“…First, as long as |M| ≥ 2, Games 1 and 2 both have multiple perfect Bayesian equilibrium outcomes not Pareto ranked. This creates rooms for equilibrium selection, and yet Pareto 9 Other studies that adopt between-subject manipulations of message spaces include Blume et al (1998), who document that restricting the message space expedites convergence in the presence of meaningless messages; Lai, Lim, and Wang (2015), who highlight the role of message spaces in facilitating information transmission in a multidimensional setting; and Serra-Garcia, van Damme, and Potters (2013), who study the effect of limiting the message space in public good games with communication of private information. Our novel contribution relative to these studies lies in providing a new channel-a refinement of equilibria-through which message spaces may play a role in information transmission.…”