Income Distributions versus Lotteries Happiness, Response-Mode Effects, and Preference

This paper provides a comparative experimental study of risky prospects (lotteries) and income distributions. The experimental design consisted of multi-outcome lotteries and n-dimensional income distributions arranged in the shapes of ten distributions which were judged in terms of ratings and valuations, respectively. Material incentives applied. We found heavy response-mode effects, which cause inconsistent behavior between rating and valuation of lotteries and income distributions in more than 50% of all cases. This means that ethical inequality measures lack support in peoples' perceptions. In addition to classical preference reversals between generalized P-bets and $-bets we observed three additional patterns of preference reversal, two of which apply only to income distributions. Dominating Lorenz curves and Lorenz curves cutting others from below receive decidedly higher ratings (which implies risk and inequality aversion), but lower valuations. The transfer principle is largely violated. The rating of lotteries is a decreasing function of skewness, the rating of income distributions is a decreasing function of standard deviation. The valuation of lotteries is an increasing function of standard deviation and kurtosis, and the valuation of income distributions is an increasing function of standard deviation, skewness, and kurtosis.