Illia Pasichnichenko, University of Sussex

"Value of Partial Information"

Abstract

Blackwell’s theorem relates the value of information to the “informativeness” of the information structure. His analysis applies to decision makers who are expected utility maximizers and know the information structure of the decision problem. When decision makers do not know the information structure precisely, the signal generating process and the posterior distributions are often only partially known. This paper studies preferences of decision makers with partial knowledge about signals and posterior probability distributions. The partial information approach allows us to relate the value of information to the decision maker’s attitude towards ambiguity. We introduce a new concept of informativeness based on the centroid and prove a theorem in the spirit of Blackwell. Furthermore, we characterize the value of information in terms of the preference relation over information structures. Depending on ambiguity attitude the value of information may be negative.

Contact person: Frikk Nesje