In rational beliefs (RB) models there is an observed empirical distribution for the stochastic process of state variables. Many different weakly asymptotic mean stationary (WAMS) processes could have generated this empirical distribution, i.e. are consistent with it, and each of them are therefore called a rational belief.We provide a general framework for using RB in general equilibrium models. Individual rational beliefs are assumed correlated by means of sunspots which at the aggregate level lead to excess volatility.The application adapts the proof by Duffie et al (1994) of the existence of a stationary ergodic RE equilibrium to the case where agents hold rational beliefs.