Friedman's (1960) proposal to pay interest on (required) reserves is considered in a setting that eliminates the indeterminacy of steady-state equilibrium discussed by Sargent and Wallace (1985). In an overlapping-generations model where the real rate of interest is technologically determined, the payment of interest on reserves results in a determinate, Pareto optimal steady-state equilibrium. However, interest payments on reserves reduce the steady-state welfare of all young agents and, for many economies, result in the existence of stationary sunspot equilibria. This is the case even if such equilibria cannot exist when reserves do not earn interest.

This article presents a monetary growth model where spatial separation and limited communication create a role for banks. Monetary policy interacts with the financial system's liquidity provision to affect the existence, multiplicity, and dynamical properties of equilibria. Moderate levels of risk aversion and tight monetary policy can lead to multiple steady states. Dynamical equilibria can be indeterminate, with oscillatory paths. Thus financial market frictions are a source of indeterminacies and endogenous volatility. Under plausible conditions, tight monetary policy raises the nominal interest rate and inflation rate and reduces long run output. Thus, a central bank's liquidity provision can promote growth.

An endogenous growth model with multiple assets is developed. Agents who face random future liquidity needs accumulate capital and a liquid, but unproductive asset. The effects of introducing financial intermediation into this environment are considered. Conditions are provided under which the introduction of intermediaries shifts the composition of savings toward capital, causing intermediation to be growth promoting. In addition, intermediaries generally reduce socially unnecessary capital liquidation, again tending to promote growth.