Elements of a theory of design limits to optimal policy

This paper presents a framework for understanding the limits that exist in optimal policy design in dynamic contexts. We consider the design of policies in the context of dynamic linear models. Fundamental design limits exist for policy rules in such environments in the sense that any policy rule embodies tradeoffs between the magnitudes of different frequency-specific components of the variance. Hence policies that are effective in eliminating low frequency variance components of a state variable can only do so at the cost of exacerbating high frequency variance components, and vice versa. Examples of the implications of such tradeoffs are considered.