How experts decide:identifying preferences versus signals from policy decisions

A large theoretical literature assumes that experts di ffer in terms of preferences and the distribution of their private signals, but the empirical literature to date has not separately identi ed them. This paper proposes a novel way of doing so by relating the probability a member chooses a particular policy decision to the prior belief that it is correct. We then apply this methodology to study diff erences between internal and external members on the Bank of England's Monetary Policy Committee. Using a variety of proxies for the prior, we provide evidence that they di ffer significantly on both dimensions.