Experimental Cost of Information

We relate two main representations of the cost acquiring information: a that depends on experiment performed, as in statistical decision theory, and distribution posterior beliefs, applications rational inattention. show many cases interest, posterior-based costs are inconsistent with primitive model costly experimentation. The inconsistency is at core known limits to application inattention games and, more broadly, equilibrium analyses where beliefs endogenous; we an experiment-based approach helps understand overcome these difficulties. (JEL D82, D83)