Impact of Overoptimism and Overconfidence on Economic Behavior

Table of ContentS

1. Introduction

2. Literature Review
2.1. Definitions
2.2. Characteristics
2.2.1. Inherent Bias Structure
2.2.1.1. Interconnectivity and Endogeneity
2.2.1.2. Shape of Distribution
2.2.2. Agent Characteristics
2.2.2.1. Biological Properties
2.2.2.2. Mental Abilities
2.2.3. Project Characteristics
2.2.4. Agent-Project Relationship
2.3. Impact on Economic Behavior
2.3.1. Impact on Private Behavior
2.3.2. Impact on Business Behavior
2.3.2.1. Bias and Hierarchy
2.3.2.2. Bias and Performance
2.3.2.3. Bias and Investment Valuation
2.3.2.3.1. Overconfidence and Overoptimism
2.3.2.3.2. Risk Aversion
2.3.2.3.3. Time Preference
2.3.2.4. Bias and Financing Decisions
2.3.2.4.1. Asymmetric Information
2.3.2.4.2. Misalignment of Interests
2.3.2.4.3. Deviating Expectations
2.4. Impact on Company Value and Overall Welfare
2.5. Preliminary Conclusion

3. Measurement Methods
3.1. Methods based on Qualitative Expressions
3.2. Methods based on Subjective Valuation
3.2.1. Subjective Valuation and Overoptimism
3.2.1.1. Comparing Estimates to Exogenous Benchmarks
3.2.1.2. Comparing Estimates to Expected Value of Peer Group
3.2.1.3. Comparing Estimate of Individual Success to Estimate of Peer-Group Success
3.2.1.4. Conclusion on Measuring Overoptimism
3.2.2. Subjective Valuation and Overconfidence
3.2.2.1. Comparing Correctness of Intervals to Requested Confidence Level
3.2.2.2. Comparing Participants Confidence Level to Their Answer Correctness
3.2.2.3. Conclusion on Measuring Overconfidence
3.3. Action-based methods
3.4. Preliminary Conclusion

4. Empirical Evidence
4.1. Definitions
4.2. Data
4.3. Bias and Gender
4.4. Bias and Industry Affiliation
4.5. Bias and Company Life Cycle
4.6. Bias and Remuneration Risk Profile
4.7. Bias and Individual Success
4.8. Preliminary Conclusion

5. Conclusion

6. Appendix

7. Bibliography

List of Figures

Figure 1: Organization of Literature Review

Figure 2: Summarized Findings of Literature Review

List of Tables

Table 1: Definition of Variables

Table 2: Variables Characteristics

Table 3: Correlation Matrix

Table 4: Bias according to Gender

Table 5: Significance of Bias Differences amongst Genders

Table 6: Bias according to Industry Affiliation

Table 7: Significance of Bias Differences amongst Industries

Table 8: Bias according to Company Life Cycle

Table 9: Significance of Bias Differences amongst Company Life Cycle

Table 10: Regression Analysis Risk Profile

Table 11: Regression Analysis Success

List of Abbreviations

illustration not visible in this excerpt

List of Symbols

illustration not visible in this excerpt

1. Introduction

Economic theory normally focuses on rational agents optimizing individual utility.^[1] Since the second half of the 20th century, this viewpoint has been enriched by findings from the field of psychology. A new trait of research was created called “behavioral economics”^[2]. It takes into account subjective characteristics such as asymmetric preference and judgment, or limits of rational processing, willpower, and greed.

This paper aims to give an overview of two related human traits that have attracted particularly wide interest, namely overconfidence and overoptimism. The two are closely related to each other, and often used synonymously. Broadly speaking, overconfidence results in underestimation of future risks, e.g. the riskiness of future cash flows, whilst overoptimism leads to an overestimation of future positive outcomes, e.g. the future returns of a company.^[3] Besides, the paper wants to deduct suggestions for further research, by systematically identifying uncovered topics in existing literature.

Usually Alpert and Raiffa (1969) are credited with the first discovery of overconfidence.^[4] However, the most influential study is probably Russo and Schoemaker (1992).^[5] It was published in the Sloan Management Review and communicated the topic to a broader audience for the first time. In particular, it revealed that assumingly rational managers were prone to overconfidence, too. This challenged traditional management doctrines and generated interest in a better understanding of the topic and further research. To exemplify overconfidence, Russo and Schoemaker (1992) asked managers to give numerical intervals for ten general-knowledge questions, such that nine out of the ten answers would be correct. On average participants included the correct value within their interval only 5 out of 10 times, i.e. they underestimated potential errors in their estimations.^[6]

Svenson (1981) is probably the most influential source regarding overoptimism.^[7] He made the subject intuitively understandable and established a standard measurement method that could be easily used for subsequent research. To give an example of overoptimism: Svenson (1981) asked students to compare their driving skills to those of their classmates. Roughly 80% believed they belonged to the top 50%, i.e. they overestimated their abilities.^[8]

This paper also provides a closer look at the empirical methods normally applied in field studies. Although the phenomena are intuitively understandable, empirical research still presents itself as a mosaic of fragmented testing rather than a coherent framework. One may assume that this is mainly caused by the difficult measurability of overconfidence and overoptimism: On the one hand, the decision maker, convinced of his own rationality, contributes zero overconfidence or overoptimism to his actions. On the other hand, even a neutral observer cannot specify any degree of biasedness a priori, as stochastic outcomes per definition do not allow for perfect prediction. Therefore, scientists frequently rely on proxy variables that at least allow for measuring a group’s average overoptimism or overconfidence.

Furthermore, this paper empirically examines several considerations regarding existing research and measurement methods. It particularly aims to connect biasedness with certain personal and economic characteristics, namely participants’ gender, industry affiliation, company life cycle, success and risk preferences. Additionally, different methods are employed for measuring overoptimism. By comparing the strength of bias indicated by each scaling, one gets interesting insights into the influence that question design has on test results.

The remainder of the paper is organized as follows: In part two the definitions of overoptimism and overconfidence found in earlier studies are described and evaluated. The most appropriate definition then helps to sort and synthesize existing research, and to discover blank fields for future research (0). In part three, methods used to measure biasedness are compared regarding their reliability and significance. Findings allow for recommending certain test designs for future surveys (3). Part four builds on these findings and empirically examines the relationship between biasedness and several personal and economic characteristics. Additionally, the reliability and significance of several scalings capturing overoptimism is tested (4). Part five concludes (5).

2. Literature Review

The following chapters aim to convey a broad understanding of overoptimism and overconfidence in a bottom-up fashion. This includes definitions and theoretical models as well as empirically-verified correlations with certain circumstances, characteristics and decisions. As overconfidence and overoptimism are closely related to each other and follow similar structures, both biases are mostly described in parallel.

The paper suggests structuring existing research on overoptimism and overconfidence as visualized in figure 1. On the one hand, this hierarchy allows for differentiating between very specific bias patterns and broad economic implications. On the other hand, it helps to identify influencing factors and the impacts of their horizontal relationships. However, as the selection of papers remains limited and subjective, comments and suggestions on structure and contents are welcome.

First of all, the phenomena are defined building on existing research (2.1). Second, characteristics of overoptimism and overconfidence are described. These include inherent structures, dependencies on agent and project characteristics, and influence of agent-project relationship (2.2). Third, the impact of biasedness on economic behavior is looked at in both private as well as professional decisions (2.3). Finally, the influence of decisions on company value and overall welfare is analyzed (2.4).

illustration not visible in this excerpt

Figure 1: Organization of Literature Review

(Source: Own representation)

2.1. Definitions

In existing research, definitions of overoptimism and overconfidence usually differ according to the dimension looked at: One group of papers concentrates on individuals’ justification for biasedness, i.e. individual skills versus external development. The other group refers to the parameter underlying individuals’ expectations, i.e. risk estimations versus return estimations.

Camerer and Lovallo (1999) or Malmendier and Tate (2005) for example belong to the first group.^[9] Accordingly, an overconfident person expects his own behavior to be abnormally successful, which is often referred to as the “better-than-average effect”^[10]. An overoptimistic person estimates events beyond his control to be outstandingly positive.^[11]

Other sources such as Kyle and Wang (1997), Odean (1998), Hackbarth (2004) and Brettel and Kasch (2006) belong to the second group.^[12] They define overoptimism and overconfidence looking at the underlying development characteristics of growth and volatility. An overoptimistic person predicts that favorable events are more likely or more positive than they actually are. In investment decisions for example he overestimates a project’s future returns or growth rates. An overconfident person believes they have more precise knowledge about unknown events than they actually have. In investment decisions for example he underestimates the riskiness or volatility of the project’s future cash flows.^[13] Hackbarth refers to overoptimism as “growth perception bias” and to overconfidence as “risk perception bias”^[14].

Both definitions described above are frequently used in literature, as each perspective contributes to the understanding of certain behavior nuances:

If one follows group one and differentiates according to bias justification, one can better argue for the extent or amplitude of biasedness. According to Weinstein (1980) for example, an overoptimistic person has even more positive expectations if they personally contributes to the project’s outcome.^[15] Similarly Adler (1980) finds that managers believe they are able to modify the risk of projects they are in control of.^[16] He calls such overconfident managers “risk makers”^[17]. These findings build on general psychological studies by Langer (1975).^[18] She finds that people normally believe they can influence outcomes that they demonstrably have no influence over. Langer (1975) calls this tendency an “illusion of control”^[19]. According to her studies, this bias is particularly strong in situations where certain skills come into play.

If one applies the definition of group two and differentiates according to underlying parameters, one can more precisely predict the type or effect direction of biasedness. Hackbarth (2004), for example, models that an overoptimistic manager might prefer other sources of financing than his overconfident counterpart.^[20] The resulting pecking order of financing might even contradict the traditional one justified by asymmetric information. This contradiction mainly rests in the different impacts that growth and risk expectations have on debt and equity valuations.

To capture the contributions made by both definitions, this paper combines both ideas and applies the following wording. As with group two, the term “overoptimism” describes a growth perception bias, whilst “overconfidence” stands for a risk perception bias. Where it is not possible or suitable to differentiate between these two, the paper talks of “expectation bias”, “bias”, or “biasedness”. Hereby, variable wording does not imply different meanings, but simply tries to avoid unaesthetic repetitions. As with group one, deviations from rational growth or volatility expectations are named “endogenous” if the expectation bias is justified by the agent’s (perceived) ability to influence outcomes, and “exogenous” if the bias relates to developments of the environment beyond the agent’s reach.

2.2. Characteristics

The following chapters explain and sort existing research on overoptimism and overconfidence. First of all, general bias patterns are described (2.2.1). Second, the influence of certain agent characteristics on biasedness is looked at (2.2.2). Third, the effect of project characteristics is examined (2.2.3). Fourth, the relationship between agent, project and biasedness is analyzed (2.2.4).

2.2.1. Inherent Bias Structure

Researchers have found several patterns in biasedness that remain independent of situational factors. The most prominent ones are described in this chapter. Although it cites papers on overoptimism only, it seems reasonable to assume that similar mechanisms are in place amongst overconfident individuals, too. First, the strength of bias is linked to heterogeneous beliefs and self-enhancing perception (2.2.1.1). Second, the impact of psychological traits on bias distribution is examined (2.2.1.2).

2.2.1.1. Interconnectivity and Endogeneity

Van den Steen (2002) models the connection between individual perception differences, overoptimism and further self-enhancing interaction with other biases.^[21] Accordingly, even in a group of rational agents with symmetric information, each member might result as being overoptimistic. Van den Steen (2002) observes that subjects disposing of the same information and cognitive abilities might naturally attribute varying subjective probabilities to the same outcomes. Based on their randomly different “priors”^[22], i.e. based on their individual beliefs, agents choose the project they expect to be most successful. Thus, each agent views their project more optimistically than the other agents do, and expects their future success to be above average as compared to their peer group. In other words, each member of the group is overoptimistic. Furthermore, when in a joint venture setting each agent chooses the work they consider to be most relevant to the project’s outcome, everyone will consider their own contribution to be more important than that of their colleagues. Miller and Ross (1975), for example, empirically verify that generally people expect their behavior to produce success.^[23]

Taylor and Brown (1988) explain the wide existence of overoptimism by its positive effect on mental health and psychological well-being.^[24] They find overoptimism with healthy participants, but rather accurate perceptions with depressed people. Their assertion contradicts traditional psychological theory which views accurate perception of the self as a precondition for mental balance, such as Maslow (1950) and Jahoda (1953).^[25] According to Taylor and Brown (1988), various modern studies find overly optimistic expectations amongst a broad range of social groups. This general pattern can best be explained by the positive effect overoptimism has on a person’s happiness, compassion and creativity.

In their field study, Miller and Ross (1975) find that an overoptimistic person contributes failure to bad luck rather than to his personal bad choice of action or missing skill.^[26] That person discerns a closer covariation between behavior and outcomes in the case of increasing success than in the case of constant failure. This argumentation helps people to strengthen their conviction and is therefore often called “self-serving bias”^[27].

According to Miller and Ross (1975), such self-serving bias also leads to insufficient learning as people see no need to alter their behavior.^[28] One can deduct that this predicts a relatively stable persistence of bias over age and job experience.

Summing up, biasedness is a natural psychological behavior which evolves from individually different character traits even amongst rational individuals. As it both contributes to well-being as well as dampens learning mechanisms it is probably rather constant over time and different social groups.

2.2.1.2. Shape of Distribution

Camerer and Malmendier (2004) point out that endogenous overoptimism seems to be unevenly distributed in itself.^[29] As described above, Svenson (1981) asks students to rank themselves within their peer group in terms of driving abilities.^[30] Interestingly, overoptimism is rather strong around the average categories, but weaker at the top: 24% of students believe they belong to the 71-80% best drivers, whilst only 2% place themselves into the 91-100% category. Camerer and Malmendier (2004) view this as a sign of modesty that keeps people from boasting that they are the best.^[31] Accordingly, this suggests lower biasedness in smaller, more socially-integrated groups. If verified, this would have important implications for organizational design.

Puri and Robinson (2007) examine the correlation between biasedness and private decisions.^[32] Within the group of overoptimistic people, they find two subgroups which they call “moderate optimists“ and “extreme optimists”^[33]. Their behavior patterns differ significantly: Moderate optimists show prudent financial behavior, whilst extreme optimists tend to make what one would normally consider irrational financial decisions.

Overall, these studies report a high amount of mildly biased individuals and a small amount of highly biased individuals within one population. The big, moderate group makes rather constructive decisions, whilst members of the small, extreme group tend towards destructive ones.

When further examining the impact of biasedness on (economic) behavior, researchers should pay special attention to this group of extremely biased individuals. Mild degrees of overoptimism or overconfidence seem to yield only neutral or maybe slightly deviating behavior. Extreme biasedness however significantly alters behavior and induces negative outcomes.

2.2.2. Agent Characteristics

Several studies have looked into the correlation between estimation bias and certain agent characteristics, especially gender, age, and intelligence. This chapter follows this focus and summarizes the findings on first, the influence of biological properties (2.2.2.1), and second, mental abilities (2.2.2.2). Again, where research exists on one bias only, it still seems reasonable to assume that overoptimism and overconfidence have similar impacts.

2.2.2.1. Biological Properties

Kovalchik et al. (2005) empirically study the relationship between aging and economic decision behavior.^[34] They observe more endogenous overconfidence amongst younger than amongst older participants.^[35] This means that older people are better able to assess the limits of their cognitive abilities. One explanation could be that they build on their more profound life experience, which helps them to countercheck their expectations.

Barber and Odean (2001) empirically test the relationship between biasedness and gender.^[36] They find higher estimation bias amongst male participants than amongst female. To measure biasedness, they rely on Odean (1998) who predicts that overconfident investors trade irrationally often and incur below-average returns.^[37] Actually, Odean only bases his model on endogenous overconfidence. Yet, higher growth expectations and lower risk discount rates both increase return estimations. Therefore, it seems reasonable that Barber and Odean (2001) view their results as evidence for overoptimism, too.

Overall, biasedness is found to be stronger amongst male individuals, and to diminish with increasing age.

2.2.2.2. Mental Abilities

Biais et al. (2005) cannot verify a statistically significant correlation between endogenous overconfidence and participants’ intelligence quotient (IQ) test score.^[38] This is in line with other studies, which found overconfidence amongst very different professions and social groups. Murphy and Brown (1984) and Malmendier and Tate (2005) for example find biasedness amongst weather forecasters as well as Fortune 500 CEOs.^[39]

Overall, biasedness seems to depend more on psychological characteristics than on cognitive abilities. It is independent of IQ, but – as mentioned in chapter 2.2.1.1 above – positively correlates to mental well-being.

However, no study has explicitly looked at whether biasedness indeed is similar throughout different industries. Therefore, further research should try to empirically validate this proposition.

2.2.3. Project Characteristics

The following chapter looks at the connection between project characteristics and biasedness. It does not seem necessary to explicitly distinguish between overoptimism and overconfidence, as findings on both topics seem to resemble each other.

According to Dunnig et al. (1989), endogenous overoptimism occurs because the meaning of most characteristics is ambiguous.^[40] This allows people to use self-serving definitions when assessing their own personality. When Dunning et al. (1989) offer a higher number of more precisely defined criteria, participants endorse both positive and negative characteristics as self-descriptive to a greater degree. Additionally, participants’ average degree of self-appraisal is reduced towards more neutral levels.

Alicke et al. (1995) show a similar connection between benchmark precision and biasedness.^[41] In their study, people on average reduce their endogenous overoptimism when they have to compare themselves to a well-known benchmark rather than to a vague peer group.

As mentioned in the introduction, Russo and Schoemaker (1992) find that individuals may be consistently endogenously overconfident when answering single questions.^[42] Yet, they may realistically asses their ability when estimating their performance on a portfolio of questions. Asking managers to give 90% confidence intervals on numeric questions, they scored roughly 6 out of 10 on average instead of the 9 to be expected. Yet surprisingly, they correctly expected a success rate of ca. 60%, instead of the 90% they were asked to aim at.^[43]

Alternatively, Camerer and Malmendier (2004) state any accompanying external validation mechanism to decrease estimation bias.^[44]

Summing it up, biasedness is particularly strong in rather vaguely defined, stand-alone decisions without supplemental external validation mechanisms.

Camerer and Malmendier (2004) put this into a business context. They expect biasedness to be stronger at the top levels of a firm.^[45] First of all, tasks at the top levels of a firm are more vaguely defined than those encountered in daily operations. This gives room to individual interpretation and evaluation. Second, single tasks touch rather different topics which impedes assessing them as a portfolio. However, one frequently sees that organizations try to foster rationality by introducing validation mechanisms. Team members or a supervisory board for example have to approve the final decision.

Such bias restricting organizational settings and processes have hardly been looked at systematically and would be an interesting topic for further research. One facet could be how individual expectations evolve once group interaction starts.

2.2.4. Agent-Project Relationship

In the chapter to come, again findings on overoptimism and overconfidence relate to and complement each other. Therefore, both biases are described in parallel.

As mentioned above, Langer (1975) states that people frequently believe they can influence outcomes that they demonstrably have no influence over.^[46] She coins this tendency an “illusion of control”^[47]. Accordingly, this bias is particularly strong when skills come into play. From this, one can deduct a self-selection process similar to the one described in chapter 2.2.1.1: Imagine a group of people with differing priors regarding their individual abilities. Then each member of the group chooses the project they think they can most positively influence. As a result each of them overestimates the success of the project they are in charge of. Several studies build on this: Weinstein (1980) shows that exogenous overoptimism is stronger amongst people assessing their own project’s performance.^[48] March and Shapira (1987) show that managers exhibiting endogenous overconfidence generally expect their own projects to produce success .^[49]

Following a similar logic, Weinstein (1980) concludes that people tend to be overly exogenously overoptimistic when outcomes improve their own well-being, i.e. when they are committed to the project’s outcome.^[50]

Camerer and Malmendier (2004) look at the relationship between individuals and the characteristics they are to assess. Accordingly, people tend to be particularly biased in situations they are not experienced in.^[51] Similarly, Murphy and Brown (1984) and Camerer et al. (2002) show that exogenous overoptimism is lower amongst more experienced agents when asked about their fields of expertise.^[52]

In parallel, Camerer and Malmendier (2004) also report estimation bias for situations without precise feedback on the project’s outcome.^[53]

Overall, biasedness is positively correlated to illusion of control, commitment, little experience and no precise feedback.

These findings can be translated into the world of business. Obviously, the considerations can be applied to entrepreneurs, too: Cooper et al. (1988) amongst others identify overoptimism with venture founders.^[54]

As mentioned above, Weinstein (1980) states a positive correlation between commitment and biasedness.^[55] Additionally, Gilson (1989) verifies that company underperformance has strong negative effects on managers’ private wealth.^[56] Combining these two findings therefore offers an additional explanation for higher bias amongst executives.

Murphy and Brown (1984) as well as Camerer et al. (2002) view experience as an important factor for alleviating biasedness.^[57] However, executives’ main trait of character is being a generalist and seeing the broad picture instead of focusing on single tasks. This impedes from learning effects for single acts and therefore predicts higher biasedness amongst top levels of a firm.

Feedback is important for learning and bias reduction. In economic terms it could be delivered via market pressure and competition intensity. However, Bebchuck et al. (2002) and Bertrand and Mullainathan (2003) report frequent entrenchment of CEOs.^[58] According to Camerer and Malmendier (2004), such defense mechanisms could decrease feedback precision significantly, thus supporting overconfidence and overoptimism amongst managers.^[59]

Overall, these characteristics, i.e. illusion of control, commitment, little experience on single task and no precise feedback, predict higher biasedness amongst top executives.

However, this relationship has not been verified empirically. Therefore, it would be interesting for further research to examine the development of biasedness along a company’s hierarchy.

2.3. Impact on Economic Behavior

The following chapter presents how the facets of overconfidence and overoptimism discussed above influence economic decisions, first, by private individuals (2.3.1), and second, by business managers (2.3.2).

2.3.1. Impact on Private Behavior

Puri and Robinson (2007) look at the correlation of endogenous overoptimism with a broad range of private work and life decisions.^[60] They find that more-optimistic people tend to work harder and expect to retire later. Besides, they save more and invest more in individual stocks. As already mentioned in chapter 2.2.1.2, Puri and Robinson (2007) also discover that moderate optimists show prudential financial behavior, whilst extreme optimists make rather destructive decisions.

Overall, few research explicitly examines the influence of biasedness on private behavior. Probably, this can be explained by the following factors: First of all, business decisions generally have a higher impact which leads to better measurability. Second, professional decisions are better documented and more easily available in databases. Furthermore, one may assume that insights into biasedness in business can be transferred to a private context: Agents potentially are the same in both spheres and decisions cover similar topics or mechanisms. However, no survey has empirically verified this proposition.

Therefore, this would be an interesting field for further research. Differences between private and professional behavior could have important implications for household finances or family businesses for example.

2.3.2. Impact on Business Behavior

The following chapter discusses the influence of biasedness on a broad range of business decisions. First of all, it is important to determine how overconfidence and overoptimism develop along a firm’s hierarchy (2.3.2.1). Second, one can then examine how decisions in the fields of managing (2.3.2.2), investing (2.3.2.3), and financing (2.3.2.4) are bended by these biases. Here, financing and valuation decisions differ depending on whether overoptimism or overconfidence is in place.

2.3.2.1. Bias and Hierarchy

According to Camerer and Malmendier (2004), an organization can only exploit estimation biases if the person leading and supervising is more rational than their subordinates.^[61]

Yet as mentioned above in chapters 2.2.2, 2.2.3, and 2.2.4, several mechanisms predict higher biasedness amongst company leaders: Ambiguous benchmarks, stand-alone decisions, illusion of control, and commitment to outcome. Biasedness could be reduced by previous validation mechanisms, expertise, experience and feedback. However, upper management’s decisions are more general in nature and cover specific topics rather infrequently. Therefore, it seems more difficult to institutionalize the creation of expertise, experience and feedback at a company’s top levels.

Following Goel and Thakor (2000), exogenous overconfidence could also increase along a company’s hierarchy due to the promotion process.^[62] Assume that of two job applicants the one with better past performance will be promoted. Then each does not look at the net present value of their decisions, but only at the probability of reaching a better outcome than their rival. This means that both seek the project with the highest volatility, and the applicant who by mere chance achieves the highest outcome is promoted. As a result the one who is most lucky having constantly bet on the “right” riskiest projects along their career will finally become CEO.

De Meza and Southey (1996) predict systematic biasedness amongst entrepreneurs due to a similar self-selection process.^[63] Only someone who is very positive about the probability of their idea’s success ends up creating a new venture.

Bertrand and Schoar (2003) empirically verify that a manager’s individual character has a strong influence on the organization he leads.^[64] They can empirically trace back a significant extent of the heterogeneity in investment, financial and organizational practices of firms to individual managers.

Summing up, several patterns of biasedness described in chapter 2.2 predict higher biasedness amongst company leaders than amongst shop-floor workers. This is further supported by self-selection processes that promote the career of biased individuals. Besides, biasedness translates from company leaders to their organization’s conduct.

Again, as no study has empirically verified the development of biasedness along a company’s hierarchy, this could be an interesting field for further research.

2.3.2.2. Bias and Performance

The following chapter describes the direct impact of biasedness on effort and performance. A clear-cut prediction is not possible, as arguments for both positive as well as negative effects exist. As described below, the opposite causality is possible, too, namely that outperformance induces overoptimism and overconfidence.

On the one hand, Camerer and Malmendier (2004) model how endogenous overoptimism may lead agents to reduce their effort in order to maximize personal utility.^[65] As a result, both principal and agent incur a loss of profit. Camerer and Malmendier (2004) assume decreasing marginal utility of wage and increasing marginal disutility of effort. Under these circumstances, an agent who overestimates his marginal productivity of effort or skill maximizes his expected utility at lower effort levels. However, if one assumes real productivity to remain constant, this leads to a drop in output. Therefore, the principal suffers from lower profits, whilst the agent is paid less (performance-based) wage. Overall, overoptimism leads to value destruction.

On the other hand, Camerer and Malmendier (2004) argue that biasedness may indeed have positive effects on payoffs, as it is often correlated with charisma, the ability to convince investors or motivate employees, and other value-enhancing behavior.^[66]

According to Taylor and Brown (1988), endogenous overoptimism may bring about a better project outcome.^[67] They state that endogenous overoptimism may lead to promote the ability for creative, productive work in two ways: First, these illusions may directly facilitate intellectually creative functioning. Yet, only little evidence supports direct effects of positive illusions on intellective processes. However, Taylor and Brown (1988) report a generally enhancing impact of positive mood on memorization, problem-solving strategies and complex thinking. Second, positive illusions strengthen motivation, persistence and thus performance.

Talyor and Brown (1988) also cite evidence suggesting a self-enhancing process amongst positively-biased individuals.^[68] More than their rational peers, they tend to view success as an acknowledgement of their abilities. This makes them even more motivated and confident when assessing forthcoming challenges. Harless and Peterson (1998) see the same effect in their field study: Portfolio managers tend to extrapolate recent performance and therefore overestimate future success.^[69]

Summing up, findings on the correlation between biasedness, effort and performance remain unambiguous. Some predict that biased agents reduce their efforts as they believe they will achieve their goals anyway. However, others see a linkage between positive attitude, higher productivity and better performance.

However, these argumentations are based on mosaic research and do not draw a coherent picture. Therefore, further research should characterize and empirically verify the relationship between biasedness and performance.

[...]

^[1] See Camerer et al. (2004), p. 3-9, also relevant for the rest of the paragraph above.

^[2] Camerer et al. (2004), p. 4.

^[3] See chapter 2.1 for more details on the definitions of overoptimism and overconfidence used in this paper.

^[4] See Alpert and Raiffa (1969), p. 294-305, as referred to by Cesarini et al. (2006), p. 454.

^[5] See Russo and Schoemaker (1992), p. 7-17, as referred to by Cesarini et al. (2006), p. 454.

^[6] See Russo and Schoemaker (1992), p. 8.

^[7] See Svenson (1981), p. 143-148, as referred to by Cesarini et al. (2006), p. 454.

^[8] See Svenson (1981), p. 144.

^[9] See Camerer and Lovallo (1999), p. 306-318, or Malmendier and Tate (2005), p. 2661-2700.

^[10] See Alicke et al. (1995), p. 804.

^[11] See Camerer and Lovallo (1999), p. 306, or Malmendier and Tate (2005), p. 2662.

^[12] See Kyle and Wang (1997), p. 2073-2090, Odean (1998), p. 1887-1934, Hackbarth (2004), p. 1-36, or Brettel and Kasch (2006), p. 1-32.

^[13] See Kyle and Wang (1997), p. 2073, Odean (1998) p. 1892, Hackbarth (2004), p. 2, or Brettel and Kasch (2006), p. 2.

^[14] Hackbarth (2004), p. 2.

^[15] See Weinstein (1980), p. 806-820.

^[16] See Adler (1980), as referred to by March and Shapira (1987), p. 1410.

^[17] Adler (1980), as referred to by March and Shapira (1987), p. 1410.

^[18] See Langer (1975), p. 311-328, also relevant for the rest of the paragraph above.

^[19] Langer (1975), p. 313.

^[20] See Hackbarth (2004), p. 14-15, also relevant for the rest of the paragraph above.

^[21] See Van den Steen (2002), p. 7-11; also relevant for the rest of the paragraph above.

^[22] Van den Steen (2002), p. 2.

^[23] See Miller and Ross (1975), p. 213-225.

^[24] See Taylor and Brown, (1988), p. 193-210, also relevant for the rest of the paragraph above.

^[25] See Maslow (1950), p. 11-34, and Jahoda (1953), p. 349, as referred to by Taylor and Brown (1988), p. 193-194.

^[26] See Miller and Ross (1975), p. 213-225, also relevant for the rest of the paragraph above.

^[27] For example Larwood and Whittaker (1977), p. 194, or Camerer and Malmendier (2004), p. 30.

^[28] See Miller and Ross (1975), p. 213-225.

^[29] See Camerer and Malmendier (2004), p. 66.

^[30] See Svenson (1981), p. 145, also relevant for the rest of the paragraph above.

^[31] See Camerer and Malmendier (2004), p. 66, also relevant for the rest of the paragraph above.

^[32] See Puri and Robinson (2007), p. 1-49, also relevant for the rest of the paragraph above.

^[33] Puri and Robinson (2007), p. 5.

^[34] See Kovalchik et al. (2005), p. 79-94.

^[35] See. Kovalchik et al. (2005), p. 82-83: Examining both students as well as neurologically healthy elderly people, they found that older participants exhibited overconfidence in only 17% of their total responses, whilst younger subjects were overconfident in 48% of all responses.

^[36] See Barber and Odean (2001), p. 261-292, also relevant for the rest of the paragraph above.

^[37] See Odean (1998), p. 1888.

^[38] See Biais et al. (2005), p. 305-306.

^[39] See Murphy and Brown (1984), p. 369-393, and Malmendier and Tate (2005), p. 2661-2664.

^[40] See Dunning et al. (1989), p. 1082-1090, also relevant for the rest of the paragraph above.

^[41] See Alicke et al. (1995), p. 804-825, also relevant for the rest of the paragraph above.

^[42] See Russo and Schoemaker (1992), p. 7-17, also relevant for the rest of the paragraph above.

^[43] See Russo and Schoemaker (1992), p. 14.

^[44] See Camerer and Malmendier (2004), p. 27.

^[45] See Camerer and Malmendier (2004), p. 20-27, also relevant for the rest of the paragraph above.

^[46] See Langer (1975), p. 311-328, also relevant for the rest of the paragraph above.

^[47] Langer (1975), p. 313.

^[48] See Weinstein (1980), p. 806-820.

^[49] See March and Shapira (1987), p. 1410-1411.

^[50] See Weinstein (1980), p. 806-807.

^[51] See Camerer and Malmendier (2004), p. 27.

^[52] See Murphy and Brown (1984), p. 369-393, and Camerer et al. (2002), p. 137-188.

^[53] See Camerer and Malmendier (2004), p. 26-27.

^[54] See Cooper et al. (1988), p. 97-108.

^[55] See Weinstein (1980), p. 806-820.

^[56] See Gilson (1989), p. 241-262.

^[57] See Murphy and Brown (1984), p. 369-393, and Camerer et al. (2002), p. 137-188.

^[58] See Bebchuck et al. (2002), p. 751-846, and Bertrand and Mullainathan (2003), p. 1043-1075.

^[59] See Camerer and Malmendier (2004), p. 27.

^[60] See Puri and Robinson (2007), p. 1-49, also relevant for the rest of the paragraph above.

^[61] See Camerer and Malmendier (2004), p. 26.

^[62] See Goel and Thakor (2000), p. 3-6, also relevant for the rest of the paragraph above.

^[63] See de Meza and Southey (1996), p. 375-376, also relevant for the rest of the paragraph above.

^[64] See Bertrand and Schoar (2003), p. 1169-1209, also relevant for the rest of the paragraph above.

^[65] See Camerer and Malmendier (2004), p. 4-5, and 22, also relevant for the rest of the paragraph above.

^[66] See Camerer and Malmendier (2004), p. 34.

^[67] See Taylor and Brown, (1988), p. 198-199, also relevant for the rest of the paragraph above.

^[68] See Taylor and Brown, (1988), p. 201-203, also relevant for the rest of the paragraph above.

^[69] See Harless and Peterson (1998), p. 258.