Experimental evidence on the evolution of the meaning of messages in sender-receiver games : An experiment

TLDR: In this article, Blume et al. consider communication between a privately informed sender and a receiver when the sender's message does not affect payoffs directly and experimentally investigate the evolution of message meanings when talk is cheap.

Abstract: This paper experimentally investigates the evolution of message meanings when talk is cheap. We consider communication between a privately informed sender and a receiver when the sender's message does not affect payoffs directly. Sender-receiver games, introduced by Jerry Green and Nancy Stokey (1980) and Vincent P. Crawford and Joel Sobel (1982), provide the simplest stylized environment in which communication is essential; communication is the only way to link the receiver's action to the sender's private information. In addition, sender-receiver games have been widely used to study communication in accounting, economics, finance, and political science. Intuition strongly suggests that effective communication will evolve when the players' interests are closely aligned. However, there is always a Nash equilibrium in which the sender randomizes uniformly across all messages. Such "babbling equilibria" are proper (Roger B. Myerson, 1978; Blume, 1994), and even strategic stability (Elon Kohlberg and JeanFrancois Mertens, 1986) does not rule out uninformative equilibria in general. In response, a literature developed that formulates equilibrium selection criteria by appealing to the existence of a commonly understood language (Matthew Rabin, 1990; Steven A. Matthews et al., 1991; Joseph Farrell, 1993) that can be used to formulate credible deviations from equilibrium. Unfortunately, the selection criteria from this literature are frequently strong enough to remove all equilibria from consideration. In addition, these arguments are vulnerable to Stiglitz's observation (In-Koo Cho and David M. Kreps, 1987) that successful deviations from equilibrium may induce further deviations, which in turn may undermine the credibility of the original deviations. Both problems can be avoided by a dynamic approach as in Rabin and Sobel (1996). We also adopt a dynamic perspective but do not require messages to have a priori meanings; instead, distinct meanings may emerge endogenously. This is important for communication across cultures and for language change.2 We are guided by recent results of * Blume: Department of Economics, University of Iowa, Iowa City, IA 52242; DeJong: Department of Accounting, University of Iowa, Iowa City, IA 52242; Kim: Department of Economics, Sung Kyun Kwan University, Seoul 110-745, Korea; Sprinkle: Department of Accounting, University of Wisconsin, Madison, WI 53706. We thank Joyce Berg, Vince Crawford, Robert Forsythe, Joel Horowitz, Deirdre McCloskey, Forrest Nelson, Larry Samuelson, Gene Savin, Joel Sobel, and especially two anonymous referees for their advice and suggestions. We benefitted from comments and discussions resulting from presentations at the 1994 summer meetings of the Econometric Society, the 1996 Social Learning Workshop at SUNY-Stony Brook, the IO/ME/GT Workshop at the University of Iowa, the University of Bonn, the University of Bielefeld, the CentER for Economic Research at Tilburg University, Ohio State University, Seoul National University, and Texas A&M University. We are grateful for financial support from the College of Business Administration at the University of Iowa, the Ira B. McGladrey Institute, and the National Science Foundation-