Pension Funds Have Been Poorly Advised on Climate Risks
a calendar icon
August 4, 2023
10
min read

Pension Funds Have Been Poorly Advised on Climate Risks

pension plans are at greater risk then most are aware

Pension Funds Rely on Flawed Climate Economics Models

A new report from Carbon Tracker reveals that pension funds worldwide have relied upon fundamentally flawed economic models to assess the risks of climate change to retirement savings. As a result, pension fund members have been systematically misled about the scale of financial risks that climate change poses to their future prosperity.

The report shows that the consulting firms which pension funds pay for advice on climate change impacts have largely based their assessments on mainstream peer-reviewed economic research. However, this academic research makes entirely unrealistic assumptions which lead to conclusions that global warming will only have modest impacts on the economy even at temperature rises of 3, 4 or even 6°C.

For example, one survey of top climate economists showed the median prediction was that 7°C of global warming would only reduce global GDP in the year 2200 by 20% compared to a world without any climate change. Other economic models predict that 4°C of warming by 2100 will reduce global GDP by between 10-23%.

These models starkly contrast with the warnings coming from climate scientists, who argue that global warming of even 2°C poses the risk of triggering dangerous “tipping points” like the melting of Arctic permafrost which could accelerate further warming. Multiple scientific papers have characterized warming of 3°C as “catastrophic” and stated that 5°C or more is likely to pose an “existential threat” to human civilization.

Yet pension funds, relying on the consultants who in turn have depended upon these mainstream climate economics models, have dismissed such risks. For example:

  • The Australian pension fund Unisuper said that a “worst case scenario” of 4.3°C warming by 2100 was an “acceptable” risk.
  • The UK’s Shropshire County Pension Fund estimated that a 4°C rise by 2100 would only dent its annual returns by 0.06% over the period to 2050.
  • Hymans Robertson consultants modeled a 4°C rise by 2042 would mean a manageable 35% hit to assets for its client Hampshire Pension Fund.

This complacent view lulls pension fund members into a false sense of security about the resilience of their retirement incomes to climate change. It is therefore vitally important to understand how pension funds and their consultants have arrived at such misleading conclusions about the economic impacts of global warming.

Mainstream Climate Economics Based on Unrealistic Assumptions  

The Carbon Tracker report identifies that the root cause of pension funds underestimating climate risks lies in fundamental flaws within mainstream climate change economics research. This field is dominated intellectually by a small group of interconnected academics centered around Professor William Nordhaus of Yale University.

Nordhaus created the first, and still most influential, climate economics model back in 1991. This is called the DICE model (“Dynamic Integrated model of Climate and the Economy”). DICE models the economy using standard economic growth theories, while modeling the climate system in a simple way based only on global average temperatures.

In DICE, the impact of warming temperatures on the economy are calibrated by making assumptions about the “damages” in terms of GDP loss caused by differing degrees of global warming. The model can then optimize the “ideal” level of carbon taxes or emissions reductions to limit climate damages to a socially optimal level. This encapsulates the “cost-benefit” approach to climate change favored by most mainstream economists.

However, the DICE model, and the many descendant models that build upon its methodology, embed fundamentally unrealistic assumptions about the climate system and climate damages. These include:

Assuming most economic activity occurs “indoors” and is immune to climate change impacts

In his foundational 1991 paper introducing the DICE model, William Nordhaus asserted:

“The major categories not directly affected by climate change include manufacturing, services, trade, and most other human activities which are undertaken in carefully controlled environments... Activities such as agriculture, forestry, fisheries, construction, insurance, and human life and health are directly exposed to climate change.”

This led Nordhaus to claim an extraordinary 87% of the US economy occurred indoors and would be “negligibly affected by climate change”. This overlooks the fact that the “indoors” sectors rely upon critical infrastructure services like electricity that depend sensitively on weather patterns.

Yet this unsupported claim underpins why most climate economics models only attribute damages to unusual weather-exposed sectors like agriculture, instead of the economy as a whole.

Assuming damages scale with temperature alone, ignoring impacts of changing rainfall patterns

Nearly all climate economics models like DICE consider only changes in temperature, and ignore the effects of changes in precipitation and rainfall patterns resulting from climate change. The models implicitly assume rainfall changes will be geographically uniform and benign.

Yet the impacts of increasing droughts, changing monsoon patterns and alterations to agriculture viability from rainfall changes could be more devastating than the temperature increases themselves. Models ignoring this dimension cannot offer credible insights into the true economic consequences.

Assuming damages can be accurately captured by a “damage function” based on limited current data

Mainstream climate economists frequently attempt to base their damage functions on statistical analysis of the observed relationship between temperature and GDP across different geographies.

For instance, a paper may look at how annual GDP fluctuates with the average temperature across countries, and extrapolate this to predict how future warming will influence future GDP. This is highly problematic:  

  • Today’s spatial relationship does not emulate how the world economy would evolve over time in a sustainably warmer world.
  • Tipping points mean the future relationship may be non-linear and suddenly shift if dangerous climate thresholds are crossed. But models based on limited current data cannot capture these nonlinearities.

Despite these shortcomings, much climate economics research relies upon such “bottom up” damage functions inappropriately calibrated by current empirical observations.

Assuming economic damages are proportional to the square of temperature change

Most climate economics models since Nordhaus have assumed, without compelling empirical evidence, that climate damages to GDP are proportional to the change in global average temperature squared.

This means (for example) that 4°C of warming causes 16 times the GDP loss as 1°C of warming. Again, this appears to be an arbitrary assumption chosen primarily for mathematical convenience when calibrating damage functions.

Since higher temperatures increase the likelihood of triggering devastating tipping points in a nonlinear way, the true “damage curve” is likely to be far steeper. But simplistic quadratic damage functions overlook these risks.

No Scientific Basis for Assumptions in Models

The Carbon Tracker report emphasizes that these crucial assumptions which underlie mainstream climate economics models have no scientific basis grounded in an understanding of the realities of the climate system.

Rather, they appear to be convenient choices made by economists to enable generating mathematical damage functions requiring minimal computational complexity. While this may be understandable from the disciplinary silo of economics, these arbitrary assumptions produce outputs severely detached from climate realities.

For instance, a recent survey ↗ showed that 11 percent of top economists believe that 6°C of warming would inflict zero economic damages, while the median prediction was just 10% GDP loss.

This contrasts markedly with the warnings of climate scientists that 6°C of warming likely poses an existential threat to human civilization due to risks like melting ice sheets displacing over a billion people living in coastal cities. Clearly, the blithe assumptions made by economists have led their models to overlook real risks that natural scientists are acutely concerned about.  

The Carbon Tracker report delves into climate science research to illustrate the depths of the divide between economists and scientists. For example, scientists estimate that collapse of the Atlantic Meridional Overturning Circulation (AMOC) - a crucial ocean current system - would have catastrophic impacts such as plunging Europe into extreme cold weather.

Yet a climate economics paper modeled AMOC collapse and absurdly predicted it would boost global economic growth. This unscientific but peer-reviewed result then gets baked into climate damage models, further contributing to their unreliability.

Unless the assumptions underlying climate economics models are re-evaluated to better reflect scientific realities, the outputs will continue to provide a false sense of security to pension funds and other financial institutions.

Groupthink and Lack of Scientific Input Exacerbates Problem

Problematically, the Carbon Tracker report reveals that the small clique of academics who produce climate change economic forecasts operate in an insular “bubble” which promotes groupthink and fails to challenge dubious assumptions.

There is a close-knit professional and academic network comprised of a few dozen researchers who control the bulk of peer-reviewed publications on the economics of climate change. This creates ripe conditions for collective biases to become entrenched.

Flawed assumptions get propagated from one paper to another without ever being rigorously scrutinized. Prior research is cited as “evidence” to justify retaining assumptions, even if the original basis was arbitrary rather than empirical. There is a tendency to uncritically rely upon simplified mathematical shortcuts like quadratic damage functions due to familiarity, overlooking conceptual inconsistencies.  

The tiny size of the field limits thorough critical introspection. Since similar approaches using familiar models proliferate, this gets misinterpreted as confirmatory consensus. In reality, there are deep problems with the foundations of the models that never get addressed