This report relies on a survey designed to collect the perceptions of climate experts on the Paris Agreement and climate change mitigation action. The study draws on the views of policymakers and other climate experts with direct experience in the design and delivery of climate policy. Their responses reflect institutional knowledge, technical expertise and a deep understanding of domestic and international climate governance. This chapter describes the survey set up and the main questions and themes that it addresses. It also provides an overview of the response collection and expert sampling process, as well as the methodology used to treat responses. Finally, it highlights the novelty of the survey with respect to the existing literature.
The Paris Agreement at Ten Years
Abstract
The survey was designed around four parts, illustrated in Figure 2.1 and referred to as parts A, B, C and D.1 They cover different aspects of the study and employ slightly different methods of elicitation, which are described in this section. The full survey is displayed in Annex A.
Source: Figure generated by the authors.
2.1.1. The added value of the Paris Agreement
The first part of the survey (part A) aims at understanding the contribution of the Paris Agreement. It mostly relies on a series of counterfactual thought experiments that aim at eliciting the contribution of the Paris Agreement. Within this part, the survey focuses on understanding how respondents perceive the value of climate change as a policy priority (part A.1) and the degree of stringency of climate change mitigation policies (part A.2). Respondents are asked to state the extent to which they agree with these statements at 5 points in time: 2000, 2015, (i.e. the year in which the Paris Agreement was signed), 2024 (i.e. the year in which the survey was launched), 2030 and 2040. For 2024, 2030 and 2040, questions are posed in two scenarios: a reference scenario in which the Paris Agreement is present and constitutes the main framework of global climate action; and a counterfactual scenario in which the Paris Agreement is absent. This counterfactual scenario leaves space for each respondent to make hypotheses on what might have happened after 2015 in absence of the Paris Agreement. As such, this scenario embraces several opposing views without judging their plausibility.
Isolating the net impact of the Paris Agreement on domestic climate action requires comparing two scenarios, each representing a distinct pathway. The first scenario accounts for the factual pathway, i.e. for all the relevant policy changes that have actually taken place between the introduction of the Paris Agreement in 2015 and 2025 and the changes that will take place between 2025 and 2040 in the presence of the Paris Agreement. The second scenario, here referred to as the counterfactual scenario, accounts for all the policy changes that would occur between 2015 and 2040 in the absence of the Paris Agreement. The pathway of the counterfactual scenario is not possible to observe: it reflects events and policy developments that would have occurred, either under a business-as-usual continuation of the status quo before 2015 or under any setting other than the Paris Agreement.
Methods based on the “wisdom of the crowd”, i.e. methods relying on averaging the judgments, opinions, or predictions of a large group of people, are commonly used to assess situations in which key data are unavailable and future events are hard to predict. Such methods may be subject to considerable limitations. The most important one is that evaluating how policy stringency would have evolved in a counterfactual situation requires knowledge that may not be available to non-experts. Methods based on expert judgment are warranted to evaluate counterfactual pathways and to make predictions about the future in general absence of data. Expert elicitation methods, e.g. Delphi surveys, may help overcome the limitations of methods based on the “wisdom of the crowd”.
An additional component (part A.3) focuses on the shift from the Kyoto Protocol to the Paris Agreement. This part of the survey explores how the shift may have influenced climate policy and broader environmental outcomes. The questions investigate whether the Paris Agreement led to stronger emissions targets, more ambitious mitigation policies, and ultimately lower emissions. It also considers whether the Agreement helped bring environmental issues into the public spotlight (e.g. by supporting action on other challenges, like plastic pollution) or shifted attention away from them. Further, the section examines the extent to which climate action has been mainstreamed across national and local governments and within the political system. To assess these effects, the survey introduces a specific counterfactual scenario in which no new international climate agreement was adopted between 2015 and 2024, allowing the Kyoto Protocol to remain the main international reference throughout that period. In this part of the survey, all questions refer to 2024 and do not extend into the future, as the assumption of the Kyoto Protocol continuing far into the future is considered less plausible.
A final component (part A.4) focuses on the importance that respondents attribute to specific characteristics and broader impacts of the Paris Agreement. It explores how respondents perceive the Agreement’s role in accelerating global climate action and shaping international cooperation. The questions measure beliefs regarding the ratcheting mechanism and peer pressure, which constitute the basic levers to scale up ambition for climate action. Part A.4 assesses how the Agreement is viewed as a diplomatic platform, i.e. whether it enhances countries’ international influence and stimulates private sector engagement. The questions also explore whether respondents agree that the Paris Agreement sets a precedent for tackling other global environmental challenges, including plastic pollution, and creates a blueprint to solve other international challenges beyond the environment. Finally, part A.4 considers whether the Agreement has helped make climate cooperation more solution-oriented and whether its implementation has been affected by recent geopolitical tensions.
Throughout part A, the survey allows for six levels of agreement in the Likert-type scale, from ‘strongly disagree’ to ‘strongly agree’. Neutral options, such as 'Neither agree nor disagree' and 'Don’t know', were excluded to avoid encouraging passive responses or providing an escape option in a context that requires reflective expert judgment, even under uncertainty.2 Results from part A are presented in Chapter 3 of this report.
2.1.2. Current efforts and barriers
The second part of the survey (part B) focuses on the current situation of climate change mitigation. Part B.1 takes stock of the beliefs regarding the current stringency of mitigation policies, their compatibility with economic objectives, and their alignment with the ambition inscribed in the Nationally Determined Contributions (NDCs). Part B.1 also explores views on whether national mitigation efforts are on track to meet stated targets on time. Responses in part B.1 follow the Likert scale, as in part A. Parts B.2 to B.4 examine policy, economic, and institutional challenges to achieving the goals of the Paris Agreement. Part B.2 focuses on policy obstacles such as low public awareness, limited acceptability of climate policies, weak incentives for business uptake of clean technologies, affordability constraints for households, and infrastructure gaps. Part B.3 turns to economic barriers, including inflation-driven reluctance to adopt green products, scarcity of public funds, private investment shortfalls, and broader macroeconomic risks. Part B.4 addresses institutional issues such as weak vertical and horizontal coordination, insufficient continuity of policy across electoral cycles and inadequate monitoring and compliance systems. Responses in parts B.2 to B.4 do not follow the Likert scale, but respondents are instead asked to allocate a total of 100 points across the proposed options. These sections include an “Other” option, allowing respondents to assign points and provide qualitative input for additional challenges they consider important. This addition was necessary, given that the predefined options may not fully capture all relevant barriers. Results from part B are presented in Chapter 4 of this report.
2.1.3. Enabling factors
The third part of the survey (part C) takes stock of beliefs regarding the enablers for the effective implementation of the Paris Agreement. As in part B.2-B.4, in parts C.1 and C.2 respondents are asked to distribute 100 points across seven channels, reflecting their expectations about where substantial change is most likely or needs to occur. Part C.1 focuses on how respondents believe the gap between current emissions and the current national targets inscribed in countries’ NDCs will be closed. It asks participants to allocate 100 points across seven channels, evaluating this way the expected contribution of each channel. The seven channels include market-based instruments (e.g. carbon taxes, trading schemes), regulatory measures (e.g. bans on carbon-intensive goods), policy support for green innovation (such as R&D subsidies), and autonomous technological progress without policy support. The options also include behavioural interventions (e.g. public awareness campaigns) and voluntary adjustments by individuals and businesses, such as corporate climate initiatives, individual lifestyle change. The scoring exercise captures perceptions on the relative weight of each channel in delivering the required mitigation outcomes. Part C.2 uses the exact same channels, but measures the weight respondents place in them when it comes to supporting more ambitious NDC targets compatible with the 1.5°C goal.
Part C.3 of the survey takes stock of how policymakers and other climate experts prioritise policy interventions across the full causal chain of emissions. Participants were invited to distribute 100 points across a diverse set of policy focus areas. These include decarbonising energy use, improving energy efficiency, shifting consumption patterns, reducing total demand via the promotion of circular economy principles, deploying carbon removal technologies, and addressing emissions not related to energy. This part operationalises the Kaya identity (Raupach et al., 2007[1]; Kaya and Yokobori, 1997[2]) by breaking down the drivers of emissions (energy carbon intensity, energy efficiency, and consumption volume), and extends beyond this to include post-emission carbon removal and emissions not related to energy. Results from part C are presented in Chapter 4 and Chapter 5 of this report.
2.1.4. Transformative change
The final part of the survey (part D) investigates expert perceptions of transformative change by evaluating the theoretical potential and actual domestic advancement of a broad set of climate policy directions. The listed policy directions span all major emission sectors and instruments including phasing out fossil fuels, carbon pricing, land-use regulations, carbon capture, utilisation and storage (CCUS) deployment, behavioural interventions, and agricultural reforms.
For the listed measures, respondents are asked to score two dimensions: (i) the extent to which a policy direction holds transformative potential to drastically reduce emissions, and (ii) the degree of progress made in implementing that policy since 2015. Unlike part C, which uses weighted distributions, this section employs fixed Likert-type scales. These scales range from 1 to 5 (e.g. from “No transformative potential at all”, to “Very strong transformative potential, essential component in the policy package”) as shown in Table 2.1.
Throughout many components of the survey, there is a fundamental difference in the framing of the questions posed to governments and to non-government experts. Whereas for governments, the survey refers to the national context (e.g. “In your country, climate change mitigation…”), the survey for non-government experts adopts an international perspective (“Internationally, climate change mitigation…”). This difference in framing, combined with the differing nature of the respondents, forms the basis for treating the two samples as strictly complementary, rather than directly comparable, in the analysis.
Source: Table generated by the authors.
The listed policy directions span all major emission sectors and instruments: from phasing out fossil fuels and carbon pricing, to land-use regulations, carbon capture, utilisation and storage (CCUS) deployment, behavioural interventions, and agricultural reforms. Results from part D are presented in Chapter 5 of this report.
The survey was circulated in two branches, resulting in two independent samples of responses. The first branch of the study collects and analyses responses from government experts and focuses on national climate action and progress. While most of the answers by governments were institutional responses,3 for some countries, individual experts filled in the survey, therefore not representing the official government position. The second branch of the study collects and analyses individual responses from non-government experts from institutions including academia and international organisations, as well as from the private sector. This second branch relies on a slightly differentiated version of the survey, with focus on international climate action. The two-branch approach allows to take stock of diverse perspectives and to compare the perception of the impact of the Paris Agreement between policymakers and other climate change experts.
Responses from governments have been collected primarily through OECD delegations.4 To collect responses from non-government institutions, initial invitations targeted a large group of experts from leading academic and research institutions specialising in climate change, as well as ones where climate change is a prominent part of the research agenda.5 Invitations were also sent to experts from Inter-Governmental Organisations (IGOs), Non-Governmental Organisations (NGOs), climate think-tanks and advocacy institutions, and the private sector (see Annex 2.A for a full list of institutions). Invitations were also sent via scientific networks of environmental economists, sustainability experts and climate change scientists (e.g. the Integrated Assessment Modelling Consortium, IAMC). The survey dissemination also incorporated elements of network chain sampling,6 often referred to as snowball or referral sampling. In particular, invitation recipients were encouraged to share the survey with other professionals possessing relevant expertise. To help address regional imbalances in the expert sample, follow-up nudges were sent more frequently to participants from non-OECD countries, which tend to be underrepresented in such surveys.7
The final set of replies from government experts includes 70 responses from 64 countries. Among these, most of OECD countries (33 out of 38 countries) are represented, alongside 31 non-OECD countries. The sample of experts from non-governmental institutions includes a total of 178 responses from 46 different countries, with 26 OECD countries represented. This sample is strongly skewed toward academia (79% of responses) and international organisations (12% of responses), with the remaining responses originating from the private sector and non-governmental organisations (NGOs). This composition reflects a strong academic and institutional engagement on the subject.
The professional background of non-government respondents spans economists, policy analysts, modelers, natural and social scientists, communication specialists, finance and legal experts, diplomats and engineers. To obtain an estimate of the degree of expertise of each respondent, three filtering questions were added. First, respondents were asked to state whether their primary area of expertise lies within the field of climate change (e.g. Climate change economics, climate policy, climate diplomacy), in fields closely related to Climate Change (e.g. Energy Policy, Environmental Economics, Sustainable Development) or in peripheral yet relevant areas (e.g. Urban Planning, Public Health, International Law, Agriculture, Risk Management). Second, respondents were asked to report whether their primary area of work is in the area of climate change mitigation, adaptation, resilience, or whether it cuts across multiple of those areas. Both questions contained the option “Other”, in which a respondent that could not identify with any of the prescribed options could provide their own description. Finally, respondents were asked to state the number of years of expertise they have in Climate Change, closely related fields, and relevant areas.
The responses collected from government officials and non-government climate change experts feed into the statistical and econometric analyses presented in Chapter 3. These analyses are conducted separately for the two groups. For government officials, all responses are equally weighted. For non-government experts, two weighting approaches are applied: (i) an equal-weight setting, where each participant is considered to contribute equally regardless of professional expertise and background; (ii) in the expertise-adjusted setting, which assigns greater weight to responses from experts whose contribution is considered more critical based on a range of criteria, including seniority and domain relevance. In the latter case, higher weights are attributed to responses from experts with more experience and demonstrated professional relevance to core areas of climate change, and specifically to mitigation. This approach allows to explore the sensitivity of the key findings to the weight given to observations collected from the various categories of experts.
Such sensitivity checks are important because it was not feasible to objectively measure expertise or eligibility using anchoring questions, i.e. items with verifiably correct answers. While such questions can sometimes be used to gauge a respondent’s ability to make accurate forward-looking projections, this approach was not applicable in the context of this study. Most meaningful climate-related outcomes, such as the long-term impact of the Paris Agreement, unfold over extended timescales and cannot be empirically validated within the operational window of the study. For example, questions like “What will global emissions be in 2025?”, “How many countries will strengthen their NDCs by end-2025 or early 2026?”, or “By 2030, how many countries will have reduced CO₂ emissions by 50% compared to 1990 levels?” could have been used as pre-screening tools in a multi-year study, but not in a study that spans months. Alternatively, including near-term prediction items with verifiable outcomes (e.g. “Will Bitcoin’s price exceed €110,000 two weeks from now?”) could help gauge the general forecasting ability of respondents, but would introduce a relevance bias. The inclusion of verifiable pre-screening items would have introduced additional complexity and potential barriers to participation. Specifically, adding them would have extended the survey, originally designed to require 20-25 minutes, and could have discouraged participation from time-constrained professionals. In such a context, pre-screening questions might also increase the perceived difficulty of the survey and elevate the dropout risk, particularly for respondents whose contributions are highly influential but not strictly technical. Moreover, cognitive assessments and knowledge checks may cause discomfort in some cultural settings, potentially introducing regional biases and undermining one of the core objectives of the survey: to achieve broad geographic representation. Due to these complications, the survey does not use anchoring questions, but a subjective response weighting system, whose role on the findings is carefully examined.
The study also accounts for response patterns that may indicate limited engagement or uncertainty. Flat or uniform responses, such as straight-lining across Likert items may reflect respondent fatigue, low confidence, or minimal cognitive effort during the time of response. To address this, the expertise-adjusted weights used in the analysis incorporate response quality indicators. This helps ensure that the analysis gives greater weight to responses that are both substantively grounded and thoughtfully completed. The weights used in the analysis can be found in Annex 2.A.
The survey powering the study does not directly build upon or extend any pre-existing survey or experimental setting on the Paris Agreement. However, it is highly related to a series of contributions utilising individual and expert elicitation surveys, as well as experiments to cast light on different aspects of the Paris Agreement.
2.4.1. Expert elicitation surveys
While there are several expert surveys on climate policies, very few climate-relevant expert elicitation studies focus on the Paris Agreement, and none focuses on its impacts. Despite a growing body of surveys examining the Paris Agreement, none provides a direct comparative assessment of the outcomes that the experts expect with and without the Paris Agreement. The present study is linked to pre-existing studies that focus on credibility and ambition, as well as the market sentiment of the Paris Agreement. These studies do not provide systematic evidence on how the Paris Agreement may have altered the trajectory of climate policy priorities, stringency, or governance relative to a counterfactual world where it had never been adopted. The present study addresses this gap.
A study by Victor, Lumkowsky and Dannenberg (2022[3]) uses expert elicitation to evaluate the ambition and credibility of the Paris Agreement pledges. It draws on the judgments of a pool of 829 climate experts, which constitute UNFCCC negotiators and IPCC scientists from more than 150 countries. The survey asked the experts to assess how ambitious the NDCs of their countries are, relative to their economic capacity. It also asked the experts to evaluate the likelihood that their country, but also other countries, deliver the pledges inscribed in their NDCs. The study by Victor, Lumkowsky and Dannenberg (2022[3]) is a largest systematic effort to capture insights from expert directly engaged in negotiations and scientific assessments. Their results challenge the long-standing assumption of a trade-off between ambition and credibility: experts rated the boldest pledges as also the most credible, especially in Europe. Moreover, institutional quality (not economic capacity) emerged as the main predictor of how credible pledges are. Most importantly, supplementary analyses show that expert assessments do not strongly correlate with indicators like Climate Action Tracker ratings or the number of climate-relevant laws. In that sense, the study by Victor, Lumkowsky and Dannenberg (2022[3]) underscored the added value of expert elicitation in understanding how the Paris Agreement operates in practice.
Another expert elicitation survey (Wynes et al., 2024[4]) collected the expectations of IPCC report authors regarding long-term outcomes under current policies. A majority of the 211 experts that participated in the study believed the world is not likely to achieve the goals of the Paris Agreement without more action: 86% of experts expect global warming to exceed 2 °C by 2100, with a median estimate at around 2.7°C. However, most experts did express a belief that net-zero emissions could eventually be reached in the second half of the 21st century, showing some optimism that current policies are bending the curve. The findings in the study by Wynes et al. (2024[4]) are in line with those from parts C.1 and C.2 of the present work, which are presented in the next section.
The study by Kornek et al. (2020[5]) surveyed 917 experts affiliated with the IPCC and UNFCCC to assess perceptions of key obstacles and enablers to achieving the 2 °C goal. On average, respondents identified opposition from special interest groups as the most significant barrier, an assessment that aligns with the findings of the present study. The survey also highlights a strong expert belief in the central role of technological R&D as the most important pathway forward.
A study by Sandalow, Benes and Augustin (2016[6]) explores the views of the private sector on how the Paris Agreement could influence investment in climate finance. Conducted just before and after COP21, their survey sampled 278 respondents, the majority of which were investors, energy and infrastructure experts from the US and the UK. The study found that national policies, such as support for renewable energy and phasing out fossil fuel subsidies, are seen by the experts as the most powerful levers for scaling up climate finance. Elements related to the Paris Agreement ranked second. These included: (i) The existence of a long-term global agreement with participation from nearly all countries; (ii) Regular reporting and review mechanisms (transparency framework) that make national performance visible; (iii) Nationally determined contributions (NDCs) that provide forward-looking policy guidance; (iv) A long-term temperature goal (below 2°C) as a credible anchor for expectations. By contrast, international climate finance pledges and carbon pricing coordination ranked lower. While the scope and the objective of the survey by Sandalow, Benes and Augustin (2016[6]) were different, the findings are in alignment with those from several components of the present survey.
2.4.2. Public opinion surveys
Public opinion surveys play a crucial role in gauging awareness of climate issues and support for climate action. Several of these surveys track public opinion on a range of climate-related topics. A substantial part of this research focuses on public support for the Paris Agreement. Public sentiment worldwide seems to have favoured international climate cooperation even before the Agreement was adopted. For instance, a 2015 survey conducted by Pew Research Center (Stokes, Wike and Carle, 2015[7]) interviewed more than 45000 individuals in 40 countries. It found a global median of 78% in support of signing an agreement to limit greenhouse gas emissions. Notably, this support often exceeded the level of personal concern about climate change. For example, 71% of Chinese respondents backed an emissions treaty despite only 18% being “very concerned” about climate change. In Europe, North America, and other major economies, large majorities likewise supported joining a climate pact, indicating broad public mandate for the Paris Agreement’s goals. On the way to COP21, another large public opinion poll was conducted by the Danish Board of Technology Foundation and the French National Commission for Public Debate (Bedsted, Mathieu and Leyrit, 2015[8]) sampling views from 10,000 citizens in 79 countries. This survey yielding one of the most comprehensive global opinion exercises on climate. Results showed strong appetite for an ambitious Paris outcome. When asked how urgently the world should act, 63% said that negotiators in Paris should do “whatever it takes” to cap warming at 2 °C. However, 68% of participants believed the Paris Agreement should include a legally binding goal of reaching net-zero emissions by 2100, applicable to all countries. Large majorities also felt nations should act on climate change even if others do not (80%) and viewed climate action as more of an opportunity than a threat. In 2022, a joint survey by the OECD and Harvard University (Dechezleprêtre et al., 2022[9]; 2025[10]) reported comparable support rates in a series of middle-income and high-income countries. The largest global climate opinion poll has been conducted by UNDP, the University of Oxford and Browning Environmental Communications (2024[11]). It involved 1.2 million respondents via mobile phones. The study found that four in five people worldwide demand more climate action from their country, 86% agree that nations should set aside differences to collaborate on climate change, and 89% call for stronger climate commitments (NDCs). Therefore, the general public is broadly calling for the Paris Agreement’s targets to be ramped up and implemented more vigorously. The set of public opinion surveys at the national level is much broader and its thorough review goes beyond the scope of this study.
Public opinion surveys are not well suited to address most of the questions explored in this report. The survey underpinning the study investigates complex issues, whose analysis requires expert knowledge and professional judgment. For instance, assessing the relative importance of levers (e.g. market-based instruments, regulations and innovation support) to close the implementation gap demands familiarity with how policies work, and the factors that could make a policy more effective in a national context than in others. Also, the questions of the survey explore how mitigation would have evolved in the absence of the Paris Agreement and whether the Agreement’s mechanisms have effectively raised policy ambition. Providing reliable answers to such questions involves counterfactual reasoning, an understanding of the international negotiation processes and an ability to recall the evolution of policies and policy priorities since 2000 (institutional memory). Other questions, such as assessing whether mitigation policies are compatible with economic objectives, require basic understanding of the economic trade-offs that may come into play in the pathway to net zero. Due to these requirements, public opinion surveys are likely to fail to reliably evaluate the effectiveness, credibility, or the transformative potential of the Paris Agreement. Expert elicitation, by contrast, leverages informed judgment from practitioners and specialists, providing a more grounded basis for understanding how international climate governance operates and evolves in practice. For these reasons, while the public opinion literature on climate change is extensive most studies do not extend to institutional or policy aspects. Therefore, their scope has minimum overlap with the questions examined in this expert survey. To keep the literature review compact, the discussion that follows refers selectively to public opinion studies that address aspects of the Paris Agreement that are also explored in the present expert elicitation survey. These include ambition, credibility, or implementation.
The Paris Agreement is likely to have had an impact on public attitudes and public support for domestic climate action. Tingley and Tomz (2020[12]) examined this possibility with the use of survey experiments on nationally representative U.S. samples. Their study tested whether international commitments can increase the willingness to support emission-reduction policies. They randomly assigned participants to scenarios in which the U.S. either joined or abstained from the Paris Agreement, with varying policy costs and emission targets. In scenarios in which the U.S. participated in the Paris Agreement they observed increased support for climate policies (14-29 percentage points). However, that increased support was observed only when commitments entailing moderate cost options, and overly ambitious pledges reduced enthusiasm.
Another critical question is whether visible failures to meet Paris commitments erode public faith in the agreement. A study by Beiser-McGrath and Bernauer (2019[13]) examined this through survey experiments involving more than 6000 participants in the United States and China. They found that domestic support for international climate agreements is not undermined by failures of other countries to meet their climate goals. This suggests the credibility of the Paris Agreement could remain resilient to possible commitment failures.
2.4.3. Laboratory experiments
While no single experimental study directly compares Paris and Kyoto in a side-by-side lab experiment, multiple works simulate their core design features.8 This strand of experimental economics simulates climate treaty negotiations in a laboratory role-playing setting. In the experiments governments are typically represented by single players. In most of the studies, each player is endowed with a virtual currency, which can either be cashed out or invested via a voluntary contribution to a global common good, such as climate. The return to investment from a contribution of a single player is ex-ante uncertain and becomes known only after all players have made their investment decisions. Typically, laboratory settings are co-ordination games: if all players invest a sufficient share of their endowments, their pay-off to cash out at the end of the game is larger than their initial endowment. Some studies, such as Barrett and Dannenberg (2016[14]) simulate NDCs, in which each round allows players to initially share their intended contribution. These initial commitments materialise only in a subsequent round, in which players may be given the option to deviate from their initial commitments.
Some experimental studies examine the strengths and weaknesses of multilateral voluntary agreements. Using repeated games that partially mimic the NDC cycles, these studies explore the stability of agreements via the build and decline of trust. This way, they provide useful insights about voluntariness, enforcement, conditionality and fairness. Barrett and Dannenberg (2016[14]) conducted a lab experiment mimicking the voluntary pledge-and-review system of the Paris Agreement. In the experiment, players selected group targets, pledged a contribution, and voluntarily contributed to a public good representing mitigation. The study found that the addition of a review process affected pledges much more than it affected actual contributions. The idea that a pledge-and-review mechanism can encourage higher pledges, but struggles to induce commensurate emissions cuts in practice is also partially reflected in the work of Cherry et al. (2021[15]). Tavoni et al. (2011[16]) argue that an unequal distribution of resources reduces the prospects of reaching a common goal, but communication via intended contributions increases them dramatically.
A few experiments go further by comparing setting that closely resemble the Paris Agreement, the Kyoto Protocol and other alternatives. A recent work by Del Ponte, Masiliūnas and Lim (2025[17]) compares a decentralized voluntary agreement against designs with additional enforcement and monitoring. In their experiment, subjects could voluntarily pledge emissions reductions, as in the Paris Agreement, with options to add monetary penalties (Kyoto-style) or peer evaluations. Strikingly, most participants did join voluntary agreements and kept their pledges, but the pledges were so modest that emissions barely fell. The experiment by Dannenberg (2016[18]) explores the effectiveness of voluntary contributions in promoting multilateral cooperation, like climate agreements. The study compares first-best agreements, which requiring full contributions from all parties, and second-best agreements, requiring only a minimum level. Without peer-to-peer punishment, second-best agreements outperform first-best ones. Dannenberg (2016[18]) finds that agreements with minimum contributions entail lower compliance costs, are more likely to be accepted and are easier to maintain. In contrast, first-best agreements may fail because early non-compliance by some players leads to frustration and breakdown of cooperation. Therefore, voluntary agreements with modest goals tend to be more stable, while overly stringent agreements (analogous to Kyoto’s binding targets) can fail if parties get discouraged by initial free riding. The experimental setting by Schmidt and Ockenfels (2021[19]) argues that shifting the focus on a single commitment, such as a minimum carbon price, could potentially foster more ambitious co-operation. The comparison is made vis-à-vis: (i) a setting based on multiple individual commitments, such as the Paris Agreement, and (ii) a setting based on multiple common goals, such as the Kyoto Protocol. None of the experiments explored here offers a setting that sufficiently resembles a comparison between the Paris Agreement and alternative approaches.9
Another strand of laboratory experiments examines the requirement that each country’s successive NDC be more ambitious than the last (“ratchet-up” mechanism). The evidence here is mixed. Initially, a study by Dorsey (1992[20]) found that when revisions of voluntary contributions were only allowed upwards, and especially when there was a clear target to be met, contributions to the public good were higher. By contrast, if participants could both increase and decrease their contributions, cooperation weakened, as people tended to withdraw their support and free-ride on others’ efforts. Gallier and Sturm (2021[21]) test what happens when people are required to increase their contributions over time, similar to how the Paris Agreement asks countries to raise climate ambition. They find that when a “ratchet-up” rule requires each round’s effort to be at least as high as before, players hold back at the start of the game. This occurs deliberately, to avoid locking themselves into ever-higher obligations in later rounds. Although contributions eventually rise over time, the early caution may reduce overall efficiency compared to a system without ratcheting. The results suggest that while the Paris Agreement’s “ratchet mechanism” aims to build ambition, it can also trigger strategic hesitation if participants fear being locked into high future costs. The idea that high voluntary contributions at the beginning of the game increase exposure risk to a subsequent coordination failure has also been pointed out in Ye et al. (2020[22]). Alt et al. (2023[23]) examine the role of introducing a minimum contribution on top of a ratchet-up mechanism. The study finds that this mechanism may create a ratchet effect, but this requires that collective minimum contributions are binding.
The present study directly contributes to this discussion from more practical viewpoint, by directly asking experts and policymakers to evaluate the degree to which they expect the Paris Agreement to generate a virtuous cycle of ambition and action.
Laboratory experiments provide useful insights into how cooperation, trust, and enforcement might work in climate agreements, but their realism is inevitably limited. They often simplify international negotiations into games where single players represent governments, played in small groups over short timeframes. Such settings cannot reflect the complexity of political systems, diverse national interests, or the long-term nature of climate policy. Moreover, experiments typically rely on student participants, reducing their relevance for real-world decision-making. These limitations mean that lab findings should be seen as illustrative rather than predictive. To complement them, expert elicitation can offer more grounded perspectives Beyond surveys, a series of contributions employ game theory to compare the Kyoto Protocol and the Paris Agreement.
References
[23] Alt, M. et al. (2023), “Collective minimum contributions to counteract the ratchet effect in the voluntary provision of public goods”, Journal of Environmental Economics and Management, Vol. 122, p. 102895, https://doi.org/10.1016/j.jeem.2023.102895.
[14] Barrett, S. and A. Dannenberg (2016), “An experimental investigation into ‘pledge and review’ in climate negotiations”, Climatic Change, Vol. 138/1-2, pp. 339-351, https://doi.org/10.1007/s10584-016-1711-4.
[8] Bedsted, B., Y. Mathieu and C. Leyrit (2015), World Wide Views on Climate and Energy: Frome the World’s Citizens to the Climate and Energy Policymakers and Stakeholders, Danish Board of Technology Foundation, Missions Publiques and the French National Commission for Public Debate.
[13] Beiser-McGrath, L. and T. Bernauer (2019), “Commitment failures are unlikely to undermine public support for the Paris agreement”, Nature Climate Change, Vol. 9/3, pp. 248-252, https://doi.org/10.1038/s41558-019-0414-z.
[29] Chaudhuri, A. (2011), “Sustaining cooperation in laboratory public goods experiments: a selective survey of the literature”, Experimental Economics, Vol. 14/1, pp. 47-83, https://doi.org/10.1007/s10683-010-9257-1.
[15] Cherry, T. et al. (2021), “Can the Paris Agreement deliver ambitious climate cooperation? An experimental investigation of the effectiveness of pledge-and-review and targeting short-lived climate pollutants”, Environmental Science & Policy, Vol. 123, pp. 35-43, https://doi.org/10.1016/j.envsci.2021.05.004.
[18] Dannenberg, A. (2016), “Non-binding agreements in public goods experiments”, Oxford Economic Papers, Vol. 68/1, pp. 279-300, https://doi.org/10.1093/oep/gpv048.
[25] Dassa, C. et al. (1997), “Effects of a Neutral Answer Choice on the Reliability and Validity of Attitude and Opinion Items”, Canadian Journal of Program Evaluation, Vol. 12/2, pp. 61-80, https://doi.org/10.3138/cjpe.12.004.
[10] Dechezleprêtre, A. et al. (2025), “Fighting Climate Change: International Attitudes toward Climate Policies”, American Economic Review, Vol. 115/4, pp. 1258-1300, https://doi.org/10.1257/aer.20230501.
[9] Dechezleprêtre, A. et al. (2022), “Fighting climate change: International attitudes toward climate policies”, OECD Economics Department Working Papers, No. 1714, OECD Publishing, Paris, https://doi.org/10.1787/3406f29a-en.
[17] Del Ponte, A., A. Masiliūnas and N. Lim (2025), “Decentralized voluntary agreements do not reduce emissions in a climate change experiment”, Ecological Economics, Vol. 227, p. 108438, https://doi.org/10.1016/j.ecolecon.2024.108438.
[20] Dorsey, R. (1992), “The voluntary contributions mechanism with real time revisions”, Public Choice, Vol. 73/3, pp. 261-282, https://doi.org/10.1007/BF00140922.
[26] Farquharson, K. (2005), “A Different Kind of Snowball: Identifying Key Policymakers”, International Journal of Social Research Methodology, Vol. 8/4, pp. 345-353, https://doi.org/10.1080/1364557042000203116.
[21] Gallier, C. and B. Sturm (2021), “The ratchet effect in social dilemmas”, Journal of Economic Behavior & Organization, Vol. 186, pp. 251-268, https://doi.org/10.1016/j.jebo.2021.03.022.
[27] Goodman, L. (1961), “Snowball Sampling”, The Annals of Mathematical Statistics, Vol. 32/1, pp. 148-170, https://doi.org/10.1214/aoms/1177705148.
[24] Kankaraš, M. and S. Capecchi (2025), “Neither agree nor disagree: use and misuse of the neutral response category in Likert-type scales”, Metron, Vol. 83/1, pp. 111-140, https://doi.org/10.1007/s40300-024-00276-5.
[2] Kaya, Y. and K. Yokobori (1997), Environment, energy, and economy : strategies for sustainability, United Nations University Press, Tokyo.
[5] Kornek, U. et al. (2020), “What is important for achieving 2 °C? UNFCCC and IPCC expert perceptions on obstacles and response options for climate change mitigation”, Environmental Research Letters, Vol. 15/2, p. 024005, https://doi.org/10.1088/1748-9326/ab6394.
[1] Raupach, M. et al. (2007), “Global and regional drivers of accelerating CO <sub>2</sub> emissions”, Proceedings of the National Academy of Sciences, Vol. 104/24, pp. 10288-10293, https://doi.org/10.1073/pnas.0700609104.
[6] Sandalow, D., K. Benes and C. Augustin (2016), The Paris Agreement and Market Signals: A Survey, Columbia SIPA (Center on Global Energy Policy).
[19] Schmidt, K. and A. Ockenfels (2021), “Focusing climate negotiations on a uniform common commitment can promote cooperation”, Proceedings of the National Academy of Sciences, Vol. 118/11, https://doi.org/10.1073/pnas.2013070118.
[7] Stokes, B., R. Wike and J. Carle (2015), Global Concern about Climate Change, Broad Support for Limiting Emissions, PEW Research Center.
[16] Tavoni, A. et al. (2011), “Inequality, communication, and the avoidance of disastrous climate change in a public goods game”, Proceedings of the National Academy of Sciences, Vol. 108/29, pp. 11825-11829, https://doi.org/10.1073/pnas.1102493108.
[12] Tingley, D. and M. Tomz (2020), “International commitments and domestic opinion: the effect of the Paris Agreement on public support for policies to address climate change”, Environmental Politics, Vol. 29/7, pp. 1135-1156, https://doi.org/10.1080/09644016.2019.1705056.
[11] UNDP, University of Oxford and Browning Environmental Communications (2024), Peoples’ Climate Vote 2024, United Nations Development Programme (UNDP).
[3] Victor, D., M. Lumkowsky and A. Dannenberg (2022), “Determining the credibility of commitments in international climate policy”, Nature Climate Change, Vol. 12/9, pp. 793-800, https://doi.org/10.1038/s41558-022-01454-x.
[4] Wynes, S. et al. (2024), “Perceptions of carbon dioxide emission reductions and future warming among climate experts”, Communications Earth & Environment, Vol. 5/1, p. 498, https://doi.org/10.1038/s43247-024-01661-8.
[22] Ye, M. et al. (2020), “One Step at a Time: Does Gradualism Build Coordination?”, Management Science, Vol. 66/1, pp. 113-129, https://doi.org/10.1287/mnsc.2018.3210.
[28] Zelmer, J. (2003), “Linear Public Goods Experiments: A Meta-Analysis”, Experimental Economics, Vol. 6/3, pp. 299-310, https://doi.org/10.1023/A:1026277420119.
Number of responses obtained by country
Note: A response refers to a completed questionnaire submitted by one or more government officials, participating either in an official or personal capacity. In some cases, multiple responses were received from the same country.
Source: OECD survey on the transformative effect of the Paris Agreement, 2025.
Number of responses obtained by institution type and institution
Note: A response refers to a completed questionnaire submitted by a single individual, who is not employed by the government and is not asked to participate in a particular capacity.
Source: OECD survey on the transformative effect of the Paris Agreement, 2025.
Relative weights adjusting for the level of expertise and flat responding
Technical notes: The prior weight that corresponds to respondent is the product of the four prior weights characterising the respondent and any weight that applies to the survey item. For example, the responses of a policy analyst (1.00) with more than 25 years of experience (1.00), whose primary area of expertise is climate change (1.00) whose primary area of work is mitigation (1.00) who in Section A.1 provides answers with some minimal variation (1.00) receive a prior weight . Now, consider a respondent with expertise in energy policy (0.75), working on decarbonisation (1.00), with 15 years of experience (0.85) and an economic background (1.00). Suppose that this expert strongly agrees that mitigation is a policy priority in 2024, strongly agrees and it will be a priority in 2030 and 2040, while somewhat agrees that it would be a policy priority in 2024, 2030 and 2040 without the Paris agreement (0.95). The prior weight of respondent in item A.1 is . In a sample of respondents, the sum of the prior weights in a survey item (e.g. A.1) is: and it is smaller than as long as the sample contains observations from non-benchmark categories. To ensure that the sum of the final weights assigned to observations will still be equal to the sample size, the final weight assigned to each observation is: .
Source: Table generated by the authors.
← 1. To avoid confusion, the terms “part” and “parts” in this document refer exclusively to compartments of the survey questionnaire, while “section” and “sections” refer exclusively to compartments of this document. The grouping of questionnaire items into categories (A, B, C, D) and the numbering of items (e.g. B.1, B.2, B.3) reflect the way results are organised and discussed in the paper, and do not necessarily match the order in which these items appear in the questionnaire.
← 2. The use and misuse of passive and neutral options in surveys employing Likert-type scales has been extensively studied in the context of general population surveys. Kankaraš and Capecchi (2025[24]) find that only a minority of respondents use the neutral response as an “escape” option, primarily in the case of socially sensitive questions. Dassa et al. (1997[25]) showed that including a neutral option has little impact on the overall reliability and validity of responses. While the evidence for general surveys remains mixed, research focused on expert surveys is limited. In the context of this study, it is acknowledged that the option "Neither agree nor disagree" would offer respondents a perfectly neutral stance. However, concerns arose that some respondents might select this option not as a reflection of genuine neutrality, but due to a lack of a well-formed opinion. While this may be acceptable in general population surveys, it would be less desirable in the context of this expert survey, which is designed to capture informed judgment, even under uncertainty. Moreover, although the Likert-type scale used is categorical and does not include the midway category, the statistical analysis that utilises the data converts responses to numerical values. This allows for midpoint values to emerge at the aggregate level (i.e. the sample mean), effectively capturing neutral or intermediate views in the overall response distribution.
← 3. An institutional response does not reflect the views of the person(s) undertaking to complete the survey, but rather the collective position of the authority they represent.
← 4. The survey was circulated to delegates of the Working Party on Climate Change (WPCC). For missing delegations, the delegates of the Environment Policy Committee (EPOC) were contacted. Non-OECD countries have been contacted through the OECD Global Relations and Cooperation Directorate and with support of project-specific contacts.
← 5. In the latter case, invited researchers were selected based on their engagement in climate-related topics.
← 6. Network chain sampling mobilises existing survey participants to recruit additional participants from their networks. It is commonly used when the target population is hard to access, e.g. experts in a specialised field.
← 7. Goodman (1961[27]) discusses the possibilities and limitations of snowball samples. Studies using variations of snowball sampling include Farquharson (2005[26]).
← 8. The literature on experiments on public good contribution is voluminous. Zelmer (2003[28]) provides a meta-analysis, arguing that cooperation is greater when returns are higher. Chaudhuri (2011[29]) provides a selective survey of the literature, focusing on sustaining co-operation.
← 9. In a laboratory environment, such an experiment would entail at least two games, whose rules would widely differ to simulate the key differences of interest. The Paris Agreement would be simulated via voluntary, non-punitive commitments and other approaches via commitments enforced top-down and penalties for non-compliance.