Unbounded probabilistic sophistication

I extend Machina and Schmeidler’s (1992) model of probabilistic sophistication to unbounded uncertain prospects (acts) and derive risk preferences over the induced probability distributions (lotteries) with unbounded support. For example, risk preferences can be derived over normal, exponential, and Poisson families of probability distributions. My extension uses a version of Arrow’s (1970) Monotone Continuity, which implies countable additivity for subjective beliefs and a novel property of tail-continuity for the revealed risk preferences. On the other hand, I do not assume P6 (Small Event Continuity) that is used both by Savage (1954) and Machina–Schmeidler. © 2010 Elsevier B.V. All rights reserved.