Session Discussion: Dynamic Risk Preferences, Poverty Traps, and Thresholds

Abstract

These papers constitute a renewed search for an alternative to what Just and Michelson describe as the classic utility function-a concave function of current wealth plus contemporaneous income. The motivation for this search can be viewed as a response to Matthew Rabin's demonstration that uniform concavity has incredible implications for behavior (e.g., Rabin and Thaler). The remedy Just and Michelson seek is (explicitly or implicitly) an indirect utility function in first-period income, albeit with both concave and convex segments, such that the decision-maker can be risk-averse or risk-loving, depending on the range of outcomes for the decision in question. A unifying theme of the session is that nonconvexities in transaction costs and asset returns, especially capital-stock thresholds, induce the appearance of risk preferences without the contrivance of exogenous risk preferences in first-period income. Buschena and Zilberman assume that the prototypical decision maker has different concave utility functions for different status levels. Assuming that the probability distribution of being in various status groups is a function of income in the current period, one can readily derive indirect utility functions of current-period income. For the two-status case, this results in a Friedman-Savage (FS) shaped indirect utility function. This helps underscore the presumption of both concave and convex segments, albeit by assuming the existence of classic utility functions in the first place. The intuition of the convex section in the middle of the FS function is clear enough; it stems from the prospect of a discontinuous jump to the high-status utility function. But one wonders why the function should be concave at very low incomes. Indeed, Lybbert and Barrett discuss the illustrative case of two thresholds-one at subsistence income and a Micawber threshold