Linde Institute/Social and Information Sciences Laboratory (SISL) Seminar

We implement the Battaglini (2002) model of multi-sender multi-dimension cheap talk in the laboratory, analyzing the effects of sender competition on information transmission. Our results indicate that competing senders provide enough information for close to full revelation, but receiver s ability to use this information crucially depends on senders biases. Receivers are close to full extraction when biases identify an ex-ante trustworthy sender. When there is no ex-ante trustworthy source, fully exploiting sent messages requires information to be inferred across dimensions. However, receivers seem to treat dimensions as separate choices, and are much closer to best-responding within each separate dimension.

Copy of paper available at: http://www.pitt.edu/~alistair/MultiSender.pdf