Let's talk about conflicts of interest in climate change research

New papers and how I think it all fits together

My coauthors and I have a couple of papers out recently:

1. Conflicts of Interest, Funding Support, and Author Affiliation in Peer-Reviewed Research on the Relationship between Climate Change and Geophysical Characteristics of Hurricanes. Bulletin of the American Meteorological Society.

2. Conflict of Interest and financial disclosure policies of journals that publish weather and climate research. Accountability in Research.

These papers work together, and I will give highlights below. First, I provide some background.

Acknowledging the Landscape

In 2019, I published Experts, regulatory capture, and the “governor’s dilemma”: The politics of hurricane risk science and insurance in Regulation & Governance.

In that paper, I wrote about the sprawling relationship between the re/insurance industry and researchers in academia and government. It is a common enough relationship to have its own session at a major reinsurance conference.

The crux bringing everyone together is risk modeling which acts as scaffolding that brings everyone together to feed in assumptions and processes. The scale of this technology niche ranges from in-house capability, to major analytics firms, to start-ups and mid-size outfits.

In my 2019 paper, I argued that a lot of the research regulators depended on to understand catastrophic weather risk is shaped by industry involvement in that area of research. Below, is Table 1 from my paper.

These industry and risk modeling activities dovetailed with deepening politicization of climate change science. Some of the biggest public fights we have had about climate change and insurance (separately and together) regard the scientific integrity of underlying assumptions used in risk characterization.

Because much in weather extremes is underdetermined, the kaleidoscopic nature of mining epistemic uncertainty for measurable risk serves business practices in market competition, portfolio design, investor demand, etc. As Swiss Re recently explained, their small army of PhDs “expands the investable universe while maintaining a well-understood risk-return profile.”

Sprinkle in some climate change assumptions, and the imagination is the limit for risk measurements. It helps that mainstream media messaging on weather, climate, and disasters frequently misrepresents scientific observation and honest accounts of the limits of scientific knowledge.

In my opinion, the dynamic has turned many weather and climate researchers into climate quants. The literature is filled with technical documentation (ie. “look what I can do with my model”) masquerading as knowledge advancement.¹

Looking beyond reinsurance these days, financial mobilization of climate change research is nothing short of targeted effort for full scale remake of global finance- literally, not hyperbolically. Here is First Street marketing material in my inbox:

We were founded nearly a decade ago as a nonprofit research organization, built on a single conviction: climate risk is financial risk…As we mark our second anniversary, our focus is clear: embed [Climate Risk Financial Modeling] into core financial decision-making at a global scale.

How does this landscape show up in author conflicts of interest disclosures?

After the paper above, I developed an interest in the literature on researcher conflicts of interest (COI). Much of this literature has its roots in problems originating in the 1980s and 1990s, dealing with the rapid growth of biomedical industry engagement in medical research and heated political controversies around environmental risk regulation.

I read the work of David Resnik, a long time NIH scholar in research ethics, and attended the 2022 AAMC Forum on Conflicts of Interest Professional Development to hear him speak.

Eventually, I reached out to David to see about a collaboration. In spring 2024, we developed a team of researchers, including student interns and an NIH statistician, to study the relationship between researcher COI disclosures, funding sources, and research outcomes.

Our research questions and methodology is standard and not complicated. The novelty is their application to climate change and hurricane research situated in the context of detection and attribution claims.² Peer review took over a year, beginning with internal processes at NIH including the Administration freeze of anything moving out of NIH, and continuing through the diplomatic and thoughtful peer review at the Bulletin of the American Meteorological Society (BAMS).³

Our study found no disclosures of COI. Out of 82 papers and 300+ authors, no one had a single conflict of interest to disclose.

The finding is surprising in the context of the known landscape of how climate change and hurricane research are used in insurance and finance and in comparison with the literature in other disciplines.

We faced some reviewer critiques:

• Climate science does not have the same extent of COI as in other areas like clinical research.
  I agree, but even so, the extent is not zero. We named some examples that suggest there should have been at least a few disclosures within our sample. More research is needed, of course, but I’ve read a lot of climate change and hurricane literature- it is very rare that there is a disclosed COI.

• A finding we had in our submitted draft about author NGO funding suffered from confounding.
  Of the 9 examples if author NGO funding in our dataset, 7 of them included the same author. So we dropped that piece of our analysis. Instead, we noted that the relationships of researchers at the top can have far reaching impacts. Leaders in a field are often sought after for their expertise. This is a hallmark of the leading expert and everything about the research profession encourages these activities. But it is silly and irresponsible to clutch pearls when someone mentions that the situation warrants renewed discussion about COI disclosure practices.

• This is just your opinion, not research.
  This critique originates from ambiguity about COI within the climate science community (a bit more on this below). It is worth noting that non-financial COI is a current, hotly debated issue in research ethics, and there is no clear criteria for non-financial COI. Some believe non-financial COI goes too far; people have a right to their personal activities, and these do not merit disclosure. However, there are activities that are far cry from volunteering for a local chapter of the Humane Society. These include,⁴

  • Being a board member to a political and research related NGO

  • Being a founder and/or director of a political and research related NGO

  • Serving as an advisor to political and research related NGOs

  • Serving as an advisor to an election campaign or partisan organization

  • Advising agencies on regulatory actions using your research/methodologies

  • Contributions to amicus briefs and witness testimony.

My co-author David Resnik has argued that uncertainty about non-financial COI calls for open discussion in professional communities:

the appropriate response to this problem is not to dismiss non-financial interests as an irrelevant distraction, but to define them more clearly and provide effective guidance.

Are authors expected to disclose COI?

If, as in our case, we find no disclosures in a standard study of disclosures and research outcomes, perhaps journals in this field have no disclosure policies or disclosure policies are unclear.

In our follow up study, we found that nearly all journals publishing weather and climate research have disclosure policies.⁵ They vary in detail and clarity, however.

The chart below illustrates correlations between the variables we looked for in the policies. It shows the strongest correlations where journals that define COI also provide examples and also explicitly address non-financial COI.

We also found that our coders disagreed to different extents on some of the variables. Coder disagreement could indicate lack of clarity in the presentation and description of the policies.

One of the journals excluded for coding because it did not have an online disclosure policy provided a statement on publication ethics on a page about the broader family of journals. An author could begin, if not also complete, the submission process without seeing the ethics statement, which, in any case, does not provide an author COI policy.

Perhaps this journal handles its ethics policy this way because the journal only publishes review articles and not original research. But, at least in biomedical research, review or commentary articles face some of the most stringent COI policies and ethics standards because of the weight these articles carry on broader conclusions about the state of knowledge.

I have been told that COI disclosure is not a concern in commentary articles because they are clearly labeled as commentary and not research. This is naïve. I’ve seen big ways in which commentary articles are cited as if they were research articles. It is only the respective journal editors who can consistently keep straight the difference between a commentary, letter, viewpoint, article, etc.

Why do researcher COI disclosure practices not reflect journal policies and the the complex industry and political landscape of the climate research enterprise?

Some possible reasons why disclosure practices have fallen behind the pace of industry engagement and what one might consider the public’s reasonable expectations of disclosure include:

1. Slow creep.
   The relationship between the insurance and financial industries with weather and climate researchers has gone on for a long time. By my estimate, it was a relatively small collegial circle that slowly expanded in the 1990s and started to snowball about 20 years ago alongside professionalization of climate and catastrophe risk modeling.

2. Obliviousness.

   The Ivory Tower is tall enough, and the righteous hard-headed enough, that it is easy to point the finger at others’ industry involvement while not recognizing it in the mirror. My experience is that there are those who believe that non-disclosure of COI in weather and climate research is an open secret and thus, for some, hardly worth mentioning and, for others, really important to address. Then there are those who are aghast that anyone could suggest bringing up the subject in the first place.

3. Confusion.

   It seems to me that researchers clearly understand the need to disclose funding sources but become confused about what constitutes a COI. It is possible that this a product of #1 and #2 above, and the prestige that comes from securing funding and linking papers to grants.

4. Non-disclosure agreements (NDAs).

   It is possible that NDAs prevent researchers from disclosures and restrict broader discussion in the community.

What to do?

To be clear, I do not believe the lack of disclosure in hurricane and climate change research is malicious. I think the main culprit is a lack of clarity around COI disclosure requirements and a lack of responsible, courageous leadership within the learned societies.

In my view, the COI problems facing this research area are akin to the problems facing medical research that arose from increased involvement with biomedical and pharmaceutical industries in the 1980s.

In medicine, there were very unfortunate cases of patient deaths connected to researcher COI and fraud. In climate science, no one dies from intentional or unintentional distortions in projected economic losses. Instead, political discourse is grossly distorted, undermining economic growth and effective policymaking; societal conflict is exacerbated; key science and decision making institutions lose public trust.

I think the disciplinary professional groups should address the disclosure issue. Doing so would demonstrate that they take seriously the lost of public trust in science.

My vibe read: The public is not looking for scientists to be more friendly in bars, chatting it up with the local folk. They want scientists to have real accountability.

I also think Congress should make a fuss of the situation (with sincerity, not for Tik-Tok videos, though that may be wishful thinking).

Interrogating the dark corners of science could be part of a broader discussion about “climate risk,” financial markets, and public struggles with insurance.⁶

1. To be fair it is not just the financial industry; it is a bigger issue.

2. Importantly, our work is not an analysis of the strengths of detection and attributions claims.

3. Peer review has a lot of problems but it is very rewarding when it is thorough and thoughtful.

4. I am assuming these are unpaid. However, they could be paid and therefore would be disclosed as financial COI. They could also be non-financial serving to drum up business for one’s personal consulting. In any case, it’s not playing with puppies on a Sunday afternoon.

5. We looked at journals listed in the Journal Citation Reports (JCR) database operated by the bibliometrics analytic company, Clarivate, and using JCR’s Category tool to filter journals in the “Meteorology and Atmospheric Science.”

6. There is crossover between re/insurance and finance interests, and extreme event attribution and related litigation. You can see the conundrum perhaps:

• re/insurers use climate change assumptions to drive risk estimates → advocates working with the reinsurance and finance market a climate change insurability crisis → reinsurers market products for climate litigation risk → Experts move around all the parts.