When Activist Research Contradicts the Consensus

This month, the National Bureau of Economic Research (NBER) released a preprint of a new paper by Adrien Bilal and Diego R. Känzig that, ostensibly, promises to upend the existing climate economics literature. According to the authors, the economic costs associated with climate change could be six times worse than the previous largest estimates, and 21 times worse than recent values used by the US government.

Their calculation of the social cost of carbon ($1,056 per tonne CO2, or $3,872 per tonne of carbon) places it somewhat…outside the range established by pre-existing estimates:

The working paper, which has not yet been subject to formal peer review, uses complex methods, not entirely documented with data and code, and employs questionable assumptions and extrapolations to arrive at their results, as responses by Richard Tol, Roger Pielke, Matthew Kahn, and others have noted. Yet the media coverage of the paper, including at outlets like the Guardian, Bloomberg, The Globe and Mail, Carbon Brief, Table Climate, has been largely credulous. Prominent climate economist Marshall Burke (whose own work largely stands in contrast to the results of the paper) gave it a big initial boost by advertising it on Twitter. Gernot Wagner, a climate economist at Columbia University who has himself published on the social cost of carbon, said, “If the results hold up, and I have no reason to believe they wouldn’t, they will make a massive difference in the overall climate damage estimates.”

As the above figure indicates, there are, in fact, plenty of first-order reasons to doubt the paper’s results. But the extreme outlier nature of their findings isn’t the only red flag. Bilal and Känzig use a somewhat bizarre methodology to find a relationship that defies common sense all in an avowed effort to “reconcile” climate economics with the idea that climate change is an “existential threat.” These methodological gymnastics act as a feature, not a bug, even producing a defense against criticism. Carbon Brief’s Simon Evans, for example, argued that, since other economists can’t even make sense of Bilal and Känzig’s analysis, their critiques are unconvincing.

In a sense, publishing ideologically slanted research is any author’s prerogative, one that hopefully the peer review process will uncover. If anything, the more worrying question raised by this paper is why so many environmental journalists and economists leapt at the chance to endorse a finding that explicitly disavows an existing academic consensus. The whole thing speaks to the threats to substantive research posed by activist commitments, and the incentive structure facing journalists and experts within the broader issue public. When a narrative-driven understanding of science becomes predominant, the imperative to produce supportive work tends to override the production of coherent, defensible research. As we’ve seen in countless examples, from the Ecological Footprint to the Planetary Boundaries hypothesis to Howarth’s fugitive methane research to Mark Jacobson’s disgraced 100% wind/water/solar model, widespread expert debunking is no match for activist embrace of bullshit. All the more reason for experts to resist these activist narratives, no matter how powerful the incentives

The Climate Cost-Benefit Test

Climate economics is difficult. The purpose—determining the economic costs associated with global warming and its impacts—must first contend with the reality that, as anthropogenic emissions have pushed global average surface temperatures up over the past couple centuries, economic and other welfare outcomes for humanity have improved enormously.

Since 1800, global average surface temperature has risen by about 1.2 degrees Celsius. In that time, human population has expanded from around 1 billion to now over 8 billion, the energy consumption of human societies has risen by a factor of 30, total societal wealth has skyrocketed by almost a factor of 100, average lifespans have more than doubled, and democratic freedom has become the norm, not the rare exception, to human social existence.

Correlation is not causation. Indeed, no one credible would argue that global warming caused the unprecedented expansion of human well-being. But there are strong, even mainstream, reasons to believe that the industrial-scale use of fossil fuels, the main cause of global warming, did. So this is not to suggest that climate change itself is good for humanity, but rather that a) it is not at first approximation associated with absolute negative outcomes for humanity as a whole (at least not yet) and b) that establishing a negative relationship requires further empirical and statistical analysis.

And it is precisely that kind of analysis that earned William Nordhaus the Nobel Prize in Economics. Nordhaus’s integrated climate-economic model, which was the first of its kind when he published it in 1990, weighed the obvious benefits of fossil fuel exploitation (see above) against the costs of atmospheric warming that fossil fuel emissions produce. Nordhaus’s original model concluded that the social optimum—the equilibrium beyond which the costs of warming would exceed the benefits of fossil fuel usage—would be about somewhere between three and four degrees C by the end of the 21st century. Over time, Nordhaus has lowered that estimate to around 2.5 degrees C, which, of course, significantly exceeds the prominent two-degree C target, as well as more recent 1.5 C aspirations.

Nordhaus’s critics argued that his imputed social discount rate was too high, and he underestimated the long-tail risks of climate change, but these criticisms largely relied on the same integrated climate-economic modeling framework that he pioneered. In more recent years, though, Nordhaus and his approach to climate risk have become somewhat verboten among climate advocates. To an activist and scholarly community dedicated to rapidly abolishing fossil fuel use, weighing the costs against the benefits of fossil fuels is an exercise in “predatory delay.”

And yet, the academic and theoretical basis for rapid near-term abolition remained thin. Thousands of papers estimate economic climate damages, but they largely find that the costs of climate change will at most cut 20% off GDP that is expected to otherwise double to quadruple.

Juking the Climate Stats

So how does this new paper arrive at its much more severe estimates of the damage of climate change, and why should they be treated with extreme skepticism? It’s worth taking a step back to try to understand the logic of the paper. The research goes back to the basic relationship between global temperature and global GDP per capita and purports to find a huge effect that nobody else has ever found before. This is a notable claim because the barriers to analyzing time series of global temperature and global GDP are very small. Presumably, this is where the vast majority of climate-economics researchers would start looking for a relationship.

This is how the logic of the paper works.

• The authors identify that there are ~0.1°C year-to-year variations above and below the long-term warming trend in global temperature.
• When they use a very specific combination of data processing that has not been used in this context before (e.g., construct ‘temperature shocks’ with a ‘Hamilton filter’), they find an association between a ~0.1°C yearly rise in temperature above the long-term trend and a ~1.2% reduction in GDP per capita six years later. This is a reduction in GDP per capita relative to what it would be otherwise (not a reduction relative to the year of the temperature shock).
• Then, they assume that these yearly ~0.1°C variations about the long-term trend are perfect analogies for long-term climate change, whereas the actual long-term trend (which is literally long-term climate change) is not.

Why has the purported large relationship between global average temperature and GDP not been identified before? A straightforward explanation is that the results are very method-dependent in the sense that they disappear if different strategies are used to investigate the same question. That does not lend strong support for the supposed effects being real.

Using extrapolations from these 0.1°C variations and various other assumptions, they calculate, for example, that the 0.75°C warming from 1960 to 2022 has suppressed GDP per capita by 37% over that time. Also, somewhat astonishingly, their derived relationships suggest that warming is universally bad, and thus, the economy would have grown that much more had the climate been cooling instead of warming.

The paper goes on to make perplexing claims about the impact of global temperatures on local weather in an effort to explain why its analysis arrives at different results from previous work. In particular, it invokes global temperature as a causal mechanism, which impacts your country's climate in a way where it somehow largely avoids impacting your country's temperature.

“It turns out that global temperature has much more pronounced impacts on [local] economic activity than local temperature.”

For example, when discussing extreme heat at the level of an individual country, the paper says,

“Local temperature shocks lead to an increase in the share of extreme heat days. However, global temperature shocks lead to a substantially larger increase in extreme heat days.”

So the claim is that when the globe has a particularly warm year (relative to the long-term trend), it typically causes your country to see a spike in extreme heat days even when your country is not experiencing a particularly warm year. This would be possible if long-term warming were associated with major widenings of daily temperature distributions such that increases in extreme cold canceled out increases in extreme heat. But if anything, we see the opposite: extreme daily cold is rapidly becoming less extreme (milder) over land, suggesting a narrowing of daily temperature distributions. Note that the working paper completely ignores any (presumably positive) impact of extreme cold becoming milder.

The paper also makes similar claims about changes in extreme precipitation and extreme wind. But in order to assess the impact of climate change on floods (the result of extreme precipitation that would impact the economy) and extreme wind, we don't need a new convoluted methodology involving the construction of ‘global temperature shocks’ with Hamilton filters; we should instead look at long-term trends in these phenomena.

For floods, we see a substantial diversity in the direction of trends, and overall, most observational studies show no increase in floods globally and, if anything, show decreases. The IPCC essentially says that it is a wash and concludes:

“There is low confidence about peak flow trends over past decades on the global scale, but there are regions experiencing increases, including parts of Asia, Southern South America, north-east USA, north-western Europe, and the Amazon, and regions experiencing decreases, including parts of the Mediterranean, Australia, Africa, and south-western USA.”

For extreme winds, the balance of evidence suggests long-term decreases globally, not increases. Here’s what the IPCC says about historical changes in extreme winds:

“In summary, the observed intensity of extreme winds is becoming less severe in the low to mid-latitudes, while becoming more severe in high latitudes poleward of 60 degrees (low confidence).”

To remind you, a scant amount of the economy is poleward of 60 degrees: essentially only Alaska, northern Canada, Greenland, Iceland, the Nordic countries, northern Russia, and Antarctica.

Furthermore, studies on the economic damage from extreme weather consistently show that observed upward trends are the result of increased value exposed. When this is accounted for, we consistently see decreasing economic damage per unit of GDP exposed and less damage per unit of GDP exposed for higher-income countries.

Regardless of the specific methodological peculiarities, this paper suffers from a common shortcoming across climate economics, which is a complete negligence of long-term economic, technological, social, and political developments. Consider a climate economist in 1924 trying to calculate the effects of the 1.3°C of warming over the next century. At the time, even in one of the richest countries in the world, the United States, only 33% of people had electricity, 28% had an automobile, 20% had a flushing toilet, and 34% had a phone. Nobody had a refrigerator or, of course, modern communication devices. These things may seem tangential to climate adaptation, but basic standards of living and technology make societies more resilient to all aspects of a climate. In order to truly estimate the costs of climate change, that 1924 climate economist would have had to anticipate, or at least reasonably account for, all these relevant developments that took place over the century.

So yes, doing climate economics is hard.

Activist Research

Discussing their paper in a recent episode of The Climate Pod, titled Groundbreaking Economic Study Suggests Greater Climate Damages, Diego Känzig explains why they are sympathetic to finding much bigger impacts than previous studies.

“When we read the newspapers and when we listen to the scientists, climate change is always portrayed as this existential threat that poses significant risks to our lives, livelihoods, and also the economy, but then when you look at these previous estimates, it doesn't look too severe so that was a tension and it's why when we arrived at these estimates they aligned better with my priors.”

Indeed, but we can't help but wonder if the authors of this paper have their priors skewed by consuming the climate reporting of particular newspapers and particular high-profile public-facing climate scientists. This would be a straightforward explanation for why they might appear unaware of relevant IPCC findings.

With the above priors as a foundation, the authors seem to go out of their way to frame their results to be as politically expedient and salient as possible.

“Our results indicate that the impact of climate change is substantial. In welfare terms, the cost of climate change is 640 times the cost of business cycles, or ten times the cost of moving from current trade relations to complete autarky. Perhaps most strikingly, in terms of output, capital, consumption, and thus welfare, climate change is comparable in magnitude to the effect of fighting a major war domestically. However, climate change is permanent. Thus, the losses from living in a world with climate change relative to a world without it are comparable to fighting a major war domestically, forever.”

"For instance, under conventional estimates based on local shocks, the domestic cost of carbon of the United States is $30/tCO2, making unilateral emissions reduction prohibitively expensive. Under our new estimates, however, the domestic cost of carbon of the United States becomes $211/tCO2 and thus largely exceeds policy costs. In that case, unilateral decarbonization policy becomes cost-effective for the United States."

Passages like this indicate that the authors had an affinity for their calculations and sought to frame them in a way that maximized attention and support from climate activists. The desirability of these conclusions from the author’s perspective, combined with their dependence on a very specific methodology that has serious technical and conceptual problems, should elicit skepticism from anyone who knows this field or, for that matter, human nature.

But instead of skepticism, we’ve seen cheerleading from swaths of environmental journalists and even economists. What explains this?

Climate activists have a committed tendency to exaggerate the dangers of climate change, asserting that the Earth will soon become uninhabitable to humans or that runaway global warming will kick in after global average temperatures exceed 1.5 degrees of warming. And while the continued invocation of these debunked ideas is perhaps understandable from card-carrying activists, science and academia are at least supposed to enforce better standards. Though it is sometimes buried under hundreds of pages of dense research, consensus climate science synthesized by venues like the IPCC, and consensus climate economics synthesized by venues like the EPA, have offered robust correctives to these apocalyptic activist imaginaries.

In the text of their paper, Bilal and Känzig come right out and say they find these correctives unacceptable, and attempt to “reconcile” activist perceptions with the academic consensus. But in order to upend decades of scholarship, emergent from thousands of studies, they have to rely on conceptually bizarre, poorly-justified economic methods.

Episodes like this one reveal the threats to substantive research posed by activist commitments, and the incentive structure facing journalists and experts within the broader issue public. It shows how the imperative to produce work supportive of a predominant narrative can distort the aggregate output of a field, which then feeds back into the activist perceptions, creating a self-perpetuating cycle. To break the feedback loop and keep science as closely oriented towards truth-seeking as possible, experts must resist the temptation to cater to activist narratives, no matter how powerful the incentives are.