You are applying a model with fixed r No, this is not true. The whole point is that R=R(r). Its just that the relationship is ~chaotic, not linear. And it gets

You are applying a model with fixed r No, this is not true. The whole point is that R=R(r). Its just that the relationship is ~chaotic, not linear. And it gets messy as either r or R --> 0. The fact that preferences are recursive on (r) should not be a surpirse, but I don't think mainstream Economists generally go there. On the topic of "temporary" vs Uncertain changes in (r), they are not the same. One is risk, the other is uncertainty. In general risk is priced a certain way (reasonably efficient), uncertainty is something you have preferences for (ie, you dont "price" it in the normal sense) and can only model indirectly in the first instance.[1] [1] eg, http://economistsview.typepad.com/economistsview/2010/06/risk-versus-uncertainty.html http://economistsview.typepad.com/economistsview/2010/06/ris...