This paper takes the parameters in central bank loss functions as fundamental preferences to be estimated from the data. It is these preferences (along with target values) that define the policy regime in operation and that potentially change with senior central bank appointments. Optimizing central banks apply policy rules whose feedback coefficients are functions of its preferences. Consequently, under some conditions, it is possible to back out estimates of the preference parameters from estimated policy reaction functions. This paper establishes conditions under which a policy regime can be identified and illustrates these conditions using a number of popular models.