Climate change and the global economy: quantifying the impact

In the second of a series of four articles, our economists examine various methods of quantifying the impact of climate change on global economic activity.


Keith Wade

Keith Wade

Chief Economist & Strategist

Marcus Jennings


Climate damage functions: quantifying the impact on activity

Early estimates of the cost of global warming on world GDP emerged in the early 1990s and since then there have been a number of studies that have both agreed with and contradicted the initial assessments.

Covington and Thamotheram (2015) base their analysis on so called “climate damage functions” that quantify the risk the economy faces as a result of climate change.

Economic climate damage is defined as the fractional loss in annual economic output at a given level of warming compared to output in the same economy with no warming.

Climate damage functions plot the level of output lost over a range of warming estimates, with all functions predicting a greater loss in annual economic output as the level of warming rises.

However, amongst the estimated climate damage functions there is a lack of consensus as to how damages evolve as warming gradually increases.

The below figure and table summarise a number of estimated economic damage functions, named after their respective originators.

We briefly discuss each climate damage function below, focusing on the 4°C mark, given that the World Bank estimates there is a 40% chance of exceeding this level by 2100, assuming emissions follow a "medium business-as-usual pathway".

Varying estimates of climate change damage

Estimates of climate change damage vary according to whether there is a tipping point at which damage accelerates

The “N-damages” climate damage function, named after its originator Nordhaus (2013), is widely used by economists and is the least concerning of the three climate damage functions.

Climate damage under this function would be progressive whereby no tipping point is reached and the world's population has the greatest amount of time to offset any negative effects of global warming.

It can be seen that by the year in which the world is 4°C warmer, annual economic output will be just 4% lower than a base case with no warming.

The baseline case in Nordhaus's study is for warming of around 3.8% by 2100.

Nordhaus believes the economic impact of climate change is likely to be small over the next couple of decades and that agriculture is the most exposed sector to global warming.

Although the cumulative effects are reasonable at the point at which 4°C is reached, the loss in terms of average annual growth would be extremely small and difficult to distinguish given that it will take many decades to reach 4°C of warming based on current estimates.

The “W-damages” function was produced by Weitzman (2012) and estimates that by the time 4°C of warming is reached, 9% of annual economic output will be lost relative to the base with no warming effect.

Under this scenario, those industries that are largely predisposed to climate change risk globally are likely to be affected:

  • Insurance
  • Agriculture
  • Forestry.

However, Pearce et al (1996) believe that only a fraction of the market economy is vulnerable to global warming:

  • Agriculture
  • Coastal resources
  • Energy
  • Forestry
  • Tourism
  • Water

These sectors contribute just 5% of global GDP to which their share is expected to shrink overtime (Mendelsohn, 2013).

This can be seen when we translate the damage function into the effect on economic growth.

If we assume a base case of 3% annual economic growth and that 4°C warming is reached by 2080, we find that annual growth will be paired back to 2.85%.

This is based on an economy that is 9% smaller due to climate damage in 2080 relative to an economy with no warming.

An effective loss of 0.15% per annum could be seen to warrant some attention from policymakers and the government alike, but is unlikely to be sufficiently powerful to promote a significant response to climate change.

In the most severe case, global GDP growth would be some 1% lower per annum

The final climate damage function, “DS-damages”, named after Dietz and Stern (2014) is the most extreme scenario in which the global economy would suffer considerable loss as a result of climate change.

Warming unfolds over time and actions today have implications for the future.

Under this scenario, as and when warming extends to 4°C, annual economic output will be 50% lower compared to a scenario where no warming occurs.

To put this into perspective, Dietz and Stern estimate warming of approximately 3.5°C by 2100.

If we take a stricter approach however, using the same assumptions as the W-damages function above but assuming 4°C is reached in 2080, the base case 3% annual economic growth rate falls to just 1.9% a year.

Reaching the tipping point

At this rate, climate change is set to have a noticeable impact on future growth and living standards.

Reaching a tipping point at 2-3°C, as Dietz and Stern predict, could therefore be seen as a crucial stage of warming for the global economy whereby the costs of insufficient action significantly weigh on growth.

Christine Lagarde, head of the International Monetary Fund (IMF), believes the planet is "perilously close" to a climate change tipping point to the extent that climate change poses the greatest economic challenge of the 21st century.

Below we summarise some additional benchmark studies in the literature aiming to address the economic impacts of climate change.

Similar to the damage functions described above and aside from the Stern review and upper estimates from the Intergovernmental Panel on Climate Change (IPCC), the consensus is that the economic costs of marginal warming will be small up to 2°C but begin to gather pace if we move toward 4°C.

This analysis indicates that output losses accelerate once warming exceeds 2°C, but that these effects are not likely to be felt for another 30 years.

It is this threshold which is apparent in investment studies such as that recently published by Mercer which finds negative returns to diversified portfolios once warming breaches 2°C.

However, let us not forget that warming unfolds over time and that actions today have implications for the future.

Since the process is largely irreversible over the medium term, the global economy can be seen to have committed to a certain degree of future warming already.

A 2014 World Bank study titled "Turn Down the Heat. Confronting the New Climate Normal" estimates that warming of close to 1.5°C above pre-industrial times is locked into the earth's atmospheric system and is thus unavoidable.

According to the same study, without reasonable action to reduce emissions, the earth is on track for 2°C warming by mid-century and 4°C or more by the end of the century.

Stern (2006) also estimates that without action to reduce emissions, the concentration of greenhouse gases could double their pre-industrial levels as early as 2035, almost committing the world to temperature increases of over 2°C.

Investors should use a range of climate damage functions

For investors assessing the value of a stream of income, these projections are critical and we would suggest using a range of climate damage functions given the uncertainty over each.

Expressing future economic losses in today's prices requires discounting the loss in output back to the present day.

By its nature, small changes in the discount rate cause large changes in loss estimates given the very long time horizon in which climate change is expected to occur.

When attempting to quantify the impact climate change will have on a diversified portfolio, Covington and Thamotheram (2015) use a discount rate of 6.5%. In contrast the Stern review (2006) uses 1.4% (0.1% above expected consumption growth), so it is unsurprising that Stern's estimates forecast greater costs of climate change than many other studies.

Given the decline in long-term interest rates since the Global Financial Crisis, it seems using a rate toward the lower end of recent studies would be reasonable.

Please find below a link to a list of relevant references for research mentioned in this article.