Strategic Communication: An Experimental Investigation

In this paper we attempt to compare theoretically and experimentally three models of strategic information transmission. In particular we focus on the models by Crawford & Sobel (1982), Lai (2010) and Ehses-Friedrich (2011). These three models differ in the information that the receiver possesses and the sender's knowledge about these information. Lai, 2010 introduce a partially informed decision maker into Crawford & Sobel's model. Ehses-Friedrich (2011) makes the decision maker's knowledge public knowledge. The experiment replicates the results of earlier experimental studies (Dickhaut et al., 1995, Cai & Wang, 2006, Wang et al., 2010): on the one hand experts usually give a too truthful advice, they overcommunicate. On the other hand the decision makers rely too much on the received information. Moreover, communication as well as payoffs decrease with increasing preference differences. We find that when decision makers are privately informed the messages from the expert to the decision maker are less precise than in the baseline setting. In the public information treatment, the communication is less biased. In all treatments, however, the messages are more precise than theoretically predicted.