This paper argues that limited asset market participation is crucial in explaining U.S. macroeconomic performance and monetary policy before the 1980s, and their changes thereafter. We develop an otherwise standard sticky-price dynamic stochastic general equilibrium model, which implies that at low asset-market participation rates, the interest rate elasticity of output (the slope of the IS curve) becomes positive - that is, "non-Keynesian." Remarkably, in that case, a passive monetary policy rule ensures equilibrium determinacy and maximizes welfare. Consequently, we argue that the policy of the Federal Reserve System in the pre-Volcker era, often associated with a passive monetary policy rule, was closer to optimal than conventional wisdom suggests and may thus have remained unchanged at a fundamental level thereafter. We provide institutional and empirical evidence for our hypothesis, in the latter case using Bayesian estimation techniques, and show that our model is able to explain most features of the "Great Inflation."