Climate change could trigger debt crises, with adaptation providing only partial relief

For many countries, climate change is likely to lower growth and increase the frequency of natural disasters. This will strain public finances from both the revenue and expenditure sides, which will in turn lead to fears of defaults and thus raise government borrowing costs, adding more strain (Dibley et al, 2021; Zenios, 2022). How likely is it that these feedback loops will lead some countries to default? Which countries are particularly at risk?

One way to answer these questions is to overlay climate scenarios onto debt sustainability analysis, which is used to assess how well-equipped sovereigns are to meet their liabilities. Calcaterra et al (2025) developed a model to do just this and here we show results from the model for Australia, Brazil, Italy, India, Finland and Tanzania.

For climate-economy scenarios, Calcaterra et al (2025) used the RICE50+ model (Gazzotti et al, 2021). Based on Nordhaus (1993), RICE50+ projects economic growth under different ‘shared socioeconomic pathways’ (SSPs) and ‘representative concentration pathways’ (RCPs), which describe future greenhouse gas emission trends, and based on different assumptions about how rising emissions will affect growth.

Two possible future pathways from the United Nations Intergovernmental Panel on Climate Change are known as SSP2-RCP4.5 and SSP3-RCP7.0 (Nakicenovic et al, 2014). SSP2-RCP4.5 assumes that socioeconomic trends will follow historical patterns and that the impact of greenhouse gas emissions will be moderate. SSP3-RCP7.0 assumes that there will be ‘regional rivalry’ – international fragmentation as observed in current geopolitical conflicts – combined with relatively high greenhouse gas emissions, leading to a median temperature rise of up to about 3.4 degrees Celsius by 2100. For each scenario, GDP projections are derived under two different assumptions: that emissions will damage output to a small or large extent 1 . Table 1 shows how GDP will be impacted by climate damages in the two socioeconomic and warming scenarios and the two output impact assumptions.

Table 1: Changes to GDP (%) due to climate change, to the end of the century, selected countries

				Small impact				Large impact
Narrative scenario	Country	2030	2050	2070	2100	Avg.		2030	2050	2070	2100	Avg.
	World	-1	-3	-4	-5	-3		-4	-18	-35	-60	-26
	Australia	0	-2	-3	-5	-2		-2	-10	-23	-45	-19
SSP2-RCP4.5	Brazil	-1	-3	-5	-7	-4		-4	-24	-49	-77	-38
	Finland	0	0	0	0	0		+8	+56	+138	+155	+92
	India	-1	-4	-7	-10	-5		-4	-24	-53	-83	-41
	Italy	0	-2	-3	-3	-2		-1	-4	-10	-22	-9
	Tanzania	-1	-3	-5	-7	-4		-3	-23	-47	-76	-37
	World	-1	-3	-5	-9	-4		-4	-17	-37	-68	-28
	Australia	0	-2	-4	-8	-4		-2	-11	-27	-56	-22
SSP3-RCP7.0	Brazil	-1	-3	-7	-12	-6		-5	-26	-52	-84	-41
	Finland	0	0	0	-1	0		+8	+57	+139	+156	+93
	India	0	-3	-8	-16	-6		-4	-23	-53	-88	-41
	Italy	0	-2	-4	-7	-3		-1	-4	-12	-32	-11
	Tanzania	-1	-3	-7	-12	-6		-4	-25	-52	-84	-41

Source: Bruegel based on Calcaterra et al (2025). Note: changes are calculated assuming no adaptation to climate change.

Table 1 shows that in both narrative scenarios, the overall GDP loss up to the end of the century would be low if the impact of climate change on output is small. However, if the impact is large (as is predicted by current climate models), GDP losses could be extremely high, averaging about -28 percent for the world from 2030 to 2100 (though differences between the losses in different countries would be substantial). The impacts are larger for SSP3 than SSP2. 

This approach can also be used to compute the optimal level of climate adaptation policies and their impact on GDP, distinguishing between reactive adaptation (eg disaster relief) and proactive adaptation (eg building coastal protection infrastructure). It can also factor in adaptive capacity linked to a country’s level of development and R&D investments (eg early warning systems). Figure 1 shows how adaptation measures increase output relative to the no-adaptation baseline, assuming large impacts. Estimates range from little difference (Italy), to a 15 percent increase in GDP relative to the climate-damaged level (Brazil). The model also provides estimates of adaptation costs, ranging from 0.12 percent to 2.46 percent of GDP, evenly split between proactive and reactive adaptation and adaptive capacity (see the appendix). 

Figure 1: Reduced climate damages due to adaptation, selected countries

Source: Bruegel based on Calcaterra et al (2025). Note: Projections assume large impacts from climate change.

 

Borrowing costs and future borrowing needs for each of the output paths generated by the RICE50+ model are based on a debt sustainability model (Zenios et al, 2021). This makes different assumptions of who will pay the costs of the adaptation investment: the private sector entirely (no fiscal cost), the public sector entirely, or a combination. On this basis, we generated distributions (‘fan charts’) for future debt.

Figure 2 shows the distributions associated with the SSP2-RCP4.5 and SSP3-RCP7.0 scenarios for both small climate damages and large climate damages (without any climate adaptation) and compares each to a benchmark in which climate damages are assumed to be zero (grey fan charts) 2 . As expected, the small and large climate damage scenarios adversely affect countries’ debt trajectories in most cases, with the 75th percentile shifting upwards. Nevertheless, the 75th percentile remains downward sloping in the small climate damage scenarios, except for Brazil and India. In the large climate impacts scenario, debt becomes explosive at the 75th percentile for all countries except Finland. It is also explosive at the median percentile for all countries except Finland and Italy.

Figure 2: Climate-change impact on country debts in the long run

A: Small climate damages  

B: Large climate damages

Source: Bruegel based on Calcaterra et al (2025). Note: the debt ratio (% of GDP) until the end of the century with small and large climate damage is overlaid on the debt ratios without climate effects (zero-damage, grey-shaded). Lines indicate the median and interquartile ranges.

The next step is to compute the long-term primary fiscal balances (the excess of revenues over non-interest expenditures) that would stabilise the climate-impacted trajectories with 75 percent probability, assuming that climate damage is zero, small or large. We consider debt-stabilising primary surpluses of up to 3 percent of GDP to be feasible (based on Eichengreen and Panizza, 2016). Table 2 shows the results, with the first column denoting historical averages.

Table 2: Primary balance to stabilise sovereign debt, %

			SSP2-CP4.5					SSP3-RCP7.0
Country	Hist.		Zero	Small	Large			Zero	Small	Large
Australia	-1.2		-1.6	-1.6	-1.0			-1.6	-1.6	-0.2
Brazil	0.7		1.4	1.6	2.8			2.4	2.6	4.0
Finland	0.2		-1.2	-1.2	-1.0			-1.2	-1.2	-1.0
India	-4.0		0.2	0.4	2.2			1.2	1.4	3.8
Italy	1.0		0.6	0.8	1.6			1.6	2.2	3.4
Tanzania	-0.7		-0.2	-0.2	0.4			-0.2	-0.2	0.8

Source: Bruegel based on Calcaterra et al (2025). 

As one would expect from Figure 2, climate damages could have a major impact on the debt-stabilising primary balance, particularly in the large-damages scenario. However, primary balances would generally remain below the 3 percent threshold. The exceptions are Brazil, India, and Italy in the large-damages scenario, suggesting that – at least without effective adaptation – climate change could lead to unsustainable debt in these countries.

We finally ask how these results change if adaptation is taken into account, based on different assumptions about who pays the adaptation costs. Figure 3 shows the range of end-of-century debt ratios with different adaptation policies for both climate narratives. The shaded area displays the interquartile range of debt ratios – there is a U-shaped relationship between end-of-the-century debt and public investment in adaptation. Fully private adaptation reduces debt ratios compared to no adaptation, although only marginally for some countries. Government spending on reactive adaptation can reduce debt ratios, but the benefits do not cover the costs if the government fully funds adaptation.

The double arrows on top of the whiskers in Figure 3 denote the direction of the 75th percentile of the debt trajectory. Horizontal means the debt is stable; upward pointing means the debt is growing. For Australia and Brazil, privately funded adaptation would stabilise the debt; in all other cases, it would lower debt ratios, but debts would remain upward-sloping. This raises the question of how much adaptation is needed to lower the debt-stabilising primary balance and if this is below the 3 percent threshold.

Figure 3: Effects of adaptation scenarios on debt ratios

Source: Bruegel based on Calcaterra et al (2025).

Table 3 shows stabilising primary balances with and without adaptation in the worst-case scenario: SSP3-RCP7.0 scenario with large climate damages (when adaptation is most beneficial). Compared to the stabilising balance under identical climate conditions but no adaptation costs, the stabilising balance with adaptation costs is lower. For Italy, the required primary balance would be reduced from 3.4 percent to 3.0 percent, which can be considered feasible, but for Brazil and India it would remain above the 3 percent threshold.

Table 3: Primary balance to stabilise sovereign debt with adaptation

			Without adaptation	With adaptation
Australia			-0.2	-0.4
Brazil			4.0	3.7
Finland			-1.0	NA
India			3.8	3.6
Italy			3.4	3.0
Tanzania			0.8	0.6

Source: Bruegel based on Calcaterra et al (2025). Note: adaptation calculation is not done for Finland because it stands to benefit economically in the climate-change scenarios.

Public financing of reactive adaptation – with the rest of the adaptation taken on by the private sector – can positively affect debt sustainability for debt that covers its cost. However, adaptation does not necessarily render debt sustainable. There is a broad consensus that significant adaptation expenditures will have to be borne by governments, but our tests highlight the limits of adaptation in averting potential debt crises precipitated by climate change.

Conclusion

Climate change is likely to have significant adverse effects on debt sustainability. The magnitude of the impact is subject to deep uncertainty. It can be a few percentage points or extremely large, potentially precipitating debt crises, depending on the scenario. The effects would be noticeable from the mid-2040s, potentially becoming substantial from mid-century.

Maintaining debt sustainability is possible with additional fiscal effort. However, the required debt-stabilising primary balances appear challenging under extreme climate narratives involving large-scale damages and may not be feasible for all countries. Climate adaptation can help mitigate the adverse effects of debt but is not a panacea. Governments would be able to finance only up to one-third of adaptation costs and still break even. The adage that there can be no public-finance sustainability without environmental sustainability is corroborated.

References

Burke, M., S.M. Hsiang and E. Miguel (2015) ‘Global non-linear effect of temperature on economic production’, Nature 527: 235–239, available at https://www.nature.com/articles/nature15725

Calcaterra, M., A. Consiglio, V. Martorana, M. Tavoni and S.A. Zenios (2025) ‘Sovereigns on thinning ice: debt sustainability, climate impacts and adaptation’, Working Paper, Bruegel, forthcoming

Nakicenovic, N., R. Lempert, A.C. Janetos (eds) (2014) ‘Special Issue: A Framework for the Development of New Socio-economic Scenarios for Climate Change Research’, Climatic Change 122(3), available at http://link.springer.com/journal/10584/122/3

Dibley, A., T. Wetzer and C. Hepburn (2021) ‘National COVID debts: climate change imperils countries’ ability to repay’, Nature 592: 184–187, available at http://dx.doi.org/10.1038/d41586-021-00871-w

Eichengreen, B. and U. Panizza (2016) ‘A surplus of ambition: can Europe rely on large primary surpluses to solve its debt problem?’ Economic Policy 31(85): 5–49, available at https://doi.org/10.1093/epolic/eiv016

Gazotti, P., J. Emmerling, G. Marangoni, A. Castelletti, K.I. van der Wijst, A. Hof and M. Tavoni (2021) ‘Persistent inequality in economically optimal climate policies’, Nature Communications 12: 3421, available at https://www.nature.com/articles/s41467-021-23613-y

Kalkuhl, M. and L. Wenz (2020) ‘The impact of climate conditions on economic pro- duction. Evidence from a global panel of regions’, Journal of Environmental Economics and Management, 103: 102360, available at https://doi.org/10.1016/j.jeem.2020.102360

Kotz, M., A. Levermann and L. Wenz (2024) ‘The economic commitment of climate change’, Nature 628: 551–557, available at https://www.nature.com/articles/s41586-024-07219-0

Nordhaus, W.D. (1993) ‘Rolling the “DICE”: an optimal transition path for controlling greenhouse gases’, Resource and Energy Economics, 15(1): 27–50, available at https://doi.org/10.1016/0928-7655(93)90017-O

Zenios, S.A., A. Consiglio, M. Athanasopoulou, E. Moshammer, A.G. Gonzalez and A. Erce (2021) ‘Risk Management for Sustainable Sovereign Debt Financing’, Operations Research 69: 755–773, available at https://dx.doi.org/10.2139/ssrn.3250806

Zenios, S.A. (2022) ‘The risks from climate change to sovereign debt’, Climatic Change, 172: 30, available at https://link.springer.com/article/10.1007/s10584-022-03373-4

Appendix: costs of proactive and reactive adaptation and adaptive capacity (% of GDP) up to the end of the century, selected countries

SSP2-RCP4.5                                                             SSP3-RCP7.0

Country	All	Proactive	Reactive	Capacity		All	Proactive	Reactive	Capacity
Australia	0.84	0.51	0.32	0.02		0.90	0.57	0.31	0.02
Brazil	2.46	0.77	0.47	1.22		2.24	0.93	0.46	0.85
India	1.32	0.49	0.62	0.20		0.96	0.37	0.47	0.12
Italy	0.12	0.08	0.02	0.01		0.16	0.11	0.03	0.02
Tanzania	0.38	0.18	0.15	0.05		0.29	0.15	0.10	0.05

Source: Bruegel based on Calcaterra et al (2025).