Hindsight check: LL2-25 (Ch 25 Why did business not react with precaution to early warnings?)#

Source section: EEA, Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013), Ch 25, by Marc Le Menestrel and Julian Rode (pp. 607–620; text pp. 607–617). There are no panels or replies. Check window: publication (2013) to late September 2026. Checked: 26 September 2026.

Method note. - No general web search. The session’s web-search budget was used up, so I retrieved sources directly from primary repositories and databases: - Crossref (metadata and abstracts), Crossref Labs (which carries Retraction Watch update records), Europe PMC and Semantic Scholar, for peer-reviewed work; - the EU Publications Office (Cellar document store and SPARQL endpoint), for Official Journal texts, Court of Justice judgments and a July 2026 Commission evaluation; - the US Federal Register and its API, SEC EDGAR (ExxonMobil annual-meeting filings), CourtListener (state and federal opinions), the US Supreme Court docket and the Third Circuit’s opinion archive; - legislation.gov.uk (UK), Justice Laws (Canada), the French Conseil constitutionnel, ANSES, and the German federal nuclear-waste office (BASE); - IARC and NTP (monographs and the Report on Carcinogens), UNEP, WHO, GOV.UK, the European Court of Auditors, NBER, and the Manville Trust’s claims administrator. - Access limits. - Springer, Taylor & Francis and Oxford journal pages refused automated access. Several papers are therefore used at abstract or title level; I say so where it matters. - OpenAlex’s daily quota was exhausted. - Légifrance served a bot-challenge page, which I did not try to bypass. The French 2014 transfer of pesticide authorisation to ANSES is taken from ANSES’s own page. - New Hampshire’s court site returned 403, so the 2015 MTBE opinion was read from CourtListener’s copy. - Pages from ExxonMobil’s corporate site and some news sites returned 404. The company’s position is taken from its vice-president’s published comment [S2]. - Scope. I read only the section digest, the section’s source extract and web sources. Page numbers are report pages of the 2013 volume. - Annex 3. This is a new cross-cutting chapter in the 2013 volume, not a 2001 case, so no Annex 3 update applies to it directly. Annex 3 updates several of the 2001 cases the chapter draws on (asbestos, fisheries, benzene, PCBs, MTBE, BSE, the Great Lakes). I did not open it, because it lies outside the files named for this task. - Reception. Indexed citation data for this chapter are sparse. Semantic Scholar lists one citing work [S59]. I found no published critique or reply aimed at the chapter itself, so this check tests its claims against later evidence rather than against a reception literature. - Source-interest flags. - Chapter authors’ own work. - The Exxon passage (p. 610) rests on van den Hove et al. (2002), which Le Menestrel co-authored. - The consumer-disclosure claim (p. 609) rests on an unpublished mimeo by Aktar and Le Menestrel (2010). In Crossref searches (26 September 2026) I found no published version under that title or those authors. - Company-side sources. The ExxonMobil critique of Supran and Oreskes [S2] was written by a company vice-president. Supran and Oreskes say part of it drew on material from a contributor paid by the company [S3]. - Advocacy-leaning sources. The Lancet commercial-determinants series [S52] and UNEP’s leaded-petrol release [S19] argue a case as well as report evidence. - Journalism. Retraction Watch [S16] is used only for the wording of a retraction notice I could not open directly. - Trust administrator. The Manville Trust figures come from the trust’s own claims administrator [S28, S29].

Overview#

The chapter is a cross-cutting essay rather than a case study. It makes three kinds of claim: a diagnosis of why firms did not act on early warnings, a set of illustrative vignettes, and four policy “reflections”. Thirteen years on, they have fared differently.

1. The diagnosis of “political actions” and research control has been strengthened. - Private knowledge, public doubt. Since 2013, documents released through litigation and archives have repeatedly shown firms holding internal knowledge of a hazard while their public communications stressed doubt: - ExxonMobil’s internal and peer-reviewed documents mostly accepted human-caused warming while its advertorials mostly expressed doubt [S1]. Its scientists’ 1977–2003 projections were accurate, and at least as skilful as independent models [S5]. - Fluorochemical manufacturers’ own documents show the class’s toxicity was known internally by 1970, about forty years before the public-health community knew [S15]. - An Exxon memo of 1985 advised against using MTBE as a blanket additive because of its groundwater behaviour [S32]. - Sponsorship bias. A Cochrane review confirmed that sponsorship shifts results and conclusions even where standard risk-of-bias measures do not differ [S13]. - A shared playbook. A cross-industry study catalogued 28 doubt-manufacturing tactics, five of them used by all five industries or organisations it examined [S14]. - Ghost-authorship. A widely cited 2000 safety review of a herbicide was retracted in December 2025 because company employees had apparently co-written it without acknowledgement [S16]. That is the pattern the chapter describes for beryllium (p. 611).

2. Cost-shifting through liability gaps and insolvency has been strengthened. - Manville. The Manville Trust, created by the 1982 bankruptcy the chapter cites (p. 612), now pays asbestos claimants 5.6% of their claims’ scheduled value, raised from 5.1% on 3 September 2026 [S29]. - Strategic bankruptcy. Asbestos defendants used purpose-built subsidiaries and bankruptcy filings in 2017–2021. A federal appeals court dismissed one such filing in 2023 because the debtor was not in financial distress [S30]. - EU assessments. The European Court of Auditors (2021) and the Commission’s own evaluation of July 2026 both conclude that insolvent operators without financial security leave remediation costs to the public [S33, S60].

3. The remedies were taken up partially, and not for uncertain hazards. - Uptake. Several reforms target the mechanisms the chapter names: - the EU’s 2024 product-liability law creates presumptions that ease a claimant’s burden of proof where scientific complexity makes proof excessively difficult [S34]; - EU food-chain law now requires firms to notify EFSA of studies before they are carried out [S21]; - a mandatory EU transparency register (2021) counts commissioning research as lobbying [S53]; - EU sustainability-reporting standards (2023) required disclosure of lobbying positions [S54], though the 2026 “Omnibus” directive then sharply narrowed which companies must report [S55]. - No assurance bonding for uncertain hazards. I found no case of assurance bonding for an uncertain hazard. The US EPA declined to impose financial-responsibility requirements on hardrock mining (2018) and on the chemical, petroleum and power sectors (2020) [S36, S37]. Bonding was raised only for known, quantifiable liabilities such as well plugging [S38]. The EU’s liability directive still has no mandatory financial security. The Commission itself says the directive’s structure limits its ability to address “new and emerging” challenges [S60].

4. The illustrative “hope” cases are weaker than the chapter presents them. - Manville fiberglass. The one success story is double-edged, and its timeline is wrong. Glass wool stayed in IARC Group 2B from 1987 until IARC’s October 2001 re-evaluation (published 2002) [S40, S41]. The US National Toxicology Program (NTP) still lists biopersistent glass-wool fibres as “reasonably anticipated” carcinogens [S42]. The warning was not shown unwarranted within “nearly five years” (p. 615). - Fukushima. Germany completed its nuclear phase-out in April 2023 [S43]. But the 2011 decision restored a phase-out already legislated in 2002. One economic study puts the social cost at about USD 12 billion a year, mostly from air-pollution mortality [S44]. That fits the chapter’s own caution that vivid events can bring over-reaction (p. 613). - Reputation. The investor channel rose and then receded. Majorities of ExxonMobil shareholders backed climate-risk reporting (2017) and climate-lobbying reporting (2021). By 2024 climate-related proposals won 21% or less, and none reached the 2025 ballot [S8–S10]. - Voluntary disclosure. The chapter’s claim that consumers rarely reward disclosure of negative information rests on the unpublished mimeo. Published experimental work suggests that disclosing bad news oneself softens damage compared with having it exposed by others [S47].

5. Selection on the dependent variable still limits the generalisation. Later evidence shows firms differing within a single sector: - a rival refiner decided not to use MTBE because of its risks [S32]; - the beryllium producer the chapter criticises co-drafted, with the steelworkers’ union, the tenfold-tighter limit OSHA adopted in 2017 [S18].

“Virtually all” remains a description of a sample chosen for failure (p. 607), not a finding about firms in general.

Claim-by-claim#

Claim 1: “In virtually all reviewed cases it was perceived to be profitable for industries to continue using potentially harmful products or operations”; short-term economic interest is the main driver of business non-precaution (pp. 607, 608, 615)#

Original claim (pp. 607–608, 615). - Ethical values count “if and only if” they add to expected profit (p. 608). External harms enter decisions only through liability, regulation or reputation (p. 608). - “Virtually all reviewed cases have in common that early warnings about harmful effects were available, but that the prospect of short-term profit generated strong economic incentives for companies to continue with their practices” (p. 608). - Lesson 1: warnings create value conflicts in which firms expect precaution to cost them (p. 615).

Subsequent developments - Internal documents confirm the profit-versus-warning conflict in further cases. - MTBE. In State of New Hampshire v. Exxon Mobil (2 October 2015), the New Hampshire Supreme Court upheld a jury verdict of about USD 236 million [S32]. The record included: - a 1985 internal memo recommending that MTBE “not be considered as an additive to Exxon gasolines on a blanket basis” because of its contaminating properties; - evidence that Exxon knew of the extra remediation costs from the 1980s; - testimony that another refiner “decided not to use MTBE because of the unique and increased risks” it perceived [S32]. - Fluorochemicals. A 2023 analysis of a chemical-industry document archive found manufacturers knew by 1970 that these chemicals were “highly toxic when inhaled and moderately toxic when ingested”, about four decades before the public-health literature [S15]. - Climate. ExxonMobil’s scientists modelled warming accurately from 1977 onwards, while its public statements contradicted that work [S5] (see Claim 3). - Cross-industry synthesis. A 2021 comparative study of five industries or organisations found 28 tactics for manufacturing doubt, five of them used by all five [S14]. The authors link these tactics to promoting an agenda “often for monetary benefit”. - Counter-examples that the chapter’s sampling excludes. - A rival refiner avoided MTBE on risk grounds [S32]. - In February 2012 Materion (formerly Brush Wellman), the firm the chapter criticises for beryllium, jointly drafted a model standard with the United Steelworkers. OSHA found “substantial agreement” between that draft and its own proposal. The final 2017 rule cut the permissible exposure limit tenfold, from 2.0 to 0.2 µg/m³ [S18]. That was a producer-supported precautionary step, though decades after the warnings the chapter describes.

Complications - The chapter’s case base was chosen as failures, so “virtually all” is close to true by construction. None of the post-2013 evidence above samples firms at random. - The MTBE record shows that “the entire industry acted in the same manner” was itself Exxon’s defence. The court rejected it because at least one competitor had judged the risk differently [S32]. Firm heterogeneity is therefore real, and it can bear on liability. - The volume’s own false-positives chapter (vol. 2 Ch 2, noted in the digest) is not engaged. See Claim 7 for a false positive inside this chapter.

Verdict: held up (as a description of the reviewed failure cases, with additional post-2013 documentary support). The generalisation beyond a failure-selected sample remains untested.

Implications for weight. The profit-conflict mechanism can carry substantial weight as an explanation of why failures happened. It should not be used as a base rate for how firms generally respond to warnings. Pair it with the counter-examples: a rival that declined a risky additive, and a producer that co-drafted a stricter limit.

Claim 2: Reputation and consumer pressure are usually too weak to drive precaution on uncertain hazards; firms can shape opinion when that is cheaper than changing practice; consumers reward voluntary disclosure of negative information “only under very limited conditions” (p. 609)#

Original claim (p. 609). - Public concern can be powerful, as with antimicrobials and growth hormones in meat. But “in many cases” the public lacks knowledge, underrates risks or “does not feel affected”, so reputation “does not provide a sufficient economic incentive”. - Firms may influence opinion “when this appears cheaper than reducing or terminating the potentially harmful practice”. - Consumers reward disclosure of negative information “only under very limited conditions (Aktar and Le Menestrel, 2010)”, an unpublished mimeo. - Harmed third parties (inshore fishers, oyster farmers, beekeepers) are exceptions with an interest in precaution, but “seem to be a minority”.

Subsequent developments - The disclosure evidence. - I found no published version of the Aktar and Le Menestrel mimeo. - A 2014 experimental paper in the Journal of Business Ethics is titled as finding that company self-disclosure of negative information “lessens damaging effects on consumer judgment and decision making” [S47] (read at title and metadata level only). - A 2022 handbook chapter reviews both “benefits and pitfalls” of this “stealing thunder” strategy [S48] (title only). - The published literature therefore compares self-disclosure with exposure by others, not with silence that is never discovered. On that comparison, disclosure tends to help. The chapter’s pessimism holds only if firms expect their private information to stay private. - Investor pressure: a channel the chapter did not consider, which rose and then fell. ExxonMobil annual-meeting results, from SEC filings [S8–S10]:

• Public concern turned into law in the chapter’s own examples. EU law since 28 January 2022 bans antimicrobial use in animals “for the purpose of promoting growth” and restricts routine prophylaxis to exceptional cases [S49]. That continues the precautionary trajectory the chapter credits partly to public concern (p. 609).
• Harmed third parties as warners. The beekeepers the chapter cites were vindicated through law rather than markets:
• The EU banned outdoor use of imidacloprid, clothianidin and thiamethoxam in 2018. EFSA had found high acute risks to bees for most field uses [S23].
• In May 2021 the Court of Justice dismissed Bayer’s appeal against the 2013 restrictions and upheld the precautionary principle as the basis for review. Beekeeping associations were intervening parties, and Bayer was ordered to pay their costs [S24].

Complications - The investor data come from one firm and cannot separate the effects of changing investor sentiment, rule changes or firm strategy. I did not examine the causes of the decline. - The antimicrobial and neonicotinoid outcomes show public concern combined with regulators and courts, not reputation acting alone. That supports the chapter’s point that reputation alone is insufficient.

Verdict: partly held up. The weakness of market reputation as a lone driver is supported. A new channel, investors, briefly strengthened and then weakened. The specific disclosure claim rests on unpublished work and sits awkwardly with published findings.

Implications for weight. - Treat “reputation is too weak” as a moderate-strength lesson with a time dimension: reputational and investor pressure is volatile and can reverse within five years. - Do not rely on the chapter’s consumer-disclosure claim. The better-supported version is conditional: disclosure pays relative to being exposed, so its value depends on how likely exposure is. - Third-party victims matter most when they have legal standing.

Claim 3: Exxon publicly contested climate science while “publicly denying the existence of the problem that they had privately identified” (p. 610)#

Original claim (p. 610). - In the early climate debate, Exxon publicly contested the science on grounds of complexity and uncertainty. It did so while presenting itself as “a science and technology-based company”. - Its strategy was “chiefly implemented through efforts in publicly denying the existence of the problem that they had privately identified” (van den Hove et al., 2002, co-authored by Le Menestrel). - More generally, emphasising uncertainty yields “paralysis by analysis”.

Subsequent developments - Independent content analysis. Supran and Oreskes (2017) coded 187 ExxonMobil communications from 1977–2014 [S1]:

The authors found that the more publicly accessible the documents were, the more doubt they expressed. They concluded that the company misled the public. Stranded-asset risk appeared in 24 documents but in no advertorial. - The company’s dispute and the response. - ExxonMobil vice-president Vijay Swarup argued in 2020 that the study compared two companies’ documents (Exxon and Mobil before their 1999 merger), and that it analysed only 36 advertorials chosen by Greenpeace [S2]. - The authors replied with new evidence that Mobil was aware of dangerous-warming potential by the early 1980s. They also said the comment drew on undisclosed material from a paid contributor [S3]. - An addendum analysed additional documents. It reported that the advertorial result was statistically significant, that Mobil and Exxon both expressed doubt, and that doubt continued after the merger [S4]. - Projections. Supran, Rahmstorf and Oreskes (Science, 13 January 2023) evaluated Exxon’s 1977–2003 warming projections [S5]. They found: - most projections were accurate against later observations; - they were at least as skilful as independent academic and government models; - the scientists correctly foresaw when human-caused warming would be detected; - the company’s public statements contradicted its own data.

That is a direct test of “privately identified”. - Litigation. - New York’s securities case failed in December 2019. The court found that the Attorney General had not proved material misstatements to investors, calling it “a securities fraud case, not a climate change case”. It added that nothing in the opinion “is intended to absolve ExxonMobil from responsibility for contributing to climate change” [S6]. The case concerned disclosures in 2013–2016, not the historic denial. - State and municipal suits alleging deception are continuing. On 23 February 2026 the US Supreme Court granted review in Suncor Energy v. County Commissioners of Boulder County (a suit against Suncor and Exxon Mobil). The question is whether federal law precludes state-law claims over climate injuries [S7]. The case had not been argued at the time of checking. - Political arena. In February 2026 the US EPA rescinded its 2009 greenhouse-gas endangerment finding on statutory grounds [S12]. It expressly did not rely on the Department of Energy climate working group reports cited at proposal, and it states that it does not rely on any specific critique of climate science. The regulatory contest moved from the science to legal authority. I draw no causal link to any firm.

Complications - The key chapter source was co-authored by a chapter author. The later corroboration comes from different authors using different methods [S1, S5], which strengthens independence. - The company disputes the characterisation [S2]. No court has adjudicated the historic-deception question on the merits, as far as I found.

Verdict: strengthened. Independent document analysis and a quantitative test of the company’s own projections support the private-knowledge/public-doubt divergence the chapter asserted. The dispute is on record, and legal accountability remains unresolved.

Implications for weight. This is now one of the best-evidenced vignettes in the chapter. The method used to corroborate it can be applied elsewhere and deserves emphasis: compare internal, technical and public-facing communications for the same period.

Claim 4: Industry can bias the apparent weight of evidence without misconduct (choice of studies, design and publication bias, non-peer-reviewed symposia); lead-in-petrol research was funded and controlled by Ethyl and GM “for over 40 years”; Brush Wellman financed beryllium papers “under the names of well-known academics” (pp. 610–611)#

Original claim (pp. 610–611). - Firms “do not even have to manipulate results”. They can “bias research results in their own interest by inducing so-called (pro-industry) design and publication biases”, and by choosing which studies to carry out, so that vote-counting of studies is skewed (p. 611). - Symposia proceedings, “typically not peer-reviewed”, are cited as results (p. 611). - Examples: - lead research “conducted and funded exclusively by the Ethyl Corporation and General Motors for over 40 years”; - Bayer relied “almost exclusively” on its own research; - Brush Wellman financed papers “under the names of well-known academics” (p. 611). - These practices “weaken the credibility of privately sponsored research” for honest firms too (p. 611).

Subsequent developments - Systematic evidence of the funding effect. The 2017 Cochrane update covered 75 papers [S13]. Compared with other sponsorship, industry-sponsored drug and device studies:

Standard risk-of-bias measures did not explain the difference. This is the chapter’s “bias without misconduct” mechanism, measured. - Tactic typologies. Goldberg and Vandenberg (2021) catalogue 28 tactics across five industries or organisations [S14]. The fluorochemical document analysis found suppression of unfavourable research and distortion of public discourse. It did not find funding of favourable research in that archive [S15]. Not every industry uses every tactic. - Ghost-authorship confirmed and sanctioned. - In December 2025 the journal retracted Williams, Kroes and Munro (2000), a glyphosate safety review [S16]. The notice cited the apparent co-writing by Monsanto employees without authorship. Retraction Watch reports the journal’s concern that the review relied on unpublished company studies, and that company emails referred to the paper as an example of having outside scientists “edit & sign their names” [S16]. - A 2016 expert-panel review of the same herbicide received an expression of concern (September 2018) and a correction (November 2018) over conflict-of-interest and third-party involvement [S17] (metadata only). - Monsanto is the firm the chapter cites for PCBs (p. 610). The glyphosate hazard classification itself remains contested and is not at issue here. What is at issue is authorship and disclosure. - Beryllium. OSHA’s 2017 rule found that workers exposed at the previous limits “are at an increased risk” of chronic beryllium disease and lung cancer. It estimated the new limit would prevent 90 deaths and 46 new cases of disease a year [S18]. That confirms the chapter’s point that by the late 1990s the old limit was not protective (p. 610). - Lead. Leaded petrol ended worldwide in July 2021, when Algeria stopped selling it. UNEP cites estimates that the ban prevents more than 1.2 million premature deaths a year [S19]. A 2022 study estimated that over 170 million Americans alive in 2015 had high childhood lead exposure, with 824 million IQ points lost [S20]. These are consequences, not tests of the “40 years” funding claim, which I did not re-verify. - Institutional remedies aimed at study selection. - EU Regulation 2019/1381 requires business operators to notify EFSA “without delay” of any study commissioned to support an application. It requires public consultation on studies planned for renewals, allows Commission-triggered verification studies, and mandates proactive publication. It applies from 27 March 2021 [S21]. - The Court of Justice held in Blaise (1 October 2019) that authorities must not “give in all cases preponderant weight to the studies provided by the applicant” [S22].

Complications - The Cochrane evidence is from medicine. Its transfer to environmental-chemical research is plausible and is what the chapter assumes, but it is not demonstrated there to the same standard. - Ghost-authorship is arguably misconduct, not “bias without misconduct”. The chapter blends the two in its beryllium example.

Verdict: strengthened. A systematic review quantifies the funding effect, a cross-industry typology generalises the tactics, a major retraction confirms ghost-authorship, and the EU has legislated study pre-notification to counter selective commissioning.

Implications for weight. This lesson can carry substantial weight. The most transferable version: the choice of which studies exist shapes the weight of evidence as much as the conduct of any single study. The remedy that directly targets this is registering or notifying studies before their results are known.

Claim 5: Absent or limited liability, a burden of proof on victims, and insolvency shift the costs of hazards to society (fisheries, MTBE, UK asbestos; Manville’s 1982 bankruptcy was “a means of dealing with” asbestos claims). Regulators with dual promote/protect mandates (USDA, DFO, MAFF, the French Ministry of Agriculture, US DOE) sided with industry and demanded proof of causation (p. 612)#

Original claim (p. 612). - Limited or absent liability “undermined the polluter pays principle”. Costs of overfishing, MTBE and UK asbestos fell on society. - Whether the burden of persuasion lies on victims or on industry is “crucial”. - “Insolvency risk can further undermine full cost internalisation”, for example “Manville Corporation filed for bankruptcy in 1982 as a means of dealing with asbestos pollution claims”. - Dual-mandate agencies aligned with industry and demanded “proof of causal relationship”. The chapter misnames Canada’s DFO as a US department.

Subsequent developments - Insolvency as cost-shifting: Manville. - The Manville Trust began operating in 1987. It initially paid claims in full, first-in first-out. By 1990 it was judicially declared a “limited fund”, after more than 190,000 claimants had come forward by 1992 [S28]. - On 3 September 2026 the trustees raised the pro rata payment from 5.1% to 5.6% of scheduled value [S29]. Claimants against the firm the chapter cites therefore receive about a twentieth of what their claims are valued at. - The Third Circuit noted in 2023 that claims were underestimated in the Manville case. It also treated Manville’s filing as a response to genuine financial distress (a USD 1.9 billion reserve would have triggered debt acceleration), not as a manoeuvre [S30]. “Means of dealing with” claims is therefore fair, but the insolvency was real. - Insolvency as strategy. The Third Circuit dismissed a 2021 bankruptcy filed by a subsidiary created to hold one consumer-products group’s talc and asbestos liabilities, because the subsidiary “was not” in financial distress [S30]. The court noted that four similar asbestos-liability debtors created by “divisional mergers” had filed in North Carolina in the preceding years (2017–2020) [S30]. The mechanism the chapter names persisted and was contested in court. - Backstops for victims. The UK Mesothelioma Act 2014 created a payment scheme for people who cannot trace a liable employer or insurer. It is funded by a levy on active employers’-liability insurers [S31]. This is a direct legislative response to the UK asbestos cost-shift the chapter describes. - Liability can eventually bite. The New Hampshire MTBE verdict (about USD 236 million, affirmed in 2015) allocated remediation costs by market share, about ten years after the suit began [S32]. - Official EU findings on cost-shifting. - The European Court of Auditors (adopted 19 May 2021) found that the Environmental Liability Directive (ELD) had not solved “the absence of financial security in cases of insolvency”. Public budgets fund more than 42% of soil-remediation activity. In 20 of the 42 EU-funded projects it examined, the polluter could not be made to pay [S33]. - The Commission’s July 2026 evaluation concludes that the ELD “has not always been effective in ensuring that the polluter pays”. It says liable operators often lack means “due to the lack of insurance and the resulting insolvency”, and that the directive imposes no mandatory insurance [S60]. - Burden of proof. The EU’s new Product Liability Directive (2024/2853), which applies to products placed on the market from 9 December 2026: - keeps the claimant’s burden; - adds presumptions of defectiveness and causation where the claimant faces “excessive difficulties, in particular due to technical or scientific complexity” and shows likelihood [S34]; - retains the development-risks defence, under which a producer escapes liability if the state of scientific knowledge could not reveal the defect [S34]. For uncertain hazards, that defence preserves much of the cost-shift the chapter describes. - Dual mandates. - France. A 2014 law transferred authorisation of plant-protection products from the Ministry of Agriculture to ANSES. Until the end of 2014, ANSES assessed and the ministry decided [S27]. The ministry still approves active substances and grants 120-day emergency derogations [S27]. - French neonicotinoids. In 2023 the Court of Justice held that member states may not use emergency derogations to authorise neonicotinoid-treated seeds that EU law has expressly banned [S25]. In August 2025 the Conseil constitutionnel struck down a new statutory route to derogate from France’s neonicotinoid ban. It found the provision insufficiently limited in time, scope and use, given the products’ effects on pollinators, water, soil and health [S26]. - Canada. The 2019 Fisheries Act amendments require the minister to maintain major stocks and to rebuild depleted ones [S35]. Both duties carry escape clauses for “adverse socio-economic impacts” (ss. 6.1(2), 6.2(2)). The dual mandate is now written into the statute, but it is explicit and published.

Complications - Where pre-market authorisation applies, as for EU pesticides, the burden is on the applicant [S22]. The chapter’s victim-burden point concerns ex post liability, not these regimes. It does not discuss REACH or pre-market regimes (a digest caveat). - The MTBE and backstop cases show that cost-shifting can be partly reversed, but late: decades after exposure.

Verdict: strengthened. - The insolvency mechanism is confirmed by the trust’s own payment ratio and by 2017–2023 litigation over strategic bankruptcy. - Two EU audit and evaluation bodies independently confirm cost-shifting to public budgets. - Dual mandates were partly separated (France) or made explicit (Canada) rather than eliminated.

Implications for weight. - Strong lesson: when harms have long latency, liability that arrives late is diluted by insolvency, and later claimants bear most of the loss. - The burden-of-proof point is moderately strong. The EU’s 2024 fix eases proof but keeps the state-of-knowledge defence, the very gap that matters most for uncertain hazards.

Claim 6: Human risk perception (neglect of rare, unexperienced events; time discounting; self-serving bias) generally holds back both corporate precaution and public pressure; a vivid event can reverse this and speed up regulation, as “may have been the case” for Germany’s post-Fukushima nuclear phase-out (pp. 613–614)#

Original claim (pp. 613–614). - People “tend to neglect the likelihood of rare events” they have not experienced (Hertwig and Erev 2009). They discount the future, and self-serving bias turns ambiguity into a “welcome ‘excuse’”. - In exceptional cases, occurrence of a low-probability event can make people “at least temporarily — overrate” its probability. Increased concern “may lead to faster regulatory measures. This may have been the case for the German decision to phase out nuclear energy after the Fukushima accident in 2011” (p. 613).

Subsequent developments - The cognitive mechanism. A 2018 meta-analysis of more than 70,000 choices by over 6,000 participants confirmed a robust gap between decisions from description and decisions from experience. Small samples and sampling error were the largest drivers [S45]. That supports the chapter’s premise that unexperienced rare harms are underweighted. - Germany’s phase-out. According to BASE [S43]:

The vivid event reversed a recent extension and accelerated a pre-existing policy. It did not originate the phase-out. - Costs of the accelerated decision. Jarvis, Deschenes and Jha estimate a social cost of about USD 12 billion a year [S44]. Over 70% of it comes from increased mortality from local air pollution, as lost nuclear output was replaced mainly by coal and imports. Even generous estimates of reduced accident and waste risks fall well short of that cost [S44]. - Integrity of the behavioural-ethics literature. Some high-profile studies in behavioural ethics were retracted after 2013 (for example, a 2012 PNAS study retracted in 2021 [S46]). In Crossref I found no retraction or correction flags for the chapter’s key sources on self-serving bias, ethical fading, ethical blindness or wilful ignorance, but I checked only those four.

Complications - Whether Germany’s 2011 decision reflected over-rating of probability, political calculation or a pre-existing preference cannot be settled from these sources. The chapter’s hedge (“may have been”) remains appropriate. - The Germany case also shows the reverse lesson the chapter mentions only briefly: vivid events can accelerate a decision whose own costs are large and fall elsewhere.

Verdict: partly held up. The cognitive premise is supported by meta-analysis. The Fukushima example is consistent with a vivid event accelerating policy, but it was an acceleration of a 2002 plan, and its costs illustrate over-reaction as much as precaution.

Implications for weight. - Moderate weight for the psychological mechanism as a background condition. - Low weight for the claim that vivid events make regulation better. The better-supported lesson is that they make it faster, with trade-offs that may themselves go unexamined.

Claim 7: Manville’s voluntary 1986 relabelling of fiberglass as possibly carcinogenic, “despite the reluctance of their lawyers”, paid off commercially and shows that denial cultures can reverse; the cancer excess was judged not significant about five years later (p. 615)#

Original claim (p. 615). - After the asbestos bankruptcy, Manville “changed its approach towards products stewardship”. - In 1986, “shortly after learning that its fiberglass products could be related to an increase in cancer rate”, it relabelled them “as possibly carcinogenic despite the reluctance of their lawyers”. - It “benefited from this proactive strategy thanks to a successful indemnification and marketing strategy, proving that what may be perceived as a conflict of interest could well lead to a successful alignment of business and social values”. - “It took then nearly five years to realise that the excess detected in respiratory cancer … [was] not sufficiently significant to justify such a warning label” (Sells 1994; Paine and Gant 2009).

Subsequent developments (including pre-2013 facts the chapter did not report) - IARC 1987–1988. The IARC working group (Lyon, 16–23 June 1987; Volume 43, 1988) classified glass wool as “possibly carcinogenic to humans (Group 2B)” [S40]. Manville’s 1986 label therefore anticipated an international classification by about a year. - IARC 2001–2002. The re-evaluation (Working Group in Lyon, 9–16 October 2001; Volume 81, 2002) placed insulation glass wool, continuous glass filament and rock and slag wool in Group 3 (“not classifiable”). Special-purpose glass fibres (E-glass, “475”) and refractory ceramic fibres remained in Group 2B [S41]. On the epidemiology: - the US cohort showed a statistically significant 6% excess of respiratory-cancer mortality; - the excess was not significant among long-term workers, and smoking “may account” for it; - human evidence was “inadequate” [S41]. - Industry observers were not present during the evaluations [S41]. - NTP. The Report on Carcinogens first listed “Glass Wool (Respirable Size)” in 1994. In 2011 the 12th Report narrowed the listing to “Certain Glass Wool Fibers (Inhalable)”, meaning biopersistent fibres, which remain “reasonably anticipated to be a human carcinogen” [S42]. This was a narrowing, not a delisting. - EU and German classification. Both classify synthetic vitreous fibres as possibly or probably carcinogenic, but exempt fibres that pass biopersistence or bioassay tests [S42]. IARC noted newly developed, less biopersistent fibres with low apparent carcinogenic potential in animals [S41]. Precautionary classification with an exoneration route went together with product reformulation.

Complications - The timeline is wrong. Official hazard assessments kept glass wool in the “possible carcinogen” category for about 14 years (1987–2001). A US federal listing began in 1994 and, in narrowed form, continues. “Nearly five years” appears to reflect the view of Sells, who led Manville’s fiberglass business, not the assessments of regulators or IARC. - The commercial pay-off rests on a single Harvard case series and on insider testimony. I could not access either. It is unverified. - Liability, not just culture. The same firm’s asbestos history (Claim 5) makes liability avoidance a plausible motive alongside culture change. The chapter does not weigh this (a digest caveat). - Double-edged example. For insulation wool the warning was eventually judged unwarranted, which makes it a false positive. For biopersistent and special-purpose fibres, caution was vindicated.

Verdict: partly held up. The broad story survives: a voluntary warning that preceded international classification, with insulation glass wool later downgraded. The chapter’s timeline is incorrect, the commercial “proof” is unverified, and the scientific outcome was mixed by fibre type.

Implications for weight. Low weight as evidence that denial cultures reliably reverse or that precaution pays. Better use: an illustration that a precautionary label, plus a clear test for exoneration, can push reformulation toward less persistent products. Also a reminder that “the warning proved unwarranted” often holds only for part of a product class.

Claim 8 (recommendation): Legal, fiscal and financial mechanisms to internalise the external costs of uncertain hazards “still have a large potential”; “assurance bonding should be considered” (pp. 611–612, 616)#

Original claim (pp. 611–612, 616). - Liability should give firms an ex ante incentive and ensure means to compensate (p. 611). - “Alternatively, ‘assurance bonding’ can require companies to deposit a premium that would cover the costs of potential damage before undertaking the dangerous activity (Kysar, 2009)” (pp. 611–612). The reference list dates Kysar’s article 2010; it concerned nanotechnology. - “For uncertain hazards, proposals for legal, fiscal, and financial regulatory mechanisms still have a large potential … Innovative solutions such as assurance bonding should be considered” (p. 616).

Subsequent developments - Financial security declined in the US. - EPA decided “to not issue final regulations” on financial responsibility for hardrock mining under CERCLA section 108(b) (21 February 2018) [S36]. - In December 2020 it declined such requirements for electric power, petroleum and coal products, and chemical manufacturing. It judged that facilities “operating under a modern regulatory framework do not present a level of risk that warrants” them [S37]. - Bonding raised only for known liabilities. The Bureau of Land Management’s April 2024 rule raised the minimum oil-and-gas lease bond to USD 150,000 and the statewide bond to USD 500,000 [S38]. Minimum bonds had not risen since 1951 (statewide and nationwide) and 1960 (lease), and had been repeatedly found inadequate [S38]. This is assurance bonding for a quantifiable cost (well plugging and reclamation), not an uncertain hazard. - EU. - The ECA (2021) recommended that the Commission “examine the scope for legislative changes to require the use of financial security”. It found only seven member states required it for some or all ELD liabilities [S33]. - The Commission’s July 2026 evaluation reports “increased uptake” of financial security in some member states through mandatory national and voluntary schemes. It still notes no EU-level mandatory insurance. It finds that the list of “hazardous” activities is outdated and limits the directive’s “potential to address new and emerging environmental challenges” [S60]. - Fiscal internalisation for diffuse, partly uncertain hazards. The recast Urban Wastewater Treatment Directive (2024/3019) requires producers of human medicines and cosmetics to cover at least 80% of the costs of quaternary treatment to remove micropollutants, by 31 December 2028 [S39]. It is an extended-producer-responsibility tool for a diffuse chemical-mixture problem. That is the closest post-2013 analogue I found to the chapter’s internalisation proposal. I did not check for later legal challenges or amendments. - Burden-of-proof easing. The EU Product Liability Directive presumptions (Claim 5) [S34].

Complications - The chapter is right that potential remains, as the EU’s own audit and evaluation say. But where regulators weighed financial assurance, they rejected it on grounds that existing regulation sufficed [S37]. That is the opposite of the chapter’s premise that regulation rarely internalises uncertain risks. - Bonding for uncertain hazards has a design problem the chapter does not address: how to size a bond for a harm whose probability and magnitude are unknown.

Verdict: partly held up. The diagnosis that internalisation is incomplete is confirmed by official EU sources through 2026. The specific remedy, assurance bonding for uncertain hazards, was not taken up in any case I found. Financial assurance advanced only where liabilities are known and quantifiable.

Implications for weight. Moderate weight for “internalisation is incomplete”. Low weight for assurance bonding as a practical near-term tool for novel hazards: it has so far been adopted only for measurable end-of-life liabilities. Producer-responsibility levies tied to remediation costs are a more tested route.

Claim 9 (recommendation): Full alignment of business and societal interests “will not always be feasible”; public institutions should analyse and disclose sector-specific “dilemmas and temptations” of early warnings and promote participatory, stakeholder-inclusive decision frameworks rather than prescribe actions (p. 616)#

Original claim (pp. 616–617). - “It seems unrealistic to believe that complete alignment of business interests with interests of society at large will always be feasible.” - “Blaming business, in particular with hindsight … may not always be constructive.” - “Public institutions could support progressive business by analysing and publically disclosing the dilemmas and temptations entailed by early warning signals”, including temptations to use regulatory loopholes and influence science. They should promote “open, transparent, and stakeholder-inclusive participatory decision frameworks”. - Clear, non-judgemental exposition “may contribute to reducing unconscious denials” (p. 616).

Subsequent developments - Uptake in general form. - The UK Government Chief Scientific Adviser’s 2014 annual report, Innovation: managing risk, not avoiding it, cites the 2013 Late lessons volume [S50]. It argues that innovation policy should not favour a trajectory “simply because of its appeal to particular powerful vested interests”. It also argues that precaution helps “avoid simply relying on hopes that powerful vested interests will be spontaneously relinquished” [S50]. - WHO’s 2023 fact sheet on commercial determinants of health distinguishes “business actions and societal engagements”, naming “research funding, lobbying, preference shaping” [S51]. That closely parallels the chapter’s business-action/political-action distinction. - The Lancet 2023 series analyses how commercial actors externalise costs and shape policy [S52]. Its framing is structural and critical rather than the chapter’s non-judgemental “dilemmas”. - Stakeholder-inclusive frameworks in law, then narrowed. - EU sustainability-reporting standards (2023) required large companies to disclose impacts on people and the environment, including political influence (see Claim 10) [S54]. - Directive (EU) 2026/470 (24 February 2026) limited sustainability reporting to firms with more than 1,000 employees and over EUR 450 million turnover. It limited corporate due diligence to firms with more than 5,000 employees and over EUR 1.5 billion turnover [S55]. - Effect on denial. I found no evaluation of whether institutional exposition of dilemmas reduced denial or changed firm behaviour.

Complications - The premise that alignment will not always be feasible is supported by the cost-shifting and doubt-manufacturing evidence under Claims 3–5. The recommendation is untested. - Institutions that took up the analysis did so mostly in a critical or structural register, not the chapter’s sympathetic “dilemmas” register. There is no test of which register works better.

Verdict: unclear. The premise held up. The recommendation was taken up partially and in general terms (UK 2014; WHO and Lancet 2023; EU reporting rules, since narrowed), and there is no evidence either way on whether it reduced denial.

Implications for weight. Use the premise with confidence and the recommendation as a hypothesis. The chapter’s distinctive idea, public, non-blaming exposition of sector-specific temptations as a debiasing tool, has not been tried in a form that could be evaluated.

Claim 10 (recommendation/prediction): Regulation that makes corporate “political actions” (lobbying, shaping research and opinion) transparent would “sustain a sound balance of power” and preserve society’s ability to benefit from early warnings; secrecy about such actions signals “bad faith” (p. 617)#

Original claim (pp. 615, 617). - “Political actions” aim at influencing the political and regulatory context “in the pursuit of profits”, as distinct from business actions within the rules (p. 615). - “The fact that some business actors spend sophisticated efforts to hide or keep secret their political actions can be seen as a signal that their behaviour is of bad faith” (p. 617). - “Regulatory efforts that make more transparent the political actions of business can help to sustain a sound balance of power” (p. 617).

Subsequent developments - Transparency reforms that match the chapter’s definition. - The EU Interinstitutional Agreement of 20 May 2021 created a mandatory transparency register for Parliament, Council and Commission, enforced by making registration a precondition for certain lobbying activities (“conditionality”). Its covered activities include “commissioning and carrying out research” [S53]. - EFSA study notification and proactive disclosure, from 2021 [S21]. - ESRS G1-5 (2023) requires in-scope companies to disclose [S54]: - political contributions; - the main topics of their lobbying and their positions on them; - their transparency-register IDs; - board appointments of people who held comparable public or regulatory posts in the previous two years.

Directive 2026/470 then narrowed which companies report [S55].

• In the management literature, Lyon et al. (2018) argue that firms must be “as transparent about their corporate political responsibility (CPR) as their corporate social responsibility (CSR)” and that ratings should assess lobbying [S56]. This is independent convergence on the chapter’s point.
• Secrecy persisted where rules were weak. Updating Brulle’s analysis to 2003–2018, Brulle et al. (2021) found that funding of US climate-change counter-movement organisations grew 3.4% a year and that the source of 74% of it could not be identified [S11] (preprint abstract).
• Disclosure came mostly through litigation, not transparency regimes. The documentary evidence behind Claims 1, 3 and 4 came mainly from court discovery, trials and archives of previously secret industry documents [S1, S15, S16, S32].
• Transparency language can be turned toward evidence rather than influence. EPA’s January 2021 rule, “Strengthening Transparency in Pivotal Science”, would have given “greater consideration” to studies whose dose-response data were publicly available for independent validation [S57]. A federal court vacated the rule and EPA removed it in June 2021 [S57]. Critics saw this kind of rule as a way to discount health studies built on confidential records. I did not retrieve a primary source for that critique, so that reading is my inference and is unadjudicated.
• Investor-demanded lobbying disclosure rose (55.6% at ExxonMobil in 2021) and then disappeared from ballots (Claim 2) [S9, S10].

Complications - No study I found tests whether lobbying or research-funding transparency reduced manufactured doubt. The prediction is untested. - “Secrecy signals bad faith” received documentary support: undisclosed ghostwriting [S16], an undisclosed paid contributor to a published critique [S3], and untraceable funding [S11]. The inference is still a judgement; firms also keep political activity private for competitive or legal reasons.

Verdict: partly held up. The diagnosis that concealed political action undermines early warnings is supported by post-2013 evidence. EU institutions enacted transparency close to what the chapter proposed. There is no evidence yet that transparency restored a “balance of power”, and the main revelations still came from litigation.

Implications for weight. - Moderate-to-strong weight for distinguishing business actions from political actions and for treating concealment as a warning sign. - Low weight, so far, for the prediction that transparency rules alone rebalance power. - Transparency that targets influence (who funds, commissions and lobbies) is a different instrument from transparency demands that target evidence of harm. The second can cut the other way.

Summary of verdicts#

Technology-neutral lessons this check supports (for later use as a lens)#

Each is tied to the section’s pages and to the later evidence above.

1. Divergence between what an organisation’s technical staff know and what its public communications say is detectable and measurable. Compare internal, technical, peer-reviewed and public-facing documents from the same period. The divergence widens as the audience becomes more public (p. 610; [S1, S4, S5, S15]).
2. The main route by which concealed knowledge surfaces has been litigation discovery and document archives, typically decades later, not voluntary disclosure or transparency regimes (pp. 610, 617; [S1, S15, S16, S32]).
3. Which studies exist shapes the weight of evidence as much as how any single study is done. Sponsorship shifts results and conclusions without detectable methodological bias. Registering or notifying studies before results are known is the remedy aimed at this mechanism (pp. 610–611; [S13, S21]).
4. Delayed liability is diluted liability. Where harms have long latency, insolvency (genuine or engineered) means most costs fall on victims and public budgets. Later claimants against one major asbestos defendant receive about 5.6% of their claims’ scheduled value (p. 612; [S28–S30, S33, S60]).
5. Easing the burden of proof does not close the gap for uncertain hazards if a state-of-knowledge defence remains. The defence protects exactly the harms that were not knowable at launch (p. 612; [S34]).
6. Financial assurance is adopted for known, quantifiable liabilities and declined for uncertain ones. Regulators typically decline on the ground that existing regulation already manages the risk (pp. 611–612, 616; [S36–S38, S60]).
7. Promote-and-protect conflicts survive institutional separation. They reappear as emergency-derogation powers and socio-economic escape clauses, and courts rather than regulators have often enforced the protective side (p. 612; [S25–S27, S35]).
8. Harmed third parties are effective early warners when they have legal standing, more than when they rely on markets or reputation (p. 609; [S23, S24]).
9. Reputational and investor pressure is volatile. It can move from majority support for disclosure to near-absence within five years, so it is a weak foundation for sustained precaution (p. 609; [S8–S10]).
10. Vivid events make policy faster, not necessarily better. They tend to accelerate existing options, and the accelerated decision’s own costs may exceed the risk it addresses (p. 613; [S43, S44]).
11. Distinguish actions within the rules from actions to change the rules, the evidence or opinion. Concealment of the second kind is a warning sign. Transparency aimed at influence differs from transparency demands aimed at evidence of harm, which can be used to discount inconvenient studies (pp. 615, 617; [S11, S53, S54, S57]).
12. A precautionary warning can be both partly vindicated and partly overturned for different members of a product class. Pairing precautionary classification with a clear exoneration test can steer reformulation toward safer variants (p. 615; [S40–S42]).

Sources#

All retrieved 26 September 2026 unless noted. EU legal texts and judgments were read in their Official Journal or Court form via the Publications Office Cellar (https://publications.europa.eu/resource/celex/<CELEX>); ELI or Curia addresses are given for readers.

Climate communications, litigation and investor votes - [S1] Supran, G., Oreskes, N. “Assessing ExxonMobil’s climate change communications (1977–2014).” Environmental Research Letters 12, 084019 (August 2017) (abstract). https://doi.org/10.1088/1748-9326/aa815f - [S2] Swarup, V. “Comment on ‘Assessing ExxonMobil’s climate change communications (1977–2014)’.” Environmental Research Letters (16 October 2020) (abstract; author is an ExxonMobil vice-president). https://doi.org/10.1088/1748-9326/abbc91 - [S3] Supran, G., Oreskes, N. “Reply to Comment on ‘Assessing ExxonMobil’s climate change communications (1977–2014)’.” Environmental Research Letters (16 October 2020) (abstract). https://doi.org/10.1088/1748-9326/abbe82 - [S4] Supran, G., Oreskes, N. “Addendum to ‘Assessing ExxonMobil’s climate change communications (1977–2014)’.” Environmental Research Letters (October 2020) (abstract). https://doi.org/10.1088/1748-9326/ab89d5 - [S5] Supran, G., Rahmstorf, S., Oreskes, N. “Assessing ExxonMobil’s global warming projections.” Science 379 (13 January 2023) (abstract). https://doi.org/10.1126/science.abk0063 - [S6] People of the State of New York v. Exxon Mobil Corp., Index No. 452044/2018 (N.Y. Sup. Ct., 10 December 2019) (Ostrager, J.). https://www.nycourts.gov/reporter/3dseries/2019/2019_51990.htm (text read from CourtListener’s copy: https://www.courtlistener.com/opinion/4687259/people-v-exxon-mobil-corp/) - [S7] Suncor Energy (U.S.A.) Inc. v. County Commissioners of Boulder County, US Supreme Court No. 25-170, docket (certiorari granted 23 February 2026) and question presented. https://www.supremecourt.gov/search.aspx?filename=/docket/docketfiles/html/public/25-170.html ; https://www.supremecourt.gov/docket/docketfiles/html/qp/25-00170qp.pdf - [S8] ExxonMobil Corp., Form 8-K, 2017 annual meeting voting results (filed 6 June 2017). https://www.sec.gov/Archives/edgar/data/34088/000003408817000031/r8k053117.htm - [S9] ExxonMobil Corp., Form 8-K, 2021 annual meeting voting results (filed 2 June 2021). https://www.sec.gov/Archives/edgar/data/34088/000003408821000031/xom-20210526.htm - [S10] ExxonMobil Corp., Forms 8-K, annual meeting voting results 2024 (filed 31 May 2024), 2025 (filed 30 May 2025) and 2026 (filed 29 May 2026). https://www.sec.gov/Archives/edgar/data/34088/000003408824000036/xom-20240529.htm ; https://www.sec.gov/Archives/edgar/data/34088/000003408825000030/xom-20250528.htm ; https://www.sec.gov/Archives/edgar/data/34088/000003408826000078/xom-20260527.htm - [S11] Brulle, R.J., Hall, G., Loy, L., Schell-Smith, K. “Obstructing action: foundation funding and US climate change counter-movement organizations.” Climatic Change (2021). https://doi.org/10.1007/s10584-021-03117-w (abstract read from the preprint, https://doi.org/10.21203/rs.3.rs-178750/v1, February 2021). Earlier analysis: Brulle, R.J. “Institutionalizing delay.” Climatic Change (December 2013), https://doi.org/10.1007/s10584-013-1018-7 (not retrieved). - [S12] US EPA. “Rescission of the Greenhouse Gas Endangerment Finding and Motor Vehicle Greenhouse Gas Emission Standards Under the Clean Air Act.” Final rule, Federal Register (18 February 2026) (abstract and preamble passages). https://www.federalregister.gov/documents/2026/02/18/2026-03157/rescission-of-the-greenhouse-gas-endangerment-finding-and-motor-vehicle-greenhouse-gas-emission

Research influence, ghostwriting and hazard evidence - [S13] Lundh, A., Lexchin, J., Mintzes, B., Schroll, J.B., Bero, L. “Industry sponsorship and research outcome.” Cochrane Database of Systematic Reviews MR000033.pub3 (16 February 2017) (abstract). https://doi.org/10.1002/14651858.MR000033.pub3 - [S14] Goldberg, R.F., Vandenberg, L.N. “The science of spin: targeted strategies to manufacture doubt with detrimental effects on environmental and public health.” Environmental Health 20 (26 March 2021) (abstract). https://doi.org/10.1186/s12940-021-00723-0 - [S15] Gaber, N., Bero, L., Woodruff, T.J. “The Devil they Knew: Chemical Documents Analysis of Industry Influence on PFAS Science.” Annals of Global Health (1 June 2023) (abstract). https://doi.org/10.5334/aogh.4013 - [S16] Retraction notice to Williams, G.M., Kroes, R., Munro, I.C. (2000), “Safety evaluation and risk assessment of the herbicide Roundup and its active ingredient, glyphosate, for humans.” Regulatory Toxicology and Pharmacology (retraction dated 5 December 2025 in Crossref/Retraction Watch records; notice in the February 2026 issue; notice text not accessible). https://doi.org/10.1016/j.yrtph.2025.106006 . Wording of the notice as reported by Retraction Watch (journalism, 4 December 2025): https://retractionwatch.com/2025/12/04/glyphosate-safety-article-retracted-elsevier-monsanto-ghostwriting/ - [S17] Expression of concern (26 September 2018) and correction (30 November 2018) to Williams, G.M. et al. (2016), “A review of the carcinogenic potential of glyphosate by four independent expert panels…”, Critical Reviews in Toxicology (metadata and Retraction Watch reason codes only). https://doi.org/10.1080/10408444.2018.1522786 ; https://doi.org/10.1080/10408444.2018.1522175 - [S18] US OSHA. “Occupational Exposure to Beryllium.” Final rule, 82 FR 2470 (9 January 2017) (abstract and full text). https://www.federalregister.gov/documents/2017/01/09/2016-30409/occupational-exposure-to-beryllium - [S19] UNEP. “Era of leaded petrol over, eliminating a major threat to human and planetary health.” Press release (Nairobi, 30 August 2021). https://www.unep.org/news-and-stories/press-release/era-leaded-petrol-over-eliminating-major-threat-human-and-planetary - [S20] McFarland, M.J., Hauer, M.E., Reuben, A. “Half of US population exposed to adverse lead levels in early childhood.” PNAS 119(11) (7 March 2022) (abstract). https://doi.org/10.1073/pnas.2118631119 - [S21] Regulation (EU) 2019/1381 of 20 June 2019 on the transparency and sustainability of the EU risk assessment in the food chain. OJ L 231, 6.9.2019 (Articles 32b, 32c, 38; application from 27 March 2021). CELEX 32019R1381; http://data.europa.eu/eli/reg/2019/1381/oj - [S22] Court of Justice (Grand Chamber), Blaise and Others, C-616/17, judgment of 1 October 2019, ECLI:EU:C:2019:800 (paras 93–95). CELEX 62017CJ0616; https://curia.europa.eu/juris/liste.jsf?num=C-616/17

Neonicotinoids and dual mandates - [S23] Commission Implementing Regulation (EU) 2018/783 of 29 May 2018 (imidacloprid), OJ L 132, 30.5.2018 (recitals on EFSA’s findings; companion Regulations 2018/784 and 2018/785 for clothianidin and thiamethoxam). CELEX 32018R0783; http://data.europa.eu/eli/reg_impl/2018/783/oj - [S24] Court of Justice, Bayer CropScience and Bayer v Commission, C-499/18 P, judgment of 6 May 2021, ECLI:EU:C:2021:367. CELEX 62018CJ0499; https://curia.europa.eu/juris/liste.jsf?num=C-499/18 - [S25] Court of Justice, PAN Europe and Others, C-162/21, judgment of 19 January 2023 (operative part). CELEX 62021CJ0162; https://curia.europa.eu/juris/liste.jsf?num=C-162/21 - [S26] Conseil constitutionnel, Décision n° 2025-891 DC du 7 août 2025, Loi visant à lever les contraintes à l’exercice du métier d’agriculteur (paras 68–83; Article 1 of the decision). https://www.conseil-constitutionnel.fr/decision/2025/2025891DC.htm - [S27] ANSES. “Twenty questions about ANSES and the authorisation of plant protection products” (undated web page). https://www.anses.fr/en/content/twenty-questions-about-anses-and-authorisation-plant-protection-products

Liability, insolvency and burden of proof - [S28] Manville Personal Injury Settlement Trust. “History.” https://mantrust.claimsres.com/history/ - [S29] Claims Resolution Management Corporation. “Manville: Increase in the pro rata payment percentage” (3 September 2026). https://www.claimsres.com/2026/09/03/manville-increase-in-the-pro-rata-payment-percentage/ - [S30] In re LTL Management, LLC, Nos. 22-2003 et al. (3d Cir., 30 January 2023) (precedential opinion). https://www2.ca3.uscourts.gov/opinarch/222003p.pdf - [S31] Mesothelioma Act 2014 (c. 1), ss. 1, 2 and 13 (in force 31 March 2014). https://www.legislation.gov.uk/ukpga/2014/1/contents - [S32] State of New Hampshire v. Exxon Mobil Corp., 168 N.H. 211, Nos. 2013-0591 and 2013-0668 (N.H., 2 October 2015). https://www.courtlistener.com/opinion/3006720/state-of-new-hampshire-v-exxon-mobil-corporation-a/ (PDF from CourtListener storage; the court’s own site returned 403) - [S33] European Court of Auditors. Special Report 12/2021, The Polluter Pays Principle: Inconsistent application across EU environmental policies and actions (adopted 19 May 2021). https://www.eca.europa.eu/Lists/ECADocuments/SR21_12/SR_polluter_pays_principle_EN.pdf - [S34] Directive (EU) 2024/2853 of 23 October 2024 on liability for defective products, OJ L, 18.11.2024 (Articles 9–11; recital 63 on application from 9 December 2026). CELEX 32024L2853; http://data.europa.eu/eli/dir/2024/2853/oj - [S35] Fisheries Act (R.S.C. 1985, c. F-14), ss. 6.1–6.2 as amended by S.C. 2019, c. 14. https://laws-lois.justice.gc.ca/eng/acts/f-14/page-2.html

Financial security and internalisation - [S36] US EPA. “Financial Responsibility Requirements Under CERCLA Section 108(b) for Classes of Facilities in the Hardrock Mining Industry.” Final action, Federal Register (21 February 2018). https://www.federalregister.gov/documents/2018/02/21/2017-26514/financial-responsibility-requirements-under-cercla-section-108b-for-classes-of-facilities-in-the - [S37] US EPA. “Financial Responsibility Requirements Under CERCLA Section 108(b) for Facilities in the Electric Power Generation, Transmission, and Distribution Industry; the Petroleum and Coal Products Manufacturing Industry; and the Chemical Manufacturing Industry.” Final action, Federal Register (2 December 2020). https://www.federalregister.gov/documents/2020/12/02/2020-26379/financial-responsibility-requirements-under-cercla-section-108b-for-facilities-in-the-electric-power - [S38] US Bureau of Land Management. “Fluid Mineral Leases and Leasing Process.” Final rule, Federal Register (23 April 2024) (minimum bond table; preamble on 1951/1960 amounts). https://www.federalregister.gov/documents/2024/04/23/2024-08138/fluid-mineral-leases-and-leasing-process - [S39] Directive (EU) 2024/3019 of 27 November 2024 concerning urban wastewater treatment (recast), OJ L, 12.12.2024 (Article 9, Annex III). CELEX 32024L3019; http://data.europa.eu/eli/dir/2024/3019/oj - [S60] European Commission. Executive summary of the evaluation of Directive 2004/35/CE on environmental liability, SWD(2026) 401 final (Brussels, 16 July 2026); full evaluation SWD(2026) 400 final (not read). CELEX 52026SC0401; https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52026SC0401 (retrieved as PDF via the Publications Office Cellar).

Man-made vitreous fibres (Manville fiberglass) - [S40] IARC. Man-made Mineral Fibres and Radon. IARC Monographs Vol. 43 (1988; Working Group met in Lyon, 16–23 June 1987) (overall evaluation section and participant list). https://publications.iarc.who.int/61 - [S41] IARC. Man-made Vitreous Fibres. IARC Monographs Vol. 81 (2002; Working Group met in Lyon, 9–16 October 2001) (Summary of Data Reported and Evaluation). https://publications.iarc.who.int/99 ; https://publications.iarc.who.int/download/mono81-6E.pdf - [S42] US National Toxicology Program. Report on Carcinogens, 15th ed., profile “Certain Glass Wool Fibers (Inhalable)”. https://ntp.niehs.nih.gov/sites/default/files/ntp/roc/content/profiles/glasswoolfibers.pdf

Risk perception and the German nuclear phase-out - [S43] Federal Office for the Safety of Nuclear Waste Management (BASE). “Nuclear phase-out” (chronology). https://www.base.bund.de/en/nuclear-safety/nuclear-phase-out/nuclear-phase-out_content.html - [S44] Jarvis, S., Deschenes, O., Jha, A. “The Private and External Costs of Germany’s Nuclear Phase-Out.” Journal of the European Economic Association 20(3) (2022), https://doi.org/10.1093/jeea/jvac007 (abstract read from NBER Working Paper 26598, December 2019: https://www.nber.org/papers/w26598) - [S45] Wulff, D.U., Mergenthaler-Canseco, M., Hertwig, R. “A meta-analytic review of two modes of learning and the description-experience gap.” Psychological Bulletin 144(2) (published online 14 December 2017) (abstract). https://doi.org/10.1037/bul0000115 - [S46] “Retraction for Shu et al., Signing at the beginning makes ethics salient…” PNAS (13 September 2021). https://doi.org/10.1073/pnas.2115397118

Reputation, disclosure and public concern - [S47] Fennis, B.M., Stroebe, W. “Softening the Blow: Company Self-Disclosure of Negative Information Lessens Damaging Effects on Consumer Judgment and Decision Making.” Journal of Business Ethics 120, 109–120 (online 14 February 2013; issue 2014) (title and metadata only). https://doi.org/10.1007/s10551-013-1647-9 - [S48] Claeys, A.-S. “The Benefits and Pitfalls of Stealing Thunder.” In The Handbook of Crisis Communication, 2nd ed., Wiley (25 October 2022), pp. 99–111 (title and metadata only). https://doi.org/10.1002/9781119678953.ch7 - [S49] Regulation (EU) 2019/6 of 11 December 2018 on veterinary medicinal products, Article 107 (application from 28 January 2022). CELEX 32019R0006; http://data.europa.eu/eli/reg/2019/6/oj

Institutions, transparency and reporting - [S50] Government Office for Science (UK). Innovation: managing risk, not avoiding it. Government Chief Scientific Adviser’s annual report 2014 and evidence volume (19 November 2014). https://www.gov.uk/government/publications/innovation-managing-risk-not-avoiding-it ; evidence PDF: https://assets.publishing.service.gov.uk/media/5a748fb940f0b616bcb17799/14-1190b-innovation-managing-risk-evidence.pdf - [S51] WHO. “Commercial determinants of health.” Fact sheet (21 March 2023). https://www.who.int/news-room/fact-sheets/detail/commercial-determinants-of-health - [S52] Gilmore, A.B. et al. “Defining and conceptualising the commercial determinants of health.” The Lancet (23 March 2023) (abstract), https://doi.org/10.1016/S0140-6736(23)00013-2 ; Friel, S. et al. “Commercial determinants of health: future directions.” The Lancet (23 March 2023) (abstract), https://doi.org/10.1016/S0140-6736(23)00011-9 - [S53] Interinstitutional Agreement of 20 May 2021 between the European Parliament, the Council and the European Commission on a mandatory transparency register. OJ L 207, 11.6.2021 (Articles 2, 3(2)(d), 5). CELEX 32021Q0611(01); https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32021Q0611(01) - [S54] Commission Delegated Regulation (EU) 2023/2772 of 31 July 2023 supplementing Directive 2013/34/EU as regards sustainability reporting standards, ESRS G1 “Business conduct”, Disclosure Requirement G1-5 “Political influence and lobbying activities”. CELEX 32023R2772; http://data.europa.eu/eli/reg_del/2023/2772/oj - [S55] Directive (EU) 2026/470 of 24 February 2026 amending Directives 2006/43/EC, 2013/34/EU, (EU) 2022/2464 and (EU) 2024/1760 as regards certain corporate sustainability reporting and due diligence requirements (scope thresholds). CELEX 32026L0470; https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32026L0470 . See also Directive (EU) 2025/794 of 14 April 2025 (postponed application dates), CELEX 32025L0794. - [S56] Lyon, T.P. et al. “CSR Needs CPR: Corporate Sustainability and Politics.” California Management Review 60(4), 5–24 (6 June 2018) (abstract). https://doi.org/10.1177/0008125618778854 - [S57] US EPA. “Strengthening Transparency in Pivotal Science Underlying Significant Regulatory Actions and Influential Scientific Information.” Final rule, Federal Register (6 January 2021), https://www.federalregister.gov/documents/2021/01/06/2020-29179/strengthening-transparency-in-pivotal-science-underlying-significant-regulatory-actions-and ; and “…Implementation of Vacatur” (2 June 2021), https://www.federalregister.gov/documents/2021/06/02/2021-11317/strengthening-transparency-in-pivotal-science-underlying-significant-regulatory-actions-and

Reception of the chapter - [S59] Semantic Scholar record for “25 Why did business not react with precaution to early warnings?” (2013), showing one citation as of 26 September 2026. https://www.semanticscholar.org/paper/0dbe39bbcbffcd1327950fe92c355ae3d2c73944