Justice in an uncertain world: Evidence on donations to cancer research

The paper uses information on actual and hypothetical charitable contributions to cancer research in the United Kingdom to elicit information on justice principles endorsed by donors. They face a choice between fund-raising contributions for several hereditary and lifestyle-related cancers. Donors’ choices of how much to donate to different cancers reveal how they view luck vis-a-vis risky individual choices. The estimation results reveal that donations are smaller for cancers with higher prevention rates, which is the probability that the potential cancer victim can avoid the cancer in question by some choice. We also find that provision of information on lifestyle-related causes of cancer adversely affects contributions. In contrast, information on hereditary causes has a positive effect on donations. Furthermore, a large share of donors indicated in their feedback that they chose donations to a hereditary over a lifestyle-related cancer to “punish” poor individual choices. These findings suggest that many donors lean toward choice egalitarianism, which conditions donations on the potential beneficiaries’ choices.

     

FOOD SCARES IN AN UNCERTAIN WORLD

We develop a model that explains the stylized facts of food scares: an immediate and sharp decline in consumption of the product followed by a slow and frequently partial recovery of demand after the scare passes. The model is used in conjunction with the natural experiment afforded by the ‘mad‐cow’ crisis to elicit empirical information on decision makers' perception of uncertainty. Our theoretical and quantitative results suggest that observed behavior is consistent with sharp changes in beliefs and the presence of Knightian uncertainty, as measured by the degree of imprecision in our model.     

Ambiguity,pessimism, and rational religious choice

Using a subclass of the α-maximin expected-utility preference model, in which the decision maker’s degree of ambiguity and degree of pessimism are each parameterized, we present a theory of religious choice in the Pascalian decision theory tradition, one that can resolve dilemmas, address the “many Gods objection,” and address the ambiguity inherent in religious choice. Parameterizing both the degree of ambiguity and the degree of pessimism allows one to examine how the two interact to impact choice, which is useful regardless of the application. Applying this model to religious choice is a move beyond subjective expected-utility theory, allowing us to show that a change in either the degree of ambiguity or the degree of pessimism can lead a decision maker to “convert” from one religion to another.     

Is Allocation Affected by the Perception of Others' Irresponsible Behavior and by Ambiguity?

The article examines how the perception of others' irresponsible behavior and ambiguity regarding probabilities affect allocation among potential beneficiaries. To elicit these views, we conducted a survey where the participants were first asked to make an allocation of a fixed sum of money between a hereditary cancer, where chance plays a central role, and a lifestyle‐related cancer, where individual lifestyle decisions are more important. Our estimation results show that a substantial share of the respondents allocate significantly more to the hereditary cancer. This may indicate that these respondents care about others' irresponsible behavior. Then, we elicited perceptions of cancer hazards in the form of imprecise probabilities and examined the interplay between allocating behavior and risk perceptions. Finally, we investigated the effects of various socioeconomic characteristics, and of awareness of highly publicized cancer cases, on respondents' allocations.     

Eliciting beliefs

We develop an algorithm that can be used to approximate a decisionmaker’s beliefs for a class of preference structures that includes, among others, α-maximin expected utility preferences, Choquet expected utility preferences, and, more generally, constant additive preferences. For both exact and statistical approximation, we demonstrate convergence in an appropriate sense to the true belief structure.