Human Behavior and Climate Change: Risk and time preferences under voluntary and institutional mechanisms

We are interested in investigating how contribution to a green fund, designed to mitigate the risk of climate change, is related to behavioural parameters such as loss aversion and present bias as predicted by our theoretical model.
The experiment has 3 tasks. In task 1 we elicit individual time preferences using CTB method. In task 2 we elicit certainty equivalents of 2 lotteries. This enables us to estimate individual loss aversion parameters. In task 3 we use a strategy method and ask individuals to declare their contribution to the green fund through a voluntary mechanism or institutional mechanism (median income tax). Subjects will be randomly assigned to one of the 4 treatment arms.

This experiment measures individuals’ present bias and loss aversion parameters and tests their association with individuals’ contribution to a green fund through voluntary and institutional mechanisms (median tax rate). The group green fund will affect the probability of receiving some positive income in the future which is called risky income, resembling the risk of climate change.

Participants will be asked to choose their most preferred combination of a sooner payment and a later payment. We use these choices to estimate the present bias, discount factor and utility function parameters.

Subjects will evaluate lotteries involving both gains and losses through a bisection procedure. The bisection procedure determines a loss value that makes a two-outcome lottery with a given gain outcome and the chosen loss value indifferent to zero. This enables the estimation of loss aversion parameters. Subjects will not experience any losses as they will be endowed with the loss amounts.
Subjects will be asked to declare their voluntary contributions or preferred tax rate for the group green fund while various factors like a timeline of events and the amount of risky income they receive in the future change. There are 3 distinct time periods t1<t2<t3. Decisions about the allocations of an endowment received either at t1 or t2 are made at t1. The risky income whose probability is affected by contributions to the green fund is received at t3 (terminal date). Participants are asked to state their expectations of the group green fund in each question.

There are 4 treatment arms.
T1: Voluntary contribution to the green fund from income received today (t1)
T2: Voluntary contribution to the green fund from income received in the future (t2)
T3: Income tax contribution to the green fund from income received today (t1)
T4: Income tax contribution to the green fund from income received in future (t2)

2022-10-01

2023-03-31

The primary outcomes are subjects' contributions to a green fund along with subject-specific time and risk preferences.

Time preferences are estimated through Convex Time Budgets (CTB) method of Andreoni et al. (2015) in task 1. We elicit certainty equivalents of two lotteries in order to estimate the loss aversion parameter in task 2. We use a bisection procedure with 6 steps to find the value of outcome L such that x>L>0 for which the subject expresses the following indifference:
(-L,0.5;x, 0.5)~0
Since we get an estimate of the parameter of the CRRA utility function in task 1, this indifference enables the estimation of the loss aversion parameter under prospect theory.

Subjects' expectations of the group fund while various factors like a timeline of events and the amount of risky income they receive in the future change. We also collect subject demographics such as age, gender, religion, household income, location and etc.

The experiment has 3 tasks. In task one, we use the CTB method to elicit time preferences. In task 2 we use a bisection procedure to elicit two lotteries' certainty equivalents. Task 3 collects subject contributions toward the green funds.
We use three distinct time periods:t1<t2<t3 and n distinct decision makers.
All decisions are made in the current period at time t1(today). In our baseline treatment,
T1, the endowments for these time periods are (Y,0,Z) where Y>0 and Z>0. At time
t1 decision-makers decide on allocating their current endowment of Y towards
current consumption and green investment. The sum of green investments across
decision-makers in a group determines the stock of green investment, G. At time t3, the endowment of each decision maker, Z, is received with probability p(G) that is increasing in G; this is the good environmental state. However, with probability 1-p(G), a bad environmental state occurs such that each decision maker receives nothing. We collect subject choices for various amounts of Z, t2 and t3 through a strategy method. We have 4 treatment arms. T1 collects voluntary contributions to the green fund from income received today (t1). T2 collects voluntary contributions to the green fund from income received in the future (t2). T3 asks for income tax contributions to the green fund from income received today (t1). And T4 is about income tax contributions to the green fund from income received in future (t2).

Randomization is done through random draws by a computer.

The unit of randomization is individuals.

No

We have four treatments.

We aim for 100 subjects in each treatment.

We are using contribution in public good games to benchmark the standard deviation of subjects' contributions to public goods. Based on Reuben and Riedl (2013) and Andreoni (1988) we opt for the relative pooled standard deviation of 0.10. We assume an effect size of 4 across treatments. With power equal to 80% and alpha of 0.05 for a significant difference in means we require 78 subjects in each treatment. Thus, we are aiming for 100 subjects in each arm.

ClimateChangePAP07092022.docx