Why is the economic impact of climate change so uncertain?

Utkarsh Somaiya and Lewis Holden

Estimates of GDP damages from climate change vary significantly – they are highly uncertain (Chart 1). Why is there so much uncertainty and what is the source of this uncertainty? Our analysis finds the largest source of uncertainty is how climate change is translated into economic damages, rather than uncertainty about the path of future climate change itself.


Chart 1: Uncertainty in global climate change and subsequent economic damages

Notes: Adapted from Mohaddes and Raissi (2025) and IPCC (2021). SSP = Shared Socioeconomic Pathways. See description below.


Uncertainty in climate change science

How will the climate change in the future? The sources of uncertainty here are:

  1. What will future greenhouse gas emissions be?
  2. How will those greenhouse gas emissions affect the climate?

The developments in technology, demographics and consumption which will influence future emissions are unknown and to some extent unknowable. Climate scientists explore this uncertainty by developing illustrative scenarios known as Shared Socioeconomic Pathways (SSPs). SSPs are a range of plausible ways in which technology and socioeconomic factors might develop in the future. Climate scientists use these futures to create possible paths for greenhouse gas emissions and climate forcing. For example, SSP1–2.6, SSP2–4.5 and SSP5–8.5, where a higher final number indicates greater climate forcing and therefore warming. We refer to these as ‘climate scenarios’ and the uncertainty they give rise to as ‘scenario uncertainty’. Chart 1 presents the current vintage of climate scenarios used by the IPCC. Note the framework being developed for future scenarios finds that SSP5–8.5 is now implausible due to the emergence of climate policy, the cost of renewables and recent emission trends.

Climate models estimate the climate impacts of these scenarios, including on global average temperatures. There is uncertainty about different parameters in the climate system, for example the strength of certain feedback effects. Different climate models therefore use different assumptions and parameters. This results in a range of plausible temperature outcomes for a given climate scenario. We refer to this as ‘model uncertainty’.

Even if we had perfect climate models, temperature projections would still be uncertain because of the inherent randomness in the climate system. This creates further uncertainty in temperature outcomes for a given climate scenario.

The relative importance of these sources of uncertainty in climate scenarios depends on the size of the geographic area, the climate variable and the time horizon being considered. In the short term and over smaller geographical areas, randomness in the climate system is the dominant source of uncertainty in temperature projections. Over the long run, climate model uncertainty and scenario uncertainty become more dominant, with randomness becoming almost irrelevant. For global average temperatures, randomness is always less important because random local temperature shocks average out at a global level (Chart 2).


Chart 2: Sources of uncertainty in climate projections

Notes: Adapted from Lehner et al (2020). Global temperatures are 10-year rolling averages.


Uncertainty in economic damages

Economists estimate the impact of the climatic changes on the global economy using ‘damage functions’. Damage functions are simplified, reduced form, relationships which translate variables such as global average temperatures into economic damages. Damage functions don’t model the structural mechanisms through which climate affects the economy, or how economic activity can feedback into affecting the climate. This simplification means that they do not model how economies might adapt to climate change over time.

There is also uncertainty about which variables should be used in these reduced form relationships. For example, whether changes in temperature at the global or local level determine damages. Within these approaches, the specific mechanism through which climate affects economic output is also uncertain. For example, whether climate change:

• continuously supresses the level of economic output
• reduces the economy’s long-term growth rate
• also manifests as periodic catastrophic shocks to the level and/or growth rate of the economy
• affects the economy through some combination of all of the above

The consequence of this uncertainty is that there is no consensus on the basic relationship between climate change and economic damages.

The importance of different sources of uncertainty

What is the relative importance of economic uncertainty versus climate change uncertainty? Howard and Sterner (2025) derive three damage function specifications for level, growth and catastrophic effects based on a meta-analysis of the literature. This can be seen as capturing the ‘average’ effect of the level, growth, and catastrophic assumptions in the literature.

We developed a framework to quantify the relative importance of these sources of uncertainty. First, we took the full range of plausible temperature pathways from all climate scenarios. We put these temperature pathways into the ‘level effects’ specification of the damage function from Howard and Sterner. This gives us a range of economic damage estimates where the variation is driven by climate scenario uncertainty. At this stage, we have not introduced uncertainty because of the damage function. The full uncertainty range of plausible damages, denoted by the green area in Chart 3, is between 0% and -7% of GDP.

We can then decompose this uncertainty by exploring the range of damages within a single climate scenario, but considering the range of plausible temperature pathways estimated by climate models within that scenario. This captures climate model uncertainty and is denoted by the blue area in Chart 3. Finally, as we are looking at global damages, we find climate randomness has little effect (Chart 3; orange area).


Chart 3: Economic damage estimates – uncertainty decomposition


Next, we put the same temperature pathways from all climate scenarios into every possible damage function from Howard and Sterner’s meta-analysis. This allows us to quantify the additional uncertainty that is created because we don’t know the ‘correct’ damage function. This additional uncertainty increases the range of economic damages to between 1% and -37%, denoted by the purple area in Chart 3. The damage function uncertainty range expands further into the future. This reflects how damage functions are built: they tend to assume that each extra degree of warming causes more damage than the previous one.

These results suggest the largest source of uncertainty is because we don’t know how temperature changes translate into economic damages, rather than uncertainty about the path of future temperatures themselves.

Can we reduce uncertainty?

Uncertainties can be thought of as gaps in human knowledge that can be improved over time. Uncertainty is the engine of science; it drives us to build theories, develop models, and test our hypotheses against data. Some of the uncertainties we’ve identified can be reduced over time (Table A).


Table A: Reducibility of different sources of uncertainty

Source of uncertaintyCan it be reduced?
RandomnessNo.
Model uncertaintyYes. Will be reduced over time as we learn more about the climate system. For example, the ‘likely’ range of global temperature increase because of a doubling of CO2 has narrowed from 1.5oC-4.5oC to 2.5o-4oC.
Scenario uncertaintyYes. While unknown unknowns cannot be resolved, the range of plausible futures will narrow over time. For example, the range of plausible futures in the future vintage of IPCC scenarios will be narrower.
Damage function uncertaintyYes. Can be reduced to some extent as we learn more about the how climate impacts the economy, including observational data.

Reducing economic uncertainty has been characterised as particularly difficult because economists study complex social systems. These systems are characterised by constantly shifting human behaviour and expectations, where cause and effect are hard to disentangle. Despite these challenges, institutions like central banks are working together to improve our understanding of the impact of climate change on monetary and financial stability, as demonstrated by a recent report by the Network for Greening the Financial System.

Conclusion

We decomposed the sources of climate and economic uncertainty that explain the range of economic damage estimates in Chart 1. In the short term, climate uncertainty is largely random. Globally and over longer horizons, climate model and scenario uncertainty dominate. But the largest source of uncertainty in future economic damages is because we don’t know how climate change affects the global economy. This outweighs uncertainty about future climate change itself.

We have made incredible progress in understanding both the climate system and its interaction with the economy. Some of the remaining uncertainty will be reduced further over time. But reducing uncertainty will take time. Waiting for too much certainty risks delaying decision-making until it is too late.

So how can central banks and other institutions act in the face of uncertainty?

If tail outcomes are catastrophic for the economy and our models are uncertain, one response is simply precaution. Decision-making under uncertainty emphasises building resilience to plausible catastrophic outcomes. Similarly, central banks and other financial institutions may want to place more emphasis on avoiding specific, worst-case scenarios. We know enough about climate change to know that some plausible outcomes are catastrophic. So even in the face of uncertainties, precautionary actions can be taken to build resilience to these outcomes.


Utkarsh Somaiya previously worked in the Bank’s Financial Risk Management Division and currently works at Dalberg Advisors, and Lewis Holden works in the Bank’s Financial Risk Management Division.

If you want to get in touch, please email us at bankunderground@bankofengland.co.uk or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.

Leave a Reply