Conference Agenda

Inspired by the discussions in Bateman et al. (1997), Knetsch (2010), and Whittington et al. (2017), we propose a new test of loss aversion that closely follows the logic of Tversky and Kahneman (1991) and is implemented through a discrete choice experiment (DCE) survey. This test relies on two pairs of valence-based framing strategies to describe future changes in the St. Lawrence Estuary beluga whale population. It intentionally dissociates the current condition (status quo point) from the reference state perceived by respondents when answering our three-option choice sets.

More precisely, taking advantage of the fact that conservation policies often take years to become effective, both the opt-out and the two opt-in alternatives in each choice set are defined in terms of the beluga stock size at a distant future point (the year 2050). By describing future changes in the beluga stock either as gains (positive valence) or losses (negative valence), relative to either today’s population level (pair 1) or the projected population level in 2050 (pair 2), our framing strategy allows us to cover all four welfare measures for the same stock change: willingness to pay for compensating variation (WTPCV), willingness to pay for equivalent variation (WTPEV), willingness to accept for compensating variation (WTACV), and willingness to accept for equivalent variation (WTAEV).

Using an online survey (N = 1,491), we employed a split-sample design to collect approximately 300 complete responses for each of the four framing versions, as well as an additional 300 responses for a standard DCE format. Our results demonstrate that valence-based framing strategies significantly modify respondents’ perceptions of their reference state. These shifts in perceived reference states, in turn, affect preferences over the proposed scenarios. We find that decision-making in the DCE is jointly determined by respondents’ perceptions of their reference state and their sensitivity to future changes in the beluga stock. Because both factors are influenced by valence-based framing, this interaction may help explain why neither our study nor previous research has been able to fully confirm the hypothesis originally proposed in prospect theory. Finally, our results indicate that respondents in standard DCE treatments tend to interpret their reference state in a manner more consistent with equivalent variation than with compensating variation.