Information about historical emissions drives the division of climate change mitigation costs

Mitigating climate change is a global social dilemma that spans geography and generations. Despite decades of negotiations, current pledges are insufficient to limit warming below 2°C. A key hurdle is disagreement over how to divide mitigation costs between industrialized (high historical emitters) and developing countries. Developing countries argue historical responsibility implies richer, high-emitting nations should bear more costs, while many industrialized countries favor cost-effectiveness irrespective of history. This study examines whether and how information about historical emissions affects the division of mitigation costs in negotiations. The authors design an experiment modeling two generations: a first generation that creates wealth while emitting, and a second generation that inherits wealth and negotiates mitigation cost sharing. The core research question is whether making historical emissions salient changes second-generation decisions on how to split costs, potentially revealing roles for collective responsibility and informing real-world climate negotiations.

Prior experimental work linked climate negotiation outcomes to decisions in earlier stages, but typically used the same participants across stages, confounding effects with personal responsibility and stable individual traits. By separating generations, this study more closely mirrors real-world temporal separation of industrialization and current decision-makers and mitigates confounds like social preferences and changed expectations. The broader debate ranges from strong claims that high historical emitters are morally responsible and should pay most costs, to positions denying historical responsibility due to personal responsibility principles, historical ignorance of harms, or practical infeasibility of attribution. Concepts of collective guilt and responsibility have been shown to influence climate-related behaviors even in minimal group contexts. The study remains agnostic on normative claims and focuses on how providing historical emissions information affects cost-sharing behavior.

Design: A four-player, two-generation game with two fictional countries (A and B). First generation leaders (A1, B1) produce wealth by completing up to 40 real-effort slider tasks; production increases both their own earnings and the future climate mitigation costs. Piece rates per completed task for A1/B1 were randomly assigned with equal probability to one of three sets to induce heterogeneity: equal (A1=$8, B1=$8), moderately unequal (A1=$10, B1=$6), or highly unequal (A1=$12, B1=$4). Environmental costs of mitigation were convex in production. Second generation: Leaders (A2, B2) inherit wealth equal to their predecessor’s earnings plus $600. They then negotiate how to split mitigation costs using an ultimatum game with the strategy method: each participant makes decisions as both proposer (offer a split) and responder (set maximum willingness to pay). If the implemented offer is rejected, there is a 90% chance of a climate disaster that yields zero earnings and a 10% chance of keeping endowments. Each second-generation participant played four times using outcomes from four distinct first-generation pairs; roles were randomly assigned for payoff after data collection. Treatments: - History: Second-generation participants were fully informed about how endowments and mitigation costs were generated, including first-generation incentives and actual production/choices. Historical emissions were defined as the share of total mitigation costs created by a participant’s predecessor (range 0–1). - Baseline: Second-generation participants faced the same endowments and costs as matched History counterparts but received no information about their origins. Participants: 103 first-generation participants; 101 second-generation (History; 62% female; mean age 22); 103 second-generation (Baseline; 69% female; mean age 22). All experiments were conducted online (Qualtrics), recruiting NUS students via ORSEE. Currency: 100 experimental dollars = 1 SGD (≈0.75 USD at the time). Comprehension checks ensured understanding. Game-theoretic predictions: For the chosen parameters, there are equilibria with agreement; any cost division where both players exceed disagreement payoffs can be supported. Information about the origins of costs does not change incentives in equilibrium predictions (identical across treatments), though multiple equilibria exist. Additional baseline: Baseline with Predecessors (n=93; avg age 22.9; 57% female) informed second-generation participants that predecessors determined endowments and costs but did not reveal the game, decisions, or mapping from decisions to outcomes. This assessed whether merely knowing that predecessors existed (without detailed historical emissions information) affected behavior. Outcome measures: Main outcome is the share of mitigation costs covered by the proposer, computed as the offer weighted by acceptance probability, aggregated at each first-generation outcome. Historical emissions are the predecessor’s share of total climate costs. Analyses included OLS at the aggregated outcome level and random-effects GLS using individual-level proposer offers and responder thresholds (clustered by participant), with robustness checks including controls for age, gender, political orientation, and climate-action support.

- Average cost shares and agreement rates: Proposers paid on average 54% of costs in History vs 51% in Baseline. Agreement rates were 60% (History) vs 61% (Baseline); difference not significant (Mann–Whitney U, two-sided p=0.95). - Correlation with historical emissions: Share of costs covered correlated more strongly with historical emissions in History (Pearson r=0.86) than Baseline (r=0.42). - OLS regressions (Table 1, outcome-level): • History: Historical emissions strongly increased proposer’s cost share (β=0.53, t=13.49, p<0.001). • Baseline: Smaller positive effect (β=0.13, t=2.80, p<0.01). • Pooled: History×Emissions interaction β=0.40 (t=6.50, p<0.001), indicating a significantly stronger effect in History. Including relative wealth leaves the interaction significant; in Baseline, the emissions effect disappears when controlling for wealth, implying effects there are driven by wealth rather than emissions per se. • Total costs had small, non-significant effects on the proposer’s share; relative wealth was not significant. R²: 0.758 (History), 0.199 (Baseline), 0.647 (pooled). - Individual-level GLS: In History, higher historical emissions increased both proposer offers and responder willingness to pay. In Baseline, these effects weakened and vanished when controlling for wealth. Results robust to demographics and climate attitudes. - Strategy classification: 22% of proposers in History used a proportional-to-emissions strategy (coefficient ≈1), vs 0% in Baseline. Conversely, 20% in Baseline disregarded emissions (coefficient ≈0) vs 7% in History. K–S tests show distributions of individual coefficients differ significantly between treatments (p<0.001 for both roles). - Heterogeneous effects: Among high emitters (>50% of costs caused by predecessor), proposers offered to pay more in History (61%) than Baseline (53%). Among low emitters (<50%), proposers offered less in History (36%) than Baseline (44%). Both differences statistically significant (GLS). High emitters did not change responder thresholds; low emitters reduced both proposer offers and responder willingness when informed. - Ex-post rationality: Optimal offers (given responders’ thresholds) were close to actual offers. In History, ex-post payoff-maximizing proposals were near proportional to historical emissions; in Baseline, the ex-post optimal split was more even. - Obstacles to agreement: Agreements were less likely with greater inequality in historical emissions and wealth, and when total costs were higher (models in Supplementary Table A5). Inequality in predecessor productivity (piece rates) did not significantly affect agreement probability. - Additional baseline (Baseline with Predecessors): The History×Emissions interaction remained significant (β=0.31, p<0.001), confirming a stronger emissions effect in History. The emissions coefficient was larger in this baseline (β=0.22) than in the original Baseline (β=0.13), but differences between baselines were not statistically significant. Agreement probabilities were similar; in this baseline, inequality coefficients were not significant.

Providing clear information linking predecessors’ emissions to current wealth and costs made successors of high emitters more willing to pay for mitigation, consistent with collective responsibility or collective guilt mechanisms. Policy implication: public understanding of the historical link between emissions and current prosperity may increase support in developed countries for bearing larger mitigation shares. In the History treatment, offers that accounted for historical emissions were both more acceptable to responders and closer to ex-post payoff-maximizing strategies, suggesting that historical emissions served as a salient focal point or anchor in a game with multiple equilibria. Despite these shifts, average agreement rates did not change between History and Baseline because increased offers by successors of high emitters were offset by decreased offers by successors of low emitters. Agreements were more likely when wealth and historical emissions were more equal, complementing evidence that international inequality hinders cooperation. By separating generations, the design removes motives for negative reciprocity and personal responsibility confounds, explaining differences from prior studies where the same participants acted across stages.

The study demonstrates that accurate information about historical emissions meaningfully reshapes the division of climate mitigation costs: successors of high emitters offer to pay more, and successors of low emitters offer to pay less. This pattern aligns with collective responsibility and yields proposals closer to ex-post payoff-maximizing splits when history is known. For policymakers, communicating the connection between past emissions, current wealth, and present climate costs can facilitate more acceptable and potentially more effective international agreements. Future research should explore scalability to multi-country settings, dynamics in repeated negotiations, the role of real national identities, and the influence of salience and anchoring. It should also consider alternative assumptions about predecessors’ knowledge of harms, nonlinear attribution of historical emissions, and partial (rather than all-or-nothing) mitigation outcomes.

- Small-group abstraction: Only two countries per generation; results may differ with more players due to diffusion of responsibility. - One-shot negotiation: No repeated adjustment or learning; real negotiations are iterative. - Fictional identities: Real national identities and attachments could alter willingness to pay. - Assumptions about knowledge: First generation was assumed aware that production generated emissions requiring future mitigation; historically, predecessors often lacked such knowledge. - Attribution simplification: Historical responsibility was proxied by total emissions shares; real impacts may be nonlinear and depend on timing/location of emissions. - Binary outcome framing: Negotiations either averted or failed to avert disaster; real agreements can partially reduce risks. - External validity: Student sample from a single university; online setting and incentives may not capture geopolitical stakes.