Lying through their teeth: Third party advice and truth telling in a strategy proof mechanism

- We test the effect of advice on the well-known top trading cycles (TTC) matching algorithm.
- We compare three treatments involving third party advice (right, wrong, and both) to a baseline.
- Truth telling rate is higher than 70% in the B.
- Truth telling rate is significantly lower in the advice treatments.
- Subjects in matching experiments are confused, as they can be easily influenced by advice.

We test the effect of advice on the well-known top trading cycles (TTC) matching algorithm. We compare three treatments involving third party advice [right advice (R), wrong advice (W), and both right and wrong advice (RW)] to a no-advice baseline (B). In line with previous literature the truth telling rate is higher than 70% in B, but it does not even reach 30% in W. Truth telling rates are also significantly lower in the other advice treatments when compared to B. This evidence suggests that the majority of participants in matching experiments fail to understand strategy-proofness, as they can be easily influenced by advice. High truth telling rates may just be the result of participants taking a default option.