Financial incentives to poor countries promote net emissions reductions in multilateral climate agreements

Abstract

•Global emissions reduction can be achieved in the absence of any binding enforcement•The rich incentivize the poor through FIs, and the poor increase their ER accordingly•Overall,FIslead not only to higher ER, but also to higher socialwelfare•The results are robust to changes of different sets of parameters Cooperation to reduce greenhouse-gas emissions has profound impacts on sustainable development. We explore the impact on global cooperation of a financial incentive (FI) by which developed countries can incentivize developing countries to reduce carbon emissions. We perform both behavioral experiments and game theoretical analyses to clarify possible result bias stemming from bounded rationality of humans or from inherent selfish motivations among different countries. Our results show that FIs systematically contribute both to increase global contribution in emissions reduction and to effectively reduce emissions globally, significantly improving the probability of achieving the targets specified in international agreements. This happens in the absence of any binding enforcement, and strongly suggests that developed countries should divert some of their financial resources to incentivize developing countries to reduce their emissions. Reducing global greenhouse-gas emissions needs global cooperation and will have a positive and profound impact on sustainable development. Climate agreements, in line with the UNFCCC, encourage developed countries to provide funds to help developing countries adapt and mitigate. However, up to now, no financial incentive (FI) has been implemented, and it remains unclear to what extent FIs can increase net emissions reductions (ER). Here we investigate a restrictive form of FI, employing both behavioral experiments and game theoretical analysis. We show that FIs significantly increase both ER and social welfare in the absence of any binding enforcement. We also find that the more developed countries invest in FIs, the more developing countries mitigate. This induces developed countries to incentivize developing countries to adapt and mitigate via FIs, resulting in a net global increase in ER. Our results are robust to different monitoring periods, loss probabilities, and mitigation cost ratios. Reducing global greenhouse-gas emissions needs global cooperation and will have a positive and profound impact on sustainable development. Climate agreements, in line with the UNFCCC, encourage developed countries to provide funds to help developing countries adapt and mitigate. However, up to now, no financial incentive (FI) has been implemented, and it remains unclear to what extent FIs can increase net emissions reductions (ER). Here we investigate a restrictive form of FI, employing both behavioral experiments and game theoretical analysis. We show that FIs significantly increase both ER and social welfare in the absence of any binding enforcement. We also find that the more developed countries invest in FIs, the more developing countries mitigate. This induces developed countries to incentivize developing countries to adapt and mitigate via FIs, resulting in a net global increase in ER. Our results are robust to different monitoring periods, loss probabilities, and mitigation cost ratios. The Paris Agreement (PA), signed in 2015, resulted in an inclusive, binding treaty that succeeds the Kyoto Protocol and the Copenhagen Accord. It has been argued that the PA constitutes a significant breakthrough in international climate negotiations,1UNFCCCAdoption of the Paris agreement. Report No. FCCC/CP/2015/L.9/Rev.1.http://unfccc.int/resource/docs/2015/cop21/eng/l09r01.pdfDate: 2015Google Scholar, 2Falkner R. The Paris Agreement and the new logic of international climate politics.Int. Aff. 2016; 92: 1107-1125Crossref Scopus (263) Google Scholar, 3Hale T. “All hands on deck”: the Paris Agreement and nonstate climate action.Glob. Environ. Polit. 2016; 16: 12-22Crossref Scopus (142) Google Scholar, 4Klein D. Carazo M.P. Doelle M. Bulmer J. Higham A. The Paris Agreement on Climate Change: Analysis and Commentary. Oxford University Press, 2017Google Scholar reinstating the United Nations Framework Convention on Climate Change (UNFCCC) as a forum for dynamic multilateralism.5UNFCCCCancun agreements. 16th Conference of the Parties. Cancun: United Nations.http://unfccc.int/meetings/cancun_nov_2010/meeting/6266.phpDate: 2010Google Scholar,6Christoff P. The promissory note: COP 21 and the Paris climate agreement.Environ. Polit. 2016; 25: 765-787Crossref Scopus (77) Google Scholar At present, the Intergovernmental Panel on Climate Change (IPCC) has pointed out that the way to solve the threats posed by climate change is to reduce emissions by at least 50% of the 2000 level by 2050.1UNFCCCAdoption of the Paris agreement. Report No. FCCC/CP/2015/L.9/Rev.1.http://unfccc.int/resource/docs/2015/cop21/eng/l09r01.pdfDate: 2015Google Scholar, 2Falkner R. The Paris Agreement and the new logic of international climate politics.Int. Aff. 2016; 92: 1107-1125Crossref Scopus (263) Google Scholar, 3Hale T. “All hands on deck”: the Paris Agreement and nonstate climate action.Glob. Environ. Polit. 2016; 16: 12-22Crossref Scopus (142) Google Scholar, 4Klein D. Carazo M.P. Doelle M. Bulmer J. Higham A. The Paris Agreement on Climate Change: Analysis and Commentary. Oxford University Press, 2017Google Scholar,7IPCCClimate Change 2007: Mitigation of Climate Change, Summary for Policymakers. Contribution of Working Group III to the Fourth Assessment. Cambridge University Press, 2007Google Scholar To achieve this goal, widespread cooperation is required. Given that countries are heterogeneous in terms of wealth (according to the World Bank, countries are categorized as high income, middle income, and low income),8World Bankhttps://data.worldbank.org/indicator/NY.GDP.MKTP.CDGoogle Scholar emissions reduction (ER) costs (the World Bank uses the GDP per kilogram of CO2 emissions as a surrogate for ER cost and, thus, high-income countries have higher costs than do middle- and low-income countries)9World Bankhttps://data.worldbank.org/indicator/EN.ATM.CO2E.KD.GDGoogle Scholar and risks (according to the IPCC, middle- and low-income countries are more vulnerable to extreme events and disasters than high-income countries)10IPCCManaging the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation Special Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, 2012Google Scholar differ when countries face climate disaster. Given the disparities between high-income and middle- and low-income countries, cooperation on ER constitutes a highly non-trivial problem, and more so when we take into consideration that global cooperation must be achieved through international agreements whereby sanctioning mechanisms are very difficult to implement. Consequently, here we explore the impact of implementing a financial incentive (FI) mechanism inspired by what the UNFCCC designates as “Financial Mechanism” (FM), whose operating entities were included as an important aspect of the PA (see Note S1.2 for detailed discussions).1UNFCCCAdoption of the Paris agreement. Report No. FCCC/CP/2015/L.9/Rev.1.http://unfccc.int/resource/docs/2015/cop21/eng/l09r01.pdfDate: 2015Google Scholar,11Lyster R. Climate justice, adaptation and the Paris Agreement: a recipe for disasters?.Environ. Polit. 2017; 26: 438-458Crossref Scopus (17) Google Scholar,12Rogelj J. Den Elzen M. Höhne N. Fransen T. Fekete H. Winkler H. Meinshausen M. Paris Agreement climate proposals need a boost to keep warming well below 2°C.Nature. 2016; 534: 631-639Crossref PubMed Scopus (1428) Google Scholar According to the FM in the PA, developed country parties shall provide financial resources to assist developing country parties, and in our experimental design, subjects with high endowment can incentivize subjects with low endowment to contribute to ER through an FI, which mimics the FM in the PA. Unlike alternative mechanisms already proposed,13Ostrom E. Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press, 1990Crossref Google Scholar, 14Carraro C. Siniscalco D. Strategies for the international protection of the environment.J. Public Econ. 1993; 52: 309-328Crossref Scopus (569) Google Scholar, 15Nordhaus W.D. Managing the Global Commons: The Economics of Climate Change. MIT Press, 1994Google Scholar, 16Carraro C. Incentives and institutions: a bottom-up approach to climate policy.in: Aldy J. Stavins R.N. Architectures for Agreement: Addressing Global Climate Change in the Post-Kyoto World. Cambridge University Press, 2007: 161-172Google Scholar, 17Ostrom E. A polycentric approach for coping with climate change.Ann. Econ. Finance. 2014; 15: 97-134Google Scholar, 18Barrett S. Avoiding disastrous climate change is possible but not inevitable.Proc. Natl. Acad. Sci. U S A. 2011; 108: 11733-11734Crossref PubMed Scopus (12) Google Scholar, 19Stewart R.B. Oppenheimer M. Rudyk B. A new strategy for global climate protection.Climat. Chang. 2013; 120: 1-12Crossref Scopus (41) Google Scholar, 20Nordhaus W. Climate clubs: Overcoming free-riding in international climate policy.Am. Econ. Rev. 2015; 105: 1339-1370Crossref Scopus (355) Google Scholar, 21Nyborg K. Social norms as solutions.Science. 2016; 354: 42-43Crossref PubMed Scopus (269) Google Scholar, 22Hannam P.M. Vasconcelos V.V. Levin S.A. Pacheco J.M. Incomplete cooperation and co-benefits: Deepening climate cooperation with a proliferation of small agreements.Climat. Chang. 2017; 144: 65-79Crossref Scopus (14) Google Scholar such as cap and trade and a carbon tax with punishment for over-emission, the FIs discussed below do not require any binding enforcement mechanism of control or any penalties to discourage free riding. Thus, a natural question is whether FIs can help to promote ER. Following previous studies, the problem of cooperation to ensure ER may be framed as a multi-period threshold public goods dilemma game involving players with different amounts of wealth, mitigation costs, and (non-negligible) risks of future losses.23Milinski M. Semmann D. Krambeck H.J. Marotzke J. Stabilizing the Earth's climate is not a losing game: Supporting evidence from public goods experiments.Proc. Natl. Acad. Sci. U S A. 2006; 103: 3994-3998Crossref PubMed Scopus (225) Google Scholar, 24Milinski M. Sommerfeld R.D. Krambeck H.J. Reed F.A. Marotzke J. The collective-risk social dilemma and the prevention of simulated dangerous climate change.Proc. Natl. Acad. Sci. U S A. 2008; 105: 2291-2294Crossref PubMed Scopus (299) Google Scholar, 25Pacheco J.M. Santos F.C. Souza M.O. Skyrms B. Evolutionary dynamics of collective action in n-person stag hunt dilemmas.Proc. R. Soc. B. 2009; 276: 315-321Crossref PubMed Scopus (204) Google Scholar, 26Milinski M. Röhl T. Marotzke J. Cooperative interaction of rich and poor can be catalyzed by intermediate climate targets.Climatic Change. 2011; 109: 807-814Crossref Scopus (53) Google Scholar, 27Santos F.C. Pacheco J.M. Risk of collective failure provides an escape from the tragedy of the commons.Proc. Natl. Acad. Sci. U S A. 2011; 108: 10421-10425Crossref PubMed Scopus (153) Google Scholar, 28Tavoni A. Dannenberg A. Kallis G. Löschel A. Inequality, communication and the avoidance of disastrous climate change in a public goods game.Proc. Natl. Acad. Sci. U S A. 2011; 108: 11825-11829Crossref PubMed Scopus (222) Google Scholar, 29Barrett S. Dannenberg A. Climate negotiations under scientific uncertainty.Proc. Natl. Acad. Sci. U S A. 2012; 109: 17372-17376Crossref PubMed Scopus (153) Google Scholar, 30Chakra M.A. Traulsen A. Evolutionary dynamics of strategic behaviour in a collective-risk dilemma.PLoS Comput. Biol. 2012; 8: e1002652Crossref PubMed Scopus (36) Google Scholar, 31Vasconcelos V.V. Santos F.C. Pacheco J.M. A bottom-up institutional approach to cooperative governance of risky commons.Nat. Clim. Change. 2013; 3: 797-801Crossref Scopus (88) Google Scholar, 32Vasconcelos V.V. Santos F.C. Pacheco J.M. Levin S.A. Climate policies under wealth inequality.Proc. Natl. Acad. Sci. U S A. 2014; 111: 2212-2216Crossref PubMed Scopus (72) Google Scholar, 33Barrett S. Dannenberg A. Sensitivity of collective action to uncertainty about climate tipping points.Nat. Clim. Change. 2014; 4: 36-39Crossref Scopus (78) Google Scholar, 34Milinski M. Hilbe C. Semmann D. Sommerfeld R. Marotzke J. Humans choose representatives who enforce cooperation in social dilemmas through extortion.Nat. Commun. 2016; 7: 10915Crossref PubMed Scopus (31) Google Scholar, 35Perc M. Jordan J.J. Rand D.G. Wang Z. Boccaletti S. Szolnoki A. Statistical physics of human cooperation.Phys. Rep. 2017; 687: 1-51Crossref Scopus (682) Google Scholar, 36Andrews T.M. Delton A.W. Kline R. High-risk high-reward investments to mitigate climate change.Nat. Clim. Change. 2018; 8: 890Crossref Scopus (10) Google Scholar, 37Chakra M.A. Bumann S. Schenk H. Oschlies A. Traulsen A. Immediate action is the best strategy when facing uncertain climate change.Nat. Commun. 2018; 9: 2566Crossref PubMed Scopus (12) Google Scholar, 38Kline R. Seltzer N. Lukinova E. Bynum A. Differentiated responsibilities and prosocial behaviour in climate change mitigation.Nat. Hum. Behav. 2018; 2: 653-661Crossref PubMed Scopus (6) Google Scholar, 39Gross J. De Dreu C.K. Individual solutions to shared problems create a modern tragedy of the commons.Sci. Adv. 2019; 5: eaau7296Crossref PubMed Scopus (13) Google Scholar, 40Wang Z. Jusup M. Guo H. Shi L. Geček S. Anand M. Perc M. Bauch C.T. Kurths J. Boccaletti S. Schellnhuber H.J. Communicating sentiment and outlook reverses inaction against collective risks.Proc. Natl. Acad. Sci. U S A. 2020; 117: 17650-17655Crossref PubMed Scopus (15) Google Scholar In this game, players are challenged to reach a pre-defined group target evaluated at the end of a series of periodical contributions; if the target is not met, players will lose their wealth with pre-defined probabilities. In this work, we set up groups of six individuals,24Milinski M. Sommerfeld R.D. Krambeck H.J. Reed F.A. Marotzke J. The collective-risk social dilemma and the prevention of simulated dangerous climate change.Proc. Natl. Acad. Sci. U S A. 2008; 105: 2291-2294Crossref PubMed Scopus (299) Google Scholar,26Milinski M. Röhl T. Marotzke J. Cooperative interaction of rich and poor can be catalyzed by intermediate climate targets.Climatic Change. 2011; 109: 807-814Crossref Scopus (53) Google Scholar,28Tavoni A. Dannenberg A. Kallis G. Löschel A. Inequality, communication and the avoidance of disastrous climate change in a public goods game.Proc. Natl. Acad. Sci. U S A. 2011; 108: 11825-11829Crossref PubMed Scopus (222) Google Scholar,38Kline R. Seltzer N. Lukinova E. Bynum A. Differentiated responsibilities and prosocial behaviour in climate change mitigation.Nat. Hum. Behav. 2018; 2: 653-661Crossref PubMed Scopus (6) Google Scholar in which we simulate the global ER problem and investigate the impact on ER in the presence and in the absence of FIs. In each group, we randomly select a player to represent a high-income country (rich), while the remaining five players represent middle- and low-income countries (poor). Rich players are not only wealthier (they start with higher initial endowments) but also may have higher ER costs compared with poor players: indeed, in 2014, the GDP per kilogram of CO2 emissions—used as a surrogate for cost of ER—was 3.2 times higher for high-income countries compared with middle- and low-income countries.9World Bankhttps://data.worldbank.org/indicator/EN.ATM.CO2E.KD.GDGoogle Scholar It is important to point out that, in each group, only one participant assumes the role of a high-income country. This way, the interaction between “rich” countries is not explicitly included in each group, unlike what happens with players acting as low-income countries (see Note S6, where interactions among the rich in larger groups are incorporated). Two different game settings were implemented, here designated as Control and Treatment. In Control, there is no FI. Both rich and poor subjects can contribute to ER in each period (ER stage). At the end of the T periods, the total contributions are compared with the target announced (quantitative details in the experimental procedures). In Treatment, each of the T periods comprises an additional FI stage before the ER stage, in which the rich subject in the group can contribute to a fund whose resources are distributed by the poor subjects of the group after their contribution is made in the ER stage. Naturally, several possibilities exist to allocate the resources in the fund. Here, we decided to allocate the resources proportional to the contribution of the poor to ER (no contribution means no transfer, see Note S1 for more details). Thus, under the FI, the rich are allowed to contribute to the fund but not to benefit from it, while the poor benefit from the fund without contributing to it. In the following we shall designate the aforementioned fund also by FI, as it is associated with the FI stage. In our experimental design (as well as in the theoretical analysis), endowments are finite and limited and the transfers are both endogenous and budget balanced. To this end we further impose that resources allocated by the rich to ER cannot be used for FIs and vice versa. Importantly, poor subjects cannot contribute more to ER even when their wealth increases via FIs (see the experimental procedures for details). All these restrictions, as discussed below, impose stringent conditions on the FIs implemented here. In addition to inequality, our experimental design attempts to mimic the global stocktake of the PA, whereby each country's ER will be monitored every 5 years, starting in 2023,1UNFCCCAdoption of the Paris agreement. Report No. FCCC/CP/2015/L.9/Rev.1.http://unfccc.int/resource/docs/2015/cop21/eng/l09r01.pdfDate: 2015Google Scholar, 2Falkner R. The Paris Agreement and the new logic of international climate politics.Int. Aff. 2016; 92: 1107-1125Crossref Scopus (263) Google Scholar, 3Hale T. “All hands on deck”: the Paris Agreement and nonstate climate action.Glob. Environ. Polit. 2016; 16: 12-22Crossref Scopus (142) Google Scholar, 4Klein D. Carazo M.P. Doelle M. Bulmer J. Higham A. The Paris Agreement on Climate Change: Analysis and Commentary. Oxford University Press, 2017Google Scholar such that future decisions may be contingent on perceived status. As global emissions should decrease to 50% of the present level by 2050,1UNFCCCAdoption of the Paris agreement. Report No. FCCC/CP/2015/L.9/Rev.1.http://unfccc.int/resource/docs/2015/cop21/eng/l09r01.pdfDate: 2015Google Scholar, 2Falkner R. The Paris Agreement and the new logic of international climate politics.Int. Aff. 2016; 92: 1107-1125Crossref Scopus (263) Google Scholar, 3Hale T. “All hands on deck”: the Paris Agreement and nonstate climate action.Glob. Environ. Polit. 2016; 16: 12-22Crossref Scopus (142) Google Scholar, 4Klein D. Carazo M.P. Doelle M. Bulmer J. Higham A. The Paris Agreement on Climate Change: Analysis and Commentary. Oxford University Press, 2017Google Scholar,7IPCCClimate Change 2007: Mitigation of Climate Change, Summary for Policymakers. Contribution of Working Group III to the Fourth Assessment. Cambridge University Press, 2007Google Scholar this means approximately six monitoring periods. We performed behavioral experiments with the multi-period threshold public goods dilemma game just described involving 840 volunteer undergraduate and graduate students recruited from Beijing Normal University who had not taken classes on game theory and economics. A rich subject has an initial endowment five times larger than a poor subject. We considered two kinds of ER cost factors: s for a rich subject, where s = 2 and s = 3 (we always have s = 1 for every poor subject), and two values for the total number of monitoring periods, T = 6 and T = 10. At the end of each multi-period experiment, if the ER target was not reached, then rich and poor subjects lost all their savings with probabilities rR and rP, respectively (so-called risks of failure;23Milinski M. Semmann D. Krambeck H.J. Marotzke J. Stabilizing the Earth's climate is not a losing game: Supporting evidence from public goods experiments.Proc. Natl. Acad. Sci. U S A. 2006; 103: 3994-3998Crossref PubMed Scopus (225) Google Scholar,24Milinski M. Sommerfeld R.D. Krambeck H.J. Reed F.A. Marotzke J. The collective-risk social dilemma and the prevention of simulated dangerous climate change.Proc. Natl. Acad. Sci. U S A. 2008; 105: 2291-2294Crossref PubMed Scopus (299) Google Scholar,26Milinski M. Röhl T. Marotzke J. Cooperative interaction of rich and poor can be catalyzed by intermediate climate targets.Climatic Change. 2011; 109: 807-814Crossref Scopus (53) Google Scholar, 27Santos F.C. Pacheco J.M. Risk of collective failure provides an escape from the tragedy of the commons.Proc. Natl. Acad. Sci. U S A. 2011; 108: 10421-10425Crossref PubMed Scopus (153) Google Scholar, 28Tavoni A. Dannenberg A. Kallis G. Löschel A. Inequality, communication and the avoidance of disastrous climate change in a public goods game.Proc. Natl. Acad. Sci. U S A. 2011; 108: 11825-11829Crossref PubMed Scopus (222) Google Scholar, 29Barrett S. Dannenberg A. Climate negotiations under scientific uncertainty.Proc. Natl. Acad. Sci. U S A. 2012; 109: 17372-17376Crossref PubMed Scopus (153) Google Scholar,33Barrett S. Dannenberg A. Sensitivity of collective action to uncertainty about climate tipping points.Nat. Clim. Change. 2014; 4: 36-39Crossref Scopus (78) Google Scholar,34Milinski M. Hilbe C. Semmann D. Sommerfeld R. Marotzke J. Humans choose representatives who enforce cooperation in social dilemmas through extortion.Nat. Commun. 2016; 7: 10915Crossref PubMed Scopus (31) Google Scholar,36Andrews T.M. Delton A.W. Kline R. High-risk high-reward investments to mitigate climate change.Nat. Clim. Change. 2018; 8: 890Crossref Scopus (10) Google Scholar, 37Chakra M.A. Bumann S. Schenk H. Oschlies A. Traulsen A. Immediate action is the best strategy when facing uncertain climate change.Nat. Commun. 2018; 9: 2566Crossref PubMed Scopus (12) Google Scholar, 38Kline R. Seltzer N. Lukinova E. Bynum A. Differentiated responsibilities and prosocial behaviour in climate change mitigation.Nat. Hum. Behav. 2018; 2: 653-661Crossref PubMed Scopus (6) Google Scholar,40Wang Z. Jusup M. Guo H. Shi L. Geček S. Anand M. Perc M. Bauch C.T. Kurths J. Boccaletti S. Schellnhuber H.J. Communicating sentiment and outlook reverses inaction against collective risks.Proc. Natl. Acad. Sci. U S A. 2020; 117: 17650-17655Crossref PubMed Scopus (15) Google Scholar,41Nordhaus W. The Climate Casino: Risk, Uncertainty, and Economics for a Warming World. Yale University Press, 2013Google Scholar we allow rR and rP to take the values 0.5 and 0.7, see below). Twelve experimental sessions were performed: six Control sessions and six Treatment sessions. Each session, associated with specific values of T, s, rR, and rP, was repeated employing 10 different groups of six individuals. Specifically, (T,s,rR,rP)=(6,2,0.5,0.5), (6,3,0.5,0.5), (6,3,0.5,0.7), (6,3,0.7,0.7), (10,2,0.5,0.5), and (10,3,0.5,0.5) in Controls 1–6, respectively, and the parameter settings in Treatments 1–6 were the same as those of Controls 1–6. In line with previous experiments,23Milinski M. Semmann D. Krambeck H.J. Marotzke J. Stabilizing the Earth's climate is not a losing game: Supporting evidence from public goods experiments.Proc. Natl. Acad. Sci. U S A. 2006; 103: 3994-3998Crossref PubMed Scopus (225) Google Scholar,24Milinski M. Sommerfeld R.D. Krambeck H.J. Reed F.A. Marotzke J. The collective-risk social dilemma and the prevention of simulated dangerous climate change.Proc. Natl. Acad. Sci. U S A. 2008; 105: 2291-2294Crossref PubMed Scopus (299) Google Scholar,28Tavoni A. Dannenberg A. Kallis G. Löschel A. Inequality, communication and the avoidance of disastrous climate change in a public goods game.Proc. Natl. Acad. Sci. U S A. 2011; 108: 11825-11829Crossref PubMed Scopus (222) Google Scholar,29Barrett S. Dannenberg A. Climate negotiations under scientific uncertainty.Proc. Natl. Acad. Sci. U S A. 2012; 109: 17372-17376Crossref PubMed Scopus (153) Google Scholar,33Barrett S. Dannenberg A. Sensitivity of collective action to uncertainty about climate tipping points.Nat. Clim. Change. 2014; 4: 36-39Crossref Scopus (78) Google Scholar,37Chakra M.A. Bumann S. Schenk H. Oschlies A. Traulsen A. Immediate action is the best strategy when facing uncertain climate change.Nat. Commun. 2018; 9: 2566Crossref PubMed Scopus (12) Google Scholar,38Kline R. Seltzer N. Lukinova E. Bynum A. Differentiated responsibilities and prosocial behaviour in climate change mitigation.Nat. Hum. Behav. 2018; 2: 653-661Crossref PubMed Scopus (6) Google Scholar,42Rand D.G. Dreber A. Ellingsen T. Fudenberg D. Nowak M.A. Positive interactions promote public cooperation.Science. 2009; 325: 1272-1275Crossref PubMed Scopus (435) Google Scholar subjects had limited options in both FI and ER stages; in the experimental procedures we explain in detail all options available, while in Note S1 we provide details of the experimental procedures. We start by investigating possible scenarios that may result from the experiments by carrying out a game theoretical analysis of (some of) the subgame perfect Nash equilibria (SPNE; which is a refinement of Nash equilibria for a multi-period game) stemming from modeling the experimental setting, as well as an identification of their stability through an evolutionary game theoretical analysis (where subjects are assumed bounded rational and learn to play the game)43Hofbauer J. Sigmund K. Evolutionary Game and Population Dynamics. Cambridge University Press, 1998Crossref Google Scholar (full details in the experimental procedures and Notes S2–S4). Given the multi-period structure of this game, individual strategies can be very complicated, and the existence of asymmetric SPNE is possible. In the following, we consider a specific subset of SPNE that we designate as quasi-symmetric SPNE (QSPNE), at which equilibrium all poor subjects use the same strategy. Furthermore, we focus on the sustainment of cooperation through the GRIM strategy, where a subject using GRIM will contribute to ER or FI in period t only if the total ER or FI in the previous periods in not less than a pre-defined value. We emphasize that the purpose of the theoretical analysis is not to perform a full analysis of the game, but to provide intuition on whether and how FI works. We find that, in the Control case, a selfish, non-cooperative SPNE dominates for rR=rP=0.5, at which SPNE no one contributes to ER (Figure S1). In the Treatment case, if the contribution of the rich to FIs positively correlates with the contribution of the poor to ER, then there is a narrow basin of attraction toward a set of (stable) cooperative QSPNE (which we designate as incentive QSPNE) in which only the poor contribute to ER, while the rich contribute solely to FIs (Figure S2). At those incentive QSPNE, both rich and poor subjects have higher wealth compared with the selfish SPNE even if, similar to the experimental design, poor subjects cannot contribute more to ER when their wealth increases via FIs (see the experimental procedures). Moreover, higher contributions by the rich to FIs are predicted to lead to higher investments in ER by the poor, but up to a limit, beyond which the incentive QSPNE become unstable. We now focus on the behavioral experiments, whose results will also enable us to confirm or dismiss the theoretical predictions made above. Figures 1A and 1B show the success rates (fraction of groups that reach the target) and relative ER (group average ER relative to the target ER value), respectively, obtained in the experiments (for further details of relative ER at group level see Figure S3). Compared with Control (data in gray, ER stage only, per period), FIs lead to systematic improvements in success rates in Treatment (data in blue, FI stage + ER stage, per period). Furthermore, the total ER in Control is significantly below the target, whereas in Treatment it is not statistically different from the target (see also Table S1). These results indicate that FIs significantly promote ER in experimental scenarios involving diverse numbers of periods, cost factors, and risk combinations. To uncover the effect of FIs in ER, we now explore the relative contributions by both rich and poor subjects, defined as their total contributions divided by their initial endowments. Figures 2A and 2B show relative contributions in Control and Treatment, respectively, where the contributions are by poor subjects to ER and by rich subjects to both ER and FIs (for further details of relative ER at group level see Figure S3). We find that poor subjects contribute significantly more to ER in Treatment than in Control, despite being unable to use the tokens received from the rich via FIs to ER. Importantly, our experiments confirm that the investment made by poor subjects is positively correlated with the amount contributed to FIs by the rich at the group level (Pearson correlation coefficient = 0.7986, p < 0.001), which, in view of our theoretical analysis, suggests that the incentive Nash equilibria (NE) may be stable (see Figure S4 for further analysis of the correlation between ER by the poor and FI by the rich at the group level). This indicates that the availability of FI plays a key role in enhancing both the total ER achieved and the success rates in reaching the ER target. From the outset, it was not clear whether the rich would contribu