مقایسه برگشت اولویت برای قرعه کشی عمومی و توزیع درآمد

This paper investigates preference reversals both for lotteries and income distributions as well as their interrelationship. In doing so, it offers several new features: Firstly, it provides a joint analysis of lotteries and income distributions. Secondly, the stimulus material used is considerably more general than used so far in the analysis of preference reversal. Instead of relying on binary lotteries and income distributions, we used ten multiple-outcome lotteries and ten n-dimensional income distributions which corresponded exactly to the lotteries. The lotteries and income distributions were chosen so as to model generalized P-bets (negatively skewed distributions) and generalized $-bets (positively skewed distributions), but we also used other shapes (uniform, symmetrical, and bimodal distributions). Thirdly, we applied material incentives to our subjects which were recruited from students of the Universities of Bari, Italy, and Castellón, Spain. Subjects were asked both to rate and evaluate the ten lotteries or the ten income distributions, respectively. We found heavy response-mode effects which confirm classical preference reversal, and revealed new patterns of preference reversal. Further, the transfer principle was largely violated.

Many scholars have recognized the close relationship between lotteries and income distributions. For instance, Kolm (1969) and Atkinson (1970) established the concept of the equally distributed equivalent income (EDE) by analogy with the certainty equivalent (CE) of a lottery.1 There are several ways to elicit preferences between lotteries and, a fortiori, between income distributions. Preferences between lotteries may be elicited by observing subjects’ choices, or they may be elicited by asking them for their valuations. Empirical research on such preferences has evidenced significant response-mode effects: Many subjects exhibit contradictory preferences when different elicitation methods are used. This phenomenon has been termed preference reversal. It has been well evidenced with lotteries, but has also recently been evidenced with income distributions. The main thrust of this paper is the investigation of preference reversal by using more general distributions than normally used in the literature, and comparing their preference reversal patterns for lotteries with their patterns for income distributions. Moreover, our paper offers several methodological advances.

Although there is a close relationship between income distributions and lotteries, their joint analysis is much in its infancy. Multiple-outcome distributions and lotteries have hardly ever been employed systematically, material incentives have rarely been used, preference reversal has only recently been tested as regards income distributions (although the EDE represents the basis of ethical inequality measures), and the influence of general distributional shapes on the subjects’ evaluation of distributions and their perception of happiness has not been systematically investigated. Our experimental design encompassed ten distributions (three negatively skewed, four positively skewed, one rectangular, one unimodal, and one bimodal) which were used as stimuli in a lottery and an income distribution experiment, which was administered to more than 50 subjects each both in Italy and Spain. The subjects’ comprehension of the experimental setting was tested before the experiment proper and material payoffs were applied. The subjects were asked to rate the lotteries or income distributions and to supply their CEs or EDEs using a BDM incentive scheme.