* 0.091 0.296 (0.033) (0.023) (0.120) (0.185)
Usage of Internet Banking (1-5) -0.013 0.016 -0.042 0.110
(0.021) (0.012) (0.072) (0.145)
Constant 0.082 -0.276* -1.253** -4.024***
(0.173) (0.149) (0.572) (1.252)
Observations 456 285 456 285
Adjusted R2/Pseudo R2 0.134 0.136 0.122 0.183
Notes: A market participation variable is regressed on a future orientation variable, and control variables of which partner is a dummy variable, and education, urbanization, financial skills, and usage of internet banking are ordinal variables described in Table 1 in section 3 and Table A1 in the Appendix. The variable for gender is omitted. The variables for income and savings have been winsor transformed to avoid extreme outliers. Bootstrapped and clustered (for province) standard errors with 1,000 replications in parentheses. ***p<0.01, **p<0.05, *p<0.1. 1)
