Editors’ note: This is the second part of a two-column series. While the first column provided a methodological review of damage function literature (Aerts et al. 2024), this column discusses the quantitative implications for NGFS scenarios.
As we approach COP29, modelling the relationship between climate variables – such as temperature and precipitation – and economic output offers valuable insights into the potential economic damage resulting from future climate change (Bilal and Rossi-Hansberg 2023). However, researchers encounter substantial uncertainty when examining these impacts (Aerts et al. 2024), which leads to a wide range of loss projections. As the understanding of climate change gradually evolves, efforts to refine damage functions remain crucial. The Network for Greening the Financial System (NGFS) closely monitors these developments to ensure that NGFS scenarios take into account the latest advances in climate science.
Higher projections of climate-induced losses
Researchers continuously seek to improve their understanding and modelling of the economic impact of climate change, exploring various methodologies and assumptions. Yet, applying different methodologies results in a wide range of output loss projections and impact channels captured. Table 1 illustrates the variety of methodological approaches and the disparity in the results when comparing damage functions proposed in the literature.
Table 1 Loss estimates across a subset of damage functions
Notes: Loss estimates are displayed for a 1°C incremental global temperature increase, which corresponds to a global warming level of 2.2°C compared to the pre-industrial average (considering that the currently observed level of global warming stands at 1.2°C above the pre-industrial average). All calculations are based on projections in line with the NGFS ‘current policies’ scenario, which foresees that a global warming level around 3°C above the pre-industrial average is reached by 2100.
Source: Own calculations.
Table 1 shows that the projected impact of an incremental 1°C temperature increase – in addition to the 1.2°C of global warming already observed today – ranges from 1% to 25% of global GDP. In the absence of further climate mitigation action (i.e. if only currently implemented policies are maintained), the global economy could lose as little as 2% or as much as 45% of total output due to climate change by the end of the century, depending on the damage function underlying the projection. These estimates represent global losses against a hypothetical baseline without further climate change (i.e. a scenario in which no further warming occurs).
Climate change may thus significantly affect our future wellbeing: the global economy could grow by almost 300% (Nordhaus and Boyer 2000) or by ‘only’ 125% (Bilal and Känzig 2024) after accounting for climate-induced losses (Figure 1). The high uncertainty around these losses underlines the need for further academic research on the economic impact of climate change.
Figure 1 Global GDP projections across a subset of damage functions
Notes: Loss estimates calculated for a 1°C incremental global temperature increase, which corresponds to a global warming level of 2.2°C compared to the pre-industrial average (considering that the currently observed level of global warming stands at 1.2°C above the pre-industrial average).
Source: Own calculations.
A new damage function in NGFS scenarios
Since their launch in 2020, NGFS scenarios have become a cornerstone of climate risk assessments for the financial sector. These are a set of hypothetical scenarios exploring a range of plausible outcomes, providing insight into future climate-related physical and transition risks. Damage functions play a vital role in NGFS scenarios by seeking to capture (chronic) physical risks posed by climate change. The NGFS closely monitors the latest developments within this field and adapts its modelling framework accordingly.
In the forthcoming release of NGFS scenarios (NGFS 2024a), a new damage function has been implemented. This has led to a significant shift in how physical risks are represented in the scenarios.
The new damage function authored by Kotz et al. (2024), which is based on the latest evidence from climate science, projects significantly higher losses compared to estimates in earlier vintages produced using the previous model authored by Kalkuhl and Wenz (2020).
The implementation of a new damage function ensures that NGFS scenarios remain state-of-the-art and reflect the latest findings in the literature on climate change:
- First, the study by Kotz et al. (2024) is a recent extension to the literature, expanding on past studies (e.g. Kalkuhl and Wenz 2020, Kotz et al. 2021, Kotz et al. 2022). This damage function combines various elements to establish a better estimate of aggregate losses caused by climate change. Moreover, it uses more observations (i.e. additional years) and more granular historical climate and economic data (i.e. additional regions) on which the function is calibrated.
- Second, the new damage function is more comprehensive. It covers additional climate impact drivers on top of mean temperature changes (which is the core variable used in most other studies). Temperature variability, annual precipitation, number of wet days, and extreme daily rainfall are also considered. It also incorporates more persistent effects of climate shocks, which means that it not only takes into account the instantaneous impact of a climate shock when it occurs, but also the delayed effects up to ten years after the initial impact. Its geographic coverage is another strength, with its ability to provide both global and country-level loss projections.
- Third, the relatively high loss estimates produced by Kotz et al. (2024) offer a credible estimate of severe but plausible potential climate impacts and, as such, are well-suited for the risk assessment purposes of NGFS scenarios. The losses projected by this damage function are in the upper range of estimates in the literature (but not the highest). Hence, it is less likely to underestimate the potential negative consequences of climate change. NGFS scenarios are typically used in financial risk assessments, for which a precautionary approach is appropriate. Therefore, the use of damage functions that are in the lower range of estimates in the literature is likely to lead to a (potentially significant) underestimation of risks.
Caveats
The integration of a new damage function offers a valuable improvement in how physical risks are represented in NGFS scenarios. Nevertheless, some important caveats need to be kept in mind. First, climate damage functions are characterised by a high degree of uncertainty, which is demonstrated by the wide confidence interval of their loss estimates. Figure 2 displays the confidence interval of the Kotz et al. (2024) damage function based on the NGFS “Current Policies” warming trajectory. Thus, not only is the choice of damage function important, but so are the assumptions made within the damage function.
In previous vintages, NGFS scenarios employed a ‘high damage’ calibration of the damage function (95th percentile) instead of median projections (alongside high temperature increase percentiles for the most disorderly scenarios) (Figure 2). This assumption was justified as the Kalkuhl and Wenz damage function produced rather modest loss estimates compared to other estimates in the literature, reflecting its narrow calibration on changes in mean temperatures, as well as a lacking quantification of the persistence of the impact. Since Kotz et al. (2024) utilise a much more comprehensive set of climate impact drivers with associated persistence effects and produce loss projections whose magnitude is at the upper end of the literature, it is warranted to adjust previous modelling assumptions. Therefore, the most recent NGFS scenarios use median estimates for both damage and temperatures.
Figure 2 Global GDP losses due to climate change (NGFS Current Policies scenario)
Notes: Global GDP loss projections based on Kotz et al. (2024) and Kalkuhl and Wenz (2020) in line with the NGFS ‘current policies’ scenario (Phase IV). The solid red line shows the median projection by Kotz et al.; the solid black line displays the median projection by Kalkuhl and Wenz, while dotted lines indicate 95th percentiles of the damage distribution based on the 95th percentile of the temperature projection. The dark red shaded area displays the 90% confidence interval of the Kotz et al. projection based on the median temperature pathway. The light red shaded area further incorporates the 90% confidence interval of the temperature projection. Note that the confidence intervals of the two damage functions overlap.
Source: NGFS scenarios based on REMIND inputs (NGFS 2023, NGFS 2024b).
Furthermore, while the Kotz et al. (2024) damage function extensively covers climate variations (e.g. temperature, precipitation), it may still not fully capture climate impacts. In particular, extreme weather events (i.e. acute physical risks) such as droughts and cyclones are expected to intensify as a result of climate change. Yet, it is unclear to what extent losses from acute events are reflected in the Kotz et al. damage estimates. Moreover, a number of other climate-related risks are not examined in this damage function, including climate-induced socioeconomic risks (e.g. migration, armed conflict) or climate tipping points. On the other hand, an underestimation of future adaptation could lead to an overestimation of losses. Overall, despite providing significantly more adverse loss projections, the Kotz et al. damage function is not the last word on all climate-related risks.
Conclusion
Damage functions are an essential tool to assess the potential economic impact of future climate change. Within the financial sector, they play a crucial role in climate scenario analysis and climate stress tests. They also help to demonstrate the losses avoided by carrying out a timely green transition. Consequently, damage functions have been integrated into the NGFS climate scenario modelling framework.
In the most recent NGFS scenarios, a new damage function by Kotz et al. (2024), has been implemented. This allows a much more comprehensive assessment of the potential damage caused by climate change, resulting in substantially higher projected losses from physical risks compared to previous assessments. This change highlights how our understanding of climate-induced economic and financial risks is deepening over time, typically showing higher impacts as research evolves. Nonetheless, further research is needed to reduce uncertainty around the impact of climate change and provide improved quantitative estimates for better policymaking and risk analysis.
Authors note: This column reflects the views of the authors and not necessarily those of the European Central Bank or the Network for Greening the Financial System. We thank Lukasz Krebel, Theresa Löber, Franziska Piontek, and Vladimir Otrachshenko for their useful comments.
References
Aerts, S, L Stracca, and A Trzcinska (2024), “Measuring economic losses caused by climate change”, VoxEU.org, 2 October.
Bilal, A, and D R Känzig (2024), “The macroeconomic impact of climate change: Global vs. local temperature”, NBER Working Paper 32450.
Bilal, A, and E Rossi-Hansberg (2023), “The importance of anticipating climate change”, VoxEU.org, 24 July.
Burke, M, S M Hsiang, and E Miguel (2015), “Global non-linear effect of temperature on economic production”, Nature 527: 235–239.
Dell, M, B F Jones, and B A Olken (2012), “Temperature shocks and economic growth: Evidence from the last half century”, Journal of Economic Literature 52(3): 740–98.
Dietz, S, and N Stern (2015), “Endogenous growth, convexity of damage and climate risk: How Nordhaus’ framework supports deep cuts in carbon emissions”, The Economic Journal 125(583): 574–620.
Hassler, J, P Krusell, and C Olovsson (2024), “The macroeconomics of climate change: Starting points, tentative results, and a way forward”, Working Paper Series WP24-8, Peterson Institute for International Economics.
Howard, P H, and T Sterner (2017), “Few and not so far between: A meta-analysis of climate damage estimates”, Environmental Resource Economics 68: 197–225.
Howard, P H, and D Sylvan (2020), “Wisdom of the experts: Using survey responses to address positive and normative uncertainties in climate-economic models”, Climatic Change 162(2): 213–32.
Kahn, M E, K Mohaddes, R N C Ng, M H Pesaran, M Raissi, and J-C Yang (2021), “Long-term macroeconomic effects of climate change: A cross-country analysis”, Energy Economics 104: 105624.
Kalkuhl, M, and L Wenz (2020), “The impact of climate conditions on economic production. Evidence from a global panel of regions”, Journal of Environmental Economics and Management 103: 102360.
Kompas, T, V H Pham, and T N Che (2018), “The effects of climate change on GDP by country and the global economic gains from complying with the Paris climate accord”, Earth’s Future 6(8): 1153–73.
Kotz, M, L Wenz, A Stechemesser, M Kalkuhl, and A Levermann (2021), “Day-to-day temperature variability reduces economic growth”, Nature Climate Change 11: 319–25.
Kotz, M, A Levermann, and L Wenz (2022), “The effect of rainfall changes on economic production”, Nature 601: 223–27.
Kotz, M, A Levermann, and L Wenz (2024), “The economic commitment of climate change”, Nature 628: 551–57.
Network for Greening the Financial System (2023), “NGFS scenarios for central banks and supervisors – Phase IV”.
Network for Greening the Financial System (2024a), “NGFS Climate Scenarios for central banks and supervisors - Phase V”, forthcoming.
Network for Greening the Financial System (2024b), “Damage functions, NGFS scenarios, and the economic commitment of climate change: an explanatory note”, forthcoming.
Nordhaus, W D, and J Boyer (2000), Warming the world: Economic models of global warming, The MIT Press.
Nordhaus, W (2014), “Estimates of the social cost of carbon: Concepts and results from DICE-2013R model and alternative approaches”, Journal of the Association of Environmental and Resource Economists 1(1/2): 273–312.
Tol, R S J (2014), “Correction and update: The economic effects of climate change”, Journal of Economic Perspectives 28(2): 221–26.
Waidelich, P, F Batibeniz, J Rising, J S Kikstra, and S I Seneviratne (2024), “Climate damage projections beyond annual temperature”, Nature Climate Change 14: 592–99.
Weitzman, M L (2012), “GHG targets as insurance against catastrophic climate damages”, Journal of Public Economic Theory 14(2): 221–44.
