Non‐Expected Utility, Saving and Portfolios

Abstract

Despite increased stockholding opportunities, standard expected‐utility models overpredict household participation and stock holdings. It has been suggested that departures from expected utility could resolve both puzzles. We investigate three measurable departures: (i) Kreps‐Porteus preferences, (ii) Yaari's Dual Theory, and (iii) Quiggin's Rank‐dependent Utility. Improvements tend to occur in predicted portfolio composition rather than participation. They are limited under (i), questionable under (ii), and more sizeable under (iii). Contrary to priors in the literature, improvements under (iii) do not result from solutions at kinks of indifference curves. We conclude that stockholding puzzles are unlikely to be resolved through preferences alone.