Journal of Behavioral Finance

ISSN: 1542-7560 (Print) 1542-7579 (Online) Journal homepage: https://www.tandfonline.com/loi/hbhf20

Potential Underdog Bias, Overconfidence and Risk

Propensity in Investor Decision-Making Behavior

Sean Combrink & Charlene Lew

To cite this article: Sean Combrink & Charlene Lew (2019): Potential Underdog Bias,
Overconfidence and Risk Propensity in Investor Decision-Making Behavior, Journal of Behavioral
Finance, DOI: 10.1080/15427560.2019.1692843

To link to this article: https://doi.org/10.1080/15427560.2019.1692843

Published online: 03 Dec 2019.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=hbhf20
JOURNAL OF BEHAVIORAL FINANCE
https://doi.org/10.1080/15427560.2019.1692843

Potential Underdog Bias, Overconfidence and Risk Propensity in Investor

Decision-Making Behavior
Sean Combrink and Charlene Lew
Gordon Institute of Business Science, University of Pretoria

ABSTRACT KEYWORDS
In this study we investigate whether investors are prone to take risks, both in terms of how Investor decision-making;
they rate their risk propensity and their behavior in choosing between options with different Risk propensity; Underdog
risk levels, and whether they display overconfidence and underdog bias. We also investigate bias; Overconfidence;
Decision bias
the relationships among underdog bias, overconfidence and risk propensity. The results indi-
cate overconfidence levels similar to that in other populations and do not reveal underdog
bias or high levels of risk propensity. We found support for a negative predictive relation-
ship between underdog bias and overconfidence.

Introduction (Guenther and Alicke 2010); (b) perceptions of whether

life supports success or the ‘top dog versus underdog’
The investment community relies on investors to
bias (Davidai and Gilovich 2016) and (c) the tendency
make rational decisions as they are custodians of the
wealth of the community. However, rationality and to take risks or personal risk propensity (Hoffmann,
predictability in decision-making is an unattainable Post, and Pennings 2015).
ideal as decision makers may often act irrationally From a social comparison theory (SCT) perspective,
(Kahneman 2011) and sometimes even consistently so we know that individuals consistently evaluate how
(Rabin and Thaler 2001). Irrational choices may par- good they are, but fail to do so accurately. A recent
tially be explained by System 1 thinking (Evans 2008) meta-analysis of SCT spanning 60 years suggest that
which is fast, instinctive and emotional, and leads to we tend to evaluate ourselves more highly after com-
decision shortcuts (Stanovich and West 2000) as parison with people lower than our standard and
opposed to less deliberate System 2 thinking. For lower after comparison with people higher than our
Kahneman (2011) instinct, salient memories, recent standard, thus contrasting ourselves with others rather
events and external impressions lead to the develop- than assimilating traits (Gerber, Wheeler, and
ment of cognitive biases in System 1 thinking. Against Suls 2018).
this backdrop of the limitations of human thinking Therefore, SCT suggests that we are prone to poor
and decision-making it is important to consider the self-rating of our performance or tend to assess the
potential cognitive biases and behavioral propensities quality of our own abilities and performance inaccur-
of investors. ately – especially in relation to the abilities and per-
To be able to consider these biases and propensities, formance of others. If investors overrate their own
one may imagine asking investors to rate their own performance in relation to others, rather than consult-
investment competence by asking the following two ing new information or advice, this may lead them to
questions: ‘Do you believe you have overcome all odds make less optimal investment decisions.
to be a successful investor?’, and: ‘How likely are you Over and above investors’ views of their own per-
to take investment risks?’ As the answers to such ques- formance, we might also ask whether they believe they
tions hold many implications for investment outcomes, have succeeded against all odds, or stated differently,
this paper explores three potentially significant factors succeeded without that having been likely. Davidai
that may impact investment decision-making: (a) the and Gilovich (2016) postulate that people remember
level of confidence in abilities or self-rated performance the headwinds of their past experiences more

CONTACT Charlene Lew [email protected] Gordon Institute of Business Science, University of Pretoria, P.O. Box 787602, Sandton 2146, South Africa.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/hbhf.
ß 2019 The Institute of Behavioral Finance
2 S. COMBRINK AND C. LEW

poignantly than they do the tailwinds. This refers to a self-rated performance, the paper presents a portrait of
biased view that their lives had had more obstacles these tendencies and the relationships among them for
than success-enabling factors, also known as head- the South African investment community.
wind-tailwind asymmetry. Such perceptions may be
due to availability or even attention bias: since
Theoretical Background
obstacles require more attention to be resolved, and
benefits are simply enjoyed, it is much easier to The Theoretical Roots of Bias and Risk
remember the obstacles. This phenomenon is referred Ever since it has become clear to economists that
to as ‘underdog bias’, implying a biased feeling of humans are not rational but bounded in our ability to
being mistreated by life and being successful despite solve complex problems (Simon 1955), scholars have
the presence of multiple obstacles. One may postulate explored this boundedness, especially where decision
that ‘underdog bias’ will relate to overconfidence, makers cannot optimize choice, but merely satisfice,
which refers to investors’ propensity to inadequately making the best possible decision within limitations.
consulting advice or information. Economists have assumed rational decision laws
Furthermore, in the presence of a view of personal whereby preferences among multiple options can be
competence and a sense that life is a battle that has predicted (Baye and Prince 2013), or where the per-
been won, one may wonder to what extent investors ceived utility of outcomes is relative to the departure
are risk seeking in their decisions. Risk propensity is point (Nutter 2010). However, evidence suggests that
seen as “ … the general tendency of a decision maker decision behavior deviates from expected or standard
to take or avoid risks” (Sitkin and Pablo 1992, p. 18) economic models (Stanovich and West 2000). This
and forms the natural or base line of risk-taking. The means that all decision makers, including investors,
undergirding principles of prospect theory suggest are unable to assess the likelihood of outcomes prop-
that investors may be risk averse, preferring guaran- erly; do not always maximize utility; and even enforce
teed wins, but then change to risk seeking behavior their opinions on others, partially due to the dual pro-
when potential losses are at stake (Kahneman and cess reasoning (Evans 2008).
Tversky 1979). However, it’s clear that often investors As opposed to the use of deep and analytical
misjudge the future value of stocks by basing their System 2 thinking (Kahneman 2011; Stanovich and
assessments on the past performance thereof West 2000), the unconscious level of instinctive,
(Barberis, Mukherjee, and Wang 2016). In contrast, immediate and default System 1 level thinking (Evans
sometimes the loss aversion of prospect theory enable 2008) may give rise to heuristics. Such decision short
investors to predict the performance of bonds (Zhong cuts may lead to biased decision-making (Kahneman
and Wang 2018). It is known that both situational 2011) and errors in judgement (Iqbal 2013).
and personal factors determine risk behavior (Iqbal Insights into heuristics lay the foundation for
2013) and that investors, as custodians of value for understanding not only the underdog bias and over-
shareholders and customers, are expected to mitigate confidence bias constructs explored by this study, but
risk and increase shareholder value. Risk propensity even risk propensity. Based on assumptions of
also varies between individuals (King and Slovic 2014) bounded rationality, biases and heuristics, we might
and the pursuit of extrinsic goals such as fame and assume that investors may display underdog bias,
money to better one’s standing in relation to others believing that they are heroes in their life stories; may
may increase the propensity to take financial risks believe that their judgement supersedes available
(Djeriouat 2017). Knowledge of the risk propensity of information and, accordingly, are likely to take risks
investors is thus clearly relevant to help mitigate in their decision-making.
against hasty or misguided decisions. Slovic et al. (1981) earlier described heuristics as
By focusing on professionals responsible for invest- inferential rules that enable people to transform com-
ment decisions in their firms, we can begin to answer plex or unknown problems into simpler ones, which
the question of how self-rated performance, risk pro- are potentially advantageous in the decision-making
pensity and underdog bias appear within the South process, but can sometimes lead to bias when used
African investment community. This paper therefore inappropriately in different contexts (Kahneman 2011).
firstly presents an overview of the theoretical framework Biases refer to inappropriate use of information or erro-
in which these constructs occur. Then, by making use neous assumptions about self and the decision situation.
of both a conventional and new scale of risk propensity, Availability bias describes the use of readily available
as well as through assessments of underdog bias and or recalled information rather than base rate
 JOURNAL OF BEHAVIORAL FINANCE 3

information (Tversky and Kahneman 1973). When we to recall positive events, as in overconfidence, and
do not have all the necessary information we may therefore there may be an inverse relationship
assume the information we have depicts the full picture between underdog bias and how investors rate their
(Kahneman 2011), or we may simply ignore the infor- performance. This suggests that underdog bias may
mation we do not know (Slovic et al. 1981). Our recol- play a role in recollection of events and self-rated per-
lections and available information shape the worldview formance and ultimately in overconfidence bias.
we hold and which is often distorted (Ross and Sicoly
1979; Taylor 1991). This may drive the behavior of
Overconfidence Bias
investors, such as causing them to overreact to big
stock price changes in their immediate decisions The perception that one’s performance is on average
(Kliger and Kudryavtsev 2010), especially for small and better than average is known as overconfidence bias
volatile stocks (Kudryavtsev 2018). Such availability (Guenther and Alicke 2010). A proliferation of litera-
bias lies at the heart of underdog bias, overconfidence ture indicates how investors’ irrational thoughts lead to
bias and risk perception relevant to this study. overconfidence, then irrational behavior and ultimately
irregularities in the financial markets. For instance,
Merkle (2017) argues that overconfidence causes invest-
Underdog Bias
ors to take higher risks, diversify less and increase their
The sense that, compared to others, one has faced trading activities. Along the same lines, Graves and
more salient barriers than enablers, describes under- Ringuest (2018) found that overconfidence relates to
dog bias. Under this bias one remembers the difficul- investment predictions. This applied when overconfi-
ties along one’s life path more acutely that the dence was operationalized as investors’ overrating the
positive events due to the effort to overcome difficul- accuracy of their own knowledge, as well as a belief in
ties (Davidai and Gilovich 2016). Underdog bias aligns their ability to anticipate the future better than others.
closely to availability bias where the most salient Overconfidence bias has its roots in distortions
information in one’s memory gains the greatest resulting from information availability, optimism, ego-
importance in decisions. In contrast, Rozin and centric tendencies or hindsight bias (Williams and
Royzman (2001) suggest that memories are skewed in Gilovich 2008); or the tendency to attribute success to
a positive direction, resulting in negative events being oneself rather than others (Schroeder, Caruso, and
downplayed over time. More recently, Madan, Ludvig, Epley 2016; Koo and Yang 2018). Availability bias
and Spetch (2017) found that we remember extreme partially explains why people would overrate their
life outcomes, whether positive gains or negative own personal contribution to group outcomes, as it is
losses, more acutely. When memories of struggle are easier to recall one’s own actions than that of others
more readily available, and people perceive themselves (Schroeder, Caruso, and Epley 2016).
as heroes in their own life stories, having overcome Hindsight bias may be another key root of over-
the difficulties they have had (Davidai and Gilovich confidence, where people deem information on previ-
2016), we assume the presence of underdog bias. ous success to be a predictor of their future success.
This may lead to individuals’ overestimating their In the much cited book Black Swan, Taleb (2007) cau-
own contributions to work (Schroeder, Caruso, and tions that the past is not a good predictor of the
Epley 2016) and performance in teams (Davidai and future, despite our tendencies to build the rationale of
Gilovich 2016). This perception, in turn, may lead to the future on events of the past. Interestingly, per-
a sense that the playing fields need to be leveled even ceived and actual expertise is likely to increase hind-
to the point of questionable ethical behavior (Davidai sight bias (Knoll and Arkes 2017). This means that
and Gilovich 2016; Tamborski, Brown and Chowning the expertise of the investment community may
2012), which holds several implications for investment indeed provide for unrealistic confidence that their
decision-making. past successes will result in similar future gains.
Davidai and Gilovich (2016) imply a relationship When considering whether there is a relationship
between self-attribution and underdog bias by describ- between overconfidence and risk taking one may note
ing how individuals that think that they face stiffer that overconfidence bias leads to risk taking that does
headwinds than others also think that they are more not realize benefits (Lovallo and Kahneman 2003).
entitled to a larger share of benefits. At the same One having lived a specific experience may cause
time, we know that the notion that life has been diffi- availability bias that explains why some situations
cult, implied in underdog bias, contrasts the tendency are perceived as more or less risky by an individual
4 S. COMBRINK AND C. LEW

(Kahneman 2011), or alternatively lack of information making. These limitations rest in either the inappro-
may cause ignorance of risk (Stanovich and West 2000). priate use of information, or a tendency to seek or
Similarly, overconfidence may lead to an overesti- avoid risk, based on the option being evaluated, one’s
mation of one’s ability to judge the riskiness of a situ- own expectations of gain, an individual traits and
ation and their own risk mitigation actions (Hoffmann, emotions and external or situational factors. It seems
Post, and Pennings 2013). Overconfidence may thus that when investors make decisions past experience
also lead to the expectation of the best case scenario may lead to overconfidence, an awareness of past
(Lovallo and Kahneman 2003) and if that is combined obstacles and the likelihood of taking risks.
with a sense of control over the risk, the propensity to We do not yet know if investors would display
take risks increases (Gilovich and Douglas 1986). underdog bias, and whether they are then indeed
overconfident and risk-seeking. One could argue that
if investors believe that they have had a tougher time
Risk Propensity
to succeed than their peers (underdog bias), they
Due to the inadequacies of utility theory to explain would be less likely to believe that their insights
risk preferences (Baye and Prince 2013; Iqbal 2013), exceed that of their peers due to hindsight effect
Kahneman and Tversky (1979)’s seminal prospect the- (Knoll and Arkes 2017). In other words, since the rec-
ory presents an alternative viewpoint. Prospect theory ollection of positive life events is an antecedent of
suggests that individuals are risk-averse when faced overconfidence, a focus on negative past events as in
with a potential gain and risk-seeking when faced with underdog bias suggests a possible inverse relationship
a potential loss, because of seeking certainty. Naturally between these biases.
people are loss averse (Kahneman 2011), but there is It is not yet clear whether overconfidence in turn
some evidence that risk propensity is based on individ- may result in investors taking risks. Risk aversion is at
ual reference points of personal preference (Novemsky the heart of individual risk propensity where we may
and Kahneman 2005). Expected gains become the ref- expect that investors will be risk-averse when they are
erence point according to which risk aversion plays out certain of gains, but risk-seeking in an attempt to pre-
(Baye and Prince 2013). Moreover when a person owns vent losses (Kahneman and Tversky 1979).
something, the endowment effect increases the per- Apparently, no study has yet attempted to deter-
ceived value of the item over willingness to pay for the mine whether any relationships exist among underdog
same item (Thaler 1980). bias, overconfidence and risk propensity. However,
Beyond the effects of reference points and the Davidai and Gilovich (2016, p. 837) state that “[t]he
endowment effect, risk propensity can also be as a belief that they have faced stiffer headwinds than
result of individual factors such as feelings about an others can also make people feel entitled to greater
option. King and Slovic (2014) showed how affect can benefits than they’ve received”. This statement implies
impact risk propensity in investment decisions. A risk a potential relationship between underdog bias and
propensity trait of individuals can also influence their self-attribution, but not necessarily overconfidence. In
behavior (Iqbal 2013). In addition to personal factors, the present study we thus sought to answer the ques-
external factors may cause people to be more risk tion: Will an investor have overcome more obstacles
seeking or averse (Lovallo and Kahneman 2003). For than others to reach success also unduly attribute suc-
instance, Bucciol and Miniaci (2018) have shown that cess to their own efforts?
risk propensity is higher just after economic growth, Since there seems to be a relationship between risk
such as after positive stock market returns, and also perception and self-rated performance (Lovallo and
when investors have had large risk exposure in their Kahneman 2003), one may wonder whether there is also
own portfolios. For ease of understanding, we refer to any relationship between risk propensity and overconfi-
risk propensity that relates to prospect theory as “risky dence. Michailova and Schmidt’s (2016) experiment
choice propensity”, and to self-rated risk propensity as shows that markets with overconfident investors bring
“personal risk propensity”. about higher incidences of stock price bubbles. Similarly,
Merkle (2017) found that overconfidence relates to risk
taking in the form of over-precision and over-placement
Research Question and Hypotheses Development
of transactions for their UK based sample, and cites sev-
As the overview of seminal and recent literature has eral examples where overconfidence links to specific risk
shown, bounded rationality theory provides for a taking behaviors. A further pertinent question thus
number of insights into the limitations of decision- emerged: Will a South African investor who overly
 JOURNAL OF BEHAVIORAL FINANCE 5

ascribes success to their own judgement also be likely to would be the mechanism for the measurement of suc-
take on more risk in decision-making? cess in order to ensure greater homogeny of the sam-
Literature suggests that underdog bias may lead to ple and to optimally describe the risk propensities and
people taking short cuts in their decisions (Davidai bias propensities of the sample.
and Gilovich 2016). In line with this statement we Through non-probability purposive sampling, we
also asked: Will an investor who is subject to under- approached 894 investors via email and obtained 184
dog bias also have greater propensity towards risk? willing respondents, yielding a response rate of 23.2%.
Here one might caution that ascribing risk aversion or We complemented the purposive sampling with
risk seeking behavior to an individual is difficult snowball sampling by requesting the investors to rec-
(Iqbal 2013) as extraneous variables may determine ommend further suitable respondents, which yielded a
related risk perception (Lovallo and Kahneman 2003). further 8 responses. The resultant sample of 161
From the theoretical background and our prelimin- respondents, after incomplete surveys were removed,
ary questions the following research questions was sufficient to overcome the increased risk of sam-
emerged: (a) to what extent do investors display over- ple error (Wegner 2016).
confidence?; (b) to what extent do investors display Of the respondents 53% were in the younger and
underdog bias?; and (c) are investors risk seeking? 47% in the older category (when the sample was split
Furthermore we proposed hypotheses regarding pos- in half) with as many as 43% in the 30–39 year old
sible relationships among the variables as outlined: age group. Only 12% was female, and 88% held post-
H1: Underdog bias predicts self-rated performance. graduate qualifications at the time, which may give an
indication of the gender and educational composition
As we know that recollection of past experiences
of the investment community in South Africa
impact decision bias, we presume that there may be a
(Hoffman et al. 2013). Only 2.5% of the sample did
relationship between underdog bias and self-rated
not have an undergraduate degree. Eighty-four
performance.
respondents had ten or fewer years of experience (or
H2: There is a statistically significant correlation 53% of the sample), and 136 respondents had less
between underdog bias and risk propensity. than 20 years of experience (or 85% of the sample)
From Davidai and Gilovich’s (2016) experiments we (see Table 1).
inferred that a sense of being unfairly treated may result The types of investment businesses represented were
in in risky corrective action, which leads to the assump- Pension/Provident Fund Management firms (29%),
tion that risk propensity may increase along with a ten- Private Wealth Investment (27%), Private Equity (14%),
dency to display underdog bias, and consequently take Hedge Fund Management (7%), Equity Trading (7%),
risks. Also, since overconfidence may entail excessive Fund of Funds (6%), Bond Trading (3%), Stock Broking
belief in one’s own risk assessments and control
(Hoffmann, Post, and Pennings 2013), it may lead to Table 1. Demographics of participants.
greater risk propensity. Thus, the following hypothesis: Age 18–29 15 9%
30–39 70 43%
H3: There is a statistically significant positive correlation 40–49 45 28%
between self-rated performance and risk propensity. 50–59 26 16%
60þ 5 3%
Gender Male 142 88%
Female 19 12%
Method Education High school or lower 1 1%
Diploma 3 2%
Participants Undergraduate degree 15 9%
Honors or post 89 55%
graduate diploma
Given their risk related decisions and their financial Masters 47 29%
custodian responsibilities, the risk investment commu- Doctorate 6 4%
Risk perception Conservative 18 11%
nity was an ideal population for this study. We tar- of environment
geted individuals who were partially or fully Moderate 79 49%
responsible in their organizations for investment in High 46 29%
Mixed 18 11%
business or financial instruments. Respondents were Experience 0–5 40 25%
limited to decision makers within investment teams in 6–10 44 28%
11–15 23 14%
the South African market. We focused the population 16–20 29 18%
only on individuals responsible for identifying returns 21–25 14 9%
26þ 10 6%
for external investors through funds as their returns
6 S. COMBRINK AND C. LEW

(2%) and Insurance (1%). We asked the respondents to and “I am coping with most of the problems
indicate the current risk level of their funds. Forty nine at work.”
percent indicated that they operated in a moderate risk The first measure of risk propensity, or risky choice
context, 29% in a high risk context, 11% in a mixed risk propensity, was designed from prospect theory
context and 11% in a conservative context. (Kahneman and Tversky 1979) offering respondents a
choice between binary options such as: Would you
rather take an “80% chance of R4 million and a 20%
Procedure chance of nothing” versus a “100% chance of R3 mil-
Measures lion?”; or a”20% chance of losing R4 million and an
The survey’s preamble was used to introduce the 80% chance of losing nothing” versus a “25% chance
research, obtain consent and screen the participants of losing R3 million and a 75% chance of los-
based on demographics. The remainder of the survey ing nothing?”
consisted of measures of Underdog Bias, Self-Rated The validity of the scale was examined by compar-
Performance, a Risk-Propensity Measure based on ing it to the original Kahneman and Tversky (1979)
Prospect theory and a newly designed Investor Risk results and only one item did not yield comparable
Propensity Scale. The scales appear in Appendix A. results. As a consequence the certainty value of the
Underdog bias was measured on a 7-point Likert options for each of the items was calculated to assign
scale between ‘entirely agree’ and ‘entirely disagree’ a score of 0 to more certain options and of 1 to risk-
for eight items relating to the construct as described ier options, using the cumulative prospect theory for-
by Davidai and Gilovich (2016) and availability bias mula (Tversky and Kahneman 1992). A single score
literature. After Pearson Correlation calculations six was then developed for the construct using a simple
items were retained to ensure convergent validity, and average of all ten questions to create a single prospect
the Cronbach alpha scores indicated internal consist- theory scale score for each respondent. Calculation of
quartiles and standard deviations, with only slight
ency of the instrument (a ¼ 0.744). Sample items are:
positive skew of the data, indicated that there were no
“I have to work harder than others to get the recogni-
outliers that needed to be removed.
tion I deserve” or “My investors are more demanding
The second measure of risk propensity, or personal
than other investors, even when I produce the same
risk propensity, in the investment community was
results.” The average of the five items yielded the sin-
designed with consideration of the Dospert scale
gle underdog bias construct score per participant. The
(Blais and Weber 2006), and Hoffman, Post and
method is similar to the method adopted by Davidai
Penning’s assessment of risk (2013; 2015). Based on a
and Gilovich (2016) for their research amongst
7-point Likert scale that ranges from ‘entirely disagree’
accounting faculty.
to ‘entirely agree’, the instrument asked the investors
The measurement of Self-Rated Performance was a
to rate themselves on items such as: “I get a thrill by
combination of the scale used by Williams and
taking decisions that I don’t know the outcome of”; “I
Gilovich (2008) and the questions from the Core Self- tend to take large but reasonable risk in my invest-
Evaluation Scale (CSES) (Judge et al. 2003) that were ment decisions”; and “I take more risk than my fellow
adapted for the investment community. The scale pro- investors”. Lack of significant correlations (p > 0.05)
vided the investors the opportunity to compare them- of three of the items led to their removal. The
selves to their peers on a sliding scale of 0% removal of a fourth item resulted in the increase of
indicating worst performance to 100% indicating best the internal consistency measure (from a ¼ 0.632
performance and with the score for the average to a ¼ 0.792).
investor set at 50%. Descriptive statistics were calcu-
lated by assigning a score of –50 to the lowest rating, Data Gathering and Analysis
and þ50 to the highest rating, with average scores at The questionnaire was piloted among six investment
zero. Following Pearson Correlation calculations three professionals and two non-investment professionals to
questions were removed for this sample. The final make minor adjustments. The final questionnaire was
eight-item instrument was internally consistent distributed electronically to reduce time and potential
(a ¼ 0.785). Typical items for participants to compare data capturing errors. Thirty incomplete surveys were
themselves to the performance of others were: “I removed from the data set.
determine what will happen in the investments we After being tested for construct validity and
make”; “I generally generate returns above my target”; internal consistency of the measures, descriptive
 JOURNAL OF BEHAVIORAL FINANCE 7

statistics were calculated and the relevance of the sam- hypothetical population mean (p < 0.0001). The mean
ple’s demographic variables for this construct was cal- of underdog bias was significantly higher for the
culated through single factor ANOVAs. Regression female respondents (M ¼ 2.96, SD ¼ 1.00) than the
modeling was then used to determine the predictive male respondents (M ¼ 2.19, SD ¼ 0.95) (see Table 3).
values of underdog bias on self-rated performance In contrast to the findings of Davidai and Gilovich
(overconfidence) and of both self-rated performance (2016) among sports fans, American voters, siblings,
and underdog bias on the two risk propensity meas- university students, and university accounting faculty, it
ures of the investors. is interesting to note that investment decision-makers
in South Africa do not display underdog bias. In other
words, there was no clear indication that they were of
Results
the opinion that other investors have had an advantage
Table 2 presents the descriptive statistics of the data. over them in the industry. The standard deviation indi-
Table 3 presents the demographic differences for cates that this was indicative of most of the sample,
the underdog bias, self-rated performance, risk pro- without outliers. Each of the items scored under the
pensity variables. Only two significant differences midpoint which means the investors are less prone to
were found, namely gender-based differences for feel that they work harder than others to achieve simi-
underdog bias and experience-based differences for lar results, work with more demanding investors, work
overconfidence as described below. harder to get earned recognition or that other investors
get recognition beyond what they deserve.
The results may vary from previous findings on
Investor Underdog Bias
underdog bias due to the specific investment-based ques-
The mean of the sample (M ¼ 2.28, SD ¼ 1.02) was tions. In Davidai and Gilovich’s study (2016) respondents
lower than the midpoint, and based on a single sam- compared themselves to specific others, such as their
ple t-test statistically significantly lower than the own siblings, where our study referred to the investment

Table 2. Descriptive statistics of the variables.

Underdog biasi Self-rated Performanceii Risk propensity: Prospect theoryiii Risk Propensityiv
Mean 2.28 17.02 0.38 2.61
Standard Error 0.08 0.93 0.01 0.08
Median 2.40 16.38 0.40 2.67
Mode 3.00 27.88 0.40 2.00
Standard Deviation 1.02 11.78 0.15 1.00
Sample Variance 1.04 138.79 0.02 1.01
Kurtosis –0.25 –0.09 0.48 –0.35
Skewness –0.05 0.33 0.18 –0.18
Range 5.00 62.00 0.60 4.67
i
7-point Likert scale.
ii
Range from 0–100.
iii
Codes of 1 assigned to risk options based on expected probability, and 0 assigned to less risky options (closer to expected probability).
iv
7-point Likert scale.

Table 3. Demographic differences (ANOVAs) of underdog bias, self-rated performance and.

Underdog bias Self-rated performance
F-Crit F-Stat p-value F-Crit F-Stat p-value
Age 2.66 1.29 0.279 2.66 1.93 0.128
Gender 3.90 10.10 0.002 3.90 0.60 0.441
Education 2.66 0.58 0.627 2.66 0.75 0.523
Risk Level (Org) 2.66 1.63 0.186 2.66 0.75 0.523
Experience 2.27 0.69 0.629 2.27 3.84 0.003
Fund Type 1.94 1.40 0.192 1.94 0.59 0.800
Risk Propensity Risky choice propensity (prospect theory)
Age 2.66 0.77 0.515 2.66 0.73 0.535
Gender 3.90 0.72 0.396 3.90 0.17 0.679
Education 2.66 0.38 0.768 2.66 1.33 0.267
Risk Level (Org) 2.66 1.50 0.216 2.66 1.82 0.146
Experience 2.27 1.12 0.351 2.27 0.37 0.870
Fund Type 1.94 0.19 0.995 1.94 0.99 0.450
Correlation is significant at the 0.01 level.
Correlation is significant at the 0.05 level.
8 S. COMBRINK AND C. LEW

community in general. In our study the investors did not Tukey-Kramer test (Jaccard, Becker, and Wood 1984)
feel they faced more headwinds, or were hard done by, indicated that those with less than five years of invest-
compared to others in the community. ment experience displayed significantly less confidence
A boundary condition of underdog bias mentioned than those with 26 years and more experience
by Davidai and Gilovich (2106) is that people are (q ¼ 5.05; p ˂ 0.05), as did the group with only 6-
acutely aware of benefits that they receive from others. 10 years investment experience (q ¼ 4.30; p ˂ 0.05).
This could explain why the investment community Measurement of self-rated performance rested on
was aware, for instance, of the recognition received two principles: (a) an understanding of investors’
from investors (a tailwind). As the headwinds in traits of self-esteem, generalized self-efficacy, neuroti-
investment are normally market-driven, they may cism, and locus of control in investment decisions as
believe that their decisions are aligned to what other adapted from the Core-Self Evaluation Scale (CSES)
investors would have made with similar information. (Judge et al. 2003); and (b) comparison to peers in
It may be that the community neutralizes or mini- the industry as adapted from Williams and Gilovich
mizes the value of poor decisions (Taylor 1991), and (2008). When their performance across a number of
that investors have positive memory bias, due to the abilities and behaviors is compared, on average the
time gap between decisions and outcomes (Rozin and respondents rated themselves in the 67th percentile of
Royzman 2001). However, the reason for this bound- the population, comparable to previous studies among
ary condition in the underdog bias of investors still university students of 60th – 65th (Williams and
needs to be empirically examined. Gilovich 2008) or 65th – 75th percentile rankings
Given that it is uncertain that a boundary condi- (Guenther and Alicke 2010), or among working indi-
tion is at play, the finding was useful in showing that viduals in the 77th – 79th percentile range (Judge et al.
this investment community does not lean towards 2003) (see Figure 1).
believing they are the heroes in their own difficult Whereas hubris is often associated with the invest-
life stories, and that their environment is suffi- ment industry, the investors did not rate their own
ciently enabling. performance above that of other investors. It should
It is also useful to consider that the female invest- be noted that they did not compare themselves to
ors were more likely to perceive headwinds than their professionals in other industries. Still they showed
male counterparts. Possible explanations for this may above average rating of themselves for getting the suc-
be the general gender based experience of lack of sup- cess they deserve at work, generating returns above
port from the environment (Morgenroth and Ryan their targets, completing their tasks successfully, deter-
2018), which may be further amplified by the poten- mining the outcome of the investments made and
tial underrepresentation of females in the industry. coping with most problems at work. Similarly they
opined that they were below average with regard to
getting depressed with the investment industry and
Investor Overconfidence
not feeling in control of success in their careers. This
The mean of the investor overconfidence was applied to all the demographic subgroups.
slightly above the average (M ¼ 17.02, SD ¼ 11.78). Interestingly, those with the least experience rated
Overconfidence differed in terms of the years of their self-performance the lowest (63rd percentile) and
experience of the respondents (see Table 3). The those with the most experience the highest (78th

Figure 1. Histogram of self-rated performance in relation to others (overconfidence).

 JOURNAL OF BEHAVIORAL FINANCE 9

percentile). This was, however, not a progressive pat- the behavioral principles of the Dospert scale (Blais and
tern for the age groups. Although there is a large dif- Weber 2006). We also utilized the risk tolerance meas-
ference at the extreme poles of the sample it remains ure from the Hoffmann, Post, and Pennings (2013,
interesting nonetheless that overconfidence appears in 2015) studies that explored decisions influenced by the
all the groups but especially those who have been in potential or actual risk taken by investors to formulate
the profession for an extensive time. Conversely, one the items of the Risk Propensity Measurement Scale.
may argue that the participants’ high degree of confi- The investors on average scored just below the mean
dence is based on accurate perceptions of their indus- for amount of risk taken compared to fellow investors;
try-specific competence. The availability heuristic of a sense of thrill when taking decision for which the out-
past success may play a role in this self-analysis comes are not known; an openness to take somewhat
(Tversky and Kahneman 1973). According to more risk to ensure greater returns; a sense that greater
Kahneman (2011), expertise is difficult to attain in risk taking leads to higher performance and a prefer-
this industry given the volatility in the investment ence for risk taking, and for taking larger, but reason-
markets, and since repeatable tasks is a prerequisite of able, risks. The mean was lower than the mean for the
expertise (Kahneman and Klein 2009). We have also associated risk tolerance scale in a study (Hoffman et al.
noted in the literature that expertise relates to hind- 2015) that found a correlation between risk tolerance
sight bias, and thus overreliance on the past (Knoll and risk expectation. Canadian studies with the Dospert
and Arkes 2017). found mean scores for financial risk taking of 49% and
Given the above, and that past successes are not pre- 44% for two samples (Weber, Blais, and Betz 2002),
dictors of future outcomes (Fischhoff 1975), we can whereas our associated tool yielded a comparative risk
infer that the higher than average rating of the very propensity score of 44%.
experienced investors in this sample still represents a The lower than midpoint score obtained is not
degree of overconfidence. The lower rating of the most indicative of a lack of risk-seeking behavior for this
inexperienced group is indicative of a lower level of sample. Comparative studies with other population
bias rather than actual performance or competence. groups and further norms are required to evaluate
level of risk propensity. However, it is notable that the
respondents didn’t rate their risk propensity highly.
Investor Risk Propensity
We did not find differences based on demographics,
Investor risk propensity appeared as a normal distri- or as one may expect based on previous research, that
bution for this data set (M ¼ 2.61, SD ¼ 1.00) and investors with more experience (Heath and Tversky
there were no significant differences based on demo- 1991; Hoffmann, Post, and Pennings 2015) or some-
graphics. For risky choice propensity, the descriptive times less experience were likely to take on more risk
statistics indicated a mean closer to a preference for (Goetzmann and Kumar 2008), or that female invest-
certainty than risk (M ¼ 0.38, SD ¼ 0.15). On com- ors would be more risk averse (Ch’ng 2017).
paring the converted percentile scores of the two risk The risky choice propensity measure used similar
propensity scales through a matched pair t-test, we comparisons to Kahneman and Tversky (1979) but
found that there was a difference in mean of the two incorporated a further four items that contained both
measures of risk propensity between 3–9% either posi- a potential loss and a potential gain (see Table 4).
tively or negatively. We also found a weak positive We found differences from the outcomes of the
correlation between the two measures at a 10% confi- original study for our sample. One question offered
dence interval (r(161) ¼ 0.14, p ¼ 0.086). investors a choice between the following options: A: a
It was essential to develop a scale to measure the 45% chance of R6 million and a 55% chance of noth-
risk propensity of the investment community based on ing; or B: a 90% chance of R3 million and a 10%

Table 4. Comparison of prospect theory risk items for original and current study.
Options Kahneman and Tversky (1979) Current Study
A B A B A B
1 (4000, 0.8) (3000) 80% 20% 80% 20%
2 (4000, 0.2) (3000, 0.25) 65% 35% 60% 40%
3 (3000, 0.9) (6000, 0.45) 86% 14% 13% 87%
4 (4000, 0.8) (3000) 92% 8% 76% 24%
5 (4000, 0.2) (3000, 0.25) 42% 58% 66% 34%
6 (3000, 0.9) (6000, 0.45) 92% 8% 76% 24%
10 S. COMBRINK AND C. LEW

Table 5. Pearson correlation results.

Underdog Risk Risky
bias Overconfidence propensity choice Age Gender Education Risk env.
Underdog Bias
Overconfidence 0.240
Risk Propensity 0.129 0.064
Risky Choice –0.115 0.070 0.136
Age –0.153 0.187 0.045 0.046
Gender 0.244 0.064 0.060 0.044 .0228
Education –0.025 0.065 –0.076 –0.133 0.155 –0.101
Risk environment 0.033 –0.035 .045 0.049 0.159 0.034 0.009
Experience 0.121 0.229 0.018 –0.010 0.744 –0.197 0.106 –0.123
N ¼ 161.
p <.00, p <.05.

Table 6. Linear Regression Results for Underdog Bias and Overconfidence Biases (controlling for demographic variables).
Overconfidence Overconfidence Model 2
B SE B b B SE B b
Underdog bias –2.78 0.89 –0.24 –2.59 0.92 –0.22
Age 0.05 1.41 0.00
Gender 0.12 0.26 0.04
Education 1.30 2.94 –0.08
Experience –0.02 0.08 0.04
Investment environment –0.26 1.10 –0.02
R2 0.06 0.10
R2 adjusted 0.05 0.01
F change 9.76 1.49
N ¼ 161.
p < .05, p < .01.

chance of nothing. Using the original certainty equiva- We therefore found comparative results for the
lent formula in the outlier question 3 we calculated Investor Risk Propensity Scale that operationalized risk
that option A had a certainty equivalent of 2038.39 as perceived behavior and the prospect theory-based
and option B a certainty equivalent of 2089.41. measure that evaluated actual probability calculations
According to prospect theory the investors should to avoid uncertainty and optimize value or gain. Both
favor option B which our sample did, contrary to the behavior and perception did not yield a high propen-
original study (Kahneman and Tversky 1979). We rea- sity to take risks. One could consider whether the fact
son that the reason our investment professionals pre- that only 29% of the investors operated in a high risk
ferred a higher probability lower loss option was context impacted the risk propensity result, as Bucciol
because their loss-gain ratio differed from that of the and Miniaci (2018) found a relationship between per-
original population. sonal risk experience and risk taking behavior.
The outcome of a further question required investi- Overall, the descriptive statistics indicate that the
gation. This question offered the investors a choice investment decision-makers do not display underdog
between the following options: A: a 20% chance of bias but experienced investors do display a degree of
losing R4 million and an 80% chance of losing noth- overconfidence. They do not display risk propensity.
ing, or B: a 25% chance of losing R3 million and a The question remains whether there are relations
75% chance of losing nothing. Since the probabilities among these two biases and risk propensity.
of 20% and 25% are at face value negligible, it may be Table 5 displays the correlation results between the
that our investors simplified the problem by rounding variables of the study.
off the figures (Kahneman and Tversky 1979), result-
ing in them only comparing the lower outcome loss.
Underdog Bias and Self-Rated Performance
For analyzing results, options with higher expected
value and lower certainty based on certainty equiva- When testing for the relationships between underdog
lent scores, were coded as the riskier options. The bias and self-rated performance we found a weak
overall mean score confirmed fewer risky choices (or negative correlation (r(161) ¼ –0.24, p < 0.001).
lower risk propensity) for our sample. This applied Regression analysis (see Table 6) indicated that under-
across the demographic groupings. dog bias negatively predicted self-rated performance
 JOURNAL OF BEHAVIORAL FINANCE 11

(R2 ¼ 0.0578, F(1,161) ¼ 9.76, p < 0.001). Hypothesis attitudes and behavioral outcomes of those who dis-
1, indicating a relationship between underdog bias play high levels of underdog bias instead.
and self-rated performance is supported, and an
inverse relationship is confirmed.
Self-Rated Performance and Risk Propensity
We reason that the availability heuristic (Tversky
and Kahneman 1973) may be the mechanism that For this sample we also found that self-rated perform-
explains the relationship. The inverse relationship ance (and related overconfidence) did not relate to or
between underdog bias and self-rated performance predict personal risk propensity. The correlations
means that the less the investors felt that they had between these variables were also weak and insignifi-
more difficult work requirements, or ability to gain cant (r(161) ¼ 0.07, p > 0.05). As may then be
recognition, the more they rated their own ability to expected, self-rated performance did not significantly
produce investment success, succeed in their careers predict personal risk propensity (R2 ¼ 0.00, F(1,161)
and belong in the industry. The implication is that a ¼ 0.60, p ¼ 0.42). We hypothesized this relationship
sense of an enabling environment may have an oppos- based on knowing that high self-rated performance
ite effect on self-rating of performance in relation to predicts risk perception (Slovic et al. 1981), and over-
others. We caution however that the significant rela- confidence leads to risk propensity through planning
tionship found was weak and more work should be fallacy (Lovallo and Kahneman 2003).
done to understand which other factors interrelate The lack of support for hypothesis 3 means that there
was no relationship for our smaple between how they
with a perception of headwinds in life and confidence.
rated their own successes and the propensity to take risk,
An optimism bias (Lovallo and Kahneman 2003), or
or choose riskier options when faced by two choices.
positive feedback (Schroeder, Caruso, and Epley 2016)
Since King and Slovic (2014) have shown that feelings
may be among those contributing factors, which
may blur the riskiness of situations, or instinctive and
require further investigation.
irrational thought may take over (Evans 2008), other fac-
tors that impact on risk propensity should be explored.
Underdog Bias and Personal Risk Propensity It would seem that, in the option analysis of risks,
our investors fared well in taking certain decisions.
The results indicated a weak and insignificant correl-
We wonder whether the exposure to risk contexts
ation (see Table 5), and therefore no predictive rela-
have made the investors more or less cognizant of
tionships between underdog bias and personal risk
risk assessment and analysis during decision-making.
propensity (r(161) ¼ 0.129, p > 0.05) and risky choice
propensity (r(161) ¼ –0.115, p > 0.05). Consequently
underdog bias also did not relate to personal risk pro- Conclusion
pensity (R2 ¼ 0.01, F(1,161) ¼ 2.67, p ¼ 0.10) or risky We set out in this paper to examine the level and
choice propensity (R2 ¼ 0.01, F(1,161) ¼ 2.13, potential impact of underdog bias, confidence levels
p ¼ 0.15) (Table 5). Underdog bias therefore did not and perceived and choice-based risk propensity on
significantly predict the tendency to take risk. investment decision-makers, and whether these varia-
We therefore reject the hypothesized relationship bles are interrelated. Understanding how these varia-
between underdog bias and personal risk propensity bles interrelate would hold implications for the
(hypothesis 2). In other words, we did not find that a investment community on how to improve decision-
biased view of overcoming hardships to reach success making. By using a new measure of risk propensity
had any relation to the risk propensity of the invest- our study made a modest empirical contribution to
ors. We do know from literature that there is a rela- assess risk propensity for investors more easily as we
tionships between a strong sense of conviction in were able to compare the outcomes thereof with the
one’s views and the tendency to take on risk risk propensity scale based on decisions between
(Hoffmann, Post, and Pennings 2013). We also know options according to prospect theory.
that a sense of control may relate to risk taking The findings firstly indicated relatively low levels of
behavior (Gilovich and Douglas 1986). We had underdog bias for the sample. We found that the invest-
assumed that wanting to right the wrongs of the past, ment professionals did not think that past barriers were
ensuring fairness, may result in taking more chances more significant than enablers in their work and
(Davidai and Gilovich 2016), but this was not the careers. This is the only known empirical support for
case. A question remains regarding other potential the potential boundary effect that Davidai and Gilovich
12 S. COMBRINK AND C. LEW

(2016) describes, namely better recall of people based The other two hypotheses, that examined the rela-
benefits versus event based obstacles, and goes against tionships between both underdog bias and self-rated
several of their findings that confirm a tendency performance and risk propensity, were not supported.
towards underdog bias. Personal, rather than imper- As we expand insights into decision bias, this finding
sonal conditions such as market forces, may have led to makes a contribution in the understanding that percep-
a more positive experience. Our sample reports an ena- tive biases, such as underdog bias and overconfidence,
bling interpersonal environment. The study contributes can occur independent of a person’s risk propensity.
to an understanding of underdog bias of investors and Likewise it means that a tendency to seek out
future studies would need to explore whether benefits risky options does not necessarily go hand in hand
from people are better recalled than obstacles. with the tendency to overrate one’s own performance.
Secondly, we found that the experienced invest- Moreover, it means that an investor may be circum-
ment professionals rated themselves above average in spect and seek out certain options and not take on
performance in comparison to their peers, a rating risks, even while being overconfident in their perform-
similar to those of different types of populations ance. Also importantly, when investors overrate their
(Williams and Gilovich 2008; Guenther and Alicke performance, this does not necessarily predict that they
2010), and which therefore confirms the presence of will also make risky investment decisions. An interest-
overconfidence bias, but only for certain levels of ing further exploration would examine the necessary
experience. This contributes not only to literature on conditions for actual past risk behavior rather than risk
overconfidence bias, but also to a growing body of lit- propensity, in relation to decision bias. Since other
erature that investigates the relationship between research suggests a relationship between overconfidence
expertise and decision bias (Knoll and Arkes 2017). and observed risk taking behavior in practice (Merkle
Thirdly, based on both the descriptive measure of 2017), our research shows that more work is required
risk propensity and another that draws the seminal to understand the conditions under which overconfi-
Kahneman and Tversky (1979) scale, we did not find dence would predict risk propensity.
a higher risk propensity for the investors. The pro- Further research among investors may explore
spect theory-based scale yielded similar results to the aspects of the self-regulation of risk propensity. New
original work, with the exception of items that were measures of risk propensity should also account for a
explained through the certainty equivalent value calcu- potential best foot being put forward by respondents
lations (Tvesrky and Kahneman 1992). This finding (Stephens-Davidowitz 2014) by including an assessment
adds to previous literature that yielded varying results of scenario-based assessments and actual behavior.
on the risk propensity of decision makers (Djeriouat The implication for the findings in business in an
2017, Iqbal 2013, King and Slovic 2014). emerging market is reassuring as cautiousness of the
Although the measure of personal risk propensity investors, as well as their level of self-rated perform-
requires thorough validation, we established internal reli- ance are similar to what may be expected in the gen-
ability for this new tool that is different to scales which eral population. However, further interventions may
either reflect the perception of market risk (Hoffmann, be developed to help make investors aware of any bias
Post, and Pennings 2013) or individual choice of options in terms of their own knowledge, so as to make better
(Tversky and Kahneman 1992). Yet the instrument use of information sources in decision-making.
yields similar results to comparable but less focused Moreover, when investors do not see the barriers in
scales on risk (e.g., Weber, Blais and Betz 2002). We their careers as having been caused by themselves, this
also found only a 3–9% difference between the two may hold implications for their view of the causes of
assessments of risk propensity used in the study. mistakes. This again highlights the need for the devel-
Testing of the three hypotheses on the relation- opment of sound perspectives on investor performance.
ships among the three variables, we found an inverse Early research has shown that individuals with a high
relationship between underdog bias and self-rated internal locus of control, with good cognitive complex-
performance. A weak tendency to see the obstacles ity skills and who are introverts, are better at self-
over enablers would predict a higher rating of self- assessing accurately (Yammarino and Atwater 1993).
performance and related overconfidence. This implies More work should be done to determine how to
that the investors in this sample, who experienced develop accurate self-perceptions among investors.
being enabled by their interpersonal environment, Limitations of this research include the need for
did not rate their own performance higher relative to validation of the newly developed Investment Risk
their peers. Propensity Scale, possible biased responses by the
 JOURNAL OF BEHAVIORAL FINANCE 13

investors, and the limitation of the scope of the sam- Goetzmann, W. N., and A. Kumar. 2008. “Equity Portfolio
pling frame to the South African investment popula- Diversification.” Review of Finance 12 (3):433–63. doi:10.
1093/rof/rfn005
tion. These limitations call for further research on bias
Graves, S. B., and J. Ringuest. 2018. “Overconfidence and
and risk propensity of investors in different contexts. Disappointment in Venture Capital Decision Making: An
The measurement of investment decision-making was Empirical Examination.” Managerial & Decision
limited to the proxy investment examples used by Economics 39 (5):592–600.
Kahneman and Tversky (1979) and did not include Guenther, C. L., and M. D. Alicke. 2010. “Deconstructing
actual investment decisions. Future research should the Better-Than-Average Effect.” Journal of Personality
and Social Psychology 99 (5):755–70. doi:10.1037/
incorporate measures of practice-based investment a0020959
decision-making. Control variables should also be Heath, C., and A. Tversky. 1991. “Preference and Belief:
introduced better to understand the conditions that Ambiguity and Competence in Choice under
impede or enable underdog bias. Uncertainty.” Journal of Risk and Uncertainty 4 (1):5–28.
doi:10.1007/BF00057884
Hoffmann, A., T. Post, and J. Pennings. 2013. “Individual
References Investor Perceptions and Behavior during the Financial
Crisis.” Journal of Banking & Finance 37 (1):60–74. no.
Barberis, N., A. Mukherjee, and B. Wang. 2016. “Prospect doi:10.1016/j.jbankfin.2012.08.007
Theory and Stock Returns: An Empirical Test.” Review of Hoffmann, A., T. Post, and J. Pennings. 2015. “How
Financial Studies 29 (11):3068–107. doi:10.1093/rfs/ Investor Perceptions Drive Actual Trading and Risk-
hhw049 Taking Behavior.” Journal of Behavioral Finance 16 (1):
Baye, M., and J. Prince. 2013. Managerial Economics and 94–103. doi:10.1080/15427560.2015.1000332
Business Strategy, Global Edition. Berkshire: McGraw-Hill Iqbal, M. M. 2013. “Choice under Cncertainty; ’Allais
Education. Paradox’ and Its Paradoxical Implication.” Journal of
Blais, A., and E. U. Weber. 2006. “A Domain-Specific Risk- Business & Economics 5 (2):129–48.
Taking (DOSPERT) Scale for Adult Populations.” Jaccard, J., M. A. Becker, and G. Wood. 1984. “Pairwise
Judgment and Decision Making 1 (1):33–47. Multiple Comparison Procedures: A Review.”
Bucciol, A., and R. Miniaci. 2018. “Financial Risk Psychological Bulletin 96 (3):589–96. doi:10.1037/0033-
Propensity, Business Cycles and Perceived Risk 2909.96.3.589
Exposure.” Oxford Bulletin of Economics and Statistics 80 Judge, T. A., A. Erez, J. E. Bono, and C. J. Thoresen. 2003.
(1):160–83. doi:10.1111/obes.12193 “The Core Self-Evaluations Scale: Development of a
Ch’ng, K. 2017. “Effects of Cognitive Appraisal Pattern on Measure.” Personnel Psychology 56 (2):303–31. doi:10.
Probability Weighing Function and Risk Behavior 1111/j.1744-6570.2003.tb00152.x
between Genders.” International Journal of Business and Kahneman, D. 2011. Thinking Fast and Slow, First Edition.
Society 18 (1):157. London: Penguin Books.
Davidai, S., and T. Gilovich. 2016. “The Headwinds/ Kahneman, D., and G. Klein. 2009. “Conditions for
Tailwinds Asymmetry: An Availability Bias in Intuitive Expertise: A Failure to Disagree.” American
Assessments of Barriers and Blessings.” Journal of Psychologist 64 (6):515–26. doi:10.1037/a0016755
Personality and Social Psychology 111 (6):835–51. doi:10. Kahneman, D., and A. Tversky. 1979. “Prospect Theory: An
1037/pspa0000066 Analysis of Decision under Risk.” Econometrica: Journal
Djeriouat, H. 2017. “Self-Determination and Risk: The Role of the Econometric Society 47 (2):263–91. doi:10.2307/
of Life Goals and Causality Orientation in Domain- 1914185
Specific Risk Propensity.” Journal of Risk Research 20 (2): King, J., and P. Slovic. 2014. “The Affect Heuristic in Early
256–76. doi:10.1080/13669877.2015.1043573 Judgments of Product Innovations.” Journal of Consumer
Evans, J. S. B. 2008. “Dual-Processing Accounts of Behaviour 13 (6):411–28. doi:10.1002/cb.1491
Reasoning, Judgment, and Social Cognition.” Annual Kliger, D., and A. Kudryavtsev. 2010. “The Availability
Review of Psychology 59 (1):255–78. doi:10.1146/annurev. Heuristic and Investors’ Reaction to Company-Specific
psych.59.103006.093629 Events.” Journal of Behavioral Finance 11 (1):50–65. doi:
Fischhoff, B. 1975. “Hindsight is Not Equal to Foresight: 10.1080/15427561003591116
The Effect of Outcome Knowledge on Judgment under Knoll, M. A. Z., and H. R. Arkes. 2017. “The Effects of
Uncertainty.” Journal of Experimental Psychology: Human Expertise on the Hindsight Bias.” Journal of
Perception and Performance 1 (3):288–99. doi:10.1037// Behavioral Decision Making 30 (2):389–99. doi:10.
0096-1523.1.3.288 1002/bdm.1950
Gerber, J. P., L. Wheeler, and J. Suls. 2018. “A Social Koo, J.-H., and D. Yang. 2018. “Managerial Overconfidence,
Comparison Theory Meta-Analysis 60þ Years On.” Self-Attribution Bias, and Downwardly Sticky Investment:
Psychological Bulletin 144 (2):177–97. doi:10.1037/ Evidence from Korea.” Emerging Markets Finance and
bul0000127 Trade 54 (1):144–61. doi:10.1080/1540496X.2017.1398643
Gilovich, T., and C. Douglas. 1986. “Biased Evaluations of Kudryavtsev, A. 2018. “The Availability Heuristic and
Randomly Determined Gambling Outcomes.” Journal of Reversals following Large Stock Price Changes.” Journal
Experimental Social Psychology 22 (3):228–41. doi:10. of Behavioral Finance 19 (2):159–76. doi:10.1080/
1016/0022-1031(86)90026-0 15427560.2017.1374276
14 S. COMBRINK AND C. LEW

Lovallo, D., and D. Kahneman. 2003. “Delusions of success. Taleb, N. N. 2007. The Black Swan. London: Random
How optimism undermines executives’ decisions.” House.
Harvard Business Review 81 (7):56–63. Tamborski, M., R. P. Brown, and K. Chowning. 2012. “Self-
Madan, C. R., E. A. Ludvig, and M. L. Spetch. 2017. “The Serving Bias or Simply Serving the Self? Evidence for a
Role of Memory in Distinguishing Risky Decisions from Dimensional Approach to Narcissism.” Personality and
Experience and Description.” Quarterly Journal of Individual Differences 52 (8):942–6. doi:10.1016/j.paid.
Experimental Psychology 70 (10):2048–59. doi:10.1080/ 2012.01.030
17470218.2016.1220608 Taylor, S. E. 1991. “Asymmetrical Effects of Positive and
Merkle, C. 2017. “Financial Overconfidence over Time:
Negative Events: The Mobilization-Minimization
Foresight, Hindsight, and Insight of Investors.” Journal of
Banking & Finance 84:68–87. doi:10.1016/j.jbankfin.2017. Hypothesis.” Psychological Bulletin 110 (1):67–85. doi:10.
07.009 1037/0033-2909.110.1.67
Michailova, J., and U. Schmidt. 2016. “Overconfidence and Thaler, R. 1980. “Toward a Positive Theory of Consumer
Bubbles in Experimental Asset Markets.” Journal of Choice.” Journal of Economic Behavior & Organization 1
Behavioral Finance 17 (3):280–92. doi:10.1080/15427560. (1):39–60. doi:10.1016/0167-2681(80)90051-7
2016.1203325 Tversky, A., and D. Kahneman. 1973. “Availability: A
Morgenroth, T., and M. K. Ryan. 2018. “Addressing Gender Heuristic for Judging Frequency and Probability.”
Inequality: Stumbling Blocks and Roads Ahead.” Group Cognitive Psychology 5 (2):207–32. doi:10.1016/0010-
Processes & Intergroup Relations 21 (5):671–7. doi:10. 0285(73)90033-9
1177/1368430218786079 Tversky, A., and D. Kahneman. 1992. “Advances in
Novemsky, N., and D. Kahneman. 2005. “The Boundaries Prospect Theory: Cumulative Representation of
of Loss Aversion.” Journal of Marketing Research 42 (2): Uncertainty.” Journal of Risk and Uncertainty 5 (4):
119–28. doi:10.1509/jmkr.42.2.119.62292 297–323. doi:10.1080/09603107.2011.583393
Nutter, F. 2010. In Encyclopedia of Research Design, Edited Weber, E. U., A. Blais, and N. E. Betz. 2002. “A Domain-
by N. Salkind. Thousand Oaks: SAGE Publications Inc.
Specific Risk-Attitude Scale: Measuring Risk Perceptions
Rabin, M., and R. H. Thaler. 2001. “Anomalies: Risk
and Risk Behaviors.” Journal of Behavioral Decision
Aversion.” Journal of Economic Perspectives 15 (1):
219–32. doi:10.1257/jep.15.1.219 Making 15 (4):263–90. doi:10.1002/bdm.414
Ross, M., and F. Sicoly. 1979. “Egocentric Biases in Wegner, T. 2016. Applied Business Statistics: Methods and
Availability and Attribution.” Journal of Personality and Excel-based Applications (Fourth Edition). Cape Town:
Social Psychology 37 (3):322–36. doi:10.1037/0022-3514. Juta and Company Ltd.
37.3.322 Williams, E. F., and T. Gilovich. 2008. “Do People Really
Rozin, P., and E. B. Royzman. 2001. “Negativity Bias, Believe They Are above Average?” Journal of
Negativity Dominance, and Contagion.” Personality and Experimental Social Psychology 44 (4):1121–8. doi:10.
Social Psychology Review 5 (4):296–320. doi:10.1207/ 1016/j.jesp.2008.01.002
S15327957PSPR0504_2 Yammarino, F. J., and Leanne E. Atwater. 1993.
Schroeder, J., E. M. Caruso, and N. Epley. 2016. “Many “Understanding Self-Perception Accuracy: Implications
Hands Make Overlooked Work: Over-Claiming of for Human Resource Management.” Human Resource
Responsibility Increases with Group Size.” Journal of Management 32 (2/3):231–47. doi:10.1002/hrm.
Experimental Psychology: Applied 22 (2):238–46. doi:10. 3930320204
1037/xap0000080 Zhong, X., and J. Wang. 2018. “Prospect Theory and
Simon, H. 1955. “A Behavioral Model of Rational Choice.”
Corporate Bond Returns: An Empirical Study.” Journal of
The Quarterly Journal of Economics 69 (1):99–118. doi:10.
Empirical Finance 47:25–48. doi:10.1016/j.jempfin.2018.
2307/1884852
Sitkin, S. B., and A. L. Pablo. 1992. “Reconceptualizing the 02.005
Determinants of Risk Behavior.” Academy of
Management Review 17 (1):9–38. doi:10.5465/amr.1992.
4279564
Slovic, P., B. Fischhoff, S. Lichtenstein, and F. Roe. 1981.
Appendix
“Perceived Risk: Psychological Factors and Social Appendix A: Measuring Instruments Underdog Bias
Implications.” Paper Presented at the Proceedings of the
Royal Society of London A: Mathematical, Physical and 1. When I source research, I have to work harder than
Engineering Sciences 376 (1764):17–34. doi:10.1098/rspa. others to achieve the same result.
1981.0073 2. My investors are more demanding than other investors,
Stanovich, K. E., and R. F. West. 2000. “Individual even when I produce the same results.
Differences in Reasoning: Implications for the Rationality 3. I have to work harder than others to get the recogni-
Debate?” Behavioral and Brain Sciences 23 (5):645–65. tion I deserve.
doi:10.1017/S0140525X00003435 4. My investors tend to blame me more harshly than
Stephens-Davidowitz, S. 2014. “The Cost of Racial Animus others when the market takes a downturn.
on a Black Candidate: Evidence Using Google Search 5. My peers get more recognition than they deserve com-
Data.” Journal of Public Economics 118:26–40. doi:10. pared to the work that they have done.
1016/j.jpubeco.2014.04.010 Source: Authors’ own.
 JOURNAL OF BEHAVIORAL FINANCE 15

Overconfidence Bias/Self-Rated Performance 6. I tend to take large but reasonable risk in my invest-
ment decisions.
(Adapted from the Core Self-Evaluation Scale (CSES) Source: Authors’ own.
(Judge et al. 2003) and the overconfidence continuum
(Williams and Gilovich 2008)
Please answer the following questions in relation to where Risky Choice Propensity (Prospect Theory)
you perceive yourself to be when compared to your peers.
For example: Where do you rate in terms of sprinting at (Based on Kahneman and Tversky 1979)
your local running club? In this section please answer the question as to which
Top: 66% Slightly above average option you would prefer (either A or B).
For the following questions please assume you adminis-
ter a fund or investment portfolio of R10 million.

1. A: 80% chance of R4 million and a 20% chance of

nothing, or
Where do you rate in terms of weight lifting at your B: 100% chance of R3 million
local gym? 2. A: 20% chance of R4 million and a 80% chance of
Top: 32%Below average nothing, or
B: 25% chance of R3 million and a 75% chance
of nothing
3. A: 45% chance of R6 million and a 55% chance of
nothing, or
B: 90% chance of R3 million and a 10% chance
1. I achieve the success I deserve at work. of nothing
2. Sometimes I get depressed with the investment indus- 4. A: 80% chance of losing R4 million and a 20% chance
try.  of nothing, or
3. I generally generate returns above my targets. B: 100% chance of losing R3 million
4. Sometimes when I fail, I feel like I can’t choose the 5. A: 20% chance of losing R4 million and an 80%
right investments.  chance of losing nothing, or
5. I complete my tasks successfully. B: 25% chance of losing R3 million and a 75% chance
6. Overall, I am satisfied with my performance. of losing nothing
7. I do not feel in control of my success in my career.  6. A: 45% chance of losing R6 million and a 55% chance
8. I am coping with most of the problems at work. of losing nothing, or
(Reverse scored items marked with asterisk) B: 90% chance of losing R3 million and a 10% chance
of losing nothing
7. A: 50% chance of losing R2 million and a 50% chance
Investor Risk Propensity Scale of gaining R4 million, or
For each of the following statements please indicate the B: 60% chance of losing R3 million and a 40% chance
level it normally applies to you. of gaining R6 million
Please select an answer on the following 7-point Likert 8. A: 30% chance of losing R3 million and a 70% chance
scale: (Entirely Agree/ Mostly Agree/ Somewhat Agree/ of gaining R2 million, or
Neither Agree or Disagree/ Somewhat Disagree/ Mostly B: 50% chance of losing R3 million and a 50% chance
Disagree/ Entirely Disagree). of gaining R5 million
9. A: 80% chance of losing R1 million and a 20% chance
1. I take more risk than my fellow investors. of gaining R5 million, or
2. I get a thrill by taking decisions that I don’t know the B: 30% chance of losing R2 million and a 70% chance
outcome of. of gaining R1 million
3. If I could take a little bit more risk, I could secure a 10. A: 30% chance of gaining R2 million and a 70%
higher return for my fund. chance of losing R1 million, or
4. The more risk I take the better I perform. B: 20% chance of gaining R4 million and an 80%
5. I like to take risks. chance of losing R1.5 million