The world is warming, and we have this summer experienced record-breaking temperatures in many places, with forests burning, mortality rising and economic activity falling. Similarly worrisome, this was most likely not an outlier, but rather an example of summers to come. These developments also have consequences for financial markets and expected returns on financial assets. But what consequences? I organised a conference on the issue and describe here some of the conclusions from the discussions we had.
Global warming already affects most of us and will continue to do so for many years, most often in negative ways. Newspapers write a lot about climate change, but less about what it means for financial markets and long-term savings.
At the Council for Return Expectations that I chair (link), we have been discussing these issues for some time. To advance our understanding of the topic, we teamed up with the Pension Research Centre (PeRCent; link), which I also chair, to organise the conference Climate Risk and Expected Returns (link). It took place last Monday, 24 August 2026.
We had an excellent programme, with some of the leading scholars in the field presenting their research and practitioners representing different types of financial institutions sharing how they address climate change and its implications in practice.
More than 200 people had signed up. It was a great and inspiring day. I learned a lot. You can find presentations from the conference here: Link.
We first discussed what climate science tells us about the present and future weather, then the macroeconomic consequences of climate change, followed by the implications for financial markets and returns. Finally, we had a panel discussion about how financial institutions deal with climate risk in practice.
In what follows, I summarise the main takeaways, or, at least, my main takeaways.
The world has become much warmer, and it will only get worse
The first presentation was given by Adrian Lema, Director of the National Center for Climate Research at the Danish Meteorological Institute.
His starting point was straightforward: global warming is already substantial and is continuing. The long-run global average land temperature is expected to cross 1.5°C above pre-industrial levels around 2029, and some degree of overshoot now appears virtually inevitable. Temperatures have risen dramatically and persistently since the late 1970s, as Figure 1 shows, and in 2025, the global temperature was 1.2 degrees above the 1950-1980 average.
Figure 1. Global surface temperature relative to the 1951–1980 average. Source: Berkeley Earth and J.Rangvid.
The consequences are not simply higher average temperatures. Climate change increases the frequency and intensity of heatwaves, droughts, heavy precipitation and storm surges, Adrian told us. It also increases the likelihood of compound events, where several hazards occur simultaneously or consecutively, creating much larger damages than individual events.
Andrian emphasised the uncertainty around tail risks. There may be low-probability but very high-impact outcomes, including tipping points, that are difficult to incorporate into standard climate models. His conclusion was clear: climate risks will affect ecosystems, health, food production, infrastructure and, ultimately, the economy. Climate risk is also economic and financial risk.
Climate change may reduce economic growth more than previously thought
We then turned to how these dramatic weather changes affect the macroeconomy. Diego Känzig from Northwestern University presented research that reaches a striking conclusion: the economic costs of climate change may be much larger than previous estimates suggest.
Rather than examining local temperature shocks, his research uses shocks to global temperature, the argument being that global temperature better captures the state of the climate system and its effects on extreme weather around the world.
Their estimates imply that a 1°C increase in global temperature eventually reduces world GDP per capita by around 20%. This is considerably more than previous literature that studies local temperature shocks has found, as Figure 2, copied from Diego’s presentation, demonstrates.
Figure 2. Impact of a one percent increase in global temperature on global GDP per capita, compared to estimates using local temperature shocks. Source: D. Känzig presentation at conference (link).
The mechanism is important. Global temperature shocks are followed by substantial increases in extreme heat, drought, precipitation and wind events – similar to what Lema emphasized – which in turn damage economic activity. The implication is that the economic case for climate mitigation may be considerably stronger than suggested by traditional climate-economics models.
While recognising that the research is published in one of the most prestigious academic journals, I think some in the room struggled to understand the sheer magnitude of the effects. What exactly is the channel that makes these estimates so much more dramatical than those other researchers have found? I think we need more research to fully answer this question.
An optimistic voice
John Hassler reminded us that it is the accumulated level of CO₂ emissions that determines global temperatures. This is important because it implies that reaching net zero emissions will not cause temperatures to fall immediately. Rather, temperatures will remain elevated and decline only very slowly, even after reaching net zero, as the stock of accumulated emissions remains in the atmosphere for a long time.
He recognised Diego’s and other researchers’ findings, but still emphasised the large uncertainty surrounding estimates of the economic consequences of climate change. We have never experienced temperatures as high as those we face already now, and even more so in the future, so we have no historical data that can tell us exactly what these temperatures will mean for the economy. To some extent, we therefore have to ‘guess’ what the consequences will be.
Despite these rather dire predictions, John was surprisingly positive about the outlook. He argued that, globally, we are moving in the right direction—with the important exception of the US—and that, if we implement the necessary changes appropriately, adaptation costs can be spread over time. They would then be manageable rather than being borne disproportionately by any particular generation.
The discount rate matters enormously
Johannes Stroebel from NYU then turned to a more familiar finance question: How should we discount economic damages that occur many decades into the future?
Consider the important concept of the Social Cost of Carbon. This is essentially the present value of future damages from an additional tonne of CO₂. To find a present value, we discount future values. The discount rate used to calculate that present value can make a huge difference.
To illustrate, Johannes referred to the impact analyses of the Affordable Clean Energy Rule (link), showing how sensitive estimates of the Global Social Cost of Carbon are to the choice of discount rate. The impact analysis found that “average global SC- CO₂ estimates across all the models for emissions occurring over 2025-2035 range from”:
- 7 percent discount rate: $6 to $9 per metric ton of CO₂ emissions (in 2016 dollars).
- 3 percent discount rate: $53 to $63 per metric ton of CO₂ emissions (2016$).
- 2.5 percent discount rate: $77 to $90 per metric ton of CO₂ (2016$).
In other words, estimates of the Global Social Cost of Carbon vary from $6 to $90 when varying the discount rate from 7 percent to 2.5 percent, an enormous difference.
Stroebel’s key point was that there is no single discount rate. In finance, the appropriate discount rate depends on risk and, specifically, on the covariance between an investment’s payoff and economic conditions. If climate damages occur mainly in bad states of the world, when consumption is low and resources are particularly valuable, climate-related payoffs should be discounted at relatively low rates, yielding higher NPV estimates.
He also presented evidence on very long-term discount rates using housing markets in the UK and Singapore. The price difference between owning a property forever and owning it for 100 years implies a discount rate of about 2.6% for a 100-year cash flow. Since housing itself is risky, this gives an upper bound for the long-run risk-free rate. Stroebel therefore argues that the appropriate discount rate for calculating the Social Cost of Carbon should be below 2.6%, meaning large costs from climate change.
How do you put climate change into capital market assumptions?
Björn Griesbach from Allianz Investment Management showed how climate scenarios can be translated into the numbers that investors actually use when constructing portfolios and making wealth projections. This is particularly relevant for our Council of Return Expectations, as our role is precisely to provide the return assumptions that banks and pension funds use when informing their customers about how their savings can be expected to develop.
Allianz’ approach starts with the NGFS (Network for Greening the Financial System) climate scenarios. Allianz translates them into assumptions about economic growth, inflation and other macroeconomic variables, and then estimates their implications for expected returns, volatility and correlations across financial assets.
In their estimates, physical climate risks lead to greater economic volatility, higher inflation and lower real growth. This translates into lower expected returns and higher volatility in financial markets. Allianz estimates average annual real equity returns between 2025 and 2050 of around 2.5% under a “Current Policies” scenario, compared with 3.2% if the world manages to keep future temperatures below a 2°C scenario. Both numbers are substantially below the returns investors have experienced historically.
In updated estimates, Allianz incorporates expected productivity gains from AI. They find that, under Current Policies, average real equity returns drop by 0.4 percentage points, to 2.1%. However, if other policies adjust to offset AI’s climate impact, keeping global warming below 2°C, the productivity gains from AI could lift real equity returns to 3.6%, still, however, substantially below historical returns.
An equally interesting result concerns diversification. More frequent negative supply shocks from climate events make bonds less effective as a hedge against equity volatility, meaning less diversification between stocks and bonds. This makes the risk-return trade-off in a stock-bond portfolio basically flat, as illustrated in Figure 3, which is a copy from Björn’s presentation. This contrasts the 1980-2019 period where a higher allocation to equities also gave higher expected returns.
Figure 3. Returns and risk for different combinations of equities and bonds over different periods. Source: B.Griesbach’s presentation (link).
Green investments probably have lower expected returns
Lasse Heje Pedersen from CBS addressed a question very much on many investors’ minds: What should we expect green stocks to return relative to brown stocks?
There are several competing theories. One says that green firms may have better future cash flows and therefore generate higher returns. Another says that investors’ preference for green assets pushes up their prices, which lowers their future expected returns.
The empirical evidence from realised returns is surprisingly inconclusive, Lasse found. Studies have reported both large underperformance and substantial outperformance of green stocks. He presented a replication across 49 countries and 23 measures of greenness, or “greenium” – the difference between returns on green and brown stocks – and showed that, once multiple testing is taken into account, the evidence from realised returns is essentially insignificant. Figure 4, taken from Lasse’s presentation, shows that t-statistics are symmetrically distributed around zero, meaning estimates of the greenium are basically indistinguishable from zero.
As Lasse told me afterwards, these results should not really come as a surprise. We only have 10–15 years of data on green and brown investment strategies. Anyone who has worked with estimating expected equity returns knows that, with such a short history, estimates will be so noisy that they are often indistinguishable from zero.
Figure 4. T-stat distribution of greenium (difference between returns on green and brown stocks) in published and replicated studies. Source: L.H.Pedersen presentation (link).
Forward-looking measures are more informative. Using implied costs of capital, Lasse and co-authors estimate that a company that is one standard deviation greener has an expected return approximately 0.4 percentage points lower per year. In other words, there is evidence of a negative greenium: investors accept somewhat lower expected returns for greener companies. Backing this up, Lasse cited research by our CBS colleague Niels-Joachim Gormsen showing that the cost of capital is lower for green firms than for brown firms.
This illustrates an important distinction. Green finance may help the climate precisely because it lowers the cost of capital of green firms. But that does not mean green investors should expect higher returns. Quite possibly, the opposite is true.
What does all this mean for pension funds?
Rasmus Bessing, Co-CIO for responsible investments and products at the pension fund PFA, gave a practical perspective on these questions.
PFA has introduced a climate-oriented pension product. Rasmus was honest that climate screens and targets will reduce the investment universe. At the same time, strong demand for green assets may raise their prices and reduce their forward-looking returns, similarly to what Lasse argued.
PFA does what it can to mitigate these limitations, but, ultimately, PFA’s calculations suggest that the expected risk and return from their climate product will be lower than from their non-climate product, though the overall difference is relatively modest. This is also reflected in the historical returns. Since its founding in 2020, the climate product has returned 73.3% in total, while the non-climate product has returned 77.5%.
Lastly, I moderated a panel debate with Jasper Riis from P+, Sasja Beslik from SDG Impact Japan, Mette Kanstrup Petry from Danmarks Nationalbank and Ann Frank Andresen from Nordea Pension. Some great discussions arose.
One particularly important point was the distinction between scenarios and probabilities. Investors can construct different climate scenarios, but what they ultimately need for portfolio decisions are probabilities attached to those scenarios. Jasper also highlighted that climate change could make diversification more difficult and potentially more expensive, similarly to what Björn argued, while increasing the importance of security selection.
Mette illustrated how Danmarks Nationalbank incorporates climate considerations into macroeconomic analysis, financial analysis, responsible investment and climate-related data collection. She made an important point: Climate risks ultimately materialise through familiar financial risk categories rather than creating an entirely separate category of risk.
Political risk
One interesting point that came up repeatedly during the day was that political risk was perhaps the most important climate-related risk. In other words, the key uncertainty is not only how climate change will affect the economy, but also how decisively, and predictably, politicians will respond to it.
This is important because political risk is particularly difficult for investors to quantify and hedge. Will the next political shift bring another Donald Trump or another Al Gore? Will governments strengthen climate policies or reverse them? Such political choices will affect both the extent of climate change and the costs of the transition. As several participants stressed during the day, this uncertainty may ultimately matter as much for investors as uncertainty about the physical and economic effects of climate change themselves.
Conclusion
So, what did I take away from the day?
For me, three conclusions stood out.
First, climate change is not a distant tail risk. Physical climate risks are already visible, and further warming will make extreme events more frequent and severe.
Second, climate change matters for economic growth and financial returns, but not necessarily in the simple way often assumed. The economic costs may be larger than previously thought, while green assets may have somewhat lower expected returns because investors value them.
Third, we can attempt to incorporate climate change directly into capital market assumptions, but uncertainty remains enormous.
Personally, one thing I struggle with is the question whether climate risk should be singled out as a particularly important risk factor, or whether it is better viewed as one of the many risks investors face. While it is undoubtedly significant and negative, it may not be more significant for financial markets—or any easier to quantify—than uncertainty about future demographics, productivity, AI, geopolitics, global indebtedness, and so on.
That may sound like an unsatisfactory conclusion. I don’t think it is. We now understand much more about the mechanisms through which climate change affects the economy and financial markets. But the right response to uncertainty is not to pretend that we can forecast its effects to the nearest basis point.
