In 2002, Daniel Kahneman and Vernon L. Smith received the Nobel Prize in economics for their work in decision-making and prospect theory. This was a significant event in the development of behavioral finance and highlighted its crucial role in advancing the understanding of dynamics and behavior in financial decisions for the financial community. Behavioral finance has continuously contributed with unorthodox and non-traditional approaches towards a better comprehension of markets and especially of its agents. The sentiment that market efficiency is an insufficient and somewhat unrealistic concept has been present for a long time. To improve the explanatory value of the concept, the structuring of observed deviations and subsequent analysis of its applications are necessary.
The present empirical study contributes to this effort by examining some of the main hypothesized biases and behavioral patterns through practical application. For this objective, a small-scale sample has been subject to a survey on decision-making, in which central conceptual biases are tested empirically. The results may serve as additional insight into behavioral patterns and confirm or challenge widely used concepts of biases in financial decision-making.
The decision-making process in finance (conceptually speaking a social science) has been fundamentally based upon assumptions of complete rationality of behavior. However, the young but rapidly growing academic stream of behavioral finance challenges this established theory, and has introduced a paradigm shift in finance (Mauboussin, 1997; Barberis & Thaler, 2003; Spellman, 2009). Numerous psychological variables and behavioral patterns have been developed to better explain investment returns, summarized in the simplified equation: r = fundamentals + X. Hereby, r denotes returns, fundamentals refer to the traditional financial techniques and tools for explaining returns, and X describes the newly introduced psychological component. X complements the explanatory variables upon which r depends.
In the context of this empirical decision-making study related to financial investments, we limit our examination to three distinct, yet interrelated variables, labeled as behavioral biases. We aim to explore investors’ behavior with regards to the “representativeness bias”, the “herding bias”, and the “hindsight bias”͘ Those three biases have received great scholarly attention and are among the most widely known decision-making phenomena which deviate from rational behavior.
Starting with the representativeness bias, in a nutshell this irrational behavior describes investors’ propensity to interpret the past business operations of a firm and the past performance of a stock as representative of future expectations (Daniel, Hirshleifer, and Teoh, 2002; Tversky and Kahneman, 1974). It might sound self- evident that companies whose products or services an individual investor personally likes do not necessarily make good stock investments and that past business performance has already been fully priced in. Future performance should thus be detached from historical data. Yet, investors often erroneously believe that the past operating performance of a firm is representative of future performance, and they tend to ignore information that is not coherent with this expectation. This then leads to errors of extrapolation when examining past stock returns. For example, a stock that has performed badly over the past years is considered a loser. On the other hand, stocks that have done great in the recent past are considered winners. Investors assume that past returns are representative of what they can expect in the future. Glamour stocks are likely to be chased and continuously to be bought by investors if firms have trended upward in price (Lakonishok, Shleifer, and Vishny, 1994). However, good companies do not perform well forever, just as bad companies do not perform poorly forever, as eventually mean reversion occurs (DeBondt and Thaler, 1985).
Representativeness bias, moreover, describes the scenario in which decision agents (in this case stock investors) rely on the wrong predictive variables when making their decision. They likely pick information that might seem relevant, but actually does not impact the stock price heavily, or is superimposed by different information or news flow which in the end determines stock price movements (Shefrin, 2002).
The second main bias discussed is investors’ general willingness and desire to act alike without any countervailing force: it is known as “herding” (Devenow and Welch, 1996). This correlation among decision- making agents has been observed countless times in the stock market. Herding is generally related to a “momentum” element of a stock, when investors jump on the bandwagon and push the stock up even further in an upward move, or exacerbate its crash when joining in a panic sale. Kindleberger (1989) identified one of the main reasons of repetitive financial bubbles and subsequent crises in the irrational herding of investors. Wood (2006) points to the fact that this behavioral pattern not only appears among professional investment committees, but also exists among individual private investors. Herding often goes hand in hand with overconfidence, which denotes the human preference of acting in agreement with others, and to follow well-worn paths. The incentive to herd may be intensified in the case of financial investments, since outcomes are very uncertain and people do not like ambiguity͘ s the saying goes “misery loves company”, people take comfort in herding. Lastly, while herding is a common bias, the outcome may not be necessarily bad for society. For the stock market though, many crises were closely associated to it.
The last perception bias presented and examined in this paper is the “hindsight bias”, which tends to occur in situations where a person believes, after having received the outcome of a decision, that the onset of some past event was predictable and completely obvious, whereas in fact, the event could not have been reasonably predicted͘ This “I knew that from the beginning” attitude tends to be a very strong mindset, constraining investors’ rationale when evaluating different alternatives for a decision (Bazerman, 2002). Hindsight is powerful since regret, i.e. the negative emotions related to not making a right decision, basically only occurs when, by hindsight, the result of a prior decision turns out to be negative and the investor lost money. As research has shown, feelings of regret are greater with active decisions (a new investment strategy was pursued) than with non-active decisions (the usual well-known strategy was maintained) (ibid.). Thus, the regret bias keeps us from changing habits, inhibiting the investor to gauge all investment opportunities equally and based on objective criteria.
We will now turn towards the survey conducted, where the biases presented above were embedded into practical applications in order to examine distinct associated phenomena. Inevitably when examining those biases empirically through applications in our survey questions, other related biases (such as overconfidence, anchoring and adjustment, etc.) cannot be discriminated in a clear-cut manner. We are aware of the potentially resulting validity problem, but believe that our survey has been constructed in a way as to address the biases we want to measure primarily and most directly.
Methodology and Results
Choosing the right methodology is key to correctly measuring and documenting behavioral patterns. Surveys, interviews and questionnaires are standard tools for work in the field. For researching behavioral biases and their application, we decided to use a comprehensive survey as instrument. The three selected behavioral biases (representativeness, hindsight, and herding) allow to keep the survey manageable in size and for appropriate discussion within the scope of the paper. A particular emphasis was placed on the representativeness bias.
The survey was designed so that the biases would be tested interconnectedly rather than separately, as to make the bias in question less obvious to survey takers, and more applicable: Real world examples will often have more than one underlying issue/bias. We tested the three biases with seven questions. Please see the attached appendix for the survey and individual results.