The Misallocation of Climate Research Funding

The window of opportunity for mitigating climate change is narrow. Limiting global warming to 1.5 °C will require rapid and deep alteration of attitudes, norms, incentives, and politics. Some of the key climate-change and energy transition puzzles are therefore in the realm of the social sciences. However, these are precisely the fields that receive least funding for climate-related research. This article analyzes a new dataset of research grants from 333 donors around the world spanning 4.3 million awards with a cumulative value of USD 1.3 trillion from 1950 to 2021. Between 1990 and 2018, the natural and technical sciences received 770% more funding than the social sciences for research on issues related to climate change. Only 0.12% of all research funding was spent on the social science of climate mitigation.


Introduction
The natural science of climate change, starting with early discoveries in the nineteenth century and bolstered by large investments over the last three decades, is mature and well established. Thus, 97% of climate scientists agree about the basics of anthropogenic climate change [1], and the International Panel on Climate Change has concluded that it is "extremely likely" that human influence is the dominant cause of ongoing global warming [2].
In tandem with growing knowledge about climate change, a set of technological mitigation options has been widely endorsed, including energy efficiency, wind and solar power, electrification of transport, and reforestation. Moreover, the cost of these solutions is falling rapidly through expanding economies of scale and incremental technological improvements [3]. For instance, from 1975 to 2012, the cost of solar panels fell by over 99%, and since then it has continued to fall [4].
However, one of the most urgent unsolved puzzles is how to get people to act on what they know, that is to say, how to alter society to mitigate climate change [5][6][7]. Because there is a limited carbon budget, the speed of reductions in annual greenhouse gas emissions is also critical [8,9]. Limiting global warming to 1.5°C will require reaching 80% zero-emission energy by 2030 and 100% by 2050 [10]. While the impact of climate change and society's adaptation to it will unfold over decades and centuries, there is only a narrow window of opportunity for mitigation. Mitigation is therefore an urgent priority [11,12].
Despite progress in some areas, ongoing changes are too shallow and too slow to reach such targets. Solar, wind, geothermal, and modern bioenergy combined still make up only 6.7% of the world's total final energy consumption [13]. Meanwhile, in the decade from 2007 to 2017, oil, gas, and coal production grew by 13%, 25% and 8%, respectively and, consequently, CO 2 emissions grew by almost 11% [14]. During the same period, three times more money was spent on oil, gas, and coal facilities than on all forms of renewable energy infrastructure, including hydropower and biofuels [15]. Deforestation and population growth also continue at a high pace [16]. Human habits are difficult to change; doing so requires altering attitudes, norms, incentives, ethics, and politics at the personal, community, and national levels [17]. Therefore, some of the key climatechange puzzles are in the realm of the social sciences broadly defined: anthropology, economics, education, international relations, human geography, development studies, legal studies, media studies, political science, psychology, and sociology [18]. Yet, as we find here, these are precisely the fields that receive least funding for climate research.
Others have made similar points before, but they have lacked comprehensive data to back them up [17,[19][20][21][22]. To make our case, we therefore analyzed a new dataset of research grants from 1950 to 2021 spanning 4.3 million awards with a cumulative budget of USD 1.3 trillion. This includes funding awarded by 332 organizations, mostly national research councils, from 37 countries, including all major member states of the Organization of Economic Cooperation and Development (OECD) as well as Brazil, China, India, and Russia. The data were obtained by mining the new dimensions.ai database (see further information in the appendices). We examined the share of overall research funding that went to research on decarbonization and climaterelated topics, the share of this funding that went to the social sciences, and lastly the share of this funding that went to mitigation-related research.

Estimating funding allocations
There is no straightforward way to identify funding related to climate change research within such a large volume of data, so we developed alternative search strings: a short string with 9 climate-related keywords, such as "climate change" and "global warming", and a long string with 89 keywords, all combined with the Boolean operator "OR" and applied to the titles and summaries of all research grants (see the full search strings in the appendices). By using two search strings, we were able to draw up lower and upper boundaries of the possible ranges of funding granted to different fields of research (see Fig. 1), a more cautious approach than trying to make an exact estimate. The two research strings can also be useful methodological tools for future research.
A limitation of our dataset is that it only covers competitive research grants. Much researching funding, for example in China, France, and Germany, is still distributed in the form of basic grants and other noncompetitive allocations where it can be difficult to know what research topics the funding was spent on. This limitation of our data should be acknowledged, while emphasizing that our aim is to map the prioritization of funding that is purposively allocated to climate research. Such funding reflects the intentions and priorities of policymakers and may be better than non-competitive funding for supporting policy-relevant and dynamic research. Furthermore, as noted in the literature, competitive research funding is a powerful tool for influencing the general research agenda [23,24].

The paucity of social science
Our data support several findings. The first is that hardly any social science research was conducted on climate change before 1990. We therefore truncated the data pre-1990 for the rest of our analysis.
The second observation is how little funding has gone into research on climate change overall since 1990, regardless of discipline. Depending on which search string one uses, climate research accounted for between 2.38 and 4.59% of the total amount of research funding during the period from 1990 to 2018. The higher estimate errs on the high side: very few projects that are really about climate change would not include any of the 89 keywords in the long search string, whereas numerous projects that happen to mention one of those words may not really be about climate change.
Third, out of the funding for climate research, the social sciences received a small share (see Fig. 1). From 1990 to 2018, the natural and physical sciences received a total of USD 40 billion compared to only USD 4.6 billion for the social sciences and humanities (based on the means of the short and long search string results). In other words, according to our estimates, the natural and technical sciences received around 770% more funding than the social sciences and humanities for research on climate change. Furthermore, the countries that spent the most on social science climate research in absolute terms according to Table 1-the UK, the USA, and Germany-in fact spent between 500% and 1200% more on climate research in the natural and technical sciences (based on the long search string).
However, even these numbers do not tell the whole story. Within the social sciences, there is also much research that is climate-related but not about climate change mitigation, for example research on adaptation to climate change, how to manage extreme weather events and recover from disasters, or the effects of past climate change on ancient civilizations. While this research is valuable, it does not tackle head-on the most urgent question: how to change society to mitigate climate change right now.
To determine how much social science research is specifically about the mitigation of climate change, we drew a random sample of 1500 climate change-related social science grants from our data using the short search string and assessed each of them. This led to our fourth and most important observation: a mere USD 393 million of funding went to social science research on the mitigation of climate change, equivalent to 5.21% of all funding for climate change research and 0.12% of all research funding.

The need to balance natural and social science research
Natural and technical climate-related research is important. There is still a need to better understand the physical causes, trajectory, and impact of climate change, as well as the technological means of mitigation. However, there is a striking imbalance between the growing knowledge about climate change and mitigation technologies and the failure to mobilize people to contribute to mitigation efforts. This indicates that research resources are not distributed optimally.
One might argue that the natural sciences need more funding because they employ more people or require more expensive equipment and materials. However, such arguments easily become circular. The numbers of researchers in different fields is as much a consequence as a cause of the availability of funding and there could simply be more high-cost research projects in the natural sciences because more funding is available for them. It would also be possible to spend large amounts of funding on social science research, for example nationally representative surveys of large numbers of countries, large-scale multilocation field experiments, the design and monitoring of living laboratories, or human coding of large volumes of text or video as a basis for machine-learning. It is therefore difficult to argue that the natural sciences are inherently more expensive. In any case, in our data there is not a significant difference between the average size of climate research projects in the natural and social sciences; in fact, the social science projects tend to be slightly larger.
One might also argue that the social sciences get less funding because they come up with fewer interesting ideas and solutions. But many social science ideas and solutions related to the mitigation of climate change have already been put forth, such as climate clubs, carbon taxes, or grassroots mobilization [25,26]. The question is whether sufficient research funding is available to develop these and other ideas properly.
The prioritization of natural science could also be related to a perceived need to overcome climate skepticism by proving that climate change is due to human greenhouse gas emissions. However, currently, climate skepticism has almost no voice in the scientific community [20] and even fossil fuel companies acknowledge anthropogenic climate change. There remains significant climate skepticism among laypeople, including prominent politicians; however, this is not a natural science problem but one of communication, vested interests, and politics-again the realm of the social sciences.

Solutions for advancing social science
Once one realizes how little funding is spent on the social science of climate mitigation, and the related social science side of energy studies, the question arises as to how the situation can be improved. Our main answer to this question is to spread awareness of how little funding is actually going into this field of research, and to contrast it with its urgency.
While our data and analysis cannot explain why funding is distributed the way it is, or exactly how it should be distributed, they still support some simple but important policy lessons which we present in the next subsections.

Funding for climate mitigation needs to match the magnitude of the threat
Funding agencies need to better secure and prioritize funding for climate change mitigation, across all disciplines. Global annual damages from climate change have already surpassed USD 10 to 40 billion from storm surge alone, and it could surpass USD 100 trillion over the next 80 years [27]. Funding for research on climate mitigation should be increased to address the magnitude of this threat and take into account the narrow window of opportunity for dealing with it.
Such research efforts cannot necessarily be guaranteed to reduce or contain the extent or distribution of climate change impacts, and we also fully appreciate that the magnitude of required research investment is almost unparalleled. By comparison, the entire cost of the United States space shuttle program, up until 2011 was estimated to cost USD 196 billion [28,29]. But individual research programs have been known to reach into the billions of dollars annually, with the United States federal government spending USD 34.8 billion per year on HIV/AIDS research and treatment in 2019 [30]. If similar efforts were invested into energy and climate social science, they could yield substantial dividends worldwide. A first important step could be a rigorous funding gaps and scoping analysis to determine precisely how much funding is needed, and for which challenges, themes, or problems.

Improved funding transparency and coordination
There is a need for better global coordination and oversight of funding for climate research. Our data provide an unprecedented overview of funding for climate research, yet they cover only a fraction of global research funding, much of which is distributed through noncompetitive base grants for universities. The lack of oversight can cause significant overlaps in funding in some research areas, while other areas are neglected.
As a concrete fix to this problem, more research financing organizations need to make their portfolios available online with standardized tags for such things as project title, summary, and discipline. Better oversight could be facilitated by the United Nations Framework Convention on Climate Change, or United Nations Educational, Scientific and Cultural Organization, or a coalition of the willing, and could help increase the efficiency of the climate research effort. Some countries, especially those that have been critical of recent IPCC reports, such as Russia and Saudi Arabia, might not be willing to join such an effort, but such actors tend not to fund large sums of energy and climate mitigation research anyway, so their exclusion would not necessarily thwart progress.
Greater transparency of global research funding would give researchers and policymakers a better understanding of what is in the pipeline and help them efficiently allocate time and funding. It could reduce redundancy and serve as a mechanism for research teams to identify synergies and possible collaborators.

More rigorous social science research
While more funding is needed for social research on climate change, the social sciences also need to rise to the challenge. Firstly, social scientists need to do a better job of ensuring rigor and validity in their research. In their survey of the field of sustainability, for instance, Brandt et al. noted that methods were often chosen based on familiarity or specialization of the researchers involved, rather than their suitability for a given research question [31]. Moreover, in an examination of 15 years of energy research (1999-2013), it was found that almost one-third (29%) of 4,444 studies examined had no research design-or method-whatsoever [32]. Hamilton et al. similarly note that in the domain of energy efficiency and buildings, "analysis is often limited to small datasets and results are not applicable more broadly due to an absence of context or baselines" [33].
Secondly, some social science research is wishy-washy, lacking an understanding of the natural sciences and the physical world [34]. Some is caught up in obscure theoretical debates-one assessment identified no less than 96 theories deemed relevant to the fairly narrow topic of the social acceptance of new technologies [35]. Much social science deals with very small groups of people or sample sizes that are difficult to generalize from [36], and that may not be of much relevance for the large-scale mitigation of climate change.
Universities or the research councils often funding them could require remedial training in methods for all social science researchers and also mandate that such training be continuous, similar to what the legal profession does with its Continuing Legal Education (CLE) requirements. According to CLE requirements, all practicing attorneys must maintain their professional certification on a continual basis even after they pass the bar.
Fixing the weaknesses of the social sciences will not be done in a day, but it is nonetheless important to start this work so that they can strengthen their real contribution to reducing greenhouse gas emissions if more funding becomes available.

Better alignment with emissions sources and trends
Also within the social sciences themselves, there is a failure to prioritize truly problem-solving research on the most burning mitigation issues. Some of the funding for climate change-related social science research follows the thematic logic of natural science funding, which does not necessarily fit the social sciences.
For example, there has been a significant amount of climate-related social science research on the Arctic [37]. For climate research in the natural sciences, the polar regions are key, both as the world's "thermometer" and because they are the locus of much of the ice melting that drives sea level rise. For the social science of climate mitigation, the poles are less important because that is not where most emissions come from nor where carbon sinks are located.
Attempts to change the priorities of social science research funding will likely encounter resistance from some entrenched interests, but the academic community has already been fairly progressive at promoting gender diversity in research (fighting patriarchy), highlighting the value of trans-disciplinary research designs (fighting dogmatism and elitism), or arguing in favor of open access publishing regimes (fighting restrictions on information from publishers). This creates a series of precedents for challenging incumbent ways of thinking.

Do not lose sight of climate change as a global challenge
Although global solutions obviously also depend on understanding the microlevel, it is surprising how little social science research goes straight for the really big issues. Will the Paris Agreement work? What are the concrete suggestions for an alternative and more binding global solution? How could households be convinced to adopt low-carbon lifestyles? How can decarbonization be promoted across cultures and market economies as diverse as China, Russia, Saudi Arabia, Singapore, and the United Kingdom?
Part of the solution could be to organize future research efforts not around disciplines, but around urgent puzzles, which are themselves linked to pressing social challenges related to climate change mitigation and energy systems. This challenges-based approach to research has been relatively successful in other domains, notably national defense (the Defense Advanced Research Projects Agency, or DARPA) [38] and business (Mission Innovation) [39].
However, the problem, challenge, or mission-based approach is only just emerging as a platform to organizing energy and climate research. One example is the Global Challenges Research Fund in the United Kingdom, which asked "How can sustainable development be achieved for all while addressing global climate change?" The European Commission's Horizon 2020 framework program also structured its research agenda around questions such as "How can Europe achieve a resource, water efficient and climate change resilient economy and society?" and "In what way does social innovation contribute to making energy more secure, sustainable and affordable?" Putting research into the context of challenging questions in this manner can promote focused but interdisciplinary social science work and is an approach that could be replicated by other national, regional, and global funding bodies. One reason why there are not more such calls may be entrenched disciplinary divides, anchored in organizational structures. These will need to be tackled directly by leaders within universities-presidents, provosts, deans, vice deans, pro-vice chancellors, faculty senate members, department chairs, and tenure and promotion committees.

Conclusion
The funding of climate research appears to be based on the assumption that if natural scientists work out the causes, impacts, and technological remedies of climate change, then politicians, officials, and citizens will spontaneously change their behavior to tackle the problem. The past decades have shown that this assumption does not hold.
Although the natural and technical sciences often generate results that are, or are perceived to be, clearer and more concrete than the social sciences, they cannot handle issue areas-such as attitudes, norms, incentives, and politics-that are intrinsically social. The solutions are to make more funding available for social science research on climate mitigation; improve global research funding coordination and transparency; prioritize and align key questions within the social sciences and increase the rigorousness of social science research. Framing climate change more as a global social challenge that cuts across disciplines will expand the scope of research, its ability to offer critical insights, and its social legitimacy among a broader base of stakeholders.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper, apart from being social scientists and therefore having an interest in increased funding for social science. For a full overview of fields, see http://www.abs.gov.au/ausstats/abs@.nsf/Latestproducts/1297.0Contents12008?opendocument&tabname= Summary&prodno=1297.0&issue=2008&num=&view= In our research, all fields of research up to and including "Other built environment and design" (ANZSRC code 1299) were counted as natural and technical sciences, the rest as social sciences and humanities.

B. Handling of random sample and definition of mitigation
A random sample was drawn of 1500 social science climate change projects to identify which projects were about climate change mitigation, and which were about other things. The following definitions were applied: Possible mitigation projects were found by reading through all titles and abstracts in the random sample as well as by carrying out searches for the terms "mitigat*", "reduction", "reduce", "limit", "curb", "abate", "emissions", "decarbon*".
Projects were allowed to have multiple / overlapping classifications, for example they could be classified as concerning both mitigation and adaptation.
We operated with two levels of certainty about whether projects concerned mitigation: "Mitigation" and "Maybe mitigation". This fuzzy logic element enabled us to handle the ambivalence of some projects and ensured that the results were as balanced as possible. Both categories were included in the final count of social science mitigation grants for the article.
We did not assess whether we thought projects were good mitigation projects or not (e.g. wood pellets), just whether the people carrying out the projects present them as somehow contributing to mitigation of climate change.
Projects were not counted as mitigation projects if: • They aimed at general enlightenment / education on climate change issues. Although enlightening people about the mechanisms behind climate change can lay the basis for mobilizing them to contribute to mitigation, it is not the same as working for mitigation per se.
• Mitigation was a small part of the project (less than 1/3 according to the assessment of the person doing coding). This also means that if research projects just seemed to be 50% about mitigation, they were counted as mitigation projects. This is one of several methodological choices that stack the data against our own arguments.
Projects on the following topics were classified as mitigation projects to ensure that our "mitigation" category was broad enough to capture all possible mitigation projects and again to stack the data against our own arguments: climate justice, a just energy transition, the consequences of mitigation, the financial consequences of mitigation, co-benefits of mitigation After a pilot run of 300 projects categorized by the lead author, the rest of the random sample of 1500 was categorized by two research assistants. Projects they were in doubt about were discussed in plenary sessions.

C. Search string development
The purpose of the search strings was to capture all research projects related to climate change in the database. If one simply searches for "climate change" one will miss many projects focused on narrow climate change sub-topics We harvested possible keywords from several sources: • word frequency analysis of IPCC reports • climate vocabularies and dictionaries: ▪ https://www.bbc.com/news/science-environment-11833685 ▪ https://en.wikipedia.org/wiki/Glossary_of_climate_change ▪ https://climatechange.ucdavis.edu/science/climate-change-definitions/ ▪ https://www.slhd.nsw.gov.au/concord/sustainability/content/pdf/climatechangeglossary.pdf Each keyword was pre-tested separately and the most reliable ones were included in our search strings.
To be on the safe side, we developed two search strings: a short one with a small number of safe terms that are clearly relevant for climate change and neutral vis-à-vis social and natural sciences, and a long, comprehensive one to capture the broader range of projects including fields that are not directly about climate change, but directly relevant for it.
We sought to balance the number of keywords related to the natural and social sciences, to avoid biasing our results. The long search string is helpful in this regard as it is so comprehensive that there are very few climate-related projects of any kind that evade it.
The long search string includes both more words related to climate change and words related to other topics that are highly relevant for climate change, for example "renewable energy". This is because climate change is the main driver for the development of renewable energy and cutting GHG emissions by changing energy production and consumption is one of the main ways to mitigate climate change. As we are particularly interested in mitigation in our analysis, it makes sense to include such key mitigation components in the long search string.
As natural science is the starting point and foundation for concern over climate change, many natural science terms are also used in descriptions of social science projects (but we still classify those projects as social science). There are also many words that occur in both natural science and social research. Thus, there is a considerable overlap between the vocabularies, which helps reduce the risk of bias somewhat.
An advantage of the long string is that each word becomes less decisive, as there are so many other words and many of them will occur together in a given project description. Thus, the difference in search results due to addition or removal of one word is small.

D. Short search string
"climate change" OR "climate mitigation" OR "climate adaptation" OR "global warming" OR "greenhouse effect" OR "greenhouse gas" OR "GHG" OR "CO2 emissions" OR "climate policy" E. Long search string "climate change" OR "climate mitigation" OR "climate adaptation" OR "global warming" OR "greenhouse effect" OR "greenhouse gas" OR "GHG" OR "CO2 emissions" OR "decarbonization" OR "decarbonization" OR "climate policy" OR "UNFCCC" OR "United Nations Framework Convention on Climate Change" OR "Intergovernmental Panel on Climate Change" OR "IPCC" OR "Kyoto Protocol" OR "Paris Agreement" OR "nationally determined contribution" OR "INDC" OR "Bali roadmap" OR "climate negotiation" OR "climate action" OR "climate justice" OR "climate ethics" OR "climate skeptic" OR "climate sceptic" OR "climate denial" OR "climate denier" OR "climate migration" OR "climate refugees" OR "cap and trade" OR "emissions trading" OR "carbon finance" OR "carbon credit" OR "carbon tax" OR "carbon market" OR "carbon bubble" OR "CO2 equivalent" OR "carbon sequestration" OR "geological sequestration" OR "carbon capture and storage" OR "carbon sink" OR "radiative forcing" OR "climate feedback" OR "sea level rise" OR "anthropogenic aerosols" OR "carbon footprint" OR "carbon offset" OR "carbon neutral" OR "carbon intensity" OR "carbon price" OR "mitigation potential" OR "climate feedback" OR "climate model" OR "ocean acidification" OR "carbon cycle" OR "climate feedback" OR "climate sensitivity" OR "climate model" OR "carbon uptake" OR "CO2 concentration" OR "coral bleaching" OR "Greenland ice sheet" OR "Arctic sea ice" OR "ice core" OR "ice loss" OR "geoengineering" OR "renewable energy" OR "renewables" OR "wind turbine" OR "solar power" OR "geothermal energy" OR "landfill gas" OR "biofuel" OR "bioenergy" OR "tidal power" OR "solar power" OR "photovoltaic" OR "heat pump" OR "distributed generation" OR "passive house" OR "smart grid" OR "smart energy" OR "microgrid" OR "feed-in tariff" OR "grid storage" OR "demand response" OR "electric vehicle" OR "electric mobility"

G. Categorization of fields of research as natural or social sciences
The dimensions.ai database applies the ANZSCR classification system for fields of research-because it is suitable to the machine learning approach that dimensions.ai uses to classify research projects. We divided the ANZSCR fields into natural and technical sciences on the one hand, and social sciences and humanities on the other, as shown in the following table below. For simplicity, we just refer to natural sciences and social sciences most of the time, subsuming technical sciences and humanities under them.