The effect of investors’ emotional and depressive states on perceived returns and risk

Expected returns and risk, which are derived from fundamental economic factors, are two important parameters used in many financial theories (Black, 1995; Fama, 1970, 1973; Markowitz, 1952). Financial theories hold that investors have homogeneous beliefs and equal abilities to access the same set of information. Given these assumptions, investors’ perceptions and expectations should align with those of other market participants in equilibrium financing models. Traditional finance theories assume that investors form unbiased expectations of returns and risk without the influence of noneconomic factors, such as psychological responses. In contrast, behavioral finance research presents substantial evidence showing that fundamental economic factors alone may not be enough to explain returns and risk estimations.

Behavioral finance theory reveals that decision-making processes and judgments are influenced by people’s cognition (Lo, 2011), mood and emotion (Alex et al., 2007; Bodoh-Creed, 2020; Dowling and Lucey, 2008; Kaplanski et al., 2015; Kassas et al., 2022; Kourtidis et al., 2016; Whiteman, 2017; Wong et al., 2018) and psychological factors such as bipolar disorder (McGraw et al., 2008), stress (Gelman and Kliger, 2021; Mokarami and Toderi, 2019) and depressive moods (Liu et al., 2021). Kassas et al. (2022) found that induced emotions affect risk aversion. Lu et al. (2024) showed that individuals with major depressive disorders tend to make abnormally risky decisions. In conclusion, numerous studies indicate that variations in investors’ psychological and emotional states are factors to consider in studying financial phenomena and that returns and risk, which are widely used financial parameters, are influenced by psychological and emotional states.

This study examines the relationship between investors’ moods, psychological states and short- and long-term expectations of returns and risk. Unlike previous studies (Brahmana et al., 2012, 2014; Ding et al., 2021; Kaplanski et al., 2015; Xu, 2022) that used indirect measurements of moods (e.g. sports, weather, sentiment or moon phase), we directly evaluate investors’ psychological and mood states using psychological diagnostic tools, including the Patient Health Questionnaire-9 (PHQ-9) and Positive and Negative Affect Schedule (PANAS). The study first explores how investors’ moods affect levels of depression and how depression influences investors’ moods. Results reveal that investors’ moods affect their levels of depression. We subsequently determine whether investors with different psychological traits and moods have varying levels of optimism about returns and risk in both the short and long terms. We conducted an online survey to collect data from 717 investors in Thailand.

The results indicate that depressive states negatively affect risk perception. Furthermore, positive moods increase, while negative moods reduce and positive perceptions of expected returns. We include emotional intensity using the interaction term between depressive state and emotional scores and find that emotional intensity significantly influences investors’ risk perception (Lerner and Keltner, 2000; Loewenstein et al., 2001; Mittal and Ross Jr, 1998). We further tested hypotheses on the relationships between depression traits, emotions and forecast directions, either positive or negative, using logistic regression. The results indicate that negative emotions increase the probability of negatively forecasting long-term stock market returns.

This study distinguishes itself through its interdisciplinary approach, integrating principles from both behavioral finance and psychology. This study makes significant contributions to these disciplines across multiple perspectives (Apicella et al., 2015; Murphy, 2012). First, this study advances the field of behavioral finance by integrating psychological evaluations using established tools – as PHQ-9 and PANAS. Second, we provide evidence on the interplay between emotions and financial decision-making by presenting results based on emotional intensity, specifically the interaction of moods and depressive states. Third, in contrast to previous research that used proxies for expected returns or risk, we capture investors’ perceptions of returns and risk by collecting data directly from investors (Murphy, 2012; Shefrin, 2015).

The practical implications of these findings are substantial. Financial advisors and investment firms can enhance their client evaluation processes by incorporating psychological assessments, provided they obtain explicit client consent and adhere to strict ethical guidelines. For instance, clients exhibiting higher levels of depression may benefit from stable, low-risk investment strategies, while those in positive moods might be inclined toward higher-risk, higher-reward opportunities.

Moreover, investment firms can develop products tailored to investors’ emotional and psychological profiles, thereby promoting more balanced and informed decision-making. These implications are further strengthened using the study’s interdisciplinary approach, which integrates psychological insights with financial analysis. By examining the interaction between emotions and financial decisions and utilizing emotional intensity as a proxy for expected returns and risks, the study offers a nuanced understanding of investor behavior.

This paper is structured as follows: Section 2 reviews the literature, Section 3 outlines data collection and variable measurement, Section 4 presents the data analysis, Section 5 reports the regression results, and Section 6 provides the conclusion and discussion.

Psychological and mood factors both play a significant role in determining asset prices (Bodoh-Creed, 2020; Kumar and Ks, 2024; Lu et al., 2024; Merkle et al., 2015; Shefrin, 2015). Moods or emotions are measured to capture whether investors’ feelings are positive or negative. A positive mood promotes the recall of positive information, while a negative mood prompts the recall of negative information (Bodoh-Creed, 2020). Daniel and Hirshleifer (2015) presented evidence on how psychological biases influence investor behavior. They find that asset prices display patterns of predictability that deviate from the rational-expectation-based finance theories of price formation. Alti and Tetlock (2014) estimated a model in which agents’ information processing can cause predictability in firms’ asset returns and investment inefficiencies. Their model includes two biases: overconfidence and overextrapolation of trends. The results indicate that the estimated bias parameters have significant power in explaining the predictability of observed returns.

Stress, depressive states and moods are critical psychological factors that influence investors’ decision-making processes. Mood is a transient emotional state that fluctuates over short periods and directly impacts investors’ perceptions of risks and opportunities. For example, Hirshleifer and Shumway (2003) found that positive emotions can drive short-term stock market returns. In contrast, depression is a persistent psychological state with long-term effects, including reduced confidence and more negative outlooks on financial decisions, particularly investments. Research indicates that depression is associated with increased risk aversion and altered decision-making (Cobb-Clark et al., 2022; Lawlor et al., 2020).

The mood-as-information hypothesis (Schwarz and Clore, 1983) explains how moods, as temporary emotional states, serve as cues for judgment and influence decision-making. Both stress and mood have been shown to impact financial decision-making processes (Schwarz and Clore, 1983; Dibb et al., 2021; Gray et al., 2023; Sarmiento et al., 2024). For instance, stress has been associated with investment myopia (Chakeeyanun et al., 2023; Hidrobo et al., 2023). However, while the role of stress in decision-making is established, its specific effect on risk attitudes remains unclear (Gray et al., 2023).

Additionally, recent research has examined the influence of cognitive biases on both short- and long-term risk perceptions (Li et al., 2023; Lu et al., 2024). These findings underscore the complex relationship between psychological states and investment behavior, emphasizing its critical role in behavioral finance studies.

Previous studies suggest that individuals with depression may have different risk preferences, being either more risk-averse or more risk-seeking (Lu et al., 2024). Depressive individuals tend to overestimate the probability of negative outcomes and give them more consideration when making decisions. The influence of moods and depressive states on financial decision-making has been studied over different time frames, including short- and long-term periods. For example, Lu et al. (2024) established that patients with major depressive disorder (MDD) exhibit greater risk-seeking behavior and often make irrational decisions that result in long-term losses.

Kaplanski et al. (2015) examined whether individuals with positive moods predict future risk and returns differently from those with negative moods. Their findings reveal that noneconomic sentiment factors, such as the success of one’s favorite sports team, influence expectations. Happier investors tend to be more optimistic and sentiment (mood) impacts expected returns more significantly than expected risk. The study also finds that investors are generally consistent in their short- and long-term expectations. Similarly, Shefrin (2015) suggests that market sentiment mediates investors’ judgments of risk.

Empirical evidence from Kumar and Ks (2024) shows that investor sentiment has a significant positive effect on portfolio returns. Studies using mood proxies indicate that positive moods correlate with positive short-term returns, while negative moods correlate with negative short-term returns (Brahmana et al., 2014; Hirshleifer and Shumway, 2003). In addition, research demonstrates that moods can influence long-term investment decisions. For instance, Pyles (2009) finds that seasonal depression impacts investment in real estate investment trusts (REITs).

Dibb et al. (2021) showed that emotions can both enhance and impair decision-making. Mittal and Ross Jr. (1998) suggest that individuals in positive moods are generally less likely to take risks than those in negative moods. Using panel data from UK brokerage clients, Merkle et al. (2015) demonstrated that investors who anticipate happiness produce more accurate forecasts, indicating that emotional states can shape economic expectations.

Using a survey adapted from the PANAS to measure individuals’ emotional factors, Kassas et al. (2022) revealed that emotions influence individuals’ risk preferences. The literature collectively indicates that psychological traits shape investors’ perceptions of risk and returns. Negative psychological states, such as depression or negative moods, tend to lead to more pessimistic attitudes toward risk and returns. Based on these findings, we propose the following:

Biased beliefs contribute to numerous return anomalies that are typically unexplained in empirical finance (Alti and Tetlock, 2014; Rezaei and Elmi, 2018). Previous research shows that individuals often exhibit biased judgment (Daniel and Hirshleifer, 2015b; Daniel et al., 1998). Financial analysts are often optimistically biased in their earnings forecasts (Al-Thaqeb, 2018; Ji et al., 2024; Jiang et al., 2022). Moreover, when investors are highly optimistic in their trading views, they become overconfident (Zahera and Bansal, 2018). In this study, we investigate whether investors’ perceptions of changes in short- and long-term market returns are influenced by their psychological traits and moods. Thus, we hypothesize that

An online survey was conducted to gather data from Thai stock market investors. The Thai stock market is primarily composed of individual investors, making it suitable for obtaining results related to individual psychology. Participants were informed that participation in the study was voluntary, and they could withdraw at any time. They were also informed that their data would be collected anonymously, with no way to identify participants.

Thai investors were chosen as the sample in this study for several reasons. First, the Capital Market Development Fund (CMDF), the funding organization for this research, specifically aimed to study the behavioral aspects of Thai investors, which made this population our primary focus. Second, Thailand presents a distinctive cultural, economic and regulatory environment, offering valuable insights into the behavioral patterns of investors. Unlike other developed stock markets dominated by institutional investors, the Thai stock market predominantly consists of individual investors. This characteristic enables us to obtain more robust results regarding individual psychology and decision-making, which are central to our study. Third, by reaching a diverse group of individuals across Thailand, our data collection process enhances the potential for generalizing our findings to other markets with similar characteristics. Although cultural and economic factors may vary across regions, the study’s design can be adapted for future research in other contexts.

The survey was distributed through various online platforms to reach a wide audience. Since the Stock Exchange of Thailand, under the CMDF, is the supporting organization for this research, we secured collaborative assistance from the Association of Thai Securities Companies (ASCO), which represents a network of financial entities in Thailand, including securities firms and investment companies.

We contacted ASCO’s member companies to help distribute the survey by inviting their clients to participate. About 741 ASCO-member security companies volunteered to assist in distributing the online survey to their clients. This collaboration significantly enhanced the survey’s reach, resulting in a diverse and representative sample of participants from Thailand’s financial sector.

A total of 952 participants responded. To ensure data quality, three screening criteria were applied. First, to confirm that participants read the survey questions carefully, responses completed in under 20 min were discarded. This threshold was based on testing, which indicated that at least 20 min was needed for thoughtful responses. Second, the survey included two attention-check questions; incorrect responses were excluded. Third, incomplete responses were omitted.

After applying these criteria, 717 responses were included for further analysis. Established guidelines recommend a sample size of 200–300 for general survey research (Krejcie and Morgan, 1970; Bartlett et al., 2001). Our sample size of 717 exceeds these recommendations, enhancing the precision and reliability of our findings.

The survey included constructs for returns and risk expectations: expected returns (ERT), expected risk (ERK), short-term expected returns (STERT), short-term expected risk (STERK), long-term expected returns (LTERT) and long-term expected risk (LTERK). In this context, the short term refers to one month, while the long term refers to one year from the time respondents completed the survey.

The constructs were measured using five-item questions, each assessed with a five-point Likert scale. Higher scores on expected returns and risk indicate that investors anticipate higher returns and perceive future risk as low. To illustrate, investors were asked to rate their agreement with statements such as, “Over the next month, I expect my stock portfolio to have a high return” for expected returns. For expected risk, they were asked to rate statements like “Over the next month, I believe investing in stock has low risk.” Higher scores reflect a more positive perception of both returns and risk.

To assess patient depressive states, psychologists utilize various depression screening tools (Baryshnikov et al., 2023; Larsen et al., 2023; Lotrakul et al., 2008). Among these, the PHQ-9 is one of the most widely used for screening depressive symptoms (Baryshnikov et al., 2023; Chakeeyanun et al., 2023; Liu et al., 2024). The PHQ-9 consists of nine statements, each rated on a four-point Likert scale. These items measure the intensity of depressive symptoms, with ratings ranging from 0 (“not at all”) to 3 (“nearly every day”). Assessed individuals are asked how often they have been bothered by specific problems over the past two weeks, such as having little interest or pleasure in doing things or if they have trouble concentrating on things. The total scores for each individual can range from 0 (no depression at all) to 27 (severe depression), with higher scores indicating more severe depressive symptoms (Chakeeyanun et al., 2023).

Mood was assessed using the PANAS scale (Cantarella et al., 2023; Kassas et al., 2022), which includes ten positive emotion words (e.g. interested, alert, inspired, strong, active, attentive, excited, enthusiastic, determined and proud) and ten negative emotion words (e.g. irritable, ashamed, nervous, scared, hostile, jittery, distressed, upset, afraid and guilty). Following standard practices in the literature, participants rated the extent of their emotions over the past seven days on a scale from 1 (“not at all”) to 5 (“extremely”).

A reliability test was performed to assess the internal consistency of the survey constructs. Reliability scores were calculated for each construct in the research model and are presented in Table A1^[1]. The results of the reliability test exceeded the recommended value of 0.70 (Nunnally, 1978), indicating acceptable internal consistency.

The Kaiser–Meyer–Olkin (KMO) and Bartlett’s tests were conducted to evaluate the degree of unidimensionality of the scales. The sphericity test produced a p-value of <0.001. The sampling adequacy was confirmed with a KMO value of 0.9346.

The survey consists of 30 items. To evaluate the factors’ convergent validity, factor loadings were verified to determine that each survey item loaded onto the appropriate factor. The results provide evidence that 26 survey items loaded onto six factors, ERT, ERK, STERT, STERK, LTERT and LTERK, explaining 72.449% of the total variance. Four items with factor loadings below 0.7 were removed before the data were analyzed. The loaded items are shown in Table A1^[1].

Regression analyses were conducted to test the hypotheses regarding psychological traits and perceptions of returns and risk. Although regression analysis is not inherently designed to establish direct cause-and-effect relationships, it can still provide meaningful causal interpretations under certain conditions. Specifically, regression allows for the evaluation of how changes in independent variables influence dependent variables. This method becomes particularly robust when sufficient control variables are incorporated to mitigate potential confounding factors (Wooldridge, 2016). By incorporating control variables, the analysis can isolate the effects of psychological traits and return/risk perceptions, controlling for other factors such as experience, income and portfolio status. This approach enhances the validity and robustness of the results, enabling more accurate inferences about the relationships being tested.

In this study, the heteroscedasticity-robust regression method was employed to address issues of heteroscedasticity, which may arise from inconsistent variance in the unobserved error term. Various control variables were included to ensure that the observed relationships between psychological traits and perceptions of returns and risk were not biased by extraneous factors. The control variables consisted of experience (DMoexper and DHiexper), gender (Dgender), generation (DGenX, DGenY and DGenZ), income (DMedIncom, DHighincom and DHNWincom), portfolio value (DmodPV, DHighPV and DXHighPV) and portfolio status (DHighloss, DMedloss, Dlowloss, DHighprofit, DMedprofit and Dlowprofit).

Given that the survey utilized multiple-choice range options, dummy variables were employed to capture the categorical nature of the responses. This approach ensures that the model accurately accounts for each distinct range selected by the respondents, thereby satisfying the ceteris paribus criteria. A summary of the variables used in the analyses is provided in Table A2^[1].

To study the relationship between moods, depressive states and positive perception of future returns and risk, we used regression analysis to test the hypothesis. The regression model is expressed in Equation (1) as follows:

In Equation (1), the $PceivedVarsi$ represents the dependent variables that capture perceived expected returns and risks, including ERT, ERK, STERT, STERK, LTERT and LTERK. To enhance conciseness, we simplify the equation by consolidating all control variables into a single representation within Equation (1).

We conducted additional analyses to determine whether depression and emotional states have a joint effect on investors’ positive perceptions of expected returns and risk. Investors with negative moods coupled with higher states of depression can be viewed as mood intensity (Lerner and Keltner, 2000; Loewenstein et al., 2001; Mittal and Ross Jr., 1998). This test employs regression analysis; the model is expressed in Equation (2):

In Equation (2), we add two interaction terms (⁠$DEPRESS×POSPANASiandDEPRESS×NEGPANASi$⁠) to test the joint effects of depressive states and moods.

Furthermore, logistic regression analysis is performed to test investors’ views on market changes in the next month and next year. The binary variable, Dopt_fcast, whose value is 1, captures the forecasted market returns in the next month (Dopt_fcastM) and next year (Dopt_fcastY) to be positive and 0 otherwise. The logistic regression model is written in Equation (3) as follows:

This study examines the impact of investors’ psychological traits on their positive perceptions of expected returns and risk. Table A3^[1] presents the descriptive statistics for investors’ psychological traits.

The mean value of depressive symptoms is 3.746, while the maximum value is 26, indicating that most investors exhibit low levels of depressive symptoms. For investors’ emotional state, the mean positive emotion state is 38.000, while the mean negative emotion state is 28.499, indicating that investors’ emotional states are generally more positive than negative.

Dependent variables are classified into six variables to capture investors’ perceptions of short-term (one month) and long-term (one year) expectations of returns and risk. Table A4^[1] presents descriptive data summarizing the six dependent variables.

Most investors have a positive perception of future returns (ERT) in the Thai stock market, with a mean value of 4.015. However, positive perception of one-month returns (STERT) has a lower average score than expected returns (ERT) at 3.588, suggesting that investors expect a lower return in the next month. For long-term expected returns (LTERT), the mean value is 4.062, indicating that investors remain confident in market recovery, as reflected in their positive perception of future one-year returns.

Additionally, positive perception of future risk (ERK) has a mean value of 2.906, suggesting that investors anticipate high future stock market risk. The mean short-term expected risk in the next month (STERK) is 3.317, and the mean long-term expected risk over the next year (LTERK) is 3.505. These figures reflect investors’ expectations that risk in the Thai stock market will decrease, indicating their positive outlook on future risk.

To control for additional factors influencing perceptions of expected returns and risk, we included demographic variables (i.e. gender, generation, education and income), stock market experience and portfolio value, with descriptive statistics presented in Table A5^[1].

We performed a regression analysis based on Equation (1) to examine how an investor’s depressive state and emotions influence their perceptions of expected returns and risk. The results of the regression analysis are shown in Table 1.

Investors experiencing higher depressive states show a significant negative correlation with positive perceptions of ERK, STERK and LTERK at the 99% confidence level. This indicates that an increase in investors’ depressive states reduces their positive perceptions of expected risk.

A significant positive relationship was observed between positive moods and positive perceptions of expected returns and risk in all models at the 1% significance level. This emphasizes that investors with positive moods tend to have a positive view of returns in both the short and long term. In contrast, negative moods exhibit a significantly negative correlation with positive perceptions of ERT at the 5% significance level. Furthermore, negative emotions show a significant positive relationship with positive perceptions of ERK, STERK and LTERK.

Based on the regression analysis in Equation (2), we examine whether the combined effect of depression and emotional states influences investors’ perceptions of expected returns and risk. The results of this analysis are presented in Table A6^[1].

According to the findings in Table A6^[1], the joint test of positive emotions and depression exhibits a significantly negative association with positive perceptions of LTERK at the 5% significance level. This indicates that the combination of positive emotions and depression leads investors to develop more negative perceptions of long-term expected risk. Negative emotions and depression exhibit a significant negative correlation with positive perceptions of ERK at the 5% significance level and STERK at the 1% significance level.

From these results, we conclude that depression and emotions jointly and significantly influence perceptions of expected returns and risk. Regarding the long-term view, depressive states overshadow-positive moods, causing investors to have a negative view of risk. Regarding the short-term view, moods and depressive states have a joint and compounding effect on perceptions of returns and risk. Our results reveal that psychological well-being is a fundamental component of risk assessment and decision-making processes.

We further analyze whether depression and emotions affect investors’ long-term market forecast (Dopt_fcast) using logistic regression (Equation (3)). Table A7^[1] presents the survey data for Dopt_fcastM and Dopt_fcastY, which reflect investors’ forecasts of stock market changes over one month and one year.

Table A8^[1] presents the results of the logistic regression analyses. According to Table A8^[1], negative emotions exhibit a significant negative correlation with investors’ long-term stock market forecasts (Dopt_fcastY). Investors with negative emotions are less likely to predict an upward change in the stock market over the next year.

This study highlights the influence of depressive and emotional states on investors’ perceptions of expected returns and risk, emphasizing their interconnectedness. Specifically, the research suggests that not only do emotional states shape risk perceptions and return expectations individually, but they also interact in ways that jointly influence investment decisions. For example, investors with depressive symptoms may avoid risk and prefer stable investments, while those in positive emotional states may seek higher-risk and higher-reward opportunities.

These findings highlight the complex interplay between emotional states and investment behavior, emphasizing the importance of considering psychological factors when analyzing investor decisions. Financial decision-making is influenced not only by rational calculations but also by emotions and perceptions of risks and rewards. Understanding this dynamic requires considering emotional factors alongside traditional financial theories.

For practical application, financial advisors and investment firms can improve their services by incorporating psychological assessments into their client self-evaluations. These assessments should be implemented with explicit client consent and strict adherence to ethical guidelines to protect client data. By understanding clients’ emotional and psychological profiles, advisors can provide more tailored investment strategies. For example, clients with higher levels of depression may prefer investments focused on stability and security, while those in positive emotional states may consider higher-risk, higher-reward options.

The increasing use of behavioral finance principles exemplifies how financial products can be adapted to investors’ emotional needs. Research by Morningstar emphasizes that understanding clients’ emotional states can lead to more personalized and effective investment strategies. Integrating psychological insights allows financial professionals to gain a deeper understanding of client motivations and to develop strategies aligned with these psychological factors. Nevertheless, further research is needed to evaluate the practical feasibility and long-term sustainability of such approaches.

Despite its contributions, this study has limitations. The sample of individuals with extreme depressive and emotional states is relatively small, and they are more likely to be receiving psychological treatment, which could introduce bias in the results. Future research should include a broader spectrum of emotional states to provide a more comprehensive understanding of psychological influences on investment decisions.

Additionally, the generalizability of this study is limited to a specific subgroup of participants, focusing on the Thai context where cultural attitudes toward mental health, risk-taking and financial decision-making may differ from those in other cultural or national contexts. Societal views on mental health and the stigma associated with psychological issues could influence how investors report their emotional states and how these states affect their financial perceptions. Consequently, the findings may have limited applicability to regions with differing demographics, regulatory environments or cultural perspectives. Cross-cultural comparisons in future research could determine whether similar patterns emerge across diverse contexts, offering valuable insights into the role of culture in shaping investor behavior.

This study focuses on exploring relationships rather than establishing causality. While regression and logistic regression analyses help identify patterns and provide insight into the influence of psychological traits on financial perceptions, they do not establish the temporal sequences necessary for causal claims. Future research could address this gap through experimental designs, such as randomized controlled trials or longitudinal studies that track changes over time to clarify causal mechanisms.

Further research could examine the role of stable personality traits, such as risk tolerance, financial knowledge and openness to new experiences, in conjunction with emotional states. Understanding how these traits interact with emotions could offer a more nuanced perspective on the drivers of investment decisions.

Finally, to strengthen the establishment of a direct causal relationship and enhance the robustness of the study, incorporating panel data or longitudinal studies would enable more comprehensive analyses by accounting for time-invariant factors that influence investment decisions. Fixed-effect models could help control for individual differences over time, providing deeper insights into the relationship between emotional states and investment behavior.

We are deeply thankful to the Association of Thai Securities Companies (ASCO) and the Association of Thai Investors (TIA) for their crucial role in distributing our questionnaires to their member investors as their collaboration significantly facilitated the success of our data collection process. Lastly, we extend our heartfelt appreciation to all the investors who participated in our study and whose insights and contributions were invaluable to our research.

Funding: This research was funded by the Capital Market Development Fund (CMDF) under contract number CMDF-0049_2565, covering the period from 1 July 2022, to 30 June 2024. The project has also received approval from the Association of Thai Securities Companies (ASCO).

The supplementary material for this article can be found online.