Schedule - Parallel Session 2 - Risk Attitudes 2

Regret theory (Loomes and Sugden 1982, Bell 1982) is one of the most popular descriptive theories for decision-making under uncertainty. It is an intuitive theory that captures the psychological content of anticipated regret (or rejoice) associated with decisions and naturally accommodates several violations of expected utility. Regret theory’s intuitive content and explanatory power make it suitable for real world applications (Barberis and Huang 2006, Muermann et al. 2006, Perakis and Roels 2008, Filiz-Ozbay and Ozbay 2007). Although regret theory is widely applied, the risk attitudes under that theory are not well understood. To understand risk attitudes under regret theory, the risk premium under regret should be analyzed. Bell (1983) formalized the risk premium under regret theory and showed that it consists of two distinct components: a resolution premium and a regret premium. However, Bell (1982, 1983) did not suggest an empirical method suitable for measuring these two components of the risk premium (Anand 1985). In this paper, we provide an analytical expression for both the resolution premium and the regret premium. These expressions enable”for the first time”a precise characterization of risk attitudes under regret theory and thus rigorous predictions about the risk attitudes of a regret-averse decision maker. We predict that regret-averse DMs will be risk seeking for low probabilities of gains and risk averse for high probabilities; we also postulate that risk attitudes are reinforced by feedback. We introduce a method, based on Bleichrodt et al. (2010), to measure the risk premium under regret theory empirically. This method allows to compute both the resolution and regret premiums and thereby to understand the effect of feedback on regret attitudes. Finally, we design an experiment to estimate empirically the risk premium’s components and to test our predictions about risk attitudes. The experiment serves also as a descriptive test of regret theory. The data support regret aversion as a robust empirical phenomenon. By measuring risk premium components empirically, we confirm that immediate feedback increases the regret aversion of regret-averse subjects but discover that it actually reduces regret aversion within the entire subject pool. However, we find no statistically significant support for the risk attitude predictions of regret theory. By modeling the risk attitudes and measuring resolution and regret premium empirically, we show that regret theory is a simple yet powerful framework to describe the pervasive emotion of regret and the risk attitudes associated. However the experimental results suggest that regret theory provides a partial account of risk attitudes as it captures the regret-rejoice trade-offs on the outcome scale only. Decision theorists should target their efforts in developing new models (like PRAM) to capture the regret-rejoice trade-offs also on the probability scale.

Consider an individual who has just suffered from a major financial setback or from a serious health loss like a handicap or the beginning of a long-term disease. In a normal context, he should give value to any small positive probability p of getting its money back or of healing. However, various psychological studies suggest that when the individual is uncertain about the durability of the loss, it can also impede successful psychological adaptation to the loss. Is it then possible that, in some cases, an individual prefers the certainty of a health or wealth loss, to which she can adapt, rather than a small positive probability p of getting it back? We examined this question theoretically through a modified version of the Koszegi and Rabin’s (2006) model. The result depends firstly on how uncertainty affects the weights that the individual ascribes to the two possible outcomes, temporary or permanent loss, used as reference points. In the simplest case in which these weights correspond to the probabilities of these two outcomes, we show that the possibility that the loss is only temporary can reduce the expected utility of the individual i) if loss aversion is strong enough (because the individual anticipates that he will be more disappointed if her loss becomes permanent than he will be satisfied if it disappears) and ii) if is low enough and below a probability threshold that depends on the number of future periods (Intuitively, nobody will refuse a significant probability to return to her initial level of health or wealth, in particular when she can benefit from it over many future periods). The model finds some empirical support in observations showing that the beneficial effects, in terms of adaptation, of the certainty of the loss are often explicitly incorporated in the cost-benefit calculus made by people who decide to stop their efforts to regain past health, wealth, wage or fame. By isolating the necessary conditions under which an individual may turn down a small positive probability of returning to her initial better situation, it also helps explain why this paradox is limited in practice to some specific circumstances.