Not So Cheap Talk: Costly and Discrete Communication

26 Pages Posted: 29 Jun 2010 Last revised: 16 May 2012

See all articles by John Smith

John Smith

Rutgers University-Camden

Johanna Hertel

Avadel Pharmaceuticals

Date Written: May 2, 2012

Abstract

We model an interaction between an informed sender and an uninformed receiver. As in the classic cheap talk setup, the informed player sends a message to an uninformed receiver who is to take an action which affects the payoffs of both players. However, in our model the sender can communicate only through the use of discrete messages which are ordered by the cost incurred by the sender. We characterize the resulting equilibria without refining out-of-equilibrium beliefs. Subsequently, we apply an adapted version of the no incentive to seperate (NITS) condition to our model. We show that if the sender and receiver have aligned preferences regarding the action of the receiver then NITS only admits the equilibrium with the largest possible number of induced actions. When the preferences between players are not aligned, we show that NITS does not guarantee uniqueness and we provide an example where an increase in communication costs can improve communication. As we show, this improvement can occur to such an extent that the equilibrium outperforms the Goltsman et al. (2009) upper bound for receiver's payoffs in mediated communication.

Keywords: information transmission, cheap talk, costly communication

JEL Classification: C72, D82, D83

Suggested Citation

Smith, John and Hertel, Johanna, Not So Cheap Talk: Costly and Discrete Communication (May 2, 2012). Available at SSRN: https://ssrn.com/abstract=1632129 or http://dx.doi.org/10.2139/ssrn.1632129

John Smith (Contact Author)

Rutgers University-Camden ( email )

Department of Economics
311 N. 5th St., 421 Armitage Hall
Camden, NJ 08102
United States

HOME PAGE: http://www.JohnSmithEcon.com/

Johanna Hertel

Avadel Pharmaceuticals ( email )

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
64
Abstract Views
874
Rank
627,771
PlumX Metrics