Not logged in.
Quick Search - Contribution
|Title||Decision Heuristic or Utility Driver?|
|Other Titles||On the robustness of the impact of gain and loss probabilities in risky choice|
|Institution||University of Zurich|
|Faculty||Faculty of Economics, Business Administration and Information Technology|
|Zusammenfassung||The concept of aspiration levels and the related probabilities of gaining and losing have received growing attention in economic literature. If human decision makers really care about achieving a certain level when dealing with risky prospects, such as investment managers, who only consider investments which are able to realize a predefined level of return as outlined by Payne et al. (1980, 1981), then the probabilities of achieving such a level are an important decision criterion. Previous literature has provided a variety of experimental evidence in favor of the probabilities of gaining and losing. While there seems to be consensus about the importance of the probabilities of gaining and losing, the circumstances under which they apply in risky choice is less clear. Levy and Levy (2009), incorporate Roy`s (1952) safety first criterion, an equivalent to the concept of aspiration levels, into a general Expected Utility framework and use their concept to explain parts of the equity premium puzzle. However, a recent contribution of Diecidue and van den Ven (2013) has raised doubts about a general applicability of the concept of aspiration levels in risky choice. Like other authors in the field, such as Payne (2005), they come up with the conjecture that the probabilities of success and failure may also be a decision heuristic, which is more frequently used by decision makers when the complexity of the choice task rises. This thesis aims to examine the robustness of previous findings in recent literature regarding the impact of gain and loss probabilities on risky choice, in particular under which conditions they matter. In order to do so, the thesis employs two different experiments which address robustnessconcerns expressed in the existing literature. One concern is that previous experiments had a tendency to employ artificial and unrealistic choice tasks, which do not represent the reality, but eventually enforce the usage of decision heuristics by subjects. While this problem was already addressed by the investment task in Zeisberger (2013), little evidence exists how the probabilities of gaining and losing affect a DMs choice in a market environment. A market pools different decision makers with different preferences and allows them to interact, while receiving instant feedback about their valuation of an asset. This first experiment used a ZTree software environment to model an experimental asset market, where subjects traded on two independent markets either a high or a low loss probability asset. The goal was to compare trading prices and test empirically whether the low loss probability asset would achieve a higher price than the high loss probability asset, which was higher valued for Expected Utility, Mean Variance and Cumulative Prospect Theory preferences. No significant difference in market prices was found. Detailed explanations are provided, which account for two possible scenarios. First, the effect is not present in a market or second, the effect is present, but small and overlaid by other phenomena. The main conclusion to be drawn remains that under the conditions tested, the effect of gain and loss probabilities on the valuation of a risky prospect is not robust in an experimental asset market. A further implication is that the translation of the effect from an individual choice problem to tasks, which include interaction on a market, is not straightforward. This result limits the application of the concept of the probabilities of gaining and losing on a broader framework in finance. The second experiment tasked decision makers to decide between a low and a high loss probability lottery for lottery pairs, with one lottery higher valued for standard preferences and the second equipped with a lower loss probability. The main experimental variable was the number of outcomes, which was varied in order to model different complexity levels of the choice task. Furthermore, the effect of downscaling the outcomes and the provision of additional information was included in the experiment. While the downsizing of outcomes and the provision of additional information did not significantly change the ratio of subjects deciding for the low loss probability, the number of outcomes had a highly significant positive influence on choosing a low loss probability lottery. The thesis concludes from the experiment that complexity of the choice task matters for the presence of the probabilities of gaining and losing in risky choice. This gives strong support to the heuristic understanding of the probabilities of gaining and losing in the literature. Assessing the results of both experiments together, does not need to be a contradiction. Even though the experimental asset market was an apparent difficult and complex choice environment, the choice conditions in the market allowed much less for the application of a heuristic behavior, which favors the probability of gaining and losing. The outcome distribution, which is likely one of the main influence factors of the probabilities of gaining and losing, was only revealed over time. At the same time the second experiment included a full revelation of the outcome distribution from the start, which allowed subjects to assess the probabilities of gaining and losing and take them into account when making their decision. Nevertheless, besides further experimental evidence, a better understanding of the probabilities of gaining and losing may also require the involvement of other psychological concepts, which have not found their way into economic theory so far.|