GENERALIZED EXPECTED UTILITY ANALYSIS AND PREFERENCE REVERSALS - SOME INITIAL RESULTS IN THE LOSS DOMAIN

In this paper a series of experiments are reported using human subjects choosing over distributions of negative expected value. The experiments were designed to examine preference reversals and fanning out type violations of expected utility theory. Each subject was endowed with an initial wealth position and asked to choose between lotteries in which they were forced, if they lost, to forfeit a portion of their fixed endowment. Consistent and systematic violations of expected utility theory were found within the loss domain which were a mirror reflection of those found within the gain domain. Preference reversals were also present over losses, but the reversal rate was lower than the rate for prospects with positive expected value. No relationship between fanning out type violations of expected utility theory and preference reversals was found, in contrast to recent theoretical predictions. © 1991.