Description:
This paper addresses a general problem with surveys asking agents for their assessment of the state of the economy: answers are highly dependent on information that is publicly available, while only information that is not already publicly known has the potential to improve a professional forecast. We propose a simple procedure to disentangle the private information of agents from knowledge that is already publicly known (that is common knowledge) for surveys that are structured like that for the European Commission's consumer sentiment indicator. We show that, empirically, this procedure works quite well for some economies, in particular for Germany.