Experimental Design
Our design encompasses two treatment groups, L and H, as well as a control group C. In the two main treatments L and H, subjects are divided into groups of three where individuals are randomly placed within the wealth distribution by receiving an endowment. In addition, they are informed about the endowments of their group members. Subjects in the control group C simply receive an endowment without being placed in a group or learning about the wealth distribution. Further, our experiment uses a mixed within- and between-subjects design, meaning that subjects are randomly assigned to one of the three treatments in the first place, but then go through the other two treatments in random order as well.
To elicit subjects’ risk preferences, we follow the elicitation method of Gneezy and Potters (1997). Our experiment is designed in such a way as to examine how subjects’ risk preferences change after experiencing wealth changes only at the relative level: some individuals receive identical endowments in L and H, but the endowment of their group members is different, i.e. they place differently in the wealth distribution. Subjects receive information about the wealth levels of the other group members and are explicitly informed whether they have the lowest, median or highest wealth endowment in their group. Under L and H, subjects are informed that all subjects within their group have been equipped with a different initial endowment, whereby this allocation took place randomly and the possible endowments were:
• L = {2,150 coins; 3,450 coins; 4,050 coins},
• H = {3,450 coins; 4,050 coins; 7,150 coins}.
With this setup, those subjects that receive 3,450 Taler or 4,050 Taler are endowed with the same absolute wealth compared to those in the other treatment, but experience a variation in relative wealth placement. Under treatment L, subjects have a higher wealth placement compared to treatment H; yet, absolute wealth levels are the same. Accordingly, this allows us to evaluate how variations only in relative wealth – holding the absolute wealth level constant – impact individuals’ risk-taking behavior.
Subjects in the control treatment C only receive an initial endowment of either 3,450 Taler or 4,050 Taler without any further information about their placement in the relative wealth distribution (or other participants).