Forecasts as repeated cheap talk from an expert of unknown statistical bias

For two periods an expert E announces his forecast of the state to a decision-maker D who chooses action. They disagree about the precision of the probability assessments. At the end of period 1 the state is observed. In the last period E makes announcements more extreme than his forecasts. Despite countable equilibria, full revelation is never realised. When in period 1 E is interested in reputation only, the initial equilibrium partition is finite; E makes announcements of greater uncertainty with respect to his forecasts. When E is interested in action too, reputational concerns mitigate exaggerated reports.