The experiment was organized as follows:
The 25 participanting students‘ task was to decide on a tax scheme which was needed to raise money for a public good (a hospital).
1. At a first stage (constitutional stage), all participants had to agree unanimously on a voting rule which decided the implementation of a tax scheme (post-constitutional decision process). Participants could choose between
- supermajority (2/3 of the votes needed to pick a choice)
- simple majority (where the distributional scheme with most votes is chosen)
- dicatorship (one randomly chosen person dictates the tax scheme)
2. At the second (post-constituitonal) stage, participants chose between four different tax schemes, each yielding a specific income distribution, represented by five different income groups. The only knowledge given to the participants was the income of each group, depending on the tax scheme, but no-one knew to which income group he or she would belong. Given this information, the participants chose one tax scheme, where the decision was made according to the decision rule agreed on at stage 1.
3. After the decision was made, a lottery was run, deciding to which of the five income groups any participants belonged to. By that, each participant was allocated a specific hypothetical income.
4. Each participant received a real monetary pay-out in Euros in relation to the hypothetically received income.
5. After the pay-outs were realized, the participants had to evaluate the decision. They could, now having the knowledge of their hypothetical income, communicate to the conductors of the experiment whether they were content with the reached result, or whether they would have liked to change their initial decision.
The stages 2 to 5 were run twice. In a first run, the tax-programms was designed such that hypothetical redistribution did not lower the aggregate income of our model-society. In the second run, more redistribution went along with a lower overall group income. The participants were able to make two independent decisions for each run, hence e.g. deciding for tax scheme A in the first run, but then decide for scheme B instead of A in the second one.
We would like to thank „Verband der Freunde der Universität Freiburg im Breisgau e.V.“ for their financial support provided for the related Master Thesis.