A penny for your thoughts: Inducing truth-telling in stated preference elicitation

Contingent valuation often induces hypothetical bias. In a laboratory experiment, we test three calibration mechanisms: cheap-talk, consequentialism, and a new mechanism, the Bayesian truth serum ("BTS"). We apply the BTS in a "faith-based" format: subjects are informed about the purpose and potential efficacy of the BTS, but not its theoretical foundations. We find that real and hypothetical responses differ significantly; real and consequentialist responses are statistically indistinguishable; cheap-talk and the BTS eliminate bias inconsistently; subject characteristics interact significantly with treatment.