We discuss the extent to which the expectation of a rare event, not present in the usual post-war sample data, "the peso problem" can affect the behaviour of rational agents and the characteristics of market equilibrium. To that end, we describe quantitatively the macroeconomic and financial properties of a standard equilibrium business cycle model, modified to allow for a very small probability of a depression state. We are careful to contrast what would be the stationary probability distribution descriptive of the dynamic rational expectations (RE) equilibrium, from the empirically observed behaviour of the economy under the same RE assumption when the depression does not appear in the sample. The effects of small probability events appear to be especially significant for financial market characteristics. We produce a reasonable model specification, for which both business cycle characteristics and mean financial returns are in accord with US observations. The 6.2% premium is obtained in an economy where agents are only moderately risk averse and where there are no frictions.