Abstract
Geoengineering technologies – deliberate, large-scale interventions to alter the Earth's climate – present an opportunity to ameliorate the effects of climate change; an opportunity policy-makers are beginning to consider. However, the safety and efficacy of geoengineering strategies is uncertain, and there is concern advancing these technologies engenders a range of non-physical risks; in particular, moral hazard. In economics, moral hazard occurs when insured agents no longer bear the full consequences of risk-taking, and consequentially increase their exposure to risk. However, while certainly analogous, the way the term is used in the geoengineering literature is ambiguous, describing a patchwork of mechanisms of action, hazardous behaviours, and undesirable outcomes. Importantly, as moral hazard concerns are likely to inform policy-making and regulatory responses to these technologies, this unclear specification could impede scholarly, policy, and public debate. This article charts this ambiguity, documenting the range of meanings of moral hazard in the geoengineering literature. Results suggest moral hazard is used at least eight different ways in the literature, with the three most common frames being Insurance, Unwilling and Avoid. With this, we suggest that those articulating moral hazard concerns about geoengineering technologies work to articulate as clearly as possible what the problem actually is.
Keywords
Introduction
In April 2020, news media reported on an experiment taking place on the Great Barrier Reef, Australia: scientists were spraying seawater into the atmosphere to brighten the clouds above the reef. 1 The hypothesis was that brighter clouds would deflect the sun more effectively, cooling the reef and offering a reprieve from heat-induced coral bleaching. However, when interviewed, project head Dr Daniel Harrison stressed that cloud brightening was not a solution to coral bleaching, but a strategy to buy time. Reducing emissions and tackling climate change, he highlighted, was the only way to save the Reef. 2
Harrison's comments capture a persistent concern that technological innovations to address climate change, like cloud brightening, might diminish efforts to mitigate climate change. This concern is commonly called a ‘moral hazard’.
Using this insurance concept as inspiration, Keith reframed this problem as a way to consider the non-physical risks associated with the broad suite of emerging and potential technologies designed to deliberately alter the Earth's climate, collectively termed geoengineering: Knowledge of geoengineering has been characterized as an insurance strategy; in analogy with the moral hazard posed by collective insurance schemes, which encourage behaviour that is individually advantageous but not socially optimal, we may ascribe an analogous hazard to geoengineering if it encourages suboptimal investment in mitigation.
5
The concern that knowledge of geoengineering in the public consciousness will undermine climate change mitigation efforts is at the heart of the concern about moral hazard, and has remained one of the longest standing and most persistent criticisms surrounding the development of geoengineering technologies. Yet importantly, as geoengineering technologies loom ever more in the policy-making agenda,
6
it is becoming ever more pressing that we understand – and potentially work to address – these concerns. Quite simply – as Jebari and colleagues stress – the current debates “are both unhelpful and potentially counterproductive”: … if the IPCC report is to be taken seriously, such questions are ones we no longer have the luxury to consider: the only live question is which combination of available mitigation, adaptation, and technological strategies is most likely to ensure a sustainable, climate-resilient future.
7
However, pinning down what the term ‘moral hazard’ is used to denote in the academic geoengineering literature is fraught. The term is impeded by ambiguous use and misapplications.
8
As a telling example, we can compare Keith's definition above with the conception of Aldy and Zeckhauser: Moral hazard concerns have been a major second component of that hostility: If you tell citizens that there may be some technological corrective for climate change, they will be less willing to curb emissions.
9
Or Fabre and Wagner's: Geoengineering (G) introduces an entirely new set of challenges into global climate policy negotiations. Some have focused on G crowding out ambitious mitigation, necessitating complex governance regimes… This often comes under the heading of “moral hazard”.
10
Although both latter definitions cite Keith, 11 the conception of moral hazard articulated appears to have shifted from the ‘encouragement of suboptimal investment in mitigation’ Keith originally described to citizens being ‘less willing to curb emissions’ or ‘crowding out ambitious mitigation’. On the surface, all are concerned about the same possible outcome – that geoengineering technology could negatively affect other efforts to address climate change. However, each description identifies a distinct phenomenon and a different hazardous behaviour. Although all claim to describe ‘moral hazard’, each points to a different problem.
As geoengineering research and development continues – and as regulatory and policy responses begin to be considered – disambiguating the meaning of moral hazard is increasingly necessary. Without a clear formulation of the term and a clear understanding of the problems it is used to point to, concerns about moral hazard, like those above, may be conflated, sidelined, prematurely dismissed, or subject to straw-manning and exaggeration. 12 Further, to the extent we may want to practically address moral hazard, or measure it empirically – or incorporate such critiques in regulatory approaches to geoengineering technologies – a clear problem definition is essential. 13 The lack of clarity around what moral hazard means, what behaviour it refers to, and what phenomenon it describes, confounds the results of empirical research determining its impact. Importantly, this ambiguity may frustrate public debate and policy making on the issue, an undertaking which already promises to be challenging. 14
This study seeks to address this gap, using qualitative content analysis to categorise the range of ways ‘moral hazard’ is used in the geoengineering academic literature, informed by existing literature endeavouring to disentangle the term. We hope that clarity regarding the types of non-physical – moral hazard – risks raised in the geoengineering literature can point to better ways to address these risks in regulatory approaches.
Literature review
As it becomes increasingly unlikely global warming will be contained below two degrees Celsius, 15 the imperative for geoengineering is growing. The term geoengineering has been used to broadly describe deliberate, large-scale interventions to alter the Earth's climate and ameliorate the effects of climate change. 16 Geoengineering technologies could complement current strategies to reduce emissions by providing pathways to draw down emissions or, more controversially, ameliorate the intensity of solar radiation. Here it should be noted that the term geoengineering has been deployed in a variety of ways in both academic and wider discussion (not always with precise definition), and overlaps in potentially ambiguous ways with a suite of other related terms (and technologies), including carbon dioxide removal, solar radiation management, climate engineering and negative emissions technologies. 17
Many of these technologies remain in their scientific infancy, have not been tested in large, outdoor settings, and possess significant uncertainties. At present no geoengineering technology can be reliably deployed at scale. 18
Geoengineering involves numerous physical hazards, which are explored elsewhere in the literature. Here we focus on moral hazard, a prominent, non-physical hazard discussed in the geoengineering literature.
Moral hazard
One of the longest-standing objections to geoengineering is its potential to undermine mitigation and other strategies to address climate change. This concern is a central part of the critique known as moral hazard.
The term moral hazard comes from the economics and insurance literature. In this, moral hazard describes a change in an agent's incentives when obtaining insurance. Commonly, it describes the problem where acquiring insurance reduces the incentives of the insured party to be risk-averse, because they no longer bear the full cost of the risk-taking. 19 Moral hazard has been used to describe behaviour in a wide variety of contexts, including health insurance, workers’ compensation, land planning, and economic policy. 20 A prominent example was concern the bailout of Wall Street firms in 2008 created a moral hazard. By designating these firms ‘too big to fail’, the US government effectively insured them against losses, a protection that incentivised the firms to continue irresponsible and risky lending practices. 21
In this light, moral hazard is characterised by the presence of two parties, the insurer and the insured, with divergent interests. 22 Their interests diverge because the insured party, no longer bearing the full consequences of their actions, has an incentive to take on more risk at the cost of their insurer. 23 For instance, when irresponsible (but profitable) lending practices led to collapses on Wall Street, it was the taxpayers who bore the cost of the bailout. Moral hazard is generally characterised as a market failure that results in inefficiencies and, for insurance companies, lost revenue. 24 Hence, labelling something a moral hazard tends not to confer a normative or ethical judgement about the behaviour. By virtue of moral hazard not necessarily entailing ‘immoral’ behaviour, a clear conception of the term should specify what the hazard to be avoided is.
Ambiguity of moral hazard
Moral hazard is a pervasive argument in geoengineering, to the point where it is credited with hampering the scientific advancement of geoengineering technologies. 25 Given the influence of moral hazard, it is perhaps concerning that multiple commentators have argued it is unclear and potentially misleading.
Some have noted, for example, that the term acts as a catch-all for many non-equivalent behaviours, 26 and conflates otherwise distinct concepts. 27 It may describe an increase in risk prone behaviour, political stagnation, the diversion of mitigation efforts, or the worsening of coordination problems. 28 There is also concern that the term is used in ways which may be technically incorrect but imply behaviour not conventionally associated with the term. 29
There have been efforts in the literature to disambiguate uses of the term, with responses differing depending on the perceived source of ambiguity. Three critiques of the term are offered below, with proposals for alternative conceptions of the problem in geoengineering. Below we build on these alternative conceptions.
Lin argues moral hazard is used in the geoengineering literature in ways inconsistent with the technical definition used in insurance. 30 This is because the relevant actors in geoengineering make it disanalogous with insurance, due to an overlap between the identity of the insured and the insurer. In geoengineering, the ‘insurer’ is likely to be government, which might sponsor or regulate geoengineering, while the ‘insured’ are taxpayers, the voting public, or policymakers, whose proclivity to mitigate emissions is influenced by geoengineering. In this case, the identity and interests of these two groups are likely to overlap, potentially significantly, rather than diverge as they do in the conventional insurance case.
Lockley and Coffman argue that moral hazard has been misapplied in the case of geoengineering. 31 They situate the problem in the insurance literature's distinction between moral hazard and morale hazard. Morale hazard is associated with indifference to risk once insured, the result of a lack of due diligence or care. Comparatively, moral hazard occurs when a party has a positive incentive to take on more risk once insured. Taking an example from insurance, this would be the difference between exercising less vigilance in replacing the batteries in smoke alarms after taking out fire insurance (morale hazard); and deliberately not changing (or perhaps removing) the batteries, in the hope it will result in a fire and subsequently, an insurance claim (moral hazard). 32 In geoengineering, this distinction can be used to understand the rationale and ideology of those who change their behaviour in response to geoengineering. 33 A group vulnerable to morale hazard may be those who take climate change seriously and have low investment in the fossil fuel economy; they may view geoengineering as a suitable complement to mitigation, but perhaps place undue faith in the technology or fail to appraise the risks, and put-off mitigation as a result. Comparatively, a group vulnerable to moral hazard are those with high investment in the fossil fuel economy, who advocate for geoengineering as an alternative to mitigation. To the extent that they overstate the benefits of geoengineering to advance their interests, and intentionally fail to mitigate emissions at the expense of the global commons and community which has ‘insured’ them, they are prone to moral hazard.
Finally, Hale argues moral hazard in geoengineering is ambiguous because it is often unclear what the hazardous wrong being described actually is. 34 Concerningly, it is frequently unclear whether the wrong is the result of moral hazard, or a consequence of geoengineering. Hale points out that many are concerned the introduction of geoengineering will cause the diminution of mitigation, and their concern seems to persist even if geoengineering turns out to be safe and effective. Failing to sufficiently demonstrate that safe, effective geoengineering is immoral, it seems the objection is to creating a moral hazard, when the main problem ought to be a subsequent bad outcome, or the execution of unethical acts. Hale argues vigilance about what we mean by ‘moral’ in moral hazard is key to disambiguating the term. It is not enough to assume moral hazard entails some ethically undesirable behaviour; the perpetrated wrong or harm done should be specified.
In sum, robust criticism exists in the literature regarding the ambiguity of the moral hazard presented by geoengineering. Work has been done to delineate distinct concerns and introduce peripheral concepts to disambiguate moral hazard. However, to date these reformulations of moral hazard remain conceptual, and have not been tested against the uses of the term in the broader geoengineering literature. Accordingly, this study proposes to document the range of different ways the term ‘moral hazard’ has been employed in the academic literature on geoengineering.
Methods
The aim of this study is to analyse the uses and meanings of ‘moral hazard’ in the academic geoengineering literature. This is done with a view to differentiate the characteristics of ‘moral hazard’ in the geoengineering literature, drawing on and comparing these findings with existing work to disambiguate the term.
Sample selection
The sample consisted of research articles, including theoretical work, observational studies such as meta-analyses, experimental studies such as participant research, as well as synthesis studies such as review papers. 35 This breadth of articles was chosen to analyse the widest possible usage of moral hazard in the geoengineering academic literature. Searching was conducted in databases Web of Science and Scopus, identified as having high coverage of social science research. 36
The development of the search string began with scoping, where articles about geoengineering and moral hazard were identified through simple searches, such as ‘geoengineering AND moral hazard’. Through these simple searches, combined with relevant papers identified in the literature review, a sample of 34 papers were identified. These papers, henceforth internal-check papers, were used to develop the keyword string and test the validity of the search results.
Due to the use of a range of terms to describe geoengineering technologies,
37
seed papers were used to identify keywords. Seed papers were selected from the internal-check papers and comprised every article with more than 30 citations. The search string was then expanded to include synonyms for the key words, as well as instances of specific technologies: moral hazard* AND (“climate engineering” OR geoengineering* OR geo-engineering OR weather modification* OR “earth systems engineering” OR “solar radiation management” OR carbon capture* OR “carbon dioxide removal” OR “negative emissions” OR “greenhouse gas removal” OR “solar geoengineering” OR stratospheric aerosol* OR cdr OR srm OR sai OR beccs
The sample obtained from the search string (n = 264) was exported to EndNote X8, and the ‘Delete Duplicates’ function was used to identify, manually check and remove duplicates (n = 152, with duplicates removed).
Each article was then manually checked at the title and abstract level to ensure it was relevant to the sample. A full text search was performed to ensure the paper also contained at least one instance of ‘moral hazard’.
This brought the total number of unique, eligible articles to 99. All inconsistent articles identified were then added to the full sample for analysis, bringing the total population for this analysis to 109 articles. A sub-sample of fifty papers was randomly selected (using Microsoft Excel's random function) for analysis (Figure 1).
Summary of method for sample identification. Diagram format adapted from (Haddaway, Macura, Whaley, & Pullin, 2018).
Sampling and segmenting
The data for analysis was then collected from the selected fifty papers. Collected were author name/s, year of publication, journal of publication, geographical location (based on primary author) and study type (categorised as discussion, experimental, observational, or synthesis).
The sample papers were then segmented for analysis. Every mention of moral hazard, excluding in the bibliographies and footnotes, was copied into a spreadsheet alongside the page number and section of the paper (e.g. Introduction). Segments were taken at the level of short paragraphs or, where paragraphs were long, sentence clusters, where content relevant to the discussion of moral hazard was in adjacent sentences. In instances where adjacent sentences did not add to the discussion of moral hazard, single sentences were drawn from the text. This was done to ensure contextual elements important to the articulation of moral hazard, or definitions and examples which spanned multiple sentences, were not excluded. Sentence clusters with multiple distinct ideas were further segmented into smaller clusters or single sentences, to meet the requirement of mutual exclusivity when conducting qualitative coding.
The segment with the clearest unique articulation of moral hazard in each article was identified. This definition was typically signposted, prefaced with indicators such as ‘Moral hazard is…’, or succeeded by ‘…this could be interpreted as moral hazard’. Where multiple, similar definitions were present in the paper, the best definition was chosen with reference to its length (longer definitions with explanation were preferred over cursory ones); location in the paper (definitions in the introduction or literature review were preferred over those in the abstract, results or conclusion); original material (the authors’ articulation was preferred over quotes from other works); and relevance (definitions made relevant to geoengineering were preferred over those generalised to the fields of economics or insurance). Segments which repeated the best definition, such as truncated definitions in the abstract, were excluded from analysis. Where multiple, non-similar articulations of moral hazard were present in the same article, more than one definition from that paper was included in the analysis.
Segments containing discussion of moral hazard, which did not include an articulation of the meaning of the term, were excluded from analysis.
Qualitative content analysis
Qualitative Content Analysis (QCA) was used to formulate thematic categories and code the sample. QCA is a method used to systematically describe meaning. 38 The meaning of textual content was interpreted with reference to the research question and reduced to categories formulated through an iterative process. 39
Building the coding frame
In this analysis, the coding frame consisted of main categories, which were informed by the research question, and subcategories, which reflected what the text said in more detail. The categories were mutually exclusive. 40 Categories were both concept-driven, defined tentatively using pre-existing research, and data-driven, in that they were extended and refined based on pilot testing.
Tentative categories were first described deductively, with reference to the Directed Approach to QCA. The main categories were derived from the three core papers discussed in the literature review, 41 based on their work to disambiguate moral hazard. The reformulations of moral hazard they outline, as well as their criticisms of the term, were transformed into main categories and pilot tested to see if they were applicable to a test sample of twelve papers. These categories were further refined over five rounds of testing, where it became clear this approach was deficient. The main categories captured too much detail, making the coding process difficult and time-consuming, while simultaneously failing to capture the broader picture about how the term was being used.
An inductive approach was adopted to further refine main categories. Broadly similar definitions were clustered to identify common elements. For this process, the pilot sample was expanded to 28 papers. The main categories were then reformulated and simplified in line with these findings.
The main coding frame is elaborated in the results section below. Three ancillary codes were created for segments which didn't clearly fit with the main coding frame. These are:
Other, where an articulation of moral hazard was given that could not be identified with any of the existing categories. Unclear, where an articulation of moral hazard was not given and could not be readily derived from the contextual content. Multiple, where an articulation of moral hazard was given that was identified with more than one category. This category ensured the requirement of mutual exclusivity of categories was maintained.
Results
Bibliographic and geographic data
Bibliographic data, including study type, year of publication, and journal of publication, was collected for the coded sample (n = 50). The dominant study type (see Figure 2) were discussion studies, making up almost half of the sample (23 of 50). The second largest category was experimental papers (14 of 50), all of which were participant research.
Summary chart showing paper type (total n = 50).
The earliest paper was published 2009, and the latest was published in 2020 (see Figure 3). Approximately 70% of articles were published between 2017 and 2020. This accords with the trend observed in research on negative emissions, which has seen significant increases in publications between 1991 and 2016. 42 Articles were collected from 41 journals; Earth's Future, Climatic Change, and Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, had the highest number of articles, with just three each. This suggests no journal is dominant in publishing on the geoengineering moral hazard.
Trend in number of publications by year (total n = 50). *2020 an incomplete record.
The geographic data suggests the USA dominates the study of moral hazard in geoengineering in this sample (see Figure 4). Almost half (23 out of 50) of primary authors were from institutions in the USA, followed by the UK (9) and Germany (7). Only one article had a primary author from an institution in the Global South (India). This reflects criticisms that the field of geoengineering is dominated by voices from the Global North. 43
Number of publications by region, representing the country of the primary author's institution (total n = 50).
Summary of main codes
From 50 articles, 354 segments containing mention or discussion of ‘moral hazard’ were identified. Of these 354 segments, 282 were excluded from analysis and assigned a reason for exclusion (study detail, beyond scope, other definition, or shortened definition, as detailed in the methods). The remaining 72 segments were coded in accordance with the coding frame. Of these segments, 19 were assigned ancillary codes (other, unclear, or multiple, as detailed in the methods). The remaining 53 segments were coded into one of eight main categories, or three mutually exclusive subcategories, as detailed below.
Insurance
The insurance code captures the fear having geoengineering as a safeguard will lead agents to diminish mitigation efforts. Typically: if geoengineering insures against climate change, then mitigation efforts will diminish. A range of terms are used to proxy for insurance, including safeguard or Plan B, to articulate the concern that having, or perceiving, geoengineering to be a back-up plan will undermine or supplant the main plan.
For example, Corner and Pidgeon state: In the context of geoengineering, the moral hazard is not so much an economic risk but a social, ethical and political one. Geoengineering might be perceived as an insurance policy against climate change, undermining support for existing climate policies.
44
Subcategory: False insurance
A common alternative to the insurance articulation is the concern that geoengineering supplants mitigation, but its efficacy is overestimated. Examples of this argument quote or paraphrase the widely cited Royal Society report, Geoengineering the Climate. 45
For example, Scott quotes the moral hazard argument as it appears in Geoengineering the Climate: Moral hazard objections argue that climate engineering research will undermine efforts “in mitigation and/or adaptation because of a premature conviction that climate engineering has provided ‘insurance’ against climate change”.
46
The definition of moral hazard here is distinct from the insurance main category because it is clear what the hazard is. The false insurance articulation identifies the hazard in the ineptitude of geoengineering, where mitigation is supplanted by an inadequate alternative. Comparatively, most versions of the insurance articulation leave unclear what the hazard is; it may be that geoengineering is inept, or adequate but unethical, or adequate but less preferable than the mitigation strategies it is supplanting.
Unwilling
The unwilling code captures the concern geoengineering will diminish the willingness of agents to engage with mitigation. This formulation does not specify a further mechanism, making it distinct from the other codes. Although most codes are concerned with the diminution of mitigation, unlike the unwilling code, most typically specify the accompanying change in incentives or beliefs.
The vagueness of this code means it captures a range of formulations, with varied triggers for action and outcomes. Merk, Pönitzsch and Rehdanz worry engagement with geoengineering will shift the attention of ‘the public, of economic enterprises, and of political decision-makers away from the reduction of greenhouse gas emissions and towards the use of CE’. 47 Bunzl worries geoengineering will ‘undermine other actions and lead to more not less greenhouse gas output’. 48 Bellamy, Lezaun and Palmer conceive the moral hazard in geoengineering as potentially weakening ‘the resolve to mitigate climate change through the reduction of greenhouse gas emissions’. 49
Avoid
Some use the term moral hazard to characterise the fear that geoengineering enables agents to avoid or delay emissions reductions. Enterprise and the fossil fuel industry are frequently identified in this characterisation. For instance, Carr and Preston discuss fears geoengineering provides a ‘way out’ for polluters: Much attention has been paid to the idea that climate engineering might present a ‘moral hazard’ by providing polluters with a seeming way out of reducing emissions.
50
Subcategory: Legitimise avoidance
A subset of arguments positing that geoengineering will enable agents to avoid emissions reductions express the additional concern that geoengineering will validate this behaviour. This is captured by a range of qualifiers in the explanation of moral hazard, where geoengineering is said to confer a licence, legitimisation, or justification for continuing to emit. For example, Corner and Pidgeon: …If geoengineering were perceived by the public as offering a licence to continue carbon-intensive lifestyles, the ‘rebound’ associated with geoengineering technologies could be even more significant. This behavioural version of the problem might be termed an individual moral hazard.
51
Solution
The solution code captures the concern the potential (or perceived potential) for geoengineering to ameliorate climate change could lead to diminution of mitigation and other activities. This code emphasises the role of knowledge and beliefs in behavioural change. The public, as well as policymakers, are most frequently implicated. For example, Campbell-Arvai, Hart, Raimi, and Wolske note: Policymakers, researchers, and the lay public have expressed concern, however, that further development of CDR strategies may create a sort of moral hazard: if people perceive the problem of climate change to have been solved, they may no longer feel the need to support carbon dioxide emissions reductions from fossil fuel use and agricultural processes.
52
The solution code is notably distinct from insurance, as insurance is characterised by geoengineering's status as a Plan B. In the solution code, moral hazard is characterised by the concern geoengineering could be implemented as Plan A.
Subcategory: False hopes
Moral hazard has been characterised by some as the concern geoengineering will be perceived as a solution to climate change, but its potential to achieve this will be overestimated. This is an especially pernicious worry, because the implication is mitigation and other actions have been diminished and geoengineering does not fulfill its perceived potential.
For instance, Merk and colleagues expresses moral hazard as arising from an overoptimistic evaluation of geoengineering: But not only strong public objection could be a reason for concern for climate policy, also an overoptimistic evaluation of the technologies could have negative repercussions on climate protection: experts fear that discussing climate engineering – especially SAI – as a way to counteract climate change could reduce the motivation to cut emissions via mitigation. This concern is called moral hazard or mitigation obstruction.
53
Risk framing
Drawing on its origins in economics, moral hazard is framed by some in terms of changes to risk exposure. There are three main ways this code manifests in the literature, depending on the authors’ views on the problem.
Firstly, the moral hazard problem may be characterised as an argument from economics, without generalising the problem to geoengineering. For example, Himmelsbach notes: Moral hazard denotes the assumption in economic theory that agents take more risks when the latter are insured by a third party.
54
Secondly, moral hazard may be defined in terms of risk, as in economics, but then applied to geoengineering. This ensures the definition remains broad, while specifying the risks unique to the geoengineering context. For example, Parker draws on Hale's definition to evade ‘issues of definition’: Not wanting to get bogged down in issues of definition, this paper reformulates Hale's definition of moral hazard, “the danger that, in the face of insurance, an agent will increase her exposure to risk,” to clarify its application to SRM: “the danger that, in the face of development of SRM, an agent will increase her exposure to a high CO2 world by disproportionately reducing her interest in emissions reductions.”
55
Thirdly, moral hazard may be defined in terms of risk to highlight the dissonance between its use in insurance compared to geoengineering. In these instances, a definition is given to demonstrate that moral hazard does not characterise the concerns raised in the geoengineering literature. For instance, Morrow initially characterises moral hazard in terms of lowering costs, noting this is the terminology other commentators use for this concern, before giving a definition framed in terms of risk to highlight the incongruity with economics: David Keith now argues that ‘risk compensation’ describes the phenomenon more accurately, since the effect depends not on voluntary risk-sharing between a risktaker and an insurer (as moral hazard does), but by a technology-driven reduction in risk to the risk-takers. More generically, the effect is sometimes described as ‘mitigation obstruction’.
56
Technical fix
The technical fix code describes the concern geoengineering is a technological corrective which fails to address the underlying cause of climate change. This code is often characterised by the expressed worry that moral hazard is about a failure to address the source of emissions. For instance, Preston articulates: The reluctance to talk about geoengineering pre-2006 was based largely on the concern that the prospect of a technical solution to climate change created a ‘moral hazard’ that might encourage risky behaviour or influence the willingness of parties to engage in mitigation and adaptation.
57
Unlike previous themes, technical fix expressly engages with a moral issue by implicitly or explicitly problematising the use of technology to correct climate change. This framing of the moral hazard problem suggests geoengineering remains problematic even if it is effective.
Crowding out
Crowding out specifies the concern that geoengineering will reduce the capital available for mitigation, or potentially dominate the climate policy space. Most of the characterisations captured by this code explicitly use the language ‘crowding out’, though some describe this effect without using this terminology.
For example, Fabre and Wagner discuss the concern that geoengineering will crowd out ambitious mitigation policy: Geoengineering (G) introduces an entirely new set of challenges into global climate policy negotiations. Some have focused on G crowding out ambitious mitigation, necessitating complex governance regimes. This often comes under the heading of “moral hazard”.
58
Lower cost
Moral hazard is also characterised as the concern the introduction of cost-effective geoengineering will result in lower demand for more costly options like mitigation.
This concern is articulated by Morrow, with reference to the increase in climate risk that would accompany lower investment in mitigation: Many commentators have worried that climate engineering research could have a similar effect by inducing society to slow its investment in mitigation. The basic idea is as follows: greenhouse gas (GHG) emissions increase climate risk. Emissions abatement is a costly way to reduce that risk. Climate engineering might reduce climate risk more cheaply. Thus, developing climate engineering might make society willing to emit more GHGs – that is, to mitigate less – than it otherwise would. Following [Keith] commentators often express this concern as a ‘moral hazard’ problem.
59
The lower cost code, like crowding out, gives an account of moral hazard in terms of market forces. However, lower cost is distinct from crowding out as it describes the incentive to pursue geoengineering (as a cheaper option).
Discussion
Moral hazard has been raised as a key issue in geoengineering research. However, the way the argument is made is inconsistent. This study used QCA to capture a range of meanings of ‘moral hazard’ in the academic literature on geoengineering. Using the databases SCOPUS and Web of Science, 109 articles were identified, 50 of which were analysed in this study.
The themes identified through this analysis demonstrate that the moral hazard argument in the geoengineering literature does not consistently align with the distinctions made by Hale, Lockley and Coffman, and Lin 60 or with the definition in economics. Furthermore, the analysis demonstrates that moral hazard is not always used consistently even within articles and, in some cases, is used without a clear articulation of what moral hazard is.
Consistency within articles
Some geoengineering articles contain more than one definition of moral hazard, describing different facets of the problem. Sometimes this is done explicitly to characterise different aspects of moral hazard. In this analysis, these definitions were divided into separate segments. A clear example can be seen in Corner and Pidgeon's work, which offered distinct definitions of moral hazard we coded as insurance, legitimise avoidance, and other. Corner and Pidgeon were mercifully clear in their explanation of each mode of operation, but we ought perhaps be concerned that one term can be conceived of so variously in a single paper. 61
In other instances, multiple conceptions of moral hazard are used without this clarity of expression. For instance, Scott gives an articulation spanning multiple themes: Those who raise moral hazard objections to climate engineering worry that it will be seen as insurance against climate change, which will lead to greater risk-taking behaviors. Stated differently, the possibility of a technological fix will tempt people to delay or avoid taking the difficult steps to put civilization on a Stabilized Pathway, and destine humanity instead to the harsh realities of Hothouse Earth.
62
The first sentence suggests the insurance code, but the second suggests technical fix, which is given as the mechanism for avoiding emissions reductions, which in turn would suggest the avoid code is also relevant here. Due to the dependency of the second sentence on the definition in the first, and inextricability of the themes discussed, the sentences were kept as one segment and coded multiple.
Comparing codes with criticisms in the literature
The second aim of this analysis was to determine whether the meanings of moral hazard identified here reflected critiques of the term and existing attempts to reformulate moral hazard in the literature. Here we focus on the analysis of Hale, Lockley and Coffman, and Lin. 63 Although many of the critiques in these articles resonate with the themes identified in this analysis, there are important points of divergence. Hale's morally-oriented definitions share scant alignment with the themes in this analysis, arguably demonstrating their repeated criticisms regarding the lack of engagement with moral issues in the characterisation of moral hazard. 64 Lockley and Coffman's distinction between moral and morale hazard could be a useful tool for further categorising some of the themes identified in this analysis, but may lack nuance when used alone. 65 Similarly, Lin's explication of risk compensation is a useful explanatory concept but does not alone capture the full range of ways moral hazard is used in the literature. 66
Several of the specific definitions Hale gives of moral hazard resonate with some of the codes identified in this analysis. Recalling the three views of moral hazard with reference to economics and moral theory, the efficiency, responsibility and vice view, Hale 67 offered 16 specifications of the moral hazard problem in geoengineering. This analysis found that definitions which highlight the responsibility view of moral hazard are the most resonant. Hale describes the view as ‘what is wrong with the moral hazard is that it creates the conditions for insured individuals to ignore or disregard countervailing reasons’. 68
Several definitions drawing on the responsibility view closely mirror the codes this analysis, for instance: Hale's Band-aid Objection closely resembles our technical fix category; the Free Riding Objection closely resembles the avoid code; the Cheating Objection also shares similarities with the avoid category, while the Political Noise Objection aligns with many formulations of the unwilling code.
Distinguishing self-interested from incautious behaviour, as Lockley and Coffman do, 69 is pertinent to moral hazard but does not capture the range of meanings present in the geoengineering literature. Lockley and Coffman hope to disambiguate moral hazard in geoengineering via intentions, using the distinction between moral and morale hazard from the insurance literature. The main problem in practice is that intentions can be difficult to discern. Some of the codes identified in this analysis arguably fall clearly in the realm of self-interested moral hazard behaviour, or imprudent morale hazard. For instance, as Lockley and Coffman highlight, polluters using geoengineering to continue emitting as usual, captured by the avoid code in this analysis, seems to be a case of perverse motivations. Similarly, the subcategories false insurance and false hopes both seem to be cases of morale hazard, as agents cease to take suitable caution regarding the risks of climate change after over-estimating geoengineering.
However, many of the codes capture motivations that are not necessarily clear. The two market-oriented codes, crowding out and lower costs, are examples of this. It is unclear whether following market signals and reducing investment in mitigation is a case of an incautious approach and a lack of foresight into the potentially deleterious effects of that action; or, intentionally advancing one's financial interests by increasing the risk others are exposed to.
Reformulating the moral hazard problem using the concept of risk compensation, as Lin does, 70 is a useful explanatory mechanism for many of the themes identified in this analysis. The solution and insurance codes are particularly well explained by the risk compensation phenomenon, where the provision of insurance causes agents to change beliefs about exposure to risk, and hence change behaviour. Risk compensation underpins these behavioural changes by providing a well-studied psychological mechanism.
Conclusion
This analysis demonstrates that moral hazard in the geoengineering literature is used in a range of ways, and not necessarily applied consistently within articles. This analysis improves upon previous work by comparing key critiques and alternative formulations of moral hazard with the way the term manifests in the literature. This highlights where there is good alignment with these critiques, as well as instances where moral hazard is used in a more nuanced way than previous works suggest. However, the coding frame identified in this analysis is still not a full picture of ambiguity in moral hazard. This paper sought to identify, broadly, what commentators in the literature use moral hazard to mean. Future work could analyse in finer detail what issues moral hazard pertains to, such as who is vulnerable, at what stage in geoengineering's development moral hazard is triggered, or whether the lack of clarity around moral hazard has changed over time.
This analysis does illustrate that the use of moral hazard in geoengineering constitutes a patchwork of concerns which, sometimes only faintly, echo the concern formulated in economics. Even though the phenomena described are variable, the application of the moral hazard label to geoengineering seems to have some legitimacy, at least on the surface. The moral hazard of economics and that of geoengineering share the prediction that potentially reducing our vulnerability leads to different behaviours. This is the case even without the clear presence of two distinct actors with divergent interests, or the clear articulation of changing incentive structures. For this reason, the incongruence between the geoengineering moral hazard and that of economics, and its inconsistent application, does not mean the term is always erroneously applied in geoengineering and ought to be abandoned.
However, the ambiguities documented here have the potential to be problematic. Arguably, the current greatest uncertainty facing geoengineering is physical, regarding the safety and efficacy of the technologies in question. It is perhaps easy now to use moral hazard ambiguously; the prospect of a geoengineering intervention equivalent in efficacy to current mitigation seems distant, so the fear mitigation could be supplanted is (mostly) salient. However, if, or when, geoengineering nears the delivery of accurate, fine-grained control of Earth's atmosphere, the replacement of mitigation with geoengineering may even seem reasonable. Those arguing about moral hazard will have to articulate what the actual problem is. While ambiguity and imprecision are not always the enemy of discourse, if we as academics wish to inform the scholarly, public and regulatory debates on these technologies, we ought to be as clear about the physical and non-physical risks as possible. This must include moral hazard.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship and/or publication of this article.
