|
Field
Experimental Design (Public)
|
Before
The following hypotheses are tested:
H1: In the absence of a governance policy (baseline condition of ‘laissez faire’), the appropriation externalities push players to assume a behaviour conducing to a Nash equilibrium.
H2: The introduction of governance policies based on informational instruments reduces the depletion rate of the CPR, bringing the system closer to the Social optimum.
H3: Communication among players contributes to bringing the system closer to the Social optimum.
To test these hypotheses, four Treatments were designed:
T0 (Baseline)
T1 (Simulator)
T2 (Communication + Simulator)
T3 (Expert Advice + Simulator)
In the baseline treatment (T0), players dispose only of a calculator facilitating their extraction choice at every given period.
In the simulator treatment (T1), players dispose of a calculator allowing them to make simulation over all rounds missing before the end of the session (5-t where t = current round). This simulator is a proxy of more information available to players about the consequences of dynamic extractions of the CPR on the CPR availability and subsequent extraction costs. As we can keep record of the use of the simulator by players, this treatment also allows observing the attitude by certain players to investigate dynamic externalities (sophisticated players looking at future extractions) in comparison with myopic players that are interested only in static externalities.
In the communication treatment (T2), players have the calculator like in T1 and in addition can chat (cheap talk) during 2 minutes before rounds 1, 3, and 5 in sequence 2.
In the Expert Advice treatment (T3), players have the calculator like in T1 and T2, and in addition receive every round a suggestion about the global amount of CPR to extract in order to maintain the social optimum in the group.
The model is the same as in Gardner et al., (1997), with groups of 5 players that can extract up to 25 tokens each at every round.
The experimental protocol involves three parts (sequences) of 5 rounds each per session. The first sequence is a training session not remunerated. The second session is always the baseline (T0), whilst the third sequence is a treatment (T0 to T3). In total 15 rounds are played. The experimental design is summarized as follows:
Sessions: Baseline, Simulator, Communication + Simulator, Expert Advice + Simulator
Sequence 1: Rounds 1-5 (Training)
Sequence 2: Rounds 6-10 (T0)
Sequence 3: Rounds 11-15 (T0, T1, T2, T3)
Groups & participants per session: 12 Groups 60 Participants
This experimental protocol will be run twice in the lab (Montpellier and Tunis), and once in the field, with farmers in the Jemna oasis, South Tunisia. A total of 240 x 3 = 720 participants is expected.
Empirical strategy
We use a difference-in-differences estimation to identify the effect of the introduction of policy measures relying on variation in the level of extraction both across groups and over sequences. Extraction levels are compared before and after policy implementation in treated groups with the extraction level of the baseline groups without policy.
|
After
The following hypotheses are tested:
H1: In the absence of a governance policy (baseline condition of ‘laissez faire’), the appropriation externalities push players to assume a behaviour conducing to a Nash equilibrium.
H2: The introduction of governance policies based on informational instruments reduces the depletion rate of the CPR, bringing the system closer to the Social optimum.
H3: Communication among players contributes to bringing the system closer to the Social optimum.
To test these hypotheses, four Treatments were designed:
T0 (Baseline)
T1 (Simulator)
T2 (Communication + Simulator)
T3 (Expert Advice + Simulator)
In the baseline treatment (T0), players dispose only of a calculator facilitating their extraction choice at every given period.
In the simulator treatment (T1), players dispose of a calculator allowing them to make simulation over all rounds missing before the end of the session (5-t where t = current round). This simulator is a proxy of more information available to players about the consequences of dynamic extractions of the CPR on the CPR availability and subsequent extraction costs. As we can keep record of the use of the simulator by players, this treatment also allows observing the attitude by certain players to investigate dynamic externalities (sophisticated players looking at future extractions) in comparison with myopic players that are interested only in static externalities.
In the communication treatment (T2), players have the calculator like in T1 and in addition can chat (cheap talk) during 3 minutes before every round in sequence 2. Cheap talk is based on the choice of one of eight pre-defined messages, following four principles: individual/collective management; time preferences; opinion/signal to others; increase/reduce the extractions. For each principle, a positive and a negative message in terms of water management is available. In the interface of choice, the 8 messages will appear on 4 rows and 2 columns. The order of the rows will be randomly tossed for each group and will be the same for the 5 rounds of the sequence. The order of the columns will change randomly every round. Subjects can add text on their own if they want in addition to the pre-defined messages.
In the Expert Advice treatment (T3), players have the calculator like in T1 and T2, and in addition receive every round a suggestion by an Expert about the amount of CPR to extract in order to maintain the social optimum in the group. The total amount for the group and the corresponding amount for each member is provided each round.
The model is the same as in Gardner et al., (1997), with groups of 5 players that can extract up to 25 tokens each at every round.
The experimental protocol involves three parts (sequences) of 5 rounds each per session. The first sequence is a training session not remunerated. The second session is always the baseline (T0), whilst the third sequence is a treatment (T0 to T3). In total 15 rounds are played. The experimental design is summarized as follows:
Sessions: Baseline, Simulator, Communication + Simulator, Expert Advice + Simulator
Sequence 1: Rounds 1-5 (Training)
Sequence 2: Rounds 6-10 (T0)
Sequence 3: Rounds 11-15 (T0, T1, T2, T3)
Groups & participants per session: 12 Groups 60 Participants
This experimental protocol will be run twice in the lab (Montpellier and Tunis), and once in the field, with farmers in the Jemna oasis, South Tunisia. A total of 240 x 3 = 720 participants is expected.
Empirical strategy
We use a difference-in-differences estimation to identify the effect of the introduction of policy measures relying on variation in the level of extraction both across groups and over sequences. Extraction levels are compared before and after policy implementation in treated groups with the extraction level of the baseline groups without policy.
|