This folder contains all files used for the Risk-Sharing experiment. The risk-sharing experiment was programmed in oTree and conducted with Uk-based participants recruited from the Prolific platform between June 5 and June 11. In total, 525 people showed up for these sessions (68% of those who signed up), of which 430 could be assigned to a risk-sharing group. 424 subjects completed the entire experiment. The experiment consisted of four parts. First, the subjects had to independently answer five questions of the staircase method (Falk et al. 2016) to measure risk-preferences. Secondly, they were paired to another subject and had to answer 15 questions of the Social Value Orientation Slider Measure. The third part was the main part of the experiment: the heterogeneous risk-sharing game. In this game, groups of 10 subjects had to decide for 20 rounds whether or not they wanted to share the risk of getting sick and not generating an income with each other. If they shared risk, they would pay a contribution that would be used to pay sick members each round. If they didn't, they would earn their full income or nothing at all (in case they were sick). One round was randomly drawn for payout in order to incentivize participants to share risk (and thus smooth their income). The subjects in the experiment had different risks, but in one condition this varied between 0.15 and 0.25, whereas in the other it varied between 0.07 and 0.33. The aim of the experiment was to study when people were willing to share risk despite heterogeneity in risk preferences (which drives up the costs for low-risk participants). The last part consisted of a survey that asked about motivations to join the risk-sharing group and basic socio-demographics.
This folder contains the apps created for the oTree experiment, all administrative data about the subjects that signed up and participated for the experiment, all raw datasets, a script to merge and clean these datasets and the final datafile. The access to this folder is closed, because it contains information about all subject IDs from the Prolific participants. The final dataset is publicly available.
