How Verifiable Cheap-Talk Can Communicate Unverifiable Information

This study describes a “cheap-talk” model in which sellers can credibly convey unverifiable information by choosing whether or not to exaggerate verifiable information. We find that unexaggerated claims can communicate favorable unverifiable information if buyers are not too likely to verify claims, and sellers with better information care more about future prices than sellers with worse information. However, there is always another equilibrium in which sellers exaggerate all verifiable claims. Laboratory tests show that when buyers infrequently verify the sellers' claims, players converge to the equilibria close to the example provided in instructions. When buyers are very likely to verify claims, players fail to converge to any equilibrium. Both of these results are consistent with an evolutionary learning model, but inconsistent with the intuitive criteria of Cho and Kreps (1987). We discuss the implications of our results for both consumer and financial markets. Copyright Springer Science + Business Media, Inc. 2005