Climate-induced economic risks and cooperative behavior: An experimental analysis

There are two between-subjects treatments, baseline and uncertainty.

Baseline: Groups of 4 players play the dynamic public goods game as described in Gächter et al. (2017). At the beginning of each round, players decide individually and anonymously how much of their initial endowment to invest in the public good. At the end of the round, players receive feedback and their individual payoffs become their initial endowment for the next round. The game ends after 10 rounds.

Uncertainty: Same as baseline, but with a 20% probability that an event destroys between 0% and 50% of the payout. The amount of damage depends on the group investment in the public good; the higher the investment, the lower the damage.

Strategic equivalence: The marginal per capita return (MPCR) is modified in the Uncertainty Treatment so that its expected value is identical to that of the Baseline Treatment.

The two rounds in the uncertainty treatment in which the event occurs are the same for all participants and are drawn in advance.

2025-08-25

2025-10-31

The main outcome variable is the relative share of initial endowment that is invested in the public good at the individual or group level.
In the first round, all observations are independent and can be considered at the individual level.
In the following rounds, the observations at the group level are dependent, and therefore the group average or group outcome in round 10 is considered.

Each round is a standard public goods game with the payoff function
N_(i,t+1)=N_(i,t)-c_(i,t)+MPCR_(i,t)*sum_(j) c_(j,t)-D_(i,t)

N_(i,t+1) is the payoff of round t for player i which become the initial endowment of round t+1.
N_(i,t) is the initial endowment (which is the same for all player in period 1).
c_(i,t) is his contribution to the public good.
MPCR_(i,t) is the marginal per capita return. In Baseline, it is simply 1.5/4 (4 is the group size). In Uncertainty, it is (1.5/4-p*delta*N_(i,t)/sum_(k)N_(k,t)), where p=0.2 is the ex-ante probability of an extreme event and delta=0.5 is the size of the damage.
D_(i,t) is the expected individual damage. D_(i,t)=p*f_(t)*d_{i,t], where f_(t) is the damage factor (1-sum_{j,t}*c_{j,t}/sum_{j,t}N_{j,t}) and d_(i,t) is the individual maximum damage delta*N_{i,t}.

In each round group members simultaneously and anonymously submit their individual contributions c_(i,t)s. At the end of the round, feedback is given. In the Uncertainty Treatment, an event may occur that destroys up to half of the payoff. The payoff of the current round becomes the endowment of the next round. After 10 rounds the game ends and participants receive their payoffs.

We also elicit subjects' risk attitudes and socio-demographic data.

by computer

individual/group; random assignment to one group of 4 participants/of the two treatments

No

None

400 Participants = 100 Groups

50 Groups per Treatment