# Young adults’ subjective and objective risk attitude in financial decision making: Evidence from the lab and the field

## Abstract

### Purpose

The purpose of this paper is to derive the determinants of young adults’ subjective and objective risk attitude in theoretical and real-world financial decisions. Furthermore, a comparison of the factors that influence young adults’ and older adults’ risk attitude is provided.

### Design/methodology/approach

The paper relies on an experimental setting and a cross-sectional field study using data of the German central bank’s (Deutsche Bundesbank) PHF-Survey.

### Findings

Young adults’ objective risk aversion is not constant but increases with stake sizes. Furthermore, young adults’ subjective risk attitude is a better predictor for their objective risk attitude than a set of commonly employed socio-demographics and economics like age or income. Moreover, young adults’ subjective risk attitude works as a mediator for the influence of their investable financial wealth on their objective risk attitude. Although young adults’ subjective risk attitude shows a gender effect, the influence of young adults’ gender on their objective risk attitude decreases with higher stake sizes. Compared to older adults, young adults generally show a similar degree of subjective risk aversion. However, due to stronger financial restrictions, young adults show a higher degree of objective risk aversion.

### Originality/value

Although individuals’ financial outcomes depend on the financial behavior established in young adulthood, there is no study that simultaneously analyzes the determinants of young adults’ subjective and objective risk attitude in real-world financial decisions with a focus on young adults as a separate age group. The paper closes this gap in literature and additionally provides a comparison of the subsamples of young adults and older adults. The analysis in this paper reveals that young adults’ lower engagement in financial markets is primarily driven by their tight budget and not by a fundamental different subjective risk attitude.

## Keywords

#### Citation

Oehler, A., Horn, M. and Wedlich, F. (2018), "Young adults’ subjective and objective risk attitude in financial decision making", *Review of Behavioral Finance*, Vol. 10 No. 3, pp. 274-294. https://doi.org/10.1108/RBF-07-2017-0069

### Publisher

:Emerald Publishing Limited

Copyright © 2018, Emerald Publishing Limited

## 1. Introduction

Recent studies show that individuals’ financial outcomes depend on the financial behavior established in young adulthood (e.g. Xu *et al.*, 2015; Eccles *et al.*, 2013)[1]. Consequently, concepts that aim to improve individuals’ financial well-being should target at the determinants of young adults’ financial behavior. Although risk attitude is one of the key determinants of investment behavior (e.g. Cohn *et al.*, 1975; Dorn and Huberman, 2005), young adults’ risk attitude is hardly sufficiently analyzed.

Risk attitude in the financial domain is covered by two different concepts[2]. One strand of literature relies on the neo-classical assumption that the financial risk taken by an individual through a financial decision under consideration of the individual’s socio-economics and demographics exactly mirrors the individual’s risk attitude (see, e.g. Pratt, 1964 and Arrow, 1965). Hence, the individual’s risk attitude can be measured by the chosen financial risk, which is considered as “objective risk attitude”[3]. Researchers commonly use decisions in hypothetical lotteries (e.g. Holt and Laury, 2002; Fellner and Maciejovsky, 2007) or the amount/percentage of wealth that subjects invest in risky assets (e.g. Friend and Blume, 1975; Fama and Schwert, 1977; Schooley and Worden, 1996; Oehler, 1998) to measure the objective risk attitude. The second strand of literature assumes that investment decisions are the result of a decision process, which is additionally influenced by individuals’ subjective perception, heuristics, and bounded rationality (see, e.g. the surveys of Oehler, 1995 and Hirshleifer, 2015 on the behavioral aspects of the decision process). Therefore, the investment decisions, and likewise the measured “objective risk attitude,” are most likely driven by partially unobservable factors in addition to an individual’s risk attitude and socio-economics and demographics (see, e.g. Schoemaker, 1993). In this framework, researchers consequently can only measure individual’s risk attitude by directly asking the individuals to self-assess their willingness to take financial risk (see, e.g. Chaulk *et al.*, 2003; Nosic and Weber, 2010; Dohmen *et al.*, 2011). Since individuals’ self-assessment always includes subjective components, their self-assessed risk attitude is considered as “subjective risk attitude”. Both risk attitude concepts are frequently used in financial as well as economic models, and undoubtedly have their merits. Since the concepts are not mutually exclusive, some studies combine both risk attitude concepts in one framework. For example, Nosic and Weber (2010) differentiate between subjective and objective risk attitude and find that subjective risk attitude is a significant determinant of objective risk attitude (see also Schooley and Worden, 1996; Chaulk *et al.*, 2003; Halko *et al.*, 2012; Kaustia and Luotonen, 2016).

Although many studies analyze or at least control for the influence of age on individuals’ subjective and/or objective risk attitude[4], these studies may only be partially suitable to sufficiently understand the determinants of young adults’ subjective and objective risk attitude. In studies that include respondents of all phases of the lifecycle young adults represent the tail of the age distribution and commonly a minority in the analyzed samples. If findings from, e.g. linear regression analyses that cover all groups of adults, are applied to a subsample such as young adults, possible heteroscedasticity effects or influential observations from the remaining subsamples probably skew the implications regarding the subsample of young adults. Therefore, analyses that aim to shed light on the determinants of young adults’ risk attitude should consider young adults as a separate group (see, e.g. Morin and Suarez, 1983). In addition, young adults’ subjective and objective risk attitude needs to be considered simultaneously to derive clear implications regarding the drivers of young adults’ financial decisions. If cross-sectional field data are employed to ensure high external validity of the findings, furthermore, an assessment whether young adults’ risk attitude is independent of time-invariant wealth effects is required (see, e.g. Paya and Wang, 2016)[5].

We are not aware of a study that combines the three previous aspects. The studies from Albert and Duffy (2012), Chaulk *et al.* (2003), and Morin and Suarez (1983) consider young adults as a separate age group. However, Albert and Duffy (2012) compare risk attitudes between young and older adults, but only focus on objective risk attitude. Chaulk *et al.* (2003) use two data sets consisting of 76 individuals and the 1998 Survey of Consumer Finances covering 4,305 individuals, but do not analyze individuals’ real-world financial investments. Morin and Suarez (1983) sidestep the subjective risk attitude. Nosic and Weber (2010) and Schooley and Worden (1996) consider young adults’ subjective and objective risk attitude simultaneously; however, Nosic and Weber (2010) solely rely on data from the lab and Schooley and Worden (1996) only differentiate between adults that are either older or younger than 45 years.

Therefore, the aim of our paper is to derive a more complete picture of young adults’ subjective and objective risk attitude with respect to theoretical and real-world financial decisions. Although some aspects of this research aim have been examined separately in previous papers, there is no study that covers young adults’ subjective and objective risk attitude in both experimental settings and real-world investments together. Hence, the main contribution of our paper is that it combines the focus on young adults with a simultaneous observation of their subjective and objective risk attitude and the utilization of data from an experimental setting and the field.

Using a questionnaire design “in the lab”, enables us to analyze young adults’ risk attitude under conditions of a high degree of internal validity, whereas the use of field data guarantees transferability to reality expressing a high external validity (see, e.g. Dohmen *et al.*, 2010). Furthermore, the two data sets offer different opportunities for our analysis. The questionnaire design is used to provide an initial assessment whether young adults’ risk attitude is independent of time-invariant wealth. More precisely, we analyze under consideration of young adults’ subjective risk attitude and gender whether young adults change their objective risk attitude toward a theoretical financial investment when only the absolute value of the outcome but not the expected distribution of outcomes is varied (with young adults’ financial wealth meanwhile staying stable). Moreover, we use data from the PHF-Survey performed and provided by the German Central Bank (Deutsche Bundesbank) to analyze the determining factors of young adults’ subjective and objective risk attitude regarding real-world financial decisions and to provide a comparison of the factors influencing young adults’ and older adults’ risk attitude to better connect with the existing literature and to derive clearer implications for concepts that aim explicitly on young adults’ financial behavior.

Our results are fourfold. First, young adults’ objective risk attitude is not constant over different stake sizes. Instead, young adults get more risk averse when a financial decision’s absolute amount of money rises (with young adults’ financial wealth being unchanged). Second, young men show a lower degree of subjective risk aversion than young women. Third, young adults’ subjective risk attitude is a better predictor for their objective risk attitude than a set of commonly employed socio-demographics and economics like age or income and works as mediator for the influence of young adults’ investable financial wealth on their respective objective risk attitude. Forth, young adults and older adults generally show a similar degree of subjective risk aversion. However, young adults show a higher degree of objective risk aversion due to stronger financial restrictions. We observe indications that if young adults were equally wealthy they would show a lower degree of objective risk aversion than older adults.

The results provide implications for financial advisers, academics, and policy makers. Financial advisors should take the gender dependence of young adults’ subjective risk aversion into account and be aware that young adults generally have fewer savings than older adults. This finding is also relevant for policy makers and academics when they elaborate on concepts to enhance the engagement of young adults in financial markets. If policy makers and academics tried to simultaneously affect the objective risk attitude of both young and older adults, without considering the financial restrictions of young adults, the interventions would most probably fail.

Our study is organized as follows. Section 2 addresses data and methodology of the experimental study and the PHF-Survey as well as the methodological approach to analyze young adults’ risk attitude. We report our results in Section 3. In Section 4, we discuss the implications of our findings and conclude the paper.

## 2. Data and methodology

### 2.1 Experimental questionnaire design

We employ a not incentivized questionnaire setting among undergraduate business students at Bamberg University. The questionnaire includes a subjective question to assess young adults’ subjective risk attitude and a hypothetical lottery design as objective risk attitude measure to analyze whether young adults’ risk attitude is constant over stake sizes and generally independent of wealth.

First, participants are asked to rate their attitude to take financial risks on a five-point Likert scale (see, e.g. Nosic and Weber, 2010; Dohmen *et al.*, 2011) ranging from 1 (very low willingness) to 5 (very high willingness). The questionnaire additionally includes participants’ gender (*Gender*_{j}, see, e.g. Barsky *et al.*, 1997; Eckel and Grossman, 2008; Booth and Katic, 2013; Oehler and Horn, 2016 for the influence of investors’ gender on their risk attitude) and age[6] (see, e.g. Morin and Suarez, 1983; Schooley and Worden, 1996; Albert and Duffy, 2012; Deutsche Bundesbank, 2015; Oehler and Horn, 2016 for the influence of investors’ age on their risk attitude) as potential influencing factors of participants’ subjective and objective risk attitude.

Second, a hypothetical lottery design which covers increasing stake sizes based on the design of Deutsche Bundesbank (2015) is employed. We would like to point out that even if we use—in line with previous literature (e.g. Nosic and Weber, 2010)—the lottery design to measure participants’ objective risk attitude, participants’ decisions in this design are influenced by their subjective perception and heuristics and, consequently, not literally objective. In the lottery design, participants have the option to choose between a certain payment (riskless choice) and participating in a lottery (risky choice). The lottery offers, with equal probability, the chance to either win twice the amount of the fixed payment or nothing. Moreover, the participants have the choice to indicate that they are indifferent between these two options since the certain payment and the lottery have the same expected value. The structure of the lottery is similar to the lottery design of Nosic and Weber (2010) and relies on the expected utility framework[7]. We provide three different designs with a certain payment of EUR 5, 50, and 500. The winning amounts of the lottery vary correspondingly (i.e. EUR 10 or 0, EUR 100 or 0, EUR 1,000 or 0). The different choices are independent from each other and every participant takes part in every design. Choosing the certain payment in the questionnaire indicates risk aversion while choosing the lottery states risk seeking. A risk-neutral subject is indifferent between both options.

If participants’ choices regarding the decision between the certain payments or participating in the lotteries vary across the three designs, the risk attitude of young adults is not constant (since their wealth does not change while answering the questionnaire). Wilcoxon tests are used to prove these potentially arising differences between the decisions of the three designs. Furthermore, cross-sectional regression analyses are provided to shed light on the determining factors of young adults’ objective risk attitude in the experimental questionnaire design. We follow the work of Nosic and Weber (2010) and employ young adults’ subjective risk attitude and gender as independent variables. The baseline model for the regression analyses is set up as follows:

*RiskAtt*

_{Obj,i}denotes the objective risk attitude of participant

*j*, i.e. her choice (certain payment=1, indifference=2, gamble= 3) in the design

*i*(expected value EUR 5, 50, 500). The independent variable

*RiskAtt*

_{Subj,j}is the participants’ subjective risk attitude and can take a value between 1 (very high degree of subjective risk aversion) and 5 (very low degree of subjective risk aversion).

*Gender*

_{j}is a dummy variable which takes the value 1 for women and 0 otherwise.

### 2.2 PHF-Survey

The data set of the PHF-Survey (DOI: 10.12757/PHF.01.01.01.stata) performed and provided by Deutsche Bundesbank is employed to analyze the determining factors of young adults’ risk attitude and to provide a comparison of the factors influencing young adults’ and older adults’ risk attitude. The data set includes information about the absolute amount of money a household invests per asset class. Hence, we are able to compute households’ objective risk attitude as the percentage of wealth invested in risky assets. In addition, the survey provides data on households’ subjective risk attitude (*RiskAtt*_{Subj,h} ranging from 1 (very high degree of subjective risk aversion) to 4 (very low degree of subjective risk aversion)) measured with a subjective question similar to the question used in Nosic and Weber (2010) and Dohmen *et al.* (2011). All in all, Deutsche Bundesbank interviewed 3,565 German households from September 14, 2010 to July 15, 2011 for the PHF-Survey.

In the first part of the cross-sectional analysis, solely households with young adults as financial decision makers are included. Regression analyses are provided to examine the influence of young adults’ socio-demographics and economics on their subjective risk attitude. Regarding the independent variables, we rely on socio-demographic factors that have been proven as significant in related studies: the gender (*Gender*_{h}) and age (*Age*_{h}) of the person who is mainly responsible for the financial decisions, the household’s logarithmized monthly income (*Income*_{h}, see, e.g. Tanaka *et al.*, 2010 for the influence of income on risk attitude), and its logarithmized total wealth in EUR (*TWealth*_{h}, see, e.g. Morin and Suarez, 1983; Barsky *et al.*, 1997; Oehler, 1998; Dorn and Huberman, 2005; Paravisini *et al.*, 2010; Oehler and Horn, 2016 for the influence of wealth on risk attitude). The regression model is set up as follows:

As measure of objective risk attitude, the percentage of wealth invested in risky assets is employed. However, due to their relatively low financial wealth (see, e.g. Taylor *et al.*, 2011), not all young adults are able to invest in risky financial assets. We therefore follow Oehler and Horn (2016) and compute the amount of money that is investable when all expenses for debts, insurances, and retirement provisions are taken into account for each household. This remaining amount of money and investments in the asset classes *Money market*, *Stocks*, *Bonds*, *Real estate funds*, and *Articles of great value*, always net of liabilities ascribed to these asset classes, are pooled in a so-called speculation portfolio. Portfolios with a net value of less than EUR 1,000[8] are not considered since the owners of such portfolios are hardly able to invest in risky assets. Since just being affluent enough to invest in financial assets or being able to spend money for consumption after having covered basic needs might influence young adults’ subjective risk attitude, we introduce an adaption of the first regression model to capture differences between young adult households with and without speculation portfolios. The presence of a speculation portfolio in household *h* is denoted by the dummy variable *SPortfolio*_{h} that is 1 if household *h* owns a speculation portfolio and 0 otherwise. We furthermore add interaction variables consisting of the first regression model’s independent variables times *SPortfolio*_{h}. The adapted regression model is set up as follows:

Young adults’ objective risk attitude (*RiskAtt*_{Obj,h}) is measured by their actual investments, i.e. the risky asset share of their speculation portfolio. The determinants of young adults’ objective risk attitude are analyzed with a regression model that includes the socio-demographic and economic variables of Model (2a) and the logarithmized net value of the young adults’ speculation portfolio (*ValueSP*_{h}, to control for differences in the value of speculation portfolios) as independent variables (Model (3a)). As risky assets we define stocks, bonds, real estate funds, and articles of great value as captured in the PHF-Survey. Furthermore, we examine whether young adults’ subjective risk attitude predicts their objective risk attitude, as indicated in the experimental setting by Nosic and Weber (2010) (Model (3b)). The full regression models are set up as follows:

In the second part of our cross-sectional analysis, we provide a comparison of the factors influencing young adults’ and older adults’ risk attitude. Given the obvious socio-demographic and economic differences between young adults’ and older adults’ households, we analyze if and how the subjective risk attitude of young adults and their older peers are differently influenced by their socio-demographics and economics. For this purpose, we use the complete data set of the PHF-survey and expand regression Model (2b) with the dummy variable *YoungAdult*_{h} (which is 1 for households with young adults as financial decision makers and 0 for the remaining households) and the interacted variables consisting of the regression model’s independent variables times *YoungAdult*_{h}. The full regression model is as follows:

In addition, we examine possible differences in the driving factors of young adults’ and the remaining adults’ objective risk attitude. We adjust regression Model (3b) and include *YoungAdult*_{h} and consequent interacted variables. The adapted regression model is:

## 3. Results

### 3.1 Findings from the experimental questionnaire design

Participants’ median age is 22 (mean: 22.4) years with a minimum of 19 and a maximum of 34 years (standard deviation: two years). Our data set includes more women (*n*=192) than men (*n*=148)[9]. Table I reports results on participants’ subjective risk attitude (*RiskAtt*_{Subj,j}). The mean value of all participants’ degree of subjective risk attitude is 2.6 (med: 2.5). Young adults’ responses range from the minimum (1) to the maximum(5) of the scale. The results indicate that men have a lower subjective risk aversion than women: the mean value for men is 3.1 (med: 3.0) and 2.3 (med: 2.0) for women. The difference between men and women regarding the mean and median values is statistically significant at the one percent level.

The participants’ answers in the questionnaire design are employed to analyze whether young adults’ objective risk attitude is constant over different stake sizes and to elicit the influence of young adults’ subjective risk attitude and gender on their choice in theoretical financial decisions. Table II contains the results regarding the young adults’ choices in the different lotteries. Results of the lottery with an expected value of EUR 5 are provided in Panel A. In total, 51 percent (*n*=175) of the young adults choose the certain payment instead of gambling, which is then again preferred by 37 percent (*n*=127). The remaining 40 participants (12 percent) are indifferent between the two choices. More women (60 percent; *n*=116) than men (40 percent; *n*=59) prefer the certain payment. Approximately the same amount of men and women are indifferent between the two choices. Consequently, men (47 percent; *n*=70) more strongly prefer gambling than women (29 percent; *n*=55). Panel B of Table II displays the results of the lottery with an expected value of EUR 50. Compared to the lottery with the lower expected value of EUR 5, a higher number of participants (74 percent; *n*=255) chooses the certain payment and less are willing to gamble (19 percent; *n*=65) or are indifferent (6 percent; *n*=22). This means that young adults become more risk averse when the stake of the financial decision increases. Again, less women (13 percent; *n*=25) than men (27 percent; *n*=40) prefer gambling and more women (83 percent; *n*=150) than men (62.4; *n*=93) choose the certain payment. The results of the lottery with an expected value of EUR 500 (in Panel C of Table II) reveal a further increase of participants’ objective risk aversion with increasing stake sizes. In total, 89 percent (*n*=305) of the young adults prefer the certain payment of EUR 500 compared to 8 percent (*n*=26) who like to gamble for EUR 1,000. The differences between men and women diminish. Only slightly less men (85 percent; *n*=127) than women (92 percent; *n*=176) prefer the certain payment.

We employ Wilcoxon tests to examine whether young adults’ risk attitude varies at statistical significant levels between the lotteries. We find that differences between young adults’ choices for the three lotteries being statistically significant at the 1 percent level. This finding indicates that young adults’ risk attitude is not constant over stake sizes. Since, in contrast, young adults’ wealth stays unchanged in this time, their risk attitude is independent of time-invariant wealth effects.

In general, the results of the lotteries reveal three main findings. First, the majority of young adults prefer the certain payment (indicating risk aversion). Second, young adults’ objective risk aversion increases with increasing stake sizes. And third, in all lotteries, men show a lower objective risk aversion than women.

We further provide an overview of participants’ migration regarding their decisions in the different lotteries in Figure 1. The 175 young adults who choose the certain payment in the first lottery (expected value: EUR 5) largely choose the certain payment of EUR 50 (*n*=162) and EUR 500 (*n*=157). Just a few young adults prefer the certain payment of EUR 5 and switch to the gambling option regarding the lotteries with an expected value of EUR 50 (*n*=11) and EUR 500 (*n*=2). Of the 127 participants who would gamble with an expected value of EUR 5, 72 participants switch to the certain payment in the lottery with an expected value of EUR 50 and just 51 participants further prefer gambling. Of these 51 subjects, 37 switch to the certain payment when the expected value rises to EUR 500. The participants, who are indifferent between both options in the first lottery, largely choose the certain payment in the second and third lottery. The analysis reveals that most young adults’ objective risk attitude becomes more risk averse when the decision’s absolute amount of money rises (or stays stable when young adults were already risk averse at the smallest absolute amount of money). Only a negligible percentage of young adults become less risk averse when the decision’s absolute amount of money rises.

The results of the regression analysis regarding the influence of young adults’ subjective risk attitude and their gender on the choice of the payment in the lottery using Model(1) are provided in Table III. For all lotteries and regression specifications, we find positive regression coefficients for *RiskAtt*_{Subj,j} with a statistical significance at the 1 percent level. This means that young adults with a lower subjective risk aversion are less likely to take the certain payment in the lottery. The influence of young adults’ gender is different between the respective lotteries. In the lotteries with EUR 5 and 50 as the expected value, we find *Gender*_{j} being statistically significant at least at the 10 percent level. The positive regression coefficients indicate that men choose less often the certain payment than women. Regarding the lottery with an expected value of EUR 500, the participants’ gender does not seem to have an influence on their choices. However, comparable to the influence of participants’ gender, also the predictive power of young adults’ subjective risk attitude regarding the choice of the lottery diminishes as the stake size of the lottery increases.

To sum up, the findings show that young adults’ objective risk attitude is not constant over stake sizes. Young adults who are prone to gamble in financial decisions with small absolute amounts become more risk averse when the financial decisions relate to higher absolute amounts. Additionally, men show a lower subjective and objective degree of risk aversion than women. Young adults with a higher subjective risk aversion also show, in general, a higher degree of objective risk aversion. However, the influence of young adults’ gender and subjective risk attitude on their objective risk attitude decreases with higher stake sizes.

### 3.2 Findings from the PHF-Survey

The data set of the PHF-Survey is employed to analyze the determining factors of young adults’ subjective and objective risk attitude regarding real-world financial decisions. Out of the 3,565 interviewed households, 535 have a young adult as financial decision maker of whom 271 are female and 264 are male. Descriptive statistics of young adults who are responsible for households’ financial decisions are provided in Table IV. The young adults have a mean (median) age of 28 (29) years. Their households generate a mean net monthly household income of EUR 2,309. The median net monthly income is EUR 1,900, indicating a right-skewed distribution of households’ income that emerges with a right-skewed-distribution of wealth. With EUR 123,888 the mean total wealth of young adults’ households clearly overcuts the median total wealth of EUR 10,000. This pattern is in accordance with findings of Badarinza *et al.* (2016).

We report results regarding the influence of young adults’ socio-demographics and economics on their subjective risk attitude in Table V. Out of the four independent variables *Gender*_{h}, *Age*_{h}, *Income*_{h} and *TWealth*_{h} from regression Model (2a) only gender and income have a statistically significant influence on young adults’ subjective risk attitude. The results are in line with the finding of our experimental setting that women show a higher degree of subjective risk aversion than men. Additionally, young adults’ subjective risk aversion decreases as their income rises[10].

Using regression Model (2b), we analyze whether the subjective risk attitude of young adults that are affluent enough to buy risky assets differs from young adults that are not able to make risky investments. The previous result that women have a higher subjective risk aversion than men is supported. In contrast, the influence of *Income*_{h} vanishes when young adults’ possibility to buy risky assets is considered. Instead, young adults’ age positively correlates with their degree of subjective risk aversion but only if they are able to invest in risky assets. However, having enough financial wealth to buy risky assets generally seems to decrease young adults’ subjective risk aversion. The regression coefficient of the speculation portfolio dummy is negative and statistically significant at the 5 percent level. These findings indicate that young adults’ subjective risk attitude depends on the absolute amount of money of their investment decisions. When young adults face investment decisions with a relatively high absolute amount of money (i.e. in the scope of investment decisions in their speculation portfolio), they try to avoid risks. The finding of our experimental setting, that young adults’ objective degree of risk aversion increases with stake size, can be interpreted in a similar way and supports this line of argument.

In Table VI, we present results regarding the question whether young adults’ objective risk attitude can be predicted by young adults’ socio-demographics and/or their subjective risk attitude. In general, the small adjusted *R*^{2}s (ranging from 0.008 to 0.016) of the individual socio-demographics used in regression Model (3a) convey that the socio-demographics have little explanatory power. In the light of the significant role of young adults’ gender for their subjective risk attitude, it, on the one hand, surprises that the gender does hardly influence their objective risk attitude. On the other hand, our experimental questionnaire design already indicated that the influence of young adults’ gender decreases with higher stake sizes. As a consequence, the only significant (at the one percent level) determinant in Model (3a) is the value of the speculation portfolio. The regression results indicate that the risky asset share of young adults’ portfolios rises with the value of their speculation portfolio. However, employing regression Model (3b) shows that young adults’ objective risk attitude is solely significantly influenced by young adults’ subjective risk attitude. Since the regression coefficient of 0.108 shows a positive relation between both variables, this finding, in turn, indicates that young adults have an idea of different asset classes’ risk and deploy this knowledge in investment decisions. The influence of the remaining variables diminishes when young adults’ subjective risk attitude is taken into account. This could imply a mediating effect of young adults’ subjective risk attitude on the influences of young adults’ gender and their speculation portfolio’s value on young adults’ objective risk attitude[11].

Taken together, our findings on the determining factors of young adults’ subjective and objective risk attitude support the conclusions of Nosic and Weber (2010) and Kaustia and Luotonen (2016) by showing that young adults’ subjective risk attitude serves as statistically significant predictor regarding their objective risk attitude in both theoretical (as in the experimental setting) and actual (as in the context of the PHF-survey) financial decisions. Young adults’ subjective risk aversion, in turn, most probably serves as mediator for the influence of young adults’ financial wealth that is investable in risky assets. The combined findings of the experimental design and the PHF-survey additionally show that young adults become more risk averse when the absolute value of a financial decision rises and that the influence of young adults’ gender decreases with higher stake sizes. The latter effect may also be the reason why a possible influence—which is probably observable for financial decisions with smaller stake sizes—of young adults’ gender on their objective risk attitude (also through young adults’ subjective risk attitude as mediator) is not evident in the field data.

### 3.3 Comparative analysis between young and old adults

Having identified the determinants of young adults’ risk attitude, we use the data set of the PHF-Survey to analyze whether the influence of these determinants has a different impact on young adults than their older peers. This research question is of interest since it seems possible that a habituation effect in context of financial decisions may appear when people get older. For example, people should, due to their income, get wealthier over time and, as a consequence, more frequently face investment decisions. One could assume that getting wealthier and gaining more experience regarding investment decisions reduces the fear of financial losses and consequently the degree of risk aversion with respect to financial decisions[12]. We, therefore, analyze whether young adults’ subjective and objective risk attitude are influenced differently by their socio-demographics and socio-economics compared to their older peers.

Descriptive statistics and the statistical significance of differences between socio-economics and demographics of young adults’ households and their older peers are reported in Table VII. While we find on average no statistically significant differences regarding young adults’ and older adults’ subjective risk attitude, households of young adults are clearly more limited regarding their available monthly income and their total wealth. The monthly income of young adults’ households is on average roughly EUR 900 lower and the total wealth EUR 230,000 lower compared to the remaining households. We consider this discrepancy in wealth and income as explanation why only 33 percent of the young adult households have enough financial wealth to invest in risky assets while 55 percent of the remaining households would be affluent enough to do so.

We provide results of the stepwise regression analysis using regression Model (4) regarding the influencing factors of households’ subjective risk attitude in Table VIII. Again, the pattern that male decision makers show a lower subjective risk aversion is statistically significant. Results regarding the age of the decision maker show differences between younger and older adults. In the full sample, subjective risk aversion increases statistical significantly with age. In contrast, the statistically insignificant coefficient of the interaction variable combining *Age*_{h} and the dummy variable *YoungAdult*_{h} yields nearly the same absolute value, which equalizes the influence of age for young adults. Therefore, the age of the household member responsible for the financial decisions only serves as determinant for the subjective risk attitude, when the household member is older than 35 years. This finding underlines the importance to analyze young adults as a separate age group. An increase in households’ income and/or total wealth generally goes hand in hand with a lower degree of subjective risk aversion. The interaction variables’ negative coefficients of the stepwise and the full model, however, show that this effect is significantly weaker pronounced for households with young adult decision makers. When the presence of enough financial wealth to invest in risky assets is solely addressed with a dummy variable, we cannot observe a statistical significant effect. The regression coefficient of the dummy variable *YoungAdult*_{h} is not statistically significant in the full regression model, but when the wealth measures are considered separately. We interpret this pattern as follows: young adults would generally show a lower degree of subjective risk aversion than older adults if the remaining socio-economics were similar. However, this initial discrepancy is hardly observable since older adults earn higher incomes and own more financial wealth which decreases their level of subjective risk aversion to the levels of the less wealthy young adults.

We focus on households that are able to invest in risky assets to answer the question whether young adults differ from their older peers regarding their objective risk attitude. Results of *t*-tests between young adults and older adults of this subsample are presented in Table IX. Even if young adults show a slightly lower (although not statistically significant) subjective risk aversion than older adults, older adults show a significantly lower degree of objective risk aversion[13]. Their portfolios, on average, consist of 24 percent risky assets, whereas portfolios of young adults only comprise 16 percent risky assets. As indicated by previous results, younger adults are significantly less affluent. The net value of their portfolios is on average EUR 23,000 which is only one-fifth compared to the average net value of the remaining adults’ portfolios. Furthermore, the observed young adult households have EUR 1,000 less monthly income and a EUR 320,000 smaller total wealth than their older peers.

The different preconditions of young adult households and their older peers raise the question whether both groups follow the same determinants to set the riskiness of their investment decisions. To assess the factors influencing households’ objective risk attitude regression analysis using Model (5) are employed. The results of the regression analysis are presented in Table X. Results of the stepwise approach reveal that the five variables: *Gender*_{h}, *Income*_{h}, *TWealth*_{h}, *ValueSP*_{h}, and *RiskAtt*_{Subj,h} provide explanatory power regarding the objective risk attitude. Among these five determinants, the predictive power measured by the adjusted *R*^{2} is clearly least pronounced for the influence of gender. Hence, the results are in line with our previous findings. As in our previous results, households’ objective risk aversion decreases with their degree of subjective risk aversion. This holds true for both young adults and older ones. In addition, the degree of objective risk aversion decreases with rising net value of the speculation portfolio, confirming the findings of Oehler (1998). Again, portfolios of young adults are no exception. Like portfolios of young adults, the remaining portfolios mirror a higher objective risk attitude when managed by a woman. An increase in *Income*_{h} or *TWealth*_{h} has a weaker effect on young adults’ objective risk attitude than on the remaining portfolios. In young adults’ portfolios, the share of risky assets has a higher initial level which young adults increase only hesitantly with raising income or wealth. In contrast, portfolios of older investors show a lower initial level of risky assets which the portfolio holders extend more rigorously when their income or wealth rises.

After combining all mentioned factors in the full regression model, the influence of *Gender*_{h} and *TWealth*_{h} disappears whereas *RiskAtt*_{Subj,h}, *ValueSP*_{h}, and *Income*_{h} stay significant at the 1 percent level. Tests on mediating effects, however, show that *RiskAtt*_{Subj,h} works as mediator for the influence of *Gender*_{h}, while *ValueSP*_{h} works as mediator for the influence of *TWealth*_{h}[14]. However, due to the low *R*^{2} of gender in the stepwise regression analysis, the finding of a possible mediating effect of *RiskAtt*_{Subj,h} regarding the influence of *Gender*_{h} needs to be interpreted with caution. The positive coefficient of the dummy variable *YoungAdult*_{h} indicates that young adults generally show a lower degree of objective risk aversion. However, young adults’ portfolios show on average smaller amounts of risky assets than the remaining portfolios because households of older investors are by far wealthier which drives them to take higher financial risks.

We sum up and interpret our findings as follows. Young adults do not generally show a lower degree of subjective risk aversion compared to older people. For both groups, the degrees of subjective risk aversion work as predictor for the degrees of objective risk aversion, indicating that young adults and their older peers likewise know the risk of different asset classes and furthermore deploy this knowledge in investment decisions. The latter implication supports survey evidence of Oehler (2012) that younger adults are neither less interested nor less caring about their personal finances than older adults. Even if young adults, on average, would probably show a higher degree of objective risk aversion than comparable older investors, we do hardly observe this constellation in practice. The reason is simple: young adults face stronger restrictions. They have a significantly lower income and far fewer savings to compensate financial losses than older investors. As a consequence, young adults invest less risky.

## 4. Discussion and conclusions

Our analysis derives the determinants of young adults’ subjective and objective risk attitude in theoretical and real-world financial decisions. The main results, which arise from both the experimental setting as well as in the cross-sectional data of the PHF-Survey, contribute to various strands of literature. First, young adults’ objective risk attitude is not constant over stake sizes. Instead their objective risk aversion increases with the absolute amount of money of a financial decision. The findings are generally in line with Markowitz (1952), Weber and Chapman (2005), and Fehr-Duda *et al.* (2010); however in contrast to Holt and Laury (2002, 2005) who do not find such an effect for hypothetical lotteries.

Second, young adults’ subjective risk attitude differs according to their gender, i.e. women show a higher subjective risk aversion than men. Hence, our results are consistent with Jianakoplos and Bernasek (1998), Sunden and Surette (1998), Hariharan *et al.* (2000), Barber and Odean (2001), Bernasek and Shwiff (2001), Felton *et al.* (2003), Eckel and Grossman (2008), Weber *et al.* (2013), Booth and Katic (2013), Oehler and Horn (2016) and Oehler *et al.* (2018a). However, the influence of young adults’ gender on their objective risk attitude decreases with higher stake sizes.

Third, young adults’ subjective risk attitude is a better predictor for their objective risk attitude than a set of often employed socio-demographics and economics like age or income. This finding is in line with Nosic and Weber (2010) as well as Dorn and Huberman (2005) who find that self-reported risk aversion is the most important determinant of portfolio diversification and turnover. More specifically, the results that young adults who report a lower degree of subjective risk aversion invest a higher amount of wealth in risky assets and prefer gambling instead of taking the certain payment is comparable to Dorn and Huberman’s findings that investors with a low degree of subjective risk aversion hold less diversified portfolios and trade more aggressively. Nevertheless, the significant influence of young adults’ subjective risk attitude, in turn, can be partially explained by a mediating effect on the influence of young adults’ financial wealth that is investable in risky assets on young adults’ objective risk attitude.

Forth, young adults and older adults generally show a similar degree of subjective risk aversion. For both groups, the degree of subjective risk aversion works as predictor for the degree of objective risk aversion. This effect may be partially rooted in the subjective risk attitude’s role as mediator for the influence of a gender effect on the objective risk attitude. However, due to the weak predictive power of investors’ gender in investment decisions with higher stake sizes, the latter finding needs to be interpreted with caution. Due to young adults’ stronger financial restrictions compared to older adults, i.e., significant lower income and far fewer savings, young adults show on average a higher degree of objective risk aversion. We find indications that young adults would show a lower degree of objective risk aversion than older adults if they were equally wealthy. These findings are in general in line with the lifecycle risk aversion hypothesis (e.g. Morin and Suarez, 1983; Bakshi and Chen, 1994; Palsson, 1996; Schooley and Worden, 1996; Ameriks and Zeldes, 2004) but in contrast to Bellante and Saba (1986).

The results provide implications for financial advisers, academics, and policy makers. When counseling young adults, financial advisors should take the gender dependence of young adults’ subjective risk aversion in account. In addition, financial advisors should be aware that young adults generally have less savings and income and are therefore stronger harmed by financial losses than older adults. This finding is also relevant for policy makers and academics. Previous findings which show that young adults borrow money at higher interest rates (e.g. Agarwal *et al.*, 2009), hold less diversified portfolios (e.g. Goetzmann and Kumar, 2008; Dorn and Huberman, 2005), and earn less investment incomes (e.g. Cole and Shastry 2009), may be (at least) partially explained by the financial limits of young adults. Since young adults’ objective risk attitude is in accordance with their subjective risk attitude, we follow the argumentation of Oehler (2012) and Oehler *et al.* (2018b) that the lower engagement of young adults in financial markets is not driven by, e.g., their allegedly low financial literacy (e.g., Lusardi *et al.*, 2010) but rather by their tight budget. When policy makers and academics elaborate on concepts to, e.g., enhance the engagement and performance of young adults in financial markets, they should be aware of young adults’ challenging economic situation as determining factor. If policy makers and academics tried to simultaneously affect the objective risk attitude of both young and older adults, without considering the financial restrictions of young adults, the interventions would most probably fail.

## Figures

Participants subjective risk attitude (*RiskAtt*_{Subj})

All | Men | Women | Significance | |
---|---|---|---|---|

n |
342 | 148 | 192 | |

Mean | 2.6 | 3.1 | 2.3 | *** |

Med | 2.5 | 3.0 | 2.0 | *** |

Max. | 5.0 | 5.0 | 5.0 | |

Min. | 1.0 | 1.0 | 1.0 | |

SD | 1.1 | 1.1 | 0.9 |

**Notes:** This table displays descriptive statistics of individuals who participate in the questionnaire study. We provide the number of individuals’ responses (*n*), mean value (Mean), median value (Med), maximum value (Max.), minimum value (Min.), and standard deviation (SD) for *RiskAtt*_{Subj}. In addition, we provide results of the *t*-test (mean) and U-test (median) as the test of equality between men and women. ***Statistically significant at the 1 percent level. Example: the median value of the male participants’ self-assessed risk attitude (*RiskAtt*_{Subj}) is 3.0

Participants’ choices in the three different lotteries

In % (absolute) | Men in % (absolute) | Women in % (absolute) | |
---|---|---|---|

Panel A: Lottery with expected value of EUR 5 |
|||

Certain payment | 51.0 (175) | 39.6 (59) | 60.4 (116) |

Gambling | 37.0 (127) | 47.0 (70) | 28.6 (55) |

Indifferent | 11.7 (40) | 12.8 (19) | 10.9 (21) |

Panel B: Lottery with expected value of EUR 50 |
|||

Certain payment | 74.3 (255) | 62.4 (93) | 83.3 (160) |

Gambling | 19.0 (65) | 26.8 (40) | 13.0 (25) |

Indifferent | 6.4 (22) | 10.1 (15) | 3.6 (7) |

Panel C: Lottery with expected value of EUR 500 |
|||

Certain Payment | 88.9 (305) | 85.2 (127) | 91.7 (176) |

Gambling | 7.6 (26) | 8.1 (12) | 7.3 (14) |

Indifferent | 3.2 (11) | 6.1 (9) | 1.0 (2) |

**Notes:** This table displays participants’ choices in the three different lotteries. We indicate the percentage of responses and the absolute frequency in parentheses for the certain payment, gambling and indifference, respectively. For each lottery, we report choices of all participants as well as separated for men and women. Panels A, B and C indicate the results for the lotteries with EUR 5, EUR 50 and EUR 500 as expected values, respectively. Example: 116 women (60.4 percent of all women) prefer the certain payment in the lottery with an expected value of EUR 5

Influence of subjects’ choice of lottery

Dependent variable: RiskAtt_{Obj,j}, i.e. participants’ choice (sure payment=1, indifferent=2, gamble=3) in the lottery i (expected value EUR 5, 50 and 500) |
|||
---|---|---|---|

Panel A: Lottery with expected value of EUR 5 |
|||

β_{0} |
1.119*** | 1.682*** | 1.109*** |

RiskAtt_{Subj,j} |
0.283*** | 0.254*** | |

Gender_{j} |
−0.392 | −0.177* | |

R^{2} (adj.) |
0.112 | 0.041 | 0.119 |

Panel B: Lottery with expected value of EUR 50 |
|||

β_{0} |
0.861*** | 1.297*** | 0.864*** |

RiskAtt_{Subj,j} |
0.222*** | 0.192*** | |

Gender_{j} |
−0.345*** | −0.183** | |

R^{2} (adj.) |
0.095 | 0.044 | 0.104 |

Panel B: Lottery with expected value of EUR 500 |
|||

β_{0} |
0.870*** | 1.156*** | 0.871*** |

RiskAtt_{Subj,j} |
0.120*** | 0.126*** | |

Gender_{j} |
−0.067 | 0.040 | |

R^{2} (adj.) |
0.056 | 0.001 | 0.054 |

**Notes:** We provide cross-sectional regression coefficients and adjusted *R*^{2} for the regression analysis using Equation (1) with the participants’ choice in the lottery *i* as the dependent variable. Panels A, B and C display the results for the lotteries with an expected value of EUR 5, 50 and 500, respectively. We employ univariate as well as multivariate regressions. The last column includes the results for the full regression model. Example: regressing *RiskAtt*_{Obj,j} in the lottery with an expected value of EUR 5 (Panel A) on participants’ subjective risk attitude (*RiskAtt*_{Subj,j}) as the sole independent variable yields a coefficient of 0.283 with statistical significance at the one percent level. *,**,***Statistically significant at the 10, 5 and 1 percent levels, respectively

Age distribution of young adults responsible for financial decision in households and households’ distribution of income and wealth

Age | Estimation of monthly household income | Estimation of total wealth | |
---|---|---|---|

Mean | 28 | 2,309 | 123,888 |

SD | 5 | 3,969 | 1,168,013 |

10th percentile | 21 | 650 | 0 |

20th percentile | 24 | 1,000 | 604 |

50th percentile | 29 | 1,900 | 10,000 |

80th percentile | 33 | 3,000 | 70,000 |

90th percentile | 34 | 3,940 | 200,000 |

**Notes:** This table displays descriptive statistics of the age, monthly income, and total wealth (both in EUR) of young adults’ households. We report mean and median values, the 10th, 20th, 80th and 90th percentile and the standard deviation (SD). Example: the mean value of households’ total wealth is 123,888 EUR with a standard deviation of EUR 11,168,013. The 20th percentile is EUR 604 and the median value is EUR 10,000. In total, 20 percent of the households have a higher total wealth than EUR 70,000 (80th percentile)

Influence of 317 young adults’ households’ characteristics on their subjective risk attitude

Panel A: Model 2a |
||||||

β_{0} |
1.803*** | 1.458*** | −0.140 | 1.058*** | 0.311 | |

Gender |
−0.215*** | −0.167** | ||||

Age |
0.001 | 0.001 | ||||

Income |
0.207*** | 0.155** | ||||

Twealth |
0.043*** | 0.020 | ||||

R^{2} |
0.038 | 0.000 | 0.043 | 0.027 | 0.068 | |

R^{2} (adj.) |
0.035 | −0.003 | 0.040 | 0.023 | 0.054 | |

F-test |
12.311 | 0.006 | 14.021 | 7.554 | 4.913 | |

VIF (highest value among all independent variables) | 1.420 | |||||

Panel B: Model 2b |
||||||

β_{0} |
1.853*** | 1.792*** | 0.385 | 1.206*** | 0.294 | 1.472** |

Gender |
−0.279*** | −0.165** | −0.223*** | |||

Gender×SPortfolio |
0.205 | 0.127 | ||||

Age |
−0.013 | 0.000 | −0.016* | |||

Age×SPortfolio |
0.032** | 0.034** | ||||

Income |
0.134** | 0.168** | 0.081 | |||

Income×SPortfolio |
0.184 | 0.173 | ||||

Twealth |
0.024 | 0.007 | 0.012 | |||

Twealth×SPortfolio |
0.043 | 0.020 | ||||

SPortfolio |
−0.168 | −0.750* | −1.336 | −0.369 | 0.116 | −2.646** |

R^{2} |
0.060 | 0.034 | 0.059 | 0.037 | 0.076 | 0.115 |

R^{2} (adj.) |
0.051 | 0.025 | 0.050 | 0.026 | 0.059 | 0.085 |

F-test |
6.678 | 3.674 | 6.493 | 3.490 | 4.459 | 3.826 |

VIF (highest value among all independent variables) | 1.669 |

**Notes**: We provide regression coefficients, *R*², adjusted *R*², and *F*-statistics for the stepwise regression analysis using Equation (2a) in Panel A and for the analyses using Equation (2b) in Panel B with households’ subjective risk attitude as dependent variable. Example: Regressing the households’ subjective risk attitude on the full model of Equation (2a) yields a coefficient of households’ monthly income of 0.155 with a statistical significance at the five percent level and an adjusted *R*^{2} of 0.054. *,**,***Statistically significant at the 10, 5 and 1 percent levels, respectively

Influence of 108 young adults households’ characteristics on households’ objective risk attitude

Model (3a) | Model (3b) | |||||||
---|---|---|---|---|---|---|---|---|

β_{0} |
0.269*** | −0.018 | −0.107 | −0.121 | −0.476*** | −0.442 | −0.077 | −0.172 |

Gender |
−0.076* | −0.070* | −0.057 | |||||

Age |
0.006 | 0.004 | −0.001 | |||||

Income |
0.035 | 0.006 | −0.021 | |||||

Twealth |
0.026** | −0.005 | 0.004 | |||||

ValueSP |
0.067*** | 0.065*** | 0.039 | |||||

RiskAtt_{Subj,h} |
0.133*** | 0.108** | ||||||

R² |
0.021 | 0.014 | 0.008 | 0.025 | 0.093 | 0.111 | 0.084 | 0.117 |

R^{2} (adj.) |
0.016 | 0.008 | 0.002 | 0.019 | 0.088 | 0.085 | 0.075 | 0.063 |

F-test |
3.821 | 2.487 | 1.413 | 4.395 | 18.112 | 4.235 | 9.699 | 2.152 |

VIF (highest value among all independent variables) | 1.960 | 1.662 |

**Notes:** We provide regression coefficients, *R*², adjusted *R*², and *F*-statistics for the stepwise regression analysis using Equations (3a) and (3b) with households’ objective risk attitude as the dependent variable. Example: regressing the households’ objective risk attitude on the full model of Equation (3a) yields a coefficient of households’ monthly income of 0.006 with no statistical significance and an adjusted *R*^{2} of 0.085. *,**,***Statistically significant at the 10, 5, and 1 percent levels, respectively

*t*-tests between households of young adults and older adults (full sample)

Young adults | Older adults | Sig. | |
---|---|---|---|

RiskAtt_{Subj} |
1.47 | 1.44 | 0.381 |

Age |
28 | 59 | 0.000 |

Income |
2,310 | 3,214 | 0.000 |

Twealth |
123,888 | 355,625 | 0.000 |

Share of households with speculation portfolio | 0.33 | 0.55 | 0.000 |

n |
317 | 2,256 |

**Notes:** This table displays the mean values of 317 young adults’ and 2,256 older adults’ subjective risk attitude (*RiskAtt*_{Subj}), age, monthly households income (income), total wealth (*TWealth*), and the share of households that own a speculation portfolio, i.e. are able to invest in risky assets. In addition, we provide *p*-values (sig.) of the *t*-test regarding the differences between the mean values of young adults’ households and older adults’ households

Influence of households’ characteristics on their subjective risk attitude

β_{0} |
1.568*** | 1.723*** | −1.121*** | 0.470*** | −0.330* | −0.442** |

Gender |
−0.121*** | −0.112*** | −0.104*** | |||

Gender×YoungAdult |
−0.087 | −0.061 | ||||

Age |
−0.006*** | −0.005*** | −0.006*** | |||

Age×YoungAdult |
0.007 | 0.006 | ||||

Income |
0.316*** | 0.217*** | 0.221*** | |||

Income×YoungAdult |
−0.127** | −0.053 | ||||

Twealth |
0.084*** | 0.042*** | 0.049*** | |||

Twealth×YoungAdult |
−0.050*** | −0.041** | ||||

SPortfolio |
0.082*** | 0.135*** | 0.049** | −0.015 | 0.028 | 0.018 |

SPortfolio×YoungAdult |
0.061 | 0.018 | 0.065 | 0.113 | 0.098 | |

YoungAdult |
0.177 | −0.316 | 1.079** | 0.643*** | 0.010 | 0.736 |

R² |
0.023 | 0.025 | 0.105 | 0.064 | 0.131 | 0.134 |

R^{2} (adj.) |
0.022 | 0.024 | 0.103 | 0.062 | 0.129 | 0.130 |

F-test |
12.318 | 13.398 | 60.369 | 33.234 | 61.006 | 34.235 |

VIF (highest value among all independent variables) | 1.880 |

**Notes:** We provide regression coefficients, *R*², adjusted *R*², and *F*-statistics for the stepwise regression analysis using Equation (4) with households’ subjective risk attitude as the dependent variable. Example: regressing the households’ subjective risk attitude on the full model of Equation (4) yields a coefficient of households’ monthly income of 0.221 with a statistical significance at the 1 percent level and an adjusted *R*² of 0.130. *,**,***Statistically significant at the 10, 5 and 1 percent levels, respectively

*t*-tests between households of young adults and older adults (*SpeculationPortfolio* sample)

Young adults | Older adults | Sig. | |
---|---|---|---|

RiskAtt_{Obj} |
0.16 | 0.24 | 0.000 |

RiskAtt_{Subj} |
1.57 | 1.49 | 0.114 |

ValueSP |
23,524 | 118,366 | 0.000 |

Age |
28 | 62 | 0.000 |

Income |
2,477 | 3,476 | 0.000 |

Twealth |
135,103 | 455,810 | 0.000 |

n |
108 | 1,288 |

**Notes:** We provide the mean values of 108 young adults’ and 1,288 older adults’ objective (*RiskAtt*_{Obj}) and subjective (*RiskAtt*_{Subj}) risk attitude, the net value of their speculation portfolio (*ValueSP*), age, monthly household income (*Income*), and total wealth (*TWealth*). In addition, we provide *p*-values (sig.) of the *t*-test regarding the differences between the mean values of young adults’ households and older adults’ households

Influence of households’ characteristics on their objective risk attitude

β_{0} |
0.318*** | 0.252*** | −1.019*** | −.522*** | −0.771*** | −0.054** | −1.138*** | −1.235*** |

Gender |
−0.058*** | −0.004 | 0.004 | |||||

Gender×YoungAdult |
−0.018 | −0.061 | ||||||

Age |
0.000 | 0.001 | 0.001 | |||||

Age×YoungAdult |
0.006 | −0.002 | ||||||

Income |
0.158*** | 0.049*** | 0.057*** | |||||

Income×YoungAdult |
−0.123*** | −0.078 | ||||||

Twealth |
0.063*** | 0.009 | 0.010 | |||||

Twealth×YoungAdult |
−0.037** | −0.005 | ||||||

ValueSP |
0.095*** | 0.060*** | 0.060*** | |||||

ValueSP×YoungAdult |
−0.028 | −0.022 | ||||||

RiskAtt_{Subj,h} |
0.199*** | 0.123*** | 0.126*** | |||||

RiskAtt_{Subj,h}×YoungAdult |
−0.066 | −0.018 | ||||||

YoungAdult |
−0.049 | −0.270** | 0.912*** | 0.401** | 0.294* | −0.023 | −0.012 | 1.063** |

R² |
0.015 | 0.007 | 0.095 | 0.094 | 202 | 0.138 | 0.257 | 0.261 |

R^{2} (adj.) |
0.014 | 0.006 | 0.093 | 0.092 | 0.201 | 0.136 | 0.253 | 0.254 |

F-test |
9.584 | 4.442 | 64.184 | 62.602 | 155.804 | 73.981 | 67.582 | 37.040 |

VIF (highest value among all independent variables) | 1.844 |

**Notes:** We provide regression coefficients, *R*², adjusted *R*², and *F*-statistics for the stepwise regression analysis using Equation (5) with the households’ objective risk attitude as the dependent variable. Example: regressing the households’ objective risk attitude on the full model of Equation (5) yields a coefficient of households’ monthly income of 0.057 with a statistical significance at the 1 percent level and an adjusted *R*² of 0.254. *,**,***Statistically significant at the 10, 5 and 1 percent levels, respectively

## Notes

We follow the common approach to define young adults by their age in the range between 18 and 35 years (see e.g. Brown and Taylor, 2011; Albert and Duffy, 2012). A further reaching definition of young adults could capture their income situation, e.g. persons who recently started their working life.

See Schoemaker (1993) for an overview of risk attitude concepts in different domains and from different research perspectives.

See Nosic and Weber (2010) for the terms “objective” and “subjective” risk attitude. Instead of the term “objective risk attitude,” other studies also use the terms “risk taking” (Schooley and Worden, 1996), “observed risk taking” (Schoemaker, 1993), “risk tolerance” (Wang and Hanna, 1997), or “relative risk aversion” (Riley and Chow, 1992) for measures relying on outcomes of financial decisions. Instead of the term “subjective risk attitude,” Kaustia and Luotonen (2016) also use the term “directly queried financial risk aversion” and Schoemaker (1993) uses the term “intrinsic risk attitude” for the measure that asks individuals about their willingness to take financial risk.

The relation between wealth and risk attitude is primarily discussed in the literature building on the concept of investors’ relative risk aversion independently developed by Pratt (1964) and Arrow (1965). See, e.g., Arrow (1971), Friend and Blume (1975), Cohn *et al.* (1975), Siegel and Hoban (1982), Morin and Suarez (1983), Oehler (1998) and Paya and Wang (2016).

Since age showed a very tight distribution among participants (standard deviation of two years) we do not consider age in the further analysis of the experimental questionnaire design.

Nosic and Weber (2010) additionally use a lottery design which measures objective risk attitude based on the risk-return framework. In contrast to Nosic and Weber (2010), we abstain from eliciting individuals’ certainty equivalents and apply the design of Deutsche Bundesbank (2015) as an easier understandable measure which is sufficient to assess whether young adults’ risk attitude is independent of time-invariant wealth effects.

See also von Gaudecker (2015) for this threshold.

Two participants did not answer the question.

We control for possible multicollinearity in all regression models without interacted variables in our analyses by measuring the variance inflation factor (VIF). In all regressions, the VIFs are much lower than 10. We, therefore, conclude that multicollinearity seems to not influence our results (see, e.g. Hair *et al.*, 2014).

Sobel (1982) and Goodman(1960) tests including the value of young adults’ speculation portfolio as independent variable, young adults’ objective risk attitude as dependent variable, and young adults’ subjective risk attitude as mediator reveal *t*-values of at least 2.35 indicating (considering the sample size of <1,000) that the indirect effect of the value of young adults’ speculation portfolio on young adults’ objective risk attitude is statistically significant at least at the 5 percent level. The weak influence of gender on objective risk attitude does hardly fulfill the requirements for a suitable test regarding a possible mediating effect.

However, if investors mainly experienced low stock returns, e.g., through investments during financial crises, their willingness to participate in stock markets may be impaired (see Malmendier and Nagel, 2011).

Please note that the measure of objective risk attitude needs to be interpreted inversely since a higher percentage of risky assets in the portfolio leads to a higher value of *RiskAtt*_{Obj,h}.

Both with a statistical significance at the 1 percent level. *t*-values of Sobel (1982) and Goodman-Tests (1960) are at least 3.25 for the mediating effect of *RiskAtt*_{Subj,h} regarding the influence of *Gender*_{h} and 15.24 for the mediating effect of *ValueSP*_{h} regarding the influence of *TWealth*_{h}. Results of the test are not reported in detail.

## References

Agarwal, S., Driscoll, J.C., Gabaix, X. and Laibson, D. (2009), “The age of reason: financial decisions over the life cycle and implications for regulation”, Brookings Papers on Economic Activity, Vol. 40 No. 2, pp. 51-117.

Albert, S. and Duffy, J. (2012), “Differences in risk aversion between young and older adults”, Neuroscience and Neuroeconomics, Vol. 2012 No. 1, pp. 3-9.

Ameriks, J. and Zeldes, S.P. (2004), “How do household portfolio shares vary with age”, working paper, Columbia University, New York, NY.

Arrow, K. (1965), Aspects of the Theory of Risk-Bearing, Yrjö Jahnssonin Säätiö, Helsinki.

Arrow, K. (1971), Essays in the Theory of Risk-Bearing, North-Holland Publishing, Amsterdam.

Badarinza, C., Campbell, J. and Ramadorai, T. (2016), “International comparative household finance”, Annual Review of Economics, Vol. 8, pp. 111-144.

Bakshi, G.S. and Chen, Z. (1994), “Baby boom, population aging, and capital markets”, Journal of Business, Vol. 67 No. 2, pp. 165-202.

Barber, B.M. and Odean, T. (2001), “Boys will be boys: gender, overconfidence, and common stock investment”, Quarterly Journal of Economics, Vol. 116 No. 1, pp. 261-292.

Barsky, R., Juster, F.T., Kimball, M. and Shapiro, M.D. (1997), “Preference parameters and behavioral heterogeneity: an experimental approach in the health and retirement study”, Quarterly Journal of Economics, Vol. 112 No. 2, pp. 537-579.

Bellante, D. and Saba, R.P. (1986), “Human capital and life-cycle effects on risk aversion”, Journal of Financial Research, Vol. 9 No. 1, pp. 41-51.

Bernasek, A. and Shwiff, S. (2001), “Gender, risk and retirement”, Journal of Economic Issues, Vol. 35 No. 2, pp. 345-356.

Booth, A.L. and Katic, P. (2013), “Cognitive skills, gender and risk preferences”, Economic Record, Vol. 89 No. 284, pp. 19-30.

Brown, S. and Taylor, K. (2011), “Household finances and the ‘Big Five’ personality traits”, Discussion Paper No. 6191, IZA, Bonn.

Chaulk, B., Johnson, P. and Bulcroft, R. (2003), “Effects of marriage and children on financial risk tolerance: a synthesis of family development and prospect theory”, Journal of Family and Economic Issues, Vol. 24 No. 3, pp. 257-279.

Cohn, R., Lewellen, W., Lease, R. and Schlarbaum, G. (1975), “Individual investor risk aversion and investment portfolio composition”, Journal of Finance, Vol. 30 No. 2, pp. 605-620.

Cole, S. and Shastry, G.K. (2009), “Smart money: the effect of education, cognitive ability, and financial literacy on financial market participation”, Working Paper No. 09-071, Harvard Business School, Boston, MA.

Deutsche Bundesbank (2015), “Zahlungsverhalten in Deutschland 2014”, Dritte Studie über die Verwendung von Bargeld und unbaren Zahlungsinstrumenten, Frankfurt am Main.

Dohmen, T., Falk, A., Huffman, D. and Sunde, U. (2010), “Are risk aversion and impatience related to cognitive ability?”, American Economic Review, Vol. 100 No. 3, pp. 1238-1260.

Dohmen, T., Falk, A., Huffman, D., Schupp, J., Sunde, U. and Wagner, G.G. (2011), “Individual risk attitudes: measurement, determinants, and behavioral consequences”, Journal of the European Economic Association, Vol. 9 No. 3, pp. 522-550.

Dorn, D. and Huberman, G. (2005), “Talk and action: what individual investors say and what they do”, Review of Finance, Vol. 9 No. 4, pp. 437-481.

Eccles, D.W., Ward, P., Goldsmith, E. and Arsal, G. (2013), “The relationship between retirement wealth and householders’ lifetime personal financial and investing behaviors”, Journal of Consumer Affairs, Vol. 47 No. 3, pp. 432-464.

Eckel, C.C. and Grossman, P.J. (2008), “Men, women and risk aversion: experimental evidence”, in Plott, C.R. and Smith, V.L. (Ed.), Handbook of Experimental Economics Results, Elsevier, North-Holland, Amsterdam, pp. 1061-1073.

Fama, E. and Schwert, G. (1977), “Human capital and market equilibrium”, Journal of Financial Economics, Vol. 4 No. 1, pp. 95-125.

Fehr-Duda, H., Bruhin, A., Epper, T. and Schubert, R. (2010), “Rationality on the rise: why relative risk aversion increases with stake size”, Journal of Risk and Uncertainty, Vol. 40 No. 2, pp. 147-180.

Fellner, G. and Maciejovsky, B. (2007), “Risk attitude and market behavior: evidence from experimental asset markets”, Journal of Economic Psychology, Vol. 28 No. 3, pp. 338-350.

Felton, J., Gibson, B. and Sanbonmatsu, D.M. (2003), “Preference for risk in investing as a function of trait optimism and gender”, Journal of Behavioral Finance, Vol. 4 No. 1, pp. 33-40.

Friend, I. and Blume, M.E. (1975), “The demand for risky assets”, American Economic Review, Vol. 65 No. 5, pp. 900-922.

Goetzmann, W.A. and Kumar, A. (2008), “Equity portfolio diversification”, Review of Finance, Vol. 12 No. 3, pp. 433-463.

Goodman, L. (1960), “On the exact variance of products”, Journal of the American Statistical Association, Vol. 55 No. 292, pp. 708-713.

Hair, J.F. Jr, Black, W.C., Babin, B.J. and Anderson, R.E. (2014), Multivariate Data Analysis, 7th ed., Pearson, Harlow.

Halko, M.-L., Kaustia, M. and Alanko, E. (2012), “The gender effect in risky asset holdings”, Journal of Economic Behavior and Organization, Vol. 83 No. 1, pp. 66-81.

Hariharan, G., Chapman, K.S. and Domian, D.L. (2000), “Risk tolerance and asset allocation for investors nearing retirement”, Financial Services Review, Vol. 9 No. 2, pp. 159-170.

Hirshleifer, D. (2015), “Behavioral finance”, Annual Review of Financial Economics, Vol. 7 No. 1, pp. 133-159.

Holt, C.A. and Laury, S.K. (2002), “Risk aversion and incentive effects”, American Economic Review, Vol. 92 No. 5, pp. 1644-1655.

Holt, C.A. and Laury, S.K. (2005), “Risk aversion and incentive effects: new data without order effects”, American Economic Review, Vol. 95 No. 3, pp. 902-904.

Jianakoplos, N. and Bernasek, A. (1998), “Are women more risk averse*?*”, Economic Inquiry, Vol. 36 No. 4, pp. 620-630.

Kaustia, M. and Luotonen, N. (2016), “What drives the heterogeneity in portfolio choice? The role of institutional, traditional, and behavioral factors”, Conference Paper at 2nd Research in Behavioral Conference, Amsterdam, available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2845963 (accessed June 18, 2018).

Lusardi, A., Mitchell, O.S. and Curto, V. (2010), “Financial literacy among the young”, Journal of Consumer Affairs, Vol. 44 No. 2, pp. 358-380.

Malmendier, U. and Nagel, S. (2011), “Depression babies: do macroeconomic experiences affect risk taking?”, Quarterly Journal of Economics, Vol. 126 No. 1, pp. 373-416.

Markowitz, H. (1952), “The utility of wealth”, Journal of Political Economy, Vol. 60 No. 2, pp. 151-158.

Morin, R.A. and Suarez, A.F. (1983), “Risk aversion revisited”, Journal of Finance, Vol. 38 No. 4, pp. 1201-1216.

Nosic, A. and Weber, M. (2010), “How risky do I invest: the role of risk attitudes, risk perceptions and overconfidence”, Decision Analysis, Vol. 7 No. 3, pp. 282-301.

Oehler, A. (1995), Explaining the Behavior of Retail Investors – Theoretical Approach and Empirical Analyses (Die Erklärung des Verhaltens privater Anleger – Theoretischer Ansatz und empirische Analysen), Betriebswirtschaftliche Abhandlungen, Neue Folge Bd. 100, Poeschel-Verlag, Stuttgart.

Oehler, A. (1998), “Decreasing or increasing relative risk aversion? (Abnehmende oder zunehmende relative risikoaversion?)”, ZBB, Vol. 10 No. 4, pp. 230-236.

Oehler, A. (2012), Ohne Moos nix los—Wie junge Menschen über Geld und Finanzen denken/ Typen für die Sendung, Studie zum Finanzwissen junger Erwachsener, WDR, Köln.

Oehler, A. and Horn, M. (2016), “Does households’ wealth predict the efficiency of their asset mix? Empirical evidence”, working paper, Bamberg.

Oehler, A., Wendt, S., Wedlich, F. and Horn, M. (2018a), “Investors’ personality influences investment decisions: experimental evidence on extraversion and neuroticism”, Journal of Behavioral Finance, Vol. 19 No. 1, pp. 30-48.

Oehler, A., Horn, M., Wendt, S., Reisch, L. and Walker, T. (2018b), “Young adults and their finances: an international comparative study on applied financial literacy”, Economic Notes, Vol. 47 Nos 2-3, pp. 305-330.

Palsson, A.-M. (1996), “Does the degree of relative risk aversion vary with household characteristics?”, Journal of Economic Psychology, Vol. 17 No. 6, pp. 771-787.

Paravisini, D., Rappoport, V. and Ravina, E. (2010), “Risk aversion and wealth: evidence from person-to-person lending portfolios”, Working Paper No. 16063, Cambridge.

Paya, I. and Wang, P. (2016), “Wealth fluctuations and investment in risky assets: the UK micro evidence on households’ asset allocation”, Journal of Empirical Finance, Vol. 38 Part A, pp. 221-235.

Pratt, J. (1964), “Risk aversion in the small and the large”, Econometrica, Vol. 32 Nos 1/2, pp. 122-136.

Riley, W.B. Jr and Chow, K.V. (1992), “Asset allocation and individual risk aversion”, Financial Analyst Journal, Vol. 48 No. 6, pp. 32-37.

Schoemaker, P. (1993), “Determinants of risk-taking: behavioral and economic views”, Journal of Risk and Uncertainty, Vol. 6 No. 1, pp. 49-73.

Schooley, D.K. and Worden, D.D. (1996), “Risk aversion measures: comparing attitudes and asset allocation”, Financial Service Review, Vol. 5 No. 2, pp. 87-99.

Siegel, F. and Hoban, J. (1982), “Relative risk aversion revisited”, The Review of Economics and Statistics, Vol. 64 No. 3, pp. 481-487.

Sobel, M.E. (1982), “Asymptotic confidence intervals for indirect effects in structural equation models”, in Leinhardt, S. (Ed.), Sociological Methodology, American Sociological Association, Washington, DC, pp. 290-312.

Sunden, A.E. and Surette, B.J. (1998), “Gender differences in the allocation of assets in retirement savings plans”, AEA Papers and Proceedings, Vol. 88 No. 2, pp. 207-211.

Tanaka, T., Camerer, C. and Nguyen, Q. (2010), “Risk and time preferences: linking experimental and household survey data from Vietnam”, American Economic Review, Vol. 100 No. 1, pp. 557-571.

Taylor, P., Fry, R., Cohn, D., Livingston, G., Kochhar, R., Motel, S. and Patten, E. (2011), The Old Prosper Relative to the Young. The Rising Age Gap in Economic Well-Being, PewResearchCenter, Washington, DC.

Von Gaudecker, H.M. (2015), “How does household portfolio diversification vary with financial literacy and financial advice*?*”, Journal of Finance, Vol. 70 No. 2, pp. 489-506.

Wang, H. and Hanna, S. (1997), “Does risk tolerance decrease with age?”, Financial Counseling and Planning, Vol. 8 No. 2, pp. 27-31.

Weber, B.J. and Chapman, G.B. (2005), “Playing for peanuts: why is risk seeking more common for low-stakes gambles?”, Organizational Behavior and Human Decision Processes, Vol. 97 No. 1, pp. 31-46.

Weber, M., Weber, E.U. and Nosic, A. (2013), “Who takes risks when and why: determinants of changes in investor risk taking”, Review of Finance, Vol. 17 No. 3, pp. 847-883.

Xu, Y., Beller, A., Roberts, B. and Brown, J. (2015), “Personality and young adult financial distress”, Journal of Economic Psychology, Vol. 51, December, pp. 90-100.