Preference elicitation under oath
Introduction
Cost-benefit analysis is at the economic core of public policies devoted to the production of non-market goods. Eliciting sincere preferences to guide such a cost-benefit analysis remains a challenge, however, due to the hypothetical bias – the gap between hypothetical monetary values and real economic commitments – and false zeros – people opt-out from the mechanism by stating a value equal to zero. Accumulating evidence from lab and field reveals that the average person tends to overstate real economic commitments by a substantial amount.1 These results reinforce the long-standing explanation that the bias arises because the budget constraint is not binding in hypothetical valuation exercises (see [21] for a review). Other observers, however, have argued the opposite—real bids are “too low” as bidders shave bids downward, down to a “false zero” if they use the bid to exit the auction [66]. Mechanism design requires the participation constraint to be binding: i.e., a person should be no worse off by participating in the mechanism than otherwise (see for instance [53]). Most lab or field experiments, however, do not provide people with a controlled “opt-out” mechanism to exit the auction once they are engaged in the exercise. Pressing people to state a bid in an auction that they do not want to participate could serve to pressure the person, thereby violating his or her participation constraint. Bidders can opt-out of the experiment by bidding zero, as a result of this violation. For instance, [55] observed that 50%–55% of all bids dropped to zero due to the real context of an auction (see [40] for an extensive discussion on this issue).2 Hypothetical bids without binding budget constraints or real economic commitments without binding participation constraints can both cause problems even if the institution is demand revealing. This suggests that we need to find an ancillary mechanism that commits the bidder to bid sincerely when stating their preferences.
Faced with such revelation problems, real-world courts ask witnesses to take an oath “to tell the truth and nothing but the truth”. Using the theory of commitment from social psychology [47], [42], we herein use a solemn oath as an ancillary truth-telling-commitment device, asking our bidders to swear on their honour to give honest answers prior to participating in an incentive compatible second-price auction. We examine whether this traditional real-world institution can induce subjects to reveal their preferences sincerely. Bidders are free to sign the oath or not, and participation and earnings are not conditional on signing the oath. Our experimental evidence comes from two experiments, designed jointly: an induced value (IV) second-price auction and a homegrown value (HG) second-price auction. The design and rules implemented in both are as identical as possible. This allows us to build on the results from the controlled IV auction to interpret bidding behavior for a non-market good (donate to adopt a dolphin) for which we do not know the homegrown demand curve. For each experiment, we run three main treatments: the baseline disconnects bidding decisions and subjects’ payoffs from the experiment (the traditional hypothetical context); the real setting reintroduces monetary incentives; and the oath treatment gives the subjects the option of signing an oath to tell the truth before participating in the baseline.
For induced values, our results show oath improved demand revelation—people tended to bid sincerely after taking an oath for a hypothetical auction; they did not bid sincerely with monetary incentives or under a no-oath hypothetical scenario. For homegrown values, the oath elicited preferences that seemed to reduce violations both in the experimental budget constraint (only 18.9% of bids are above the disposable income earned in the experiment compared to 47.7% in the hypothetical-only auction) and in the participation constraint (only 1.1% zero bids compared to 26.7% in the real-only auction). Put together, those results unambiguously support having subjects sign an oath to tell the truth has a significant impact on bidding behavior. We moreover confirm the robustness of the results through training subjects with the elicitation mechanism before the actual procedure takes place. We rely on further companion treatments to assess the reasons why it happens. First, we explore whether the oath works by helping subject to tell the truth, or through a change in their willingness to do so, by adding an explicit exhortation to truthfully reveal preferences—a procedure known as cheap talk scripts since [23]. Our results support the second interpretation—oath seems to make people more willing to tell the truth. We then turn to the behavioral motives underlying the observed change in behavior. We provide insight to help disentangle the two main candidate explanations: intrinsic preference to behave in accordance with past actions or aversion to guilt. In an HG setting, adding explicit warnings about the negative consequences associated with lying is neutral on the effect of the oath. This points to internal reasons underlying the effect of the oath, which favors the commitment-based interpretation. Overall, our findings suggest that the oath has potential to create the commitment needed to better link intentions and actions in demand revelation.
Section snippets
The oath as a commitment device
The social psychology theory of commitment tells us the risk of lying is diminished within an oath-taking context.3
Overview of the experimental design
The open question this paper addresses is whether an oath can induce more sincere bidding behavior, thanks to commitment, thereby improving the power of demand revelation and the accuracy of stated preference methods. We test whether a person taking an oath bids more sincerely in an incentive-compatible auction, both in induced and homegrown value auctions. Table 1 provides an overview of the experimental design. Throughout the paper, we hold constant the demand revelation mechanism—the classic
Design of the IV-Experiment
Our first experiment implements a three-treatment design of an IV auction: (i) baseline hypothetical bidding, with neither an oath nor monetary incentives; (ii) baseline coupled with an oath and (iii) bidding with binding monetary incentives.
IV-Baseline. We use a Vickrey second price auction, in which the highest bidder pays the second-highest bid. We divide each 18-subjects session into two independent groups of nine bidders to avoid too much distortion of bidding at the bottom of the
Application treatment: the oath in HG auctions
We now consider preference elicitation of homegrown values for a private good with non-market attributes: adopting a dolphin through a monetary donation to the World Wide Fund (hereafter WWF), a well-known non-governmental organization devoted to “protecting the future of nature”.18
Robustness treatments: commitment of trained bidders
Ideally, preference elicitation should be more accurate if bidders have no misconceptions about the operations and procedures of the auction mechanism (see for example, [56], [65]). While straightforward in theory, the second price auction used in our treatments is likely to be unfamiliar to many bidders. In particular, they might not immediately realize that bidding their true preferences is the weakly dominant strategy. The oath, by providing subjects with some information about what they
Does the oath reduce to a demand effect?
An alternative view of our results is that the oath induces subjects to provide their true value because of a demand effect – in the sense that they comply with the request contained in the oath – rather than thanks to a commitment of subjects to tell the truth. In this section, we disentangle between the two based on new experiments aimed at contrasting the preference elicited between having people taking an oath versus an exhortation on the importance of accurate reporting. Such an
Why the oath works: an experimental appraisal of behavioral motives
Our treatments suggest the oath pushes bidding behavior towards more truth telling behavior. Subjects seem more willing to reveal their true preferences. We now discuss two behavioral models that rationalize this effect: (i) cost of lying, an intrinsic preference to keep one's word (in psychology this is a self-attribution process), and (ii) guilt aversion. We consider a companion treatment helps disentangle these two behavioral models.
Summary of the results
Finally, we run a conditional test to confirm all seven results by pooling data from all nine HG treatments in a random effects panel Tobit model. We control for all treatments, interactions between treatments, total earnings, and individual characteristics of bidders. The referent behavior is bidding under in the HG-Baseline treatment.
Table 11 supports the main conclusions derived from unconditional statistics run on aggregate bidding behavior. First, the parameter associated with monetary
Conclusion
The oath is a centuries-old mechanism designed to align internal incentives with social goals. Herein we examine preference elicitation under oath. Based on the social psychology theory of commitment, we examine whether after taking a solemn oath to tell the truth, people are more likely to bid sincerely in induced and homegrown value auctions. For induced value, we find the oath-only treatment induced sincere bidding behavior on average within a second-price auction; the other hypothetical and
Acknowledgements
This paper is a revised and augmented version of CES WP n°2009-43. We wish to thank Romain Zeiliger and Maxim Frolov for their assistance in developing the software and Bruno Rock for his help in running the experiment. We gratefully acknowledge Russell Davidson, Edward Deci, Bernard Fortin, Alan Kirman, Danica Prelec, Drazen Prelec, Stéphane Robin, Laura Taylor, Miriam Teschl and participants in the ANR Workshop on incentives and motivation, 2008 (Paris); Tenth Annual CU Environmental and
References (69)
- et al.
Cheap talk reconsidered: new evidence from CVM
Journal of Economic Behavior & Organization
(2006) - et al.
Measuring beliefs in an experimental lost wallet game
Games and Economic Behavior
(2000) - et al.
Earned wealth, engaged bidders? Evidence from a second price auction
Economics Letters
(2009) - et al.
Calibration of the difference between actual and hypothetical valuations in a field experiment
Journal of Economic Behavior & Organization
(1998) - et al.
A random nth-price auction
Journal of Economic Behavior & Organization
(2001) - et al.
Willingness to pay for curbside recycling with detection and mitigation of hypothetical bias
American Journal of Agricultural Economics
(2003) - et al.
Social image and the 50-50 norm: a theoretical and experimental analysis of audience effects
Econometrica
(2009) - et al.
Guilt in games
American Economic Review
(2007) - et al.
A radical Dissonance Theory
(1996) - et al.
Petit traité de manipulation à l'usage des honnêtes gens
(2002)