bc3 seminars: "effective climate agreements under uncertainty" dr.mcevoy (2015-05-07)
TRANSCRIPT
Effective Climate Agreements under Uncertainty
a Center for International Climate and Environmental Research, Oslo (CICERO) b Department of Economics, Appalachian State University, Boone NC We thank the International Foundation for Research in Experimental Economics (IFREE) for funding this research
Todd L. Cherrya,b and David M. McEvoyb
Basque Centre for Climate Change May 2015
A collective action problem
Collectively countries are better off reducing greenhouse-gas emissions, but individually they are better off left unconstrained. The non-cooperative outcome is socially undesirable. There is a need to design institutions to change the underlying incentives à International Agreements
By many metrics the Kyoto Protocol is/was a failure
Part of that failure can be attributed to free riding:
Non-participation Non-compliance
Participation and compliance
Countries must voluntarily join an agreement
… AND they must find it in their interest to comply with their commitments … AND this all has to happen under a great deal of uncertainty regarding the damages from climate change
Effective agreements
“the economic impact of climate change […] is the thorniest issue in climate-change economics” - William Nordhaus
DICE (Dynamic Integrated Climate Economy)
A deposit-refund/performance bond system:
Signatories pay deposits upon ratification Deposits are paid back to compliant members
Who’s talking about this kind of institution/agreement?
Finus 2008 Gerber and Wichardt 2009 Hovi et al. 2012 McEvoy 2013; Cherry and McEvoy 2013 Hovi, Ward and Grundig 2014
The institution we consider
Countries are uncertain about the benefits of emissions abatement when deciding whether to join the agreement When making emissions decisions the uncertainty is resolved (Kolstad 2007, JEEM)
Uncertainty
We design a series of public goods experiments to test the effectiveness of an agreement that requires up front deposits. We consider the effects of:
Uncertainty in the benefit of contributing Requiring less than full participation Low deposits
Objective
The Problem of Designing Effective Institutions
We lack the data required to test the effectiveness of different agreement architectures.
The Role of Experiments
1. Speaking to Theorists
2. Searching for Facts
3. Searching for Meaning 4. Whispering in the Ears of Princes
-Alvin Roth
The Role of Experiments
The laboratory as a “wind tunnel” -Vernon Smith
Kosfeld et al. 2009* McEvoy 2010* Tavoni et al. 2011 Dannenberg et al. 2011;2012* Barrett and Dannenberg 2012; 2013 Kolstad and Burger 2011* McEvoy et al. 2011*; Cherry and McEvoy 2013* Hovi et al. 2012 Miliniski et al. 2008 Vasconcelos et al. 2014
Experiments on collective action to provide global public goods
Baseline – voluntary contribution mechanism (VCM) each player in a group of 5 has 14 tokens and can contribution 0 to 10 to a public account
Marginal return (MPCR) = 0.60
Nash equilibrium = zero contributions, each player earn 14 tokens Social optimum = full contributions (50 tokens to public account) Average contribution = 4.17 tokens 1 token = $1
Full participation – certain benefits Stage one: players decide simultaneously whether to join an agreement If all 5 players join the agreement then it enters into force. Members pay a deposit of 4 tokens and commit to contributing at least 8 tokens to the public account. If fewer than 5 join then an agreement does not form and everyone reverts to the baseline (VCM). Stage two: players decide their contribution amount. If members are compliant (when agreements form) they get the 4 token deposit back. Otherwise they lose it.
Stage two: if agreement forms: full compliance (all 5 players contribute 8 tokens) – each player earns 30 tokens
why? deposit > gains from violations if agreement fails: back to baseline – each player earns 14 tokens Stage one: since all players are critical they all join: full participation
Equilibrium (SPNE)
Full participation – uncertain benefits Stage one: when players decide whether to join an agreement they DO NOT know the MPCR. They know it will be either HIGH (0.80) or LOW (0.40) with a 50/50 likelihood. Note: expected MPCR is 0.60 Stage two: state of world is revealed, then same as before Note: Half of the time deposit amounts of 4 tokens will be TOO LOW to motivate compliance
Stage two:
When return is HIGH: full compliance (all 5 players comply and contribute 8) – each earns 36 When return is LOW: no compliance (all 5 players contribute zero and lose their deposit) – each earns 10
Stage one: all 5 (risk neutral) players join: full participation
Equilibrium (SPNE)
Partial participation – certain & uncertain benefits Stage one: players decide whether to join an agreement Like previous except that an agreement enters into force if at least 3 players join Stage two: same as before for members. Non-members make no deposits and have no commitments
Partial participation – certain & uncertain benefits Note: Agreements of size 3 are profitable: Compliant members earn = 14 – 8 + 0.60(24) = 20.4 Non-members earn = 14 + 0.60(24) = 28.4
Stage two: if agreement forms
When return is HIGH: full compliance (members comply and contribute 8 and earn 23.2 tokens) When return is LOW: no compliance (members contribute zero and lose their deposit and earn 10 tokens)
Stage one: two equilibria An agreement with 3 members and 2 nonmembers No agreement
Equilibrium (SPNE)
Low deposits – full participation Marginal return is certain = 0.60 Deposit amount = 2 SPNE:
Stage one: no participation
Stage two: no compliance Why? Benefit of violating = 8 * 0.40 = 3.20 Cost of violating = deposit = 2
Summary of group level predictions
Treatment Members Non-‐members Compliance Avg. PG
Baseline -‐-‐-‐ -‐-‐-‐ -‐-‐-‐ 0
Full-‐certain 5/5 (100%) -‐-‐-‐ 100% 8
Full-‐uncertain 5/5 (100%) -‐-‐-‐ 50% 4
Part-‐certain 3/5 (60%) 2/5 100% 4.8
Part-‐uncertain 3/5 (60%) 2/5 50% 2.4
Full-‐low deposit 0/5 (0%) 5/5 0% 0
Subgame perfect Nash equilibria for risk-neutral agents
3 sessions per treatment 20 subjects per session groups of 5, reshuffled for 20 periods Subjects participated in only one treatment Risk-elicitation exercise Stage one expectations regarding number of others joining
Protocol
Note: 1200 individual observations; 240 group-level observations per treatment
HERE YOU WILL BE REMINDED WHETHER AN AGREEMENT FORMED AND WHAT YOUR COMMITMENTS ARE (IF ANY)
HERE IS WHERE YOU ARE INFORMED WHETHER THE RETURN IS HIGH OR LOW
Percent that Joined Percentage of agreements that formed
5 of 5 required
Certainty 93.8%
Uncertainty
3 of 5 required
Certainty 55.6%
Uncertainty
79.3%
65.5%
74.1% 42.0%
78.8%
63.3%
Stage one results: participation and agreement formation
Percent compliance overall
Percentage Compliant when
Deposit is Sufficient
Percentage Compliant when
Deposit is Insufficient
5 of 5 required
Certainty 96.7% 96.7% -‐-‐-‐
Uncertainty
3 of 5 required
Certainty 94.1% 94.1% -‐-‐-‐
Uncertainty
85.8% 96.4% 77.5%
93.7% 97.8% 89.4%
Compliance when agreements form
Average ContribuIons
Average ContribuIons
when Agreements Formed
Average ContribuIons
when Agreements Failed
VCM 4.17 -‐-‐-‐ -‐-‐-‐
5 of 5 required
Certainty 7.20
Uncertainty
3 of 5 required
Certainty 4.33
Uncertainty
5.47
5.87
8.49
5.86
6.77
7.85
3.49
3.77
1.70
2.53
Crowding out Average contributions
Average contributions (non-‐members)
Average ContribuIons
Average ContribuIons by
Members
Average ContribuIons by Nonmembers
VCM 4.17 -‐-‐-‐ -‐-‐-‐
3 of 5 required
Certainty 5.86 8.27 0.883
Uncertainty 6.77 8.42 2.27
Leakage?
ParIcipaIon Agreement FormaIon
Compliance when
Agreements Form
Avg ContribuIons
Avg ContribuIons when Agreements
Failed
Low Deposit
93% 69.4% 78.9% 6.01 3.60
Compare to Nash predic]on of 0
Low deposit treatment
With certain benefits, we observe high levels of participation, and high levels of compliance with agreements. Uncertainty has competing effects in terms of participation (though insignificant with conditional tests) We observe high levels of compliance even when it is costly for members to comply (i.e., deposits are set too low). Trying to cooperate and failing triggers non-cooperative behavior
Conclusions