Bureaucrats, voters, and public investment

This paper explores the provision of a durable public good in Romer and Rosenthal's agenda setter model. It identifies a type of equilibrium, called a Romer–Rosenthal equilibrium, in which in every period the bureaucrat proposes the maximum level of public investment the voter will support. The paper establishes that such an equilibrium exists for a variety of public good benefit functions. Equilibrium public good levels converge or almost converge to a steady state. These steady states can involve a unique public good level being provided each period or may exhibit a two period cycle. Steady state public good levels exceed the voter's optimal level. More surprisingly, steady state equilibrium reversion levels can exceed the voter's optimal steady state level, meaning that reversion levels cannot be used to bound the optimal level. Reflecting the inability of the agents to commit to their future proposing and voting behavior, equilibrium paths are Pareto inefficient.