Climate policy models help governments weigh the economic cost of cutting greenhouse gases against the damage caused by a warming planet. A newly published study argues that one of the best-known models may be understating how quickly those cuts should happen because of the way it represents population and income.
Writing in Scientific Reports, researchers Danilo Liuzzi, Veronica Lupi and Simone Marsiglio extended the widely used DICE climate-economy framework so that population size can affect emissions directly, rather than only through its contribution to economic output. Their results materially changed the timing and intensity of the model’s preferred climate response.
Under their central calibration, the modified model called for an optimal emissions abatement rate of 46.7% in 2030 and 72.4% in 2050, reaching full abatement by 2070. The standard DICE benchmark used for comparison produced much slower action, with abatement of 19.6% in 2030, 29.4% in 2050 and 65.9% in 2100.
The difference comes from a deceptively simple modelling choice. Traditional integrated assessment models commonly link industrial emissions closely to total economic output. Population matters because more people can mean more production and consumption, but it is not necessarily represented as a separate, non-proportional driver of emissions.
Liuzzi and colleagues instead combined DICE with the STIRPAT framework, which allows environmental impacts to respond differently to population and affluence. Their baseline calibration used a population elasticity of 1.85 and an affluence elasticity of 0.87, drawing on empirical estimates in the existing literature. An elasticity above one means that, within the model, emissions respond more than proportionally to changes in population, while the sub-one income elasticity partly offsets that effect.
The consequence is substantial. Without abatement, gross emissions in the extended DICE-STIRPAT model were around 20% higher than standard DICE in 2030 and 101% higher by 2100. That faster emissions trajectory increased atmospheric carbon concentrations and projected climate damages, which in turn made earlier mitigation economically preferable.
Three policy settings were examined: a business-as-usual pathway, a socially optimal pathway balancing mitigation costs and climate damages, and a pathway constrained to keep warming below 2°C. In the temperature-constrained case, the modified model required roughly 60% abatement by 2030 and full decarbonisation around 2055. Under the economically optimal scenario, full abatement arrived in 2070.
The carbon price implied by the model also rose sharply. By the end of the century, the optimal scenario in DICE-STIRPAT produced a carbon price of about $350, compared with roughly $220 in the standard DICE optimal scenario. Under the 2°C pathway, the modified model’s carbon price peaked at around $450 in 2055, while the optimal pathway peaked at about $410 in 2070.
The researchers tested whether the conclusion depended too heavily on a single demographic forecast. Their central case used a United Nations population pathway reaching about 11.5 billion people by the end of the century, but they also considered lower and higher trajectories of about 9.0 billion and 12.5 billion. The precise timing and carbon prices changed, yet the central conclusion remained: explicitly modelling population as a separate emissions driver generally brought stronger mitigation forward.
Further sensitivity tests varied the population and affluence elasticities. Those exercises produced a wide range of possible dates for full decarbonisation in the optimal scenario, reflecting genuine uncertainty in empirical estimates. Even so, the authors report that the qualitative result persisted across the alternative specifications: asymmetric responses to population and income tended to increase unabated emissions relative to standard DICE and therefore increased the case for earlier abatement.
The study is particularly relevant to climate-policy debates because integrated assessment models are not forecasts in the ordinary sense. They are structured tools for exploring how assumptions about the economy, emissions, climate damages, technology and social preferences interact. Changing one relationship can therefore alter what the model regards as an economically efficient pathway.
That distinction is also an important limitation. The results do not show that global emissions will necessarily be twice the conventional DICE baseline in 2100. They show what happens when a specific set of empirically motivated population and income elasticities is incorporated into the DICE architecture. The authors acknowledge that estimates in the STIRPAT literature vary substantially, which is why their robustness analysis spans alternative parameter values.
Nor should the results be read as an argument for population control. The researchers explicitly reject that interpretation. Their policy point is about the specification of emissions inside climate-economy models: if population has a larger direct relationship with emissions than conventional frameworks assume, then decarbonising energy, production and consumption may need to proceed faster than those frameworks imply.
For economies such as South Africa, where climate policy must sit alongside electricity security, industrial development, employment and fiscal constraints, the paper illustrates why modelling assumptions matter. A pathway that shifts economically justified abatement forward by decades would affect the implied urgency of power-sector investment, transport electrification, industrial efficiency and carbon-pricing decisions. It does not prescribe a South African policy mix, but it highlights the risks of treating long-term mitigation timelines as fixed outputs rather than results that depend on how emissions drivers are represented.
The larger message is methodological rather than demographic. Climate-economy models simplify an extraordinarily complex system, and those simplifications can materially influence the apparent cost of waiting. In this case, allowing population and per-capita income to affect emissions differently moved the model toward much earlier and more aggressive decarbonisation.
Source Information
Study Title: Population driven emissions call for earlier climate mitigation
Authors: Danilo Liuzzi, Veronica Lupi and Simone Marsiglio
Journal: Scientific Reports
Year: 2026
DOI: 10.1038/s41598-026-57960-x
