Excerpt
Table of Contents
1. Introduction
2. Literature Review
3. Research Question
4. Methodology and experiment
5. Discussion
6. Conclusion
7. References
8. Appendix
8.1 Appendix A: Tables
8.2 Appendix B: Questionnaire single decision maker
8.3 Appendix C: Questionnaire single decision maker + Advisor
8.4 Appendix D: Questionnaire group decision
8.5 Appendix E: Follow up questionnaire
Abstract
Our paper looks at whether there is a difference in perception of utility with respect to our three decision settings- a) individual decision maker (the control group), b) decision maker with two advisors and c) group with three decision makers. To find ways how to de-bias decision-making, we specifically look at expected versus experienced utility between and within these groups. Our results show that there is not only a difference in utility perception with respect to the number and role of people involved with the decision but also that if sunk costs are involved the positive utility (joy) is stronger than negative utility (regret). Supported by significant results from our research we offer solutions to mitigate two of the four causes of the disposition effect discussed by Kahneman and Statman (1985): 1. Regret Aversion and 2. Prospect theory. The group decision part of our experiment shows a significant mitigation of regret aversion. We therefore propose a set of managerial implications. By forcing our participants to invest, we provoke the closing of a mental account rendering the investment a sunk cost for the investor. The resulting gain in utility is then not subject to the effects of prospect theory.
illustration not visible in this excerpt
1. Introduction
Decisions are taken every day by billions of people. No matter where they are from, what the colour of their skin is or their age, people take decisions that can either change their own life or in some cases the life of thousands or even millions of others. Of course, the decision-making process is not universal all around the world but depends on culture and heritage to some extent e.g. because of the use of different culture connected heuristics (Varnum et al; 2010). Nevertheless, at the end of the day people try to derive positive utility from decisions they make. One of the complex tasks for academics in the field of behavioural finance is to understand what influences people and how decisions that are made on the basis of subconscious biases (irrational/ less favourable) can be de-biased. Here it is interesting to not only look at the individual decision maker but also the decision-making in groups. In this research we introduce three decision settings in which decision-making is presumably either: a) individual decision maker (the control group); b) dominated by consensus in a group of three equals; c) by consultation in a group with one decision maker and two consultants. In our study we do not look as much at the real outcome of the decision people have to take, but at the difference between anticipated and experienced utility. In other words we compare the utility people think they will experience before knowing the outcome of a decision they made and the actual utility right after. We focus on anticipated utility because in the decision-making process it represents the decision maker’s interaction with his emotions at the moment he makes the decision and thus forms one of the basis pillars of the process. To validate this, Pfister and Böhm (2008) state in their paper that "the issue of rationality should be based on the validity of emotional evaluations rather than on formal coherence”. Here emotional evaluation includes making decisions with the help of e.g. regret. In the latter the weight of the anticipated utility approximation is high, even outperforming the rational expectations approximation (T Cogley and T. J. Sargent, 2008).
Human beings are not well equipped to estimate their feelings ex ante when taking the decision. Prospect theory claims that people may feel negative utility from bad decisions approximately two and a half times as strong as positive utility from good ones (Benartzi and Thaler, 1995). This is one of the reasons why people often prefer inaction to action. The other one is the correct estimation of emotions and utility in anticipation of the outcome and after. If they could, they would incorporate regret into their decision-making and limit the bias of inaction. The fact that regret looms larger than pride leads to the inaction rather than action by humans. Therefore the behavioural de-biasing techniques we are going to propose should be applied at this crucial evaluation stage at the moment of the actual decision-making when the outcome is not yet known.
In the first part of our study we are going to review the existing literature about and around the topic of the disposition effect. We then build up towards our hypothesis, covering the motivation behind our three-decision setting approach. In the methodology part, the experiment and research set up is explained. Before we draw our conclusions, real life implications and limitations, we discuss the results on a hypothesis-by-hypothesis basis.
2. Literature Review
Oftentimes, people simply decide based on decision heuristics. They rather decide based on a rule of thumb than on a systematic evaluation. However, in a disposition effect setting the behavioural bias relates to the phenomenon of investors selling winners too early and holding on to losers as elaborated by Kahneman & Tversky (1979) in their paper about myopic loss aversion. Myopic loss aversion is further explained by Shefrin & Statman (1984) with the help of prospect theory, mental accounting, regret vs. pride and self-control. Regarding prospect theory, investors experience a higher negative utility in case of losses due to stronger convexity of the utility function in the loss area than in the gain area. This implies that if a loser stock price in the negative utility domain starts rising towards the original reference point, the investor will experience a higher utility than in the positive utility domain due to stronger convexity of the decision curve. These lead to the fear of regret of not selling a winning stock when they had large profits or the fear of regret of realizing a loss and confessing a wrong investment choice. Investors are reluctant to realize losses - and therefore hold on to losers.
It is closely linked to the fact that peoples’ evaluation of expected versus experienced utility diverges. In hedonic forecasting the prediction of one’s affect in the future, individuals overestimate the gain and the loss in utility they expect (Hastie & Dawes, 2001). However, since the disposition effect is confirmed by many academics, Kahneman, Tversky and Thaler argue, the force of regret is stronger than pride. This was proven for the individual decision maker. Our research adds value in looking at the effect of a change in a decision setting as discussed above.
As a result, a single decision maker may favour inaction over action and thus procrastinates a decision. S/he fears the regret feeling of realizing the loss. Imagine a person that ordered a subscription for playing squash a year ago. This person used the squash court for the first two months and afterwards not anymore. Every month this person thought: “I have to go and play squash more often to make up for lost two months”. Instead of doing that, s/he did not just not go to play anymore but also did not end the contract with the centre. Why could that be? The moment s/he would end the contract, s/he realizes the loss by foregoing the opportunity to make up for the payments by playing a lot more. This means, although there was a withdrawal from his/her bank account of 40€ per month, a black on white loss without return, that is psychological realized only the moment the contract was ended. Therefore, one of our aims is to uncover the reasons of how decision makers can overcome their inaction, namely by making the decision in one of the two settings proposed; decision maker with two advisors and a group with three decision makers.
If we have a look at the dynamics of group decision-making or collaborative decision-making opposed to individual decision-making there are many pieces to the puzzle and it cannot simply be stated that group decision-making is better or worse opposed to individual decision-making. Examples like the failed Bay of Pigs invasion of Cuba where group consensus of homogenous CIA members was stronger than the many alarm bells that should have been ringing for that inadequate action beyond the CIA’s capacities and unreasonable risk taking. Therefore the decision-making of homogenous groups is likely to result in groupthink and thus have a negative effect on the quality of the decision (Janis, I. L.; 1972). On the other hand, a synergetic effect of
the group members could lead to a better quality decision. What we expect from group decisionmaking is that the individual members of the group carry less responsibility for the decision and therefore show less extreme positive and negative utility (Shefrin and Statman, 1984). This could also help to work around the regret aversion bias in decision-making.
Depending on how you frame the proposition, people create their mental accounts differently. The individual cuts different investments or gambles into separate pieces and assigns a different account to each. In case of a loss, individuals are reluctant to close these. Thereby, prospect theory is applied to each individual account (Shefrin and Statman, 1984). The reference point of the mental account is in a disposition effect setting the stock price. In our case, it is the value of the game.
If people would integrate the game they have only one mental account and would combine the investment they make with either outcome. However, in case they segregate, the investment is treated as a sunk cost. Therefore, the end result is always higher than in the integration case.
Finally, self-control might be another explanation for the disposition effect. In particular, the process in people's head can be seen as an agency conflict. The principal or planner can commit to rules in advance such as a stop-loss or a not changeable savings plan because the agent or doer acts more emotional and irrational (Shefrin and Statman, 1984).
Could the confirmation of the results have implications for professional decision makers and justify many existing mechanisms in the real world? For example, in the context of corporate governance Hermalin and Weisbach (2003) make the point that boards of directors are not the best alternative in a perfect world, however they are used because they are the second best alternative in the equilibrium process. Our study therefore tries to understand if people together (like in boards of directors or together with advisors) take better decisions or take them at all, than a single decision maker because they share the feeling of pride and regret. The single decision maker might take no action motivated by the anxiety of the anticipated regret at all.
In the same vein, the results might point to the fact that it is better for individual investors to invest in quorum and combine their portfolios. Of course with this form of investing other problems will arise, but at least this might overcome the tendency of inaction because of a steeper utility function. One has to weigh the costs and benefits of such a model against each other.
3. Research Question
From the above discussion, naturally the following research question derives:
What is the effect of the number of people involved in making a decision (one person, including consultant, many decision makers) on anticipated and experienced joy/ regret at the win/ loss of a game and how do sunk costs influence the decision?
In our experiment participants had to make a choice whether a graph of a stock would move up or down within the following three days. They gave an indication of their anticipated utility and the experienced utility after the outcome of the game was known.
In order to be able to make conclusions about the data we retrieved during our experiments, our first hypotheses starts with the anticipated utility of the participants before the outcome of the game is known. Our claim is that there might be a difference between our groups. Therefore our null hypothesis states:
H1a) Anticipated regret is not moderated by the decision setting. H1b) Anticipated joy is not moderated by the decision setting.
With the first hypothesis we tested for whether there is a difference in anticipated utility. For the second hypothesis we test for the direction of that difference to find out whether the decision setting from individual decision maker to the group decision-making in a decision decreases anticipated regret and pride. Our claim is that there is a decreasing difference between the selected groups because the responsibility is shared with more decision makers and therefore the individual decision maker should experience the most joy if s/he thinks that s/he will win the game and the most regret if s/he thinks that s/he will lose. It would imply that the standard S- shaped utility function of the game is steeper in the gain as well as in the loss domain for the individual decision maker opposed to the other two decision settings. Therefore our second nullhypothesis states the following:
H2a) There is not more anticipated joy, if the decision setting changes from decision setting 1. (Individual) to setting 2. or setting 3. (group)
H2b) There is not less anticipated regret if the decision setting changes from decision setting 1. (Individual) to setting 2. or setting 3 (group)
At this stage of our game the participants had to wait three days until the movement of the stock price determined the result of their choice. After the three days the participants receive the outcome of the game and are asked again to state their experienced joy/regret. Therefore we hypothesize:
H3a) Experienced regret is not moderated by the decision setting. H3b) Experienced joy is not moderated by the decision setting.
This takes place after the experiment is done and the results are disseminated to the participants. To continue the conversation about individuals overestimating the gain and the loss in utility they expect (Hastie & Dawes, 2001), we hypothesise that the expected utility measures are higher than the experienced. Therefore our null hypothesis is:
H4a) Anticipated regret is not higher than experienced regret. H4b) Anticipated Joy is not higher than experienced joy.
Based on the concept of sunk cost, our experiment is structured to push the participant to treat his/her investment as a cost that has already been incurred and cannot be recovered. The participant is obliged to invest his/her initial 100 Euro. There is no option to not invest this capital. We expect that this restriction leads to a closing of the mental account and creates a new frame: the possibility to win 200 or win 0. Because the participant in this frame segregates the possible win from his/her original sunk investment, we observe a de-biasing effect of that new frame towards the prospect theory bias. We hypothesise a new correlation between lost and gained utility in which the steeper slope for lost utility is eliminated. If we obtain significant results, the sunk cost allocation frame holds. Please note that the data is not categorized by the different decision settings discussed above as the closing of the mental account happens at a personal level. This leads us to the our final null hypothesis: H5a) Anticipated joy is not higher than anticipated regret. H5b) Experienced joy is not higher than experienced regret.
[...]