Demographics and Life Financial Outcomes

A questionnaire was administered to assess the age, marital and occupational status, level of income, number of years of education, and ethnicity of the subjects, as well as their assets and debt. Household income was measured using a scale from 1 to 12, where 1 represented “less than $15,000” and 12 represented “higher than $500,000”. Assets were assessed with the question “What are your approximate current assets? (i.e., home value, bank accounts, investments, belongings)” using a 16-category ordinal response scale ranging from <$500.00 in the lowest category to >$1,500,000.00 in the highest. Debt was assessed with the question “What are your approximate current debts? (i.e., outstanding home loans, outstanding car loans, outstanding student loans, credit card debt, medical debt)” using a 16 category ordinal response scale ranging from <$500.00 in the lowest category to >$1,500,000.00 in the highest.

Cognitive Ability and Numeracy

Subjects were also administered standard tests of cognitive ability. The Trail Making Test [14] assessed cognitive flexibility. The test has two parts (A & B) which are both timed until completion. The first part (Trails A) requires that subjects sequentially connect 25 encircled numbers (1, 2, 3, etc) that are randomly arranged on a sheet of paper. The second part (Trails B) requires that subjects connect a series of numbers and letters in an alternating pattern (1, A, 2, B, 3, C, etc.) that are randomly arranged on a sheet of paper. The score on this test is calculated as the difference between the time taken to complete Trails B versus Trails A, and indicates how easily individuals can alternate or switch between different activities. Higher scores correspond to less cognitive flexibility.

The Letter-Number Sequencing subtest from the Wechsler Adult Intelligence Scale [15] assessed memory capacity. An experimenter verbally listed a series of randomly ordered numbers and letters to the subject (e.g., C, 1, A, 6, 2) and asked the subject to repeat the series back in with the numbers listed first, in numerical order, followed by the letters in alphabetical order (e.g., 1, 2, 6, A, C). Performance on this test requires that subjects both maintain the series of randomly ordered numbers and letters in short term memory and manipulate the stored information by sorting the representation in memory before verbally repeating it. Researchers refer to this combination of short-term memory maintenance and manipulation as “working memory”. This measure of working memory correlates well with general intelligence.

A numeracy inventory assessed quantitative skills with basic number problems [16]. This 11-item measure contains questions such as: “The chance of getting a viral infection is.0005. Out of 10,000 people, about how many of them are expected to get infected?”. All of the questions are focused on computing probabilities or proportions. The measure is considered to index an individual’s ability to accurately compute numerical information about risk.

Financial Choices

To assess subjects’ willingness to take financial risk, we asked: “If you suddenly received $10,000, how much would you allocate to each of the following? (out of 100%): a. equities: – % (includes stocks, mutual funds, or other equity components); b. bonds: – % (includes government bonds, municipal or corporate bonds, bond mutual funds, or other fixed-income components), c. cash: –% (includes money market accounts).” To measure subjects’ involvement in their finances, we asked the question: “How much experience do you have with investing?”. The possible answers were:(1) “I have had a savings account, but no other investments”; (2) “I have had investments other than a savings account (e.g., stocks, bonds, or mutual funds), but I do not tend to make my own decisions about those investments”; (3) “I actively make decisions about investing my money (e.g., in stocks, bonds, and other types of investments).”

Beliefs

To index subjects’ beliefs about the risk of investing in equities, we asked: “Rate how risky you perceive this activity to be: Investing 5% of your annual income in a very risky stock”. The answer was an integer from 1 (“not at all risky”) to 7 (“extremely risky”).

Socioeconomic Status

A subjective assessment of subjects’ socioeconomic status (SES) indexed their perceived standing in society and associated perceived control over life outcomes, since prior research suggests that individuals with low subjective SES perceive the world as less controllable [17]. Subjects saw a picture of a ladder along with instructions that read: “Think of this ladder as representing where people stand in the United States. At the top of the ladder are the people who are best off - those who have the most money, the most education and the most respected jobs. At the bottom are the people who are the worst off who have the least money, least education, and the least respected job or no job. The higher up you are on this ladder, the closer you are to the people at the top; the lower you are, the closer you are to the people at the bottom. Where would you place yourself on this ladder? Place an X on the rung where you think you stand at this time of your life relative to other people in the United States.” For local status, the instructions read: “Think of this ladder as showing where people stand in their communities. People define community in different ways. Please define it in whatever way is most meaningful to you. At the top of the ladder are the people who have the highest standing in their community. At the bottom are the people who have the lowest standing in their community. Where would you place yourself on this ladder? Place an X on the rung where you think you stand at this time of your life relative to other people in your community.” The answers can range from 1 to 10, where 1 represent the lowest rung on the ladder, while 10 represents the highest.

Credit Report Data

We obtained credit reports for a subsample of 31 subjects, from which we extracted subjects’ overall FICO credit scores, and calculated the available credit amount and the percent of credit used as objective proxies of assets and debt, respectively. The correlations between these objective measures (using log values) and the subjective self-reported assets and debt were significant and robust (r = 0.56 for assets, and r = 0.85 for debt, ), supporting the validity of the self-reported information.

Financial Learning Task

This task indexed subjects’ ability to learn from financial gains or losses [18]. Gain and loss learning were separately assessed to account for the possibility that depending on subjects’ 5-HTTLPR genotype, they may differentially attend to, encode, or retrieve gain or loss information. Subjects made 24 choices between two risky assets in both gain and loss conditions. In each condition, the two assets were represented by a pair of abstract symbols. After choosing one of the assets from each pair, subjects saw the outcome associated with their choice. On average, one of the assets yielded a better outcome, while the other yielded a worse outcome. Specifically, in the gain condition the better asset had a 66% probability of yielding a $1 dividend and a 33% probability of yielding a $0 dividend. These probabilities were reversed for the worse asset (i.e., the $1 dividend had only a 33% chance of being obtained). In loss condition, the better asset had a 66% probability of yielding a $0 dividend and a 33% probability of yielding -$1.00, and outcome probabilities were reversed for the worse asset. Within each pair, assets appeared randomly and with equal frequency on the left or right side of the screen. Asset pairings with better or worse outcomes were randomly assigned by the computer at the start of the experiment and counterbalanced across subjects. Subjects were explicitly informed about the dividend distributions and instructed to try to maximize their earnings throughout the experiment by choosing the assets they believed to be the better ones. To quantify how well subjects learn from financial information, we calculated the fraction of trials in the gain condition and in the loss condition when subjects made the correct Bayesian choice, conditional on the information set available at the time. For this calculation, we excluded trials in which both options had equal chances of being optimal (including the first trial). For incentive compatibility, dividends accumulated during this task determined subjects’ payment.

Affective Traits

Although a great deal of research has focused on the influence of transient emotional states (i.e., current levels of excitement or anxiety) on decision making [19], [20], relatively less research has focused on chronic emotional traits (i.e., how excited or anxious someone is on a regular basis). We measured trait differences in positive and negative emotions related to more stable measures of the affective style of an individual. This enduring style may have cumulative influences on choice behavior over the life course, since prior work suggests that experimentally-induced transient positive aroused affect can decrease perceptions or risk, whereas negative aroused affect can increase perceptions of risk [21], [22], suggesting that these transient “anticipatory” affective states can increase or decrease financial risk taking, respectively [23].

A 60-item short form of the Neuroticism, Extroversion/Introversion and Openness to Experience Personality Inventory (NEO-SF) [24] asked subjects to indicate their level of endorsement of each of 60 statements related to commonly-assessed personality traits. These common traits are Neuroticism, Extraversion, Openness to Information, Agreeableness, and Conscientiousness. Sample items for Neuroticism include: “I am (not) a worrier.” and “I often feel tense and jittery.” In this paper we focus on neuroticism for several reasons, including the strong link between this measure and anxiety, its high level of heritability (50% of individual differences in neuroticism are due to genetic factors), and its correlation with the 5-HTTLPR genotype [10], [11], [25]. High neuroticism scores indicate a high propensity to experience anxiety, worry, vulnerability and self-doubt.

We also administered the Affect Valuation Index [26] to assess the extent to which subjects experience and value positive and negative emotions on a regular basis. Subjects saw 30 emotional descriptors referring to either positive affect (e.g., enthusiastic, excited, happy) or negative affect (e.g., fearful, nervous, sad) and were asked to rate how often they actually experienced that feeling over the course of a typical week, using a scale from 1 (“never”) to 5 (“all the time”). For each subject, we calculated their average rating across words in the positive domain, and separately in the negative domain, and used these as proxies of the intensity of positive and negative affect that characterize that individual in general.

Financial Choices

The regression models in Table 2 present the effects of having the SS or SL version of the gene, relative to having the LL version, on several financial choices and outcomes. In these regressions, we also included subjects’ age and sex as predictors of choices and outcomes. Subjects with the SS genotype choose to invest 23.84% less in stocks and choose to keep 25.60% more in cash than LL subjects (). There was no significant difference between these groups regarding the percentage invested in risky bonds. Both SS and SL subjects reported that they were less involved in actively making financial decisions compared to LL subjects. Using the 1 to 3 scale of active financial involvement, SS subjects’ average involvement score was 0.45 points lower than that of LL subjects (). In the subsample of 31 subjects for whom credit reports were available, SS and SL subjects had approximately 13 fewer credit lines compared to LL subjects, and also had higher FICO scores. The difference in FICO score was 47 points higher for the SL subjects (not statistically significant) and 93 points higher for the SS subjects ().

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Genetic effects on financial choices and outcomes.

https://doi.org/10.1371/journal.pone.0054632.t002

Wealth and Income

Findings in Table 2 suggest that 5-HTTLPR short allele carriers are more conservative in their financial decisions relative to those carrying two copies of the long allele. One possible explanation for these findings is that these groups differ in terms of their wealth, income, or credit constraints. The regression models presented in Table 3 investigate this possibility. As before, we predicted financial variables using only exogenous regressors: the 5-HTTLPR genotype, age, and sex. We did not find any significant differences between the SS, SL or LL subjects in terms of assets, debt, wealth (computed as assets minus debt), income, available credit and used credit (the latter two measures were only available for the subsample of 31 subjects whose credit reports were obtained). As expected, we found that older subjects had significantly more assets and net wealth. Also, males in the sample had higher incomes than females. The lack of an effect of genotype on these wealth-related variables indicates that previously observed differences in financial choices of SS, SL and LL subjects must be driven by other mechanisms.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. The wealth channel.

https://doi.org/10.1371/journal.pone.0054632.t003

Cognitive Abilities and Learning in Financial Markets

It is possible that the 5-HTTLPR short allele carriers are more financially conservative due to varying cognitive skills and abilities to learn in financial markets, since researchers have suggested that serotonin function may also be related to cognitive inflexibility [27]. For example, relative to LL subjects, SS or SL subjects may have lower numeracy scores or may be unable to learn in complex and dynamic environments where optimal choices change over time. These subjects may also have acquired less education and so may have a more limited understanding of financial markets. Such differences could lead short allele carriers to rationally avoid risky or complex financial strategies. Nonetheless, as shown in Table 4, the three groups of subjects did not differ in terms of numeracy, cognitive flexibility, or years of education. The only noticeable difference between groups involved working memory, with SS and SL subjects having slightly lower scores than the LL subjects. While this might present a handicap in certain life domains, we found that it did not correlate with subjects’ abilities to learn from new information in the financial investment task used in this study. Also, short allele carriers did not attend to different types of information in the financial task than did long allele carriers. These results are shown in the last two columns in Table 4. Thus, subjects’ ability to learn from either gain or loss information did not depend on their 5-HTTLPR genotype. The overall learning rate also did not significantly differ across the SS, SL and LL groups. Across the entire sample, subjects made correct (i.e., Bayesian) choices in 62% of trials in the gain condition, and in 66% of the trials in the loss condition (see Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. The cognitive ability and learning channel.

https://doi.org/10.1371/journal.pone.0054632.t004

Beliefs About Risk and Self

Another possibility is that genotype influences subjects’ beliefs about the risk of financial investing or their own ability to invest. Findings provided some support for this interpretation. As the results in Table 5 show, controlling for age and sex, SS subjects were more likely to perceive stocks as very risky, relative to LL subjects (although at weak significance levels). At the same time, SS subjects were significantly more likely to report having lower socio-economic status, either in their local community () or relative to the U.S. population as a whole (). Since there were no significant differences between SS subjects and the other groups in terms of wealth, income, or education (including sex and age as controls), these findings suggest that short allele carriers were subjectively more pessimistic when assessing their standing in society (their status, ability or success). To the extent that low socio-economic status perceptions correlate with, or induce, pessimistic beliefs regarding one’s abilities across other domains, this result suggests that 5-HTTLPR short allele carriers may be more likely to doubt their ability as investors, compared to other individuals. Also, prior findings that low socio-economic status correlates with less perceived control [17] may provide another explanation as to why long allele carriers (who score higher in perceived socio-economic status) may be more willing to invest in risky assets. The control variables used in the regressions in Table 5 confirmed this line of reasoning. Relative to female subjects, males perceived stocks to be less risky. Older adults believed that they ranked higher in terms of socio-economic status, consistent with the finding in Table 3 that they possess higher wealth relative to younger subjects.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. The beliefs channel.

https://doi.org/10.1371/journal.pone.0054632.t005

Neuroticism, Positive Affect, and Negative Affect

The link between the 5-HTTLPR gene variants and financial choices may also be related to affect, since prior work suggests that manipulations of the serotonin system in the brain can influence people’s emotional state [28]. In light of the evidence in prior section, it is possible that short allele carriers, relative to long allele carriers, are more anxious about facing risky outcomes, and as a result, chose to avoid risky financial strategies. Findings strongly supported this mechanism in several ways.

First, allelic variation correlates robustly with individual differences in subjects’ neuroticism, as shown in the regressions in Table 6. SS and SL subjects had were significantly more neurotic than LL subjects (6 points difference, ), in line with results from prior genetic association studies (e.g., [11]). This represents a large effect, given that the mean and standard deviation of neuroticism in the sample were 27.93 and 8.98, respectively (see Table 1). Subjects in the SS and SL groups also reported experiencing more negative affect than subjects in the LL group (0.2 points difference, ), and less positive affect (0.4 points difference for the SS group, ). These effect sizes were comparable to one standard deviation for each of the measures (i.e., 0.40 points for negative affect and 0.52 points for positive affect, see Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Genetic effects on anxiety and affect.

https://doi.org/10.1371/journal.pone.0054632.t006

Second, financial choices were specifically related to the component of the neuroticism score attributable to 5-HTTLPR genotype. In the first column of Table 7 the neuroticism score was regressed onto dummy variables indicating whether a subject’s genotype was SS, SL or LL. This regression yielded two orthogonal components of neuroticism: that driven by the 5-HTTLPR genotype, and a residual component (i.e., predicted values and residuals in the regression in the first column). Regression models similar to those in Table 2 were estimated to determine whether the component of neuroticism driven by the 5-HTTLPR genotype, as well as the residual neuroticism, correlated with subjects’ allocations to stocks, cash, their involvement in financial decisions, their number of credit lines, and their FICO score. Results indicated that the component of neuroticism specifically driven by subjects’ 5-HTTLPR allelic variation significantly and robustly predicted all five measures of financial choices and outcomes. The residual component of neuroticism, however, only correlated with one of the five measures – involvement in finances. The sign of all these effects was consistent with the predicted anxiety mechanism, such that subjects with higher values of the genetically-driven component of the anxiety measure invested a lower fraction of money in stocks, kept a higher fraction in cash, were less actively involved in complex financial decisions, had fewer credit lines, and had higher credit scores.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. Genetically-driven effects of anxiety on financial choices and outcomes.

https://doi.org/10.1371/journal.pone.0054632.t007

To verify the robustness of these effects, in Table 8 we estimated similar regression models as in Table 7 but included the subjects’ wealth, cognitive ability (as measured by the performance on the Trail Making Test of cognitive flexibility) and beliefs regarding the riskiness of investing in stocks as additional control variables. Even with the inclusion of these controls, the effects of the genetically-driven neuroticism on avoiding risky and complex financial choices were in line with those estimated in the main analysis in Table 7. The control variables themselves had expected effects. For example, subjects with more wealth were more actively involved in making financial decisions and had higher FICO scores, while those who had stronger beliefs that stocks were risky investments in fact allocated a smaller fraction of their money into stocks, and a larger fraction into cash. Using different proxies for cognitive ability and beliefs yielded results similar to those documented in Table 8.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 8. The role of additional controls on the genetically-driven effects of anxiety on financial choices and outcomes.

https://doi.org/10.1371/journal.pone.0054632.t008

Conclusion

The primary goal of this research was to elucidate a mechanism by which genetic factors might influence financial choice. Thus, we focused on one genetic variant that laboratory studies have related to financial risk taking – a specific polymorphism in the the promoter region of the serotonin transporter gene (5-HTTLPR). We found that individuals who have the 5-HTTLPR short allele avoid risky and complex financial decisions, even though they do not differ from others in terms of wealth, income, assets, debt, education, cognitive abilities, numeracy, or learning based on financial information. Instead, a mechanism responsible for the effects of the short allele on financial choice appears to involve people’s negative affective reactions when anticipating upcoming investment decisions. Specifically, relative to other subjects, short allele carriers perceived stocks to be riskier and were more anxious and prone to feeling negative emotions, suggesting that these individuals tend to experience higher anxiety when faced with risky investments involving potential losses.

These findings thus make several novel contributions. First, we identified a mechanism through which a specific genetic factor (5-HTTLPR) influences financial choice, rather than merely documenting correlations between genetic variants and behavior. Second, we acquired integrative genetic, survey, and experimental data in a community sample of adults, and corroborated this data with objective information from individuals’ credit reports. Together, these findings suggest that studying the biological aspects of decision making can help explain individuals’ immediate economic choices and eventual life financial outcomes.