Abstract:
Prospect theory (PT) is the dominant descriptive theory of decision making under risk today. For the modeling of choices, PT relies on a psychologically founded separation of risk attitudes into attitudes towards outcomes, captured in a value function; and attitudes towards probabilities, captured in a probability weighting function. However, while it is theoretically sound, it is unclear whether this clear separation is reflected in actual choices. To test this, we designed two experiments. In the first experiment, we elicit the value and probability weighting functions both under known and unknown probabilities. The results support PT and show that the value function is unaffected by the nature of the probabilities, which only affects probability weighting. More in general, this finding supports theories that represent ambiguity attitudes through probability transformations rather than utility transformations. In the second experiment, we examine the effects of an increase in stakes on risk attitudes. We find that the stake increase is not reflected in the value function, but rather in the weighting function, thus contradicting PT's prediction.
