Transcripts from the meetings of the Federal Open Market Committee (FOMC) show that the policy proposed by its chair is always adopted with a majority of votes and limited dissent. An interpretation of this observation is that the power of the chair vis-a-vis the other members is so large that the policy selected by the committee is basically that preferred by the chair. Instead, this paper argues that the observation that the chairís proposal is always approved is an equilibrium outcome: the proposal passes because it is designed to pass and it does not necessarily correspond to the policy preferred by the chair. We construct a model of inclusive voting where the chair has agenda-setting powers to make the proposal that is initially put to a vote but is subject to an acceptance constraint that incorporates the preferences of the median and the probability of counter-proposals. The model is estimated by the method of maximum likelihood using real-time data from FOMC meetings. Results for the full sample and sub-samples for each chair between 1974 and 2008 show that the data prefer a version of our model where the chair is moderately inclusive over a dictator model. Thus, the workings of the FOMC appear to be stable over time and no chair, regardless of personality and recognized ability, can deviate far from the median view.