LL2-23 — Part D introduction (Costs, justice and innovation) and Ch23 Understanding and accounting for the costs of inaction#

Late lessons from early warnings: science, precaution, innovation (EEA Report No 1/2013). Report pp. 561–580; PDF pp. 563–582.

How this was read. I read the full text extract page by page, through to the final marker (PDF 582 / p. 580). I also rendered the PDF and checked these pages by eye: - p. 561, the Part D title page; - p. 564, the chapter summary; - pp. 567–569, Box 23.1 and Figure 23.1; - p. 570; - pp. 572–573, Box 23.2 and Figure 23.2; - p. 574, Table 23.1; - pp. 575–576, Box 23.3 and Figures 23.3–23.4; - p. 577.

The text extraction was accurate except for subscripts and the data inside figures, which are taken here from the page images. No web sources were used. Anything added from general knowledge, such as post-2013 context or probable errors, is marked [outside knowledge — verify].

Authors and standpoint#

Part D front matter (pp. 561–563). Part D has no introductory essay. It consists of three pages: - a title page, “Part D Costs, justice and innovation”, with a photograph of a balance scale (p. 561); - a contents page covering Chapter 23 (costs of inaction), Chapter 24 (protecting early warners and late victims) and Chapter 25 (why business did not react with precaution) (p. 562); - a list of panels (p. 563). All three belong to Chapter 24: - Panel 24.1, David Gee, on scientific support for early warners (p. 584); - Panel 24.2, Owen McIntyre, on asbestos liability and “toxic torts” (p. 591); - Panel 24.3, Robert Costanza, on precautionary assurance bonds (p. 602).

The Part’s framing comes only from its title, the image and how the chapters are grouped. Chapter 23 barely touches “innovation”: only in the Japan example (p. 573) and the claim that less harmful alternatives to lead in petrol “were not patent-protected” (p. 567).

Chapter 23 authors. Mikael Skou Andersen and David Owain Clubb (p. 564). No affiliations are given. Several of the chapter’s central numbers come from the lead author’s own damage-cost work: - Andersen (2010), a report on environmental-economic unit prices from the Danish National Environmental Research Institute; - Andersen et al. (2011), nitrate valuation from the EU FP7 EXIOPOL project; - Pizzol et al. (2010), external costs of atmospheric lead, with Andersen as co-author (pp. 577–579).

That work puts Andersen in the “impact pathway” tradition of environmental economics, which traces a pollutant from emission through exposure and health effect to a money value. [outside knowledge — verify: Andersen was a professor at Aarhus University and worked with the EEA; Clubb’s role is not stated.]

Panels. Chapter 23 has no panels or commentaries. No industry or regulatory response or other dissenting voice appears in the section. Every view is either the authors’ own or that of a source they cite.

Stance. The chapter reviews methods and ends in a pro-precaution conclusion, but it reaches that conclusion through economics rather than by opposing it. - The authors take it as a common political reality that policymakers “often” want monetised “costs of inaction” before they act (p. 564). - They argue this demand can be met in precaution’s favour if estimates: - rest on good science and modelling; - use low “social” discount rates; - report ranges, and treat lower bounds as lower bounds; - are produced with expertise in “health, ecology, demography, modelling and science” rather than by “economists alone” (pp. 564, 568, 574, 577). - [My framing, not the chapter’s:] this departs from the strand of precautionary writing that treats cost-benefit analysis (CBA) as an obstacle. Here CBA can be an ally if it is done a certain way. - The tone moves between technical exposition (valuing deaths, discounting) and advocacy. Examples are the leaded-petrol industry’s “desire… to make profits regardless” (p. 567) and the statement that the abandoned nitrate guide value “deserves reconsideration” (p. 571).

Type of chapter. This is a methods chapter, not a case study. The authors say their cases “provide a generic perspective” and do not all match other chapters (p. 565). The structure is: - three short case sections: lead, nitrate, and how to value deaths from air pollution; - a section on discounting, followed by conclusions; - three boxes of monetary estimates: lead and mercury; SO2 and other air pollutants; ozone-depleting substances (ODS).

The Stern Review is the framing example (p. 565).

Section-by-section notes#

Chapter summary (p. 564)#

• Premise. “the burden of proof is often distributed such that policymakers only respond to early warning signals from environmental hazards once the costs of inaction have been estimated”. No evidence is offered for “often”.
• Stated aim. The aim is to examine the methodological challenges of producing “credible and appropriate” estimates, “rather than present specific estimates”. Yet the boxes are full of specific estimates.
• Second aim. The cases are also meant to show “how early warning signals might provide a basis for estimating the costs of inaction, when the science base is less consolidated”.
• Headline claims.
• Nitrate: a precautionary approach to the costs of inaction “is quite conceivable”.
• ODS: early-warning scientists “successfully alerted the world”. The effects on global warming, recognised later, made costs “considerably higher than initially believed”. This is “a reminder that figures for the costs of inaction have often been grossly underestimated”. In the chapter, the support for “often… grossly” amounts to the ODS climate effect, lead’s omitted cardiovascular effects (p. 568) and the biodiversity costs left out of the air pollution figures (p. 573).
• Air pollution: using both higher and lower valuations of mortality gives a range. Even lower bounds that are “perhaps too conservative, with a bias towards health effects” will usually “encourage more rather than less abatement effort”, and help biodiversity.
• Closing claim, repeated at p. 577. Estimates “should not be left to economists alone”. Interdisciplinary estimates can strengthen “scattered and diffuse interests” that struggle to be heard.

23.1 Introduction (p. 565)#

• The asymmetry at the heart of the chapter. It opens with a quote from Late Lessons 1 (EEA 2001: 3–4). Prevention costs are “tangible, clearly allocated and often short term”. The costs of failing to act are “less tangible, less clearly distributed and usually longer term”, “posing particular problems of governance”. Weighing them involves “ethical as well as economic considerations”.
• The Stern Review (2007).
• Unabated warming costs “in the magnitude of 5 % of global GDP annually”. This rises to 20 % once non-market costs are included and developing countries are counted “in an equitable way”.
• Mitigation costs about 1 % of GDP a year and is “unlikely to exceed 3.5 %”.
• Scrutiny of the review fell “mainly over the so-called ‘discount rate’”. But its “basic finding… was broadly accepted” (p. 565).
• The chapter credits the review with influencing the UK Climate Change Act (an 80 % cut by 2050) and the European Council’s March 2007 climate decision. It asserts this influence without documenting it.
• Wider use. “Costs of inaction” has since recurred in debates on climate and biodiversity. OECD (2008) surveys studies on air and water pollution, resource depletion and industrial accidents. Stern used “a risk-analysis approach” that stressed ranges and uncertainty, unlike “the more static cost-benefit approach”.
• Key clarification. “under an economics perspective the ‘costs of inaction’ are simply analogous to the benefits that can be obtained with proper controls” (OECD 2008: 49). The methods are those of ordinary benefit estimation. The term relabels benefits; it is not a new measure.
• Role of modelling. Computing and modelling are said to let analysts handle complexity and uncertainty “and hence to integrate a precautionary perspective”. This is asserted rather than shown.
• Roadmap. Lead shows the difference between proof and evidence. Nitrate shows estimation under uncertainty. Air pollution illustrates the controversy over valuing mortality. The boxes cover lead and mercury, SO2 and ODS.

23.2 Proof versus evidence: lead phase-out (pp. 565–566)#

• Source. The section draws on Heinzerling, Ackerman and Massey (2005). Needleman’s research on children’s lead levels and cognition “made it possible to state in quantitative terms the loss of IQ”. Then “economic analysis eventually came to the rescue of the lead phase-out” by linking lower IQ to lower lifetime income (also citing Nichols 1997) (p. 566).
• Change of method. This replaced willingness-to-pay studies that asked parents what they would pay to prevent exposure. Those studies “expected firm monetary preferences to emerge for a problem the dimensions of which few parents would understand”. Scientific findings could now feed “directly” into economic assessment (p. 566).
• Needleman’s contribution. He measured lead accumulated in children’s baby teeth rather than recent exposure in blood. One challenge was relating blood lead to lead in ambient air. (The same page first describes his research as being on lead “in children’s blood”, so the chapter is loosely worded here.)
• Proof versus evidence.
• Many hazards have “only limited scientific evidence”, and understanding them will take “many years”.
• Economists “will often ask whether there is ‘scientific proof’”, but there is “usually not a simple answer”.
• Analysis “should therefore consider the less certain levels of knowledge that make up scientific evidence” (p. 566).
• Model of knowledge. Research does not turn ignorance into “indisputable proof”. Knowledge has “a focused nucleus of complete understanding” surrounded by poorly understood linkages. “Early warning signals are often located in this blurred area where linkages are not fully understood” (p. 566).
• Legal standard. The EU Treaty (TFEU) “does not generally mandate a requirement for scientific proof”. Article 114(5), the environmental guarantee, lets a Member State adopt stricter national measures on the basis of “scientific evidence”. Precaution (art. 191) “is really about acting in the absence of scientific proof” (p. 566). [Outside knowledge — verify: the treaty text of art. 114(5) requires “new scientific evidence” about a problem specific to that Member State, so the chapter’s gloss simplifies it.]
• Tools exist but are rarely used. Computer models now link low-dose exposure to children’s blood lead at different ages (Pizzol et al. 2010), “But the use of such tools remains the exception”.
• PCB. Abatement costs are “now counted in billions of euros” (von Bahr and Janson 2004), while “welfare economic damage costs are largely unknown”. The costs of acting are known; the costs of harm are not.
• Proposed method. Where knowledge is thin, rely on early-warning evidence. For example, start from “already-proven impacts of high doses in the work environment, and then scaling linearly to low doses” (p. 566). This is a precautionary default that assumes no safe threshold. The chapter does not say when it would overstate or understate harm.

Box 23.1 Costs of inaction on lead and mercury (pp. 567–569)#

Lead: the harm (p. 567). - A pervasive neurotoxicant affecting “virtually every organ or system”. Children and pregnant women absorb more. - Even “extremely low” blood lead levels affect academic achievement (Miranda et al. 2011). The box cites a “2005 US Centre for Disease Control’s recommendation that there is no safe BLL for children”, but the reference given is ACCLPP 2012, so the date and the cited source do not match. [Outside knowledge — verify: a 2005 CDC statement did say no safe blood lead level had been identified, so the problem may lie in the citation rather than the date.] - The effects “are irreversible”. Prenatal exposure causes permanent harm (cited as “Lourdes et al., 2006”, which is evidently Schnaas et al. 2006). Landrigan (2002) called lead a public health “catastrophe”.

Lead: recognition (p. 567). - Known as a health problem “since Roman times”. About a century ago an Australian doctor identified children’s vulnerability (Taylor et al. 2010). Lead nonetheless stayed in paint “despite multiple early warnings”. - Petrol did the most harm, exposing “hundreds of millions of people”. Lead was added from the 1920s, and phase-out took “many decades”. - Industry motive. “a sad testament to the desire of companies to make profits regardless of the consequence to human health”. Less harmful alternatives “were not adopted because they were not patent-protected”, which would have “greatly reduced profits” (Ackerman et al. 2005, the same article as Heinzerling et al. 2005). A strong claim resting on one secondary source, stated as fact. - Bangkok’s sampled population averaged 40 µg/dl in the 1980s–1990s. Death may occur around 100 µg/dl. - EU path. - 1981: a limit of 0.4 g per litre. - Catalytic converters made lead “less attractive”. - 1985: Germany introduced a tax differential between leaded and unleaded petrol, and others copied it (Hammar and Löfgren 2004). - By the 2000 ban “there was little protest from any quarter”. - As of January 2011 only six countries still allowed leaded petrol: Iraq, Yemen, Algeria, North Korea, Burma and Afghanistan.

Lead: economics (pp. 567–568). - Economic analysis “was instrumental in convincing the Reagan administration” (US EPA 1985). - One IQ point is worth 2–3 % of lifetime income, “highest for females” (Salkever 1995). - There is no lower threshold for harm to children, “so all emissions impair cognitive function” (Schwartz 1994). - The discount rate dominates. The chapter says the rate “decisively influences the final monetary estimate” (p. 567). US economists discounted at 10 %; an EU study did not discount at all. In footnote 1, the US EPA’s private rate implied 8 US cents (1983 prices) per gram of lead, against EUR 5.90 (2004 prices) undiscounted (p. 568). The spread is well over an order of magnitude. The chapter attributes it to the discount rate, but it compares two different studies, from different years and in different currencies, which may also differ in method, so not all of the gap can be pinned on that one choice. - “IQ levels have been demonstrated to influence dramatically the relative affluence of nations, and are therefore a social good (see Figure 23.1)”. This is unreferenced; see Limitations. - Central estimate. At a social rate of 1.4 % (Stern; EC 2008), the damage is EUR 1.50 per gram emitted in urban areas (Andersen 2010). The 1.4 % matches Stern’s 0.1 % for social “mortality” plus 1.3 % average growth (p. 574). [Outside knowledge — verify: the EC (2008) guide to cost-benefit analysis recommends considerably higher social discount rates (3.5 %, or 5.5 % in cohesion countries), so citing it for 1.4 % looks odd.] The estimate is also sensitive to assumptions about the food chain (footnote 2: inhalation to ingestion 1:25) and “resuspension” of old deposits. - Historical scale. Petrol before the 1970s held “about 1 gram of lead per litre”. At EUR 1.50 per gram, this implies “annual costs of about 4–6 % of GDP in EU Member States”. This is a back-of-envelope figure: it applies the urban unit value everywhere and assumes 1 g/l. - Today. New emissions cause damage of “less than 0.1 % of GDP”. Further cuts in products such as paints and toys remain desirable. - Legacy. “All the lead emitted during the 20th century is dispersed across the environment” and some returns to the air every year. Urban topsoils justify “the resurfacing of playgrounds”. - Lower bound. Cardiovascular effects in adults (Menke et al. 2006) are excluded, so “some damage estimates are lower-bound conservative values” (Gould 2009). “This tendency for cost-benefit analysis to lean toward conservative values further strengthens the case for adopting a precautionary perspective” (p. 568).

Figure 23.1 “Diminished IQ stunts economic development” (p. 568, read visually). Two bell curves of IQ across a population:

• The labels are “mentally retarded” and “gifted”, both in quotation marks. The source is www.ourstolenfuture.org.
• It illustrates what happens to the tails when a population average shifts (the implied population is about 260 million). It is not an estimate for lead, and it does not show anything about national affluence despite its title.

Mercury (p. 569). - The harm. Mercury stays in the atmosphere for 6–18 months, so it is a “global pollutant”. It is converted to methylmercury and concentrates up the food chain into predatory fish such as tuna and shark. It crosses the placenta and enters breast milk, and reduces IQ. - Recognition lag. Minamata disease was acknowledged in spring 1956. Tracing the cause took “several years”. The Japanese government “‘announced its opinion’” that factory effluent was responsible only in 1968. An official apology came in 2006. Low-level effects on IQ continue. - Estimates. The box says mercury, like lead, has “only harmful effects”, unlike metals the body needs in small amounts. It concedes “the wide uncertainty in these figures”, but says they show the impact “is far from trivial”. - Axelrad et al. (2007) pooled cohorts from New Zealand, the Seychelles and the Faroes: a child loses 0.18 IQ points per part per million of mercury in the mother’s hair. They assume a threshold of 0.1 µg/kg body weight a day, and lifelong, irreversible effects. - Spadaro and Rabl (2008): about USD 1,500 per kg emitted with a threshold, USD 3,400 without. The chapter calls these “conservative” against a US EPA study implying more than USD 6,000 per kg (33 t → USD 210m; Griffiths et al. 2007). It notes that the EPA study “used a different methodology” and assumed no threshold for prenatal exposure, so the two figures are not like for like. - Greenland: three-quarters of newborns have elevated blood mercury despite “hardly any local emissions”. Damage is USD 59m a year, about USD 59,000 per newborn (cited as Hylander and “Goodwin”; the reference list gives Goodsite). The Arctic diet “serves as a sink for global mercury emissions”. - Projection: global emissions will rise “20 % by 2020 (relative to 2005 cf. AMAP, 2011:143)”.

23.3 Nitrate in drinking water (pp. 566, 570–571)#

• The standards. The 1980 Drinking Water Directive (80/778/EEC) set a maximum admissible concentration (MAC) for nitrate “based directly on WHO guidelines”. Water at the MAC is “suspected” of raising the risk of methaemoglobinaemia (“blue baby” syndrome) in infants. At the MAC, drinking water would “double the amount of nitrates” an average person takes in from other sources (p. 566). The MAC is 50 mg NO3/l; the directive added a “guide value” of 25 mg/l (pp. 566, 570).
• Guide value dropped. In the 1998 revision “the precautionary guide value… was deleted”. “The main reason for abandoning a guide value was the absence of scientific proof to underpin it” (p. 570).
• Related law.
• The Nitrates Directive (91/676/EEC) capped nitrogen applied to land and set manure rules. The Urban Waste Water Directive (91/271/EEC) limited nitrogen from sewage and effluent. Eutrophication was also a motive.
• “Farming interests in particular have requested derogations and exemptions”.
• No cost-benefit analysis. “No analysis has so far been carried out in Europe” comparing the benefits of nitrate regulation with its costs. This sits awkwardly with the chapter’s own citation of Van Grinsven et al. (2010), whose title (reference list, p. 580) calls it “a tentative cost-benefit assessment” of nitrate and colon cancer in the EU.
• Contested evidence.
• The gastric cancer link is disputed. WHO (2007: 12) calls a link via nitrosation in the body “possible”.
• WHO “normally requires at least three different studies” before treating a link as conclusive, and there are “few well-designed epidemiological studies”.
• Why the evidence is thin. Cohort studies are “both laborious and costly”, and historical exposures are hard to reconstruct. Only two cohorts exist, and their results are “ambiguous”:
• Weyer et al. (2001) found more bladder cancer among 10,000 Iowa women aged 55 or over, in a farming state with high nitrate in its water;
• an uncited Dutch study found no increase, “presumably because nitrate levels in Dutch public water supply are rather low” (p. 570).
• Evidence arrived after the decision. “The Iowa study was published after the EU Council of Ministers abandoned the 25 mg… guide value”. “Along with other types of studies”, it suggests effects “well below the MAC-value”, with thresholds of 15–25 mg/l. “Hence there appear to be costs of inaction at stake”, but “how significant are they?”
• Estimating potential costs, “the risks in other words” (italics on “potential” in original).
• The World Bank (2007) produced rough estimates. The reference list shows this is a study of pollution costs in China.
• Van Grinsven et al. (2010) put colon cancer costs at EUR 0.7 per kg of fertiliser nitrogen.
• Andersen et al. (2011, the lead author’s own work) estimated by river catchment, mainly from Weyer. The mean is EUR 0.3 per kg of nitrogen lost; the UK and Belgium are at EUR 1.3 per kg, and some urban areas are “much higher”.
• “Consequently”, EUR 2.6bn a year for the UK alone, “comparable to” an earlier estimate of all UK agriculture’s external costs (Pretty et al. 2001) (p. 570). The calculation is not shown; it appears to follow from the Andersen et al. unit values.
• Conclusion. Given nitrogen’s other costs (surface water, ammonia, nitrous oxide), “these figures taken together would suggest that a precautionary approach is warranted”. “The abandoned guide value for nitrate in drinking water also deserves reconsideration” (pp. 570–571).

23.4 Air pollution: valuing mortality risk (p. 571)#

• The US figure. The US uses wage-risk analysis (pay premiums for riskier jobs) to value preventing a death at about USD 5.5m (1999 prices).
• The European adjustment.
• Europe uses values from transport economics. But road victims are middle-aged and lose 30–40 years, while air pollution victims are mostly over 65 (Pope 2002) and “may stand to lose only a few years”.
• The European Commission set the value of a statistical life (VSL) at EUR 1.4m (2000 prices), reduced to about EUR 1m for air pollution because of victims’ age.
• “Similar adjustments proposed in the United States were however met with public outcry” and were branded a “‘senior death discount’” (NYT 2011).
• VSL versus VOLY. VSL “remains the conventional metric” in the economics literature, but it is being challenged by the value of a life-year (VOLY), which gives more conservative figures (Hofstetter and Hammitt 2002). The chapter itself raises the ethical question: should late-life years be valued more (“as more precious”) or less (because of reduced “health and vigour”)?
• VOLY can be derived from VSL by assuming the VSL covers “3–40 years as in road traffic”. This is probably a typo for 30–40.
• OECD (2006) recommends VSL for acute deaths and VOLY for chronic exposures such as air pollution or nitrate.
• US contrast. The chapter attributes to the “Science Advisory Board of the US EPA” the view that VSL is “the only solid value available”, but the citation is an EPA white paper (US EPA 2010: 12). The US therefore “accords a higher monetary value to environmental risk reduction than Europe”.
• Politics. “US economists are facing decreased political acceptance of the recommended approach”. The Obama administration, “acting under political pressures”, withdrew proposed ozone restrictions “despite the high costs of inaction”. No source is cited for this decision (the NYT 2011 citation belongs to the “senior death discount” sentence).
• Robustness argument. VOLY figures “provide perhaps only a lower-bound estimate”, but “in many cases these conservative values are already sufficient to justify abatement action”.
• CAFE. The EU’s Clean Air For Europe programme reported both sets of figures and “left it to policymakers” to choose. Costs came to 3–5 % of EU-25 GDP.
• “The acknowledgement that economics may not offer a mature consensus corroborates the role and significance of the precautionary principle”. Stern likewise reported ranges (p. 571).

Box 23.2 SO2 and other air pollutants (pp. 572–573)#

• Health effects. SO2 comes from burning coal. It causes acidification (see Late Lessons 1, Ch. 10) and forms secondary sulphate particles.
• Asthmatics react below 0.1 ppm when exercising. Healthy adults react at 5, 10 and 20 ppm (ATSDR 2011).
• Sulphate correlates strongly with mortality (Smith et al. 2009).
• Recognition.
• The chapter says fatal effects were reported “as early as 1929 in Wallonia”. The Meuse Valley fog was December 1930 [outside knowledge — verify].
• London 1952: “as many as 13 000 deaths” (Bell et al. 2003).
• Taller chimneys “alleviated the problem” but “caused the acidification of rain, lakes and rivers in Scandinavia”.
• Japan: more than 100,000 people were registered as affected over three decades from about 1950. “Yokkaichi Asthma” was one of the four “big pollution diseases” (p. 572).
• Contested economics.
• In the 1970s “some claimed” the problem was not “severe enough to warrant expensive action”. “Some even called it a ‘million dollar problem with a billion dollar solution’”, cited to a 1977 Nature opinion piece. That piece’s title ends in a question mark (‘Million-dollar problem—billion-dollar solution?’, p. 579), so it may have questioned the framing rather than endorsed it. The link to mortality “was only gradually accepted”.
• Counting deaths as well as illness, the costs “should indeed have been counted in billions too”.
• Pope et al. (1995, 2002). An American Cancer Society cohort of more than 500,000 people, with controls for confounders and individual death certificates. Most deaths came from chronic exposure. “The US EPA decided to require a complete reanalysis of the data by an independent research team before finally accepting the findings — it led to the same results” (p. 572).
• CAFE (2005), with a WHO review of the evidence.
• VOLY basis: EUR 276bn a year (3 % of EU-25 GDP). VSL basis: EUR 427bn (about 5 %).
• Poland and other new Member States: 15–22 % of GDP (reference year 2000).
• These figures cover combined exposure to particles from SO2, NOx and primary PM, not SO2 alone.
• Biodiversity costs are omitted, yet “the figures suffice to justify further action”.
• Damage is EUR 5–9 per kg of SO2, while marginal abatement costs “start from below EUR 1 per kg” (Rive 2010). So “there are large benefits for nearly all efforts regarding SO2 abatement” (p. 573).
• Multi-pollutant complexity. Non-linear chemistry, for instance for ozone, calls for integrated assessment models. “The technically feasible reduction of air pollution has been estimated at EUR 56–181 billion equivalent to 0.6–2 % of GDP in EU, which are the costs that could be avoided”. These appear to be avoided damages, not abatement costs; the wording is ambiguous.
• Japan.
• In 1975 abatement investment was “18 % of Japanese capital investment and 6.5 % of GDP”. OECD (1977) judged the macroeconomic impact “practically negligible”.
• The investment “accelerated technological innovation, raised product quality and lowered technical costs”. Japanese firms still hold many patents and licences, “demonstrating the economic significance of being a pioneer” (p. 573).
• The 6.5 % figure looks high if the 18 % refers to industrial investment [outside knowledge — verify].
• Figure 23.2, global SO2 emissions 1850–2005 by sector (read visually).
• Emissions reach about 60,000 Gg by 1950 and peak at about 130,000 Gg around 1970–1990. They dip to about 110,000 around 2000 and rise to about 120,000 by 2005. Energy is the largest source.
• The source, Smith et al. 2011, is missing from the reference list, and the post-2000 rise is not discussed.

23.5 Discounting (pp. 571, 574)#

• The timing problem. Many costs of inaction fall in the future. Skin cancers from ozone loss peak “only after 60 years” (pp. 571, 574).
• Why discount. Two “textbook reasons” are given: the investor “may not be alive”, and growth means they will be richer. This is simplified. It leaves out the term for how quickly the value of extra consumption falls as incomes rise, and it reduces pure time preference to mortality.
• The equity objection. The chapter attributes this objection to “many non-economists”: to them, discounting “is at odds with the core idea of intergenerational equity that is central to sustainability”. A “related problem” is that many economists apply “discount rates that are typically used in the corporate sector — rising up to as much as 10 %” without reflection.
• Stern’s approach. “society is not mortal”, so the rate for social “mortality” is 0.1 %. The consumption component should track expected growth; Stern’s PAGE model averaged 1.3 %. Avoided damages of 5–20 % of consumption “would well exceed the involved costs”.
• Table 23.1 (Stern via OECD 2008: 96).

The text says the table shows “how the social cost of carbon depends crucially on the discount rate”. It does not show a social cost of carbon; it shows costs of inaction as a share of GDP. The point stands all the same: a 1.5-point rise in the rate cuts the estimate by about 70 %, yet even 4.2 % exceeds Stern’s 1 % mitigation cost. - Shorter horizons. Scrubbers last 10–20 years, so there is less “shrinking”. For mortality, “discounting is not very important”, because VSL rises with growth and cancels the consumption component, “leaving only the 0.1 %”. This rests on unstated assumptions about how the VSL and consumption values scale with income (p. 574).

23.6 Concluding remarks (pp. 574, 577)#

• Contingent valuation. Diamond and Hausman (1994) raise two problems with asking people what they would pay (contingent valuation, CVM):
• hypothetical answers overstate, “neglecting their level of income”;
• answers are inconsistent: the value for cleaning one lake “might be similar” to that for five, and people may simply want “to contribute EUR 50 for a good purpose”.
• “These are strong methodological critiques”.
• European attitudes to CBA.
• Europe “has acted cautiously in mandating formal requirements for cost-benefit analysis”. It is seen as “a somewhat American-style approach”. The Dutch renamed it SCBA “to sweeten the pill” (RMNO 2008).
• The UK has taken it up most and has pushed it into EU processes. Commission impact assessments cite CBA findings. For 15 years, projects funded by the EU Structural Funds have required a CBA, guided by an EC (2008) manual (pp. 574, 577).
• Progress claimed. Despite “great methodological challenges”, “improvements in scientific knowledge and modelling techniques have helped to significantly influence the ratios of benefits to costs in favour of regulation”. No examples are given beyond the chapter’s own cases.
• Conclusions (p. 577).
• “Precautionary action can be justified by using credible estimates of the costs of inaction.”
• Lead shows that understanding “only some of the benefits” can be enough to prompt action.
• Nitrate shows “risk calculations can be useful and help prompt immediate action”, even though “full proof will likely take decades to materialise”.
• For air pollution, a range with conservative lower bounds “will in many situations encourage more rather than less abatement effort”.
• The interdisciplinarity and diffuse-interests message is repeated.
• ODS and mercury are not mentioned.

Box 23.3 Benefits of early action on ODS (pp. 575–576)#

• The harm. ODS reach about 25 km and destroy ozone (Molina and Rowland 1974; Late Lessons 1, Ch. 7). UV causes skin cancer, cataracts and reduced resistance to infection (WHO 1994), and UVB drives both melanoma and non-melanoma skin cancer. The box also notes that too little UV causes problems too, such as vitamin D deficiency.
• Recognition. “Scientist Richard Scorer of Britain’s Imperial College was a respected environmentalist, but sided with industry on the safety of CFC’s“. In 1975 he said: “‘The only thing that has been accumulated so far is a number of theories’” (Roan 1989: 61).
• The Montreal Protocol.
• Despite scepticism, ODS control became a “cause célèbre” as an early coordinated global phase-out.
• The Protocol came into force on 1 January 1989. Annan (2003) called it “‘perhaps the single most successful international environmental agreement to date’”. Every country (196) had ratified it by 16 September 2009.
• The ozone layer “should return to its pre-1980 levels sometime between 2050 and 2075” (UNEP 2009a) (p. 575).
• The “world avoided” (Newman et al. 2009). Without the Protocol, ozone “could have reduced in thickness by as much as 67 % by 2065”. Figure 23.4 (read visually) shows the midsummer noon UV index at 30–50°N rising from about 10 in 1980 to roughly 30 by 2065 in that scenario, while the reference future stays flat at about 10. Figure 23.3 is NASA’s image of the Antarctic ozone hole, 1 October 2005.
• Health benefits (p. 576).
• UNEP (2009b), assuming a 48 % ozone loss was prevented by 2050, puts cases prevented at more than 20 million skin cancers and 130 million cataracts.
• In north-western Europe by 2050, about 47,000 skin cancer cases will be avoided each year. 14,000 extra cases a year are still expected because of damage already done and the 60-year latency.
• About 500 deaths a year are avoided. This follows from 99.5 % of cases being non-melanoma at 1 % mortality and 0.5 % melanoma at 24 %. The arithmetic checks: 47,000 × 0.01115 ≈ 524.
• Using US ratios (Sunstein 2007), the benefits are “not less than EUR 3 billion” a year for Europe and EUR 10–11bn for the EU-27. This is a multi-step scaling, and the uncertainty is not stated.
• Climate benefits.
• Halocarbons and ODS account for 13 % of present warming (IPCC 2007). In 1990 they amounted to 7.5 ± 0.4 GtCO2-eq a year, about 33 % of fossil CO2 emissions.
• With 3 % annual growth, 2010 emissions would have been about 14 Gt. So the Protocol saved “about 215 GtCO2-equivalent” over two decades. This is “bigger than cuts required by developed countries under the Kyoto Protocol”.
• The EUR 2,150bn is framed as a counterfactual. Had ODS been counted under Kyoto rather than regulated separately, these emissions (“mainly from developed countries”) would have had to be offset. At EUR 10 per tonne that would have cost about EUR 2,150bn, about 0.5 % of OECD countries’ annual GDP over the two decades.
• Check: 215 Gt is the whole counterfactual stream (7.5 Gt growing 3 % a year for 21 years sums to about 215). It therefore treats actual emissions over the period as zero. [Outside knowledge — verify: the widely cited figure from Velders et al. (2007) is about 135 GtCO2-eq avoided over 1990–2010, so the chapter’s figure is probably an overestimate.]
• It amounts to 7–12 years of CO2 growth. Counting “the full reduction achieved since” Molina and Rowland’s 1974 warning, “as many as 30 to 45 years may have been gained” (Velders et al. 2007).
• From these estimates the authors infer that “early warning scientists must be attributed a role in curbing ODS-consumption being equally important to the Montreal protocol itself” (p. 576). This is an inference; no estimate separates the scientists’ effect from the treaty’s.

References (pp. 577–580)#

• Duplicates and mismatches.
• Heinzerling, Ackerman and Massey (2005) is listed twice.
• “Lourdes” should be Schnaas, and “Goodwin” should be Goodsite.
• Smith et al. 2011 (Figure 23.2) and the Dutch nitrate study are not cited or listed.
• Bell et al. is dated 2003, but the volume cited, EHP 112(1), is from 2004 [verify].
• Lanphear et al. (2005), De Roos et al. (2003), Morgenstern (1997) and NASA (2010) are listed but not cited.
• “IPCC, 2007” carries the title of the IPCC/TEAP special report on safeguarding the ozone layer, which is usually dated 2005 [outside knowledge — verify].
• The Reddy and Braun entry has no year, and the page range for Taylor et al. (2010) is garbled.
• Mix of sources. Mostly peer-reviewed epidemiology and economics, plus agency reports (US EPA, OECD, AEA/CAFE, UNEP, WHO, AMAP). Some advocacy and popular sources carry weight: ourstolenfuture.org, the NYT, the Japan Times, and Roan’s popular history for the Scorer quote.

Case timeline (multi-case, as stated in the chapter)#

The authors’ own lessons and conclusions#

Lessons derived from the evidence presented (with an assessment of fit): 1. Economic analysis can secure action. The lead case: IQ loss was converted to lost income and persuaded the US administration (pp. 566–567, 577). This is a well-known episode, only summarised here. 2. Estimates are often lower bounds because pathways are omitted. Examples are lead’s cardiovascular effects (p. 568), biodiversity (p. 573) and ODS climate effects (p. 576). This is well supported for these examples; “often grossly underestimated” (p. 564) goes beyond them. 3. Valuation choices (discount rate, VSL versus VOLY) drive results (pp. 568, 571, 574). The chapter’s own numbers demonstrate this. 4. Conservative VOLY bounds already justify air pollution abatement (pp. 571, 573). Supported by the CAFE figures and by damage per kg far exceeding abatement cost, although the abatement cost rests on a single citation. 5. Survey-based willingness-to-pay values are unreliable for unfamiliar hazards (pp. 566, 574). This follows from the critique cited. 6. Proof lags evidence. The nitrate guide value was dropped before the evidence arrived (p. 570). The sequence is clearly dated within the chapter (1998 deletion, 2001 Iowa study), though the evidence itself remains “ambiguous” by the chapter’s own account.

Recommendations and advocacy: - Estimate on the basis of evidence, not proof, including linear scaling down from high occupational doses (p. 566). - Use a low social discount rate (1.4 %; Stern’s 0.1 % pure time preference) rather than corporate rates (pp. 568, 574). - Resurface playgrounds and keep reducing lead in products (p. 568). - Take a precautionary approach to nitrogen and reconsider the 25 mg/l guide value (p. 571). - Report both VSL and VOLY ranges and let policymakers decide (pp. 571, 577). - Produce estimates across disciplines, not “economists alone” (pp. 564, 577). - Thesis: “Precautionary action can be justified by using credible estimates of the costs of inaction” (p. 577). This is normative and conditional. - Credit early warners as much as the Montreal Protocol (p. 576). This is an attribution claim.

Mechanisms and dynamics#

1. Monetised harm works as a de facto burden of proof. - The founding premise is that policymakers respond only once costs of inaction are priced (p. 564). Until someone monetises the harm, the status quo is presumed acceptable. - The chapter meets this burden rather than contesting it. - Nitrate is a related case, though it turns on proof rather than price. The guide value was removed because of “the absence of scientific proof”, and no cost-benefit comparison of nitrate regulation has been done in Europe (p. 570). - The US ozone decision shows that even when the costs of inaction are judged high, action is not guaranteed (p. 571).
2. The costs of acting and not acting are asymmetric. - Prevention costs are tangible and allocated; the costs of inaction are diffuse and delayed (p. 565). - Diffuse interests “have difficulty making their voices heard” (pp. 564, 577). - Concentrated interests appear in the cases: farming interests seeking derogations (p. 570), and a petrochemical industry that reportedly passed over less harmful alternatives because they “were not patent-protected” (p. 567). - The chapter illustrates this dynamic but does not analyse it.
3. Knowledge has a solid core and blurred edges; decision cultures demand proof. - Warnings sit in the blurred zone (p. 566). - Economists “will often ask whether there is ‘scientific proof’” (p. 566). The Council treated lack of proof as grounds to delete a precautionary value (p. 570). WHO’s three-study convention is a formal threshold (p. 570). - The law asks less: art. 114(5) needs only “scientific evidence” (p. 566). There is a gap between what law permits and what decision cultures demand.
4. The way evidence is produced keeps “proof” scarce. - Chronic, long-latency harms need cohort studies that are “laborious and costly”, with exposures hard to reconstruct (p. 570). So absence of proof partly reflects research cost, not safety. - Well-quantified harms (IQ) get counted; poorly quantified ones do not (PCB damage, p. 566; lead’s cardiovascular effects, p. 568). - Meanwhile abatement costs are concrete: “billions of euros” for PCB (p. 566).
5. Estimates are biased low. CBA “lean[s] toward conservative values” (p. 568). The authors turn this into a precautionary argument: the known error runs one way. They do not examine bias in estimates of the costs of action.
6. Valuation choices embed ethics and dominate outcomes. - Discount rate. Two lead estimates differ by well over a factor of ten, which the chapter attributes to the discount rate, although the studies also differ in date, currency and possibly method (p. 568, fn. 1). Stern’s figures vary 3.5-fold across 1.5 points (p. 574). Corporate rates are applied “without reflecting” on context. “Many non-economists” see equity between generations at stake (p. 574). - VSL versus VOLY. Adjusting for age cuts the value of deaths among people over 65. In the US this was branded the “senior death discount” (p. 571). The chapter itself raises the underlying question of whether late-life years are worth more or less (p. 571), but it does not resolve it. It leaves the choice to policymakers. - Valuing children’s IQ by future earnings. Losses come out “highest for females” (p. 567). In my reading this reflects labour-market differences; the chapter does not discuss it. - Across countries. Poland’s 15–22 % of GDP (p. 573) probably reflects both higher exposures and EU-wide mortality values applied to a lower-income economy. The chapter does not discuss this (my observation).
7. Harm has to be made legible in decision-makers’ terms. - The lead case worked because science produced a quantitative endpoint (IQ points) that economics could convert into a metric decision-makers already valued (p. 566). That persuaded the Reagan administration (p. 567), which I characterise as deregulatory [outside knowledge; the chapter does not say this]. - Tracing harm by modelling the exposure pathway displaced surveys asking parents to price something they could not grasp (p. 566).
8. Institutions scrutinise inconvenient findings harder. The EPA required an independent reanalysis of Pope et al. before accepting them, and the result “led to the same results” (p. 572). A higher bar for findings with big regulatory stakes delays action, but surviving it consolidates acceptance.
9. Harm is moved rather than resolved. - After the London smog, taller chimneys “alleviated the problem” locally but dispersed pollution regionally and “caused the acidification of rain, lakes and rivers in Scandinavia” (p. 572). - Global mercury ends up in Greenland, which has “hardly any local emissions” (p. 569). - ODS came “mainly from developed countries” while the harm is global (p. 576). - Displaced harm falls outside the emitter’s accounting.
10. Persistence, latency and legacy.
    • 20th-century lead is still in soils and blows back into the air (p. 568). IQ loss is irreversible (pp. 567, 569).
    • Skin cancer’s 60-year latency locks in 14,000 extra cases a year despite action (p. 576). The ozone layer recovers only by 2050–2075 (p. 575).
    • Delay builds up stocks, and stocks commit future harm.
11. How a transition is designed affects resistance. The EU lead phase-out ran in stages: a limit, then complementary technology (catalytic converters), then a tax differential. The 2000 ban met “little protest” (p. 567). The implication is that staged, incentive-aligned transitions defuse opposition. Plausible, but asserted.
12. Commercial incentives shape which technology is chosen. Lead was reportedly preferred because the less harmful alternatives “were not patent-protected” (p. 567). If so, the ability to capture profit through intellectual property steered a harmful choice. The claim has a single source.
13. Innovation and first-mover advantage. Japan’s abatement drive had “practically negligible” macroeconomic cost and brought innovation, better quality and export patents (p. 573). This is the chapter’s only real engagement with innovation, and it rests on one case.
14. Interactions within systems. Controlling ODS delivered climate benefits larger than Kyoto’s required cuts (p. 576). The nitrogen cascade links water, air and climate (pp. 570–571). Multi-pollutant chemistry is non-linear (p. 573). Single-issue assessments miss these cross-domain effects, which is one reason the authors think costs of inaction are underestimated.
15. Framing and language.
    • Scorer’s “a number of theories” (p. 575) dismisses mechanistic evidence while waiting for proof.
    • The “million dollar problem with a billion dollar solution” (p. 572) sets well-counted costs against poorly counted harms. The chapter attributes it to “some”, not to industry. The Nature piece it cites puts the phrase as a question (p. 579).
    • The Japanese government “announced its opinion” (p. 569): hedged official language about causation.
    • The Dutch relabelled CBA as SCBA “to sweeten the pill” (p. 577).
    • The “senior death discount” (p. 571) is public framing that defeated a technical adjustment.
    • The chapter’s own device, “costs of inaction”, is analytically identical to “benefits of action” (p. 565) but puts the status quo on the loss side of the ledger. The chapter does not reflect on this persuasive choice.
16. Early warners change behaviour before regulation does. The ODS box counts “the full reduction achieved since” the 1974 warning, which gains up to 30–45 years of climate delay. From this it infers that early-warning scientists deserve credit “equally important to the Montreal protocol itself” (p. 576). It does not separate the scientists’ effect from that of later regulation. This links to Chapter 24’s theme of protecting early warners.
17. The political legitimacy of the tools matters. Europe sees CBA as “American-style”, the UK promotes it, and the Structural Funds require it (pp. 574, 577). In the US, valuation based on the VSL faces “decreased political acceptance” (p. 571). Whether estimates prompt action depends on whether the tool is accepted as well as on technical quality.

Transferable insights (technology-neutral)#

1. When decision-makers wait for monetised harm estimates, the burden shifts onto those raising warnings, and the status quo prevails until harm is priced. Evidence: p. 564; nitrate (p. 570, though that case turned on the demand for proof rather than a price); US ozone (p. 571, where pricing alone was not enough). Suggestive. It is stated as a general pattern without systematic evidence, though the chapter’s cases are consistent with it.
2. Concentrated, near-term costs of acting outweigh diffuse, delayed costs of not acting, tilting decisions toward inaction and quieting those affected. Evidence: pp. 564, 565, 566 (PCB), 570, 577. Moderate. A well-established pattern in political economy, illustrated here rather than tested.
3. Early warnings sit in a zone of partial understanding. Demanding proof rather than weighing evidence postpones action, possibly for decades. Evidence: pp. 566, 570, 577. Moderate. Conceptually sound, and the nitrate case illustrates it.
4. Absence of proof can reflect how hard and costly the proving research is, not absence of harm. Protective standards may be withdrawn for “absence of scientific proof” just before the evidence arrives. Evidence: p. 570, where the guide value was deleted in 1998 and the Iowa study appeared in 2001. Moderate. Clear in one case, though the evidence is still contested.
5. Accounting for harm captures what is already well quantified, so early estimates tend to be lower bounds, and newly recognised pathways push costs up. Evidence: pp. 564, 568, 570–571, 573, 576. Moderate for the mechanism. Asserted for “often grossly underestimated”: the chapter does not test it against counter-cases.
6. Choices that look technical (discount rate, horizon, VSL versus life-years, thresholds) can shift estimates several-fold, and embed ethical judgements about future generations, older people, and whose income counts. The spreads run from about 1.5× (VOLY versus VSL, p. 573), through about 2× (mercury with and without a threshold, p. 569) and 3.5× (Table 23.1, p. 574), to well over an order of magnitude in the lead comparison, which also mixes studies (p. 568 fn. 1). Evidence: p. 568 fn. 1; p. 569; pp. 571, 573; Table 23.1 on p. 574. Strong for sensitivity, since the chapter’s own numbers demonstrate it. The ethical dimension is raised by the chapter (pp. 571, 574) but not worked through.
7. Under methodological disagreement, report the range and check whether the decision changes across it. If even the conservative bound justifies action, the dispute need not delay it. Evidence: pp. 565, 571, 573, 574. Moderate. The logic is sound and CAFE is a real example, but it depends on abatement-cost estimates that go unscrutinised.
8. Harm gains traction when science yields an endpoint translatable into something decision-makers already value. Asking the public to price an unfamiliar harm yields unstable numbers. Evidence: pp. 566–567, 574. Moderate. Well-known lead history plus a widely cited critique, but the ethics of the translation go unexamined.
9. Fixing a problem locally by dispersing it exports harm to other places and populations who did not cause it. Evidence: pp. 569, 572, 576. Strong for these well-documented historical cases. Moderate as a general rule.
10. Persistent, accumulating hazards commit future harm regardless of later action; delay builds legacy stocks. Evidence: pp. 567–569, 575–576. Strong for the physical facts. Moderate as a general inference about the cost of delay.
11. Commercial incentives such as patentability can steer a sector toward a more harmful option when safer alternatives exist. Evidence: p. 567. Suggestive. A single-source assertion. The account is common in the historical literature but its details are debated [outside knowledge].
12. Staged transitions combining limits, complementary technologies and price signals can erode opposition before a final prohibition. Evidence: p. 567. Suggestive. One case, with the causal link asserted.
13. Institutions may demand independent reanalysis of findings with large regulatory consequences. This delays action but can settle the question when the results replicate. Evidence: p. 572. Suggestive. One case.
14. Early investment in reducing harm need not be macro-economically costly and can confer innovation and market advantage on first movers. Evidence: p. 573. Suggestive. One case, a figure that needs checking, and the claim is contested in the wider literature.
15. Interventions in interconnected systems can produce large co-benefits (or co-harms) outside the domain assessed, which single-issue appraisals miss. Evidence: pp. 570–571, 573, 576. Moderate. The ODS-climate link is well established (Velders et al. 2007).
16. Early warners can shift behaviour before formal rules exist, so their contribution may rival that of the eventual regulation. Evidence: p. 576. Suggestive. It rests on a counterfactual the chapter does not examine.
17. Whether an analytic tool is seen as legitimate shapes whether its results matter. Sound estimates can be overridden by political pressure or cultural resistance. Evidence: pp. 571, 574, 577. Suggestive. Anecdotal but consistent.
18. Where evidence is thin, potential harm can be estimated by scaling down from well-documented high-exposure effects. Evidence: p. 566. Asserted. It is not validated, and it can mislead in either direction where the dose-response is not linear.
19. Estimates built within one discipline miss relevant knowledge. Interdisciplinary estimates may better represent diffuse interests. Evidence: pp. 564, 577. Asserted. No evidence is offered that they have changed outcomes.

Limitations, contestation and bias check#

Advocacy versus analysis. - The technical expositions (valuing mortality, discounting, contingent valuation) are broadly fair. - The boxes and conclusions select and frame in favour of action. - The chapter’s thesis, “Precautionary action can be justified…” (p. 577), is normative. - The claim about the leaded-petrol industry’s motive (p. 567) is presented as fact on one secondary source. The account is common in environmental histories, but how practical the alternatives were at the time is debated [outside knowledge — cross-check LL2 Ch. 3].

One-sided on the costs of action. - There are no false positives and no cases where costs of inaction were overestimated. - Costs of action are quantified and weighed only where they are small: Stern’s 1 %, SO2 abatement below EUR 1 per kg, and Japan’s “negligible” macroeconomic costs. One large abatement cost is mentioned: PCB, “counted in billions of euros” (p. 566). It is used to show that damage costs are unknown, not weighed against them. Conflicts over the costs of the nitrate directives are noted but not quantified (p. 570). - The claim that improved science and modelling have shifted benefit–cost ratios “in favour of regulation” (p. 577) is unsupported. - The chapter never asks how accurate ex-ante abatement-cost estimates have been.

Case selection and hindsight. - Lead, SO2, mercury and ozone are settled cases, so “estimates were too low” benefits from hindsight. - Nitrate is the only live case. The authors are candid that its evidence is “contested” and “ambiguous” (p. 570), yet they conclude that the figures “would suggest that a precautionary approach is warranted” (p. 571). The hedge is in the wording, but it does not carry through to the p. 577 conclusions. - The conclusion that nitrate “demonstrates that risk calculations can… help prompt immediate action” (p. 577) is not borne out by the chapter’s own account. There, no action followed: the guide value was dropped and no CBA was done. The case shows that such calculations could be made, not that they prompted anything.

Weight of the nitrate evidence. - The UK figure of EUR 2.6bn rests mainly on one US cohort (Weyer 2001), transferred to European catchments by the lead author’s own model. The calculation is not shown. - The Dutch null result is uncited and explained away by a presumption. - The chapter says no European cost-benefit analysis of nitrate exists (p. 570), yet it cites Van Grinsven et al. (2010), titled “a tentative cost-benefit assessment” for the EU (p. 580). - IARC’s 2010 Group 2A classification of ingested nitrate/nitrite under nitrosating conditions is not mentioned, although it would have strengthened the case [outside knowledge — verify].

Self-citation. The key numbers come from the lead author’s own work: the EUR 1.50 per gram for lead, the nitrate unit values, the EUR 2.6bn, and the lead modelling. This is not improper, but these figures get little independent corroboration within the chapter. The only cross-checks offered are Van Grinsven et al.’s EUR 0.7 per kg N for nitrate and a loose comparison of the UK total with Pretty et al. (2001).

Unreferenced or questionable claims. - IQ and “the relative affluence of nations” (p. 568). - The claim is unreferenced and gestures at a literature on national IQ that is contested on methodological and ethical grounds [outside knowledge]. - Figure 23.1 comes from an advocacy site, shows a hypothetical distribution shift rather than national affluence, and uses dated clinical labels. - Its statistical point, that small mean shifts change the tails substantially, is valid if IQ is normally distributed. - Lead costs of “4–6 % of GDP” (p. 568) is an extrapolation of the urban unit value. - Japan’s “6.5 % of GDP” (p. 573) needs checking [verify].

Errors and inconsistencies. - “1929” for what is probably the Meuse Valley episode of December 1930 (p. 572) [outside knowledge — verify]. - Table 23.1 is described as the social cost of carbon (p. 574). - “3–40 years” (p. 571). - The Annan quote may have had “environmental” inserted (p. 575) [outside knowledge — verify]. - A “2005” CDC recommendation is cited to a 2012 report (p. 567). - A view attributed to the “Science Advisory Board” is cited to an EPA white paper (p. 571). - “No analysis” of nitrate costs and benefits in Europe (p. 570) sits against the cited Van Grinsven et al. (2010) “tentative cost-benefit assessment”. - The ODS “215 GtCO2-equivalent” saved (p. 576) equals the whole counterfactual emission stream, so it implicitly treats actual emissions as zero [outside knowledge — verify: about 135 Gt is the commonly cited figure]. - The 1.4 % social discount rate is cited partly to EC (2008), which I believe recommends higher rates (p. 568) [outside knowledge — verify]. - The US ozone withdrawal (p. 571) has no source. - Various reference errors (see above).

Stern and discounting. - Calling Stern’s finding “broadly accepted” (p. 565) understates the dispute. - Critics accepted that mitigation was warranted but contested its magnitude and urgency, largely over the discount rate [outside knowledge: Nordhaus, Dasgupta, Weitzman]. The chapter does not engage them. - The Ramsey-rule exposition is simplified. - The claim that for mortality “discounting is not very important” (p. 574) rests on unstated assumptions about elasticities.

Use of Heinzerling, Ackerman and Massey (2005). - The chapter says economic analysis “came to the rescue” of the lead phase-out (p. 566). - My understanding is that the article’s broader thesis is sceptical of CBA: past environmental successes, including the early decisions on lead, were achieved without it or despite it [outside knowledge — verify]. - If so, the chapter uses the source selectively. It should be retrieved.

Justice is underplayed for a Part titled “justice”. - The chapter raises intergenerational equity (p. 574) and the “senior death discount” (p. 571). - It notes but does not analyse: - Poland’s 15–22 % of GDP (p. 573); - Greenland’s imported mercury burden (p. 569); - the higher income loss for females (p. 567); - Stern’s equity weighting (p. 565). - It does not discuss the ethics of valuing children’s cognition by their future earnings.

Strategic tension. - By accepting that action waits on monetised estimates (p. 564), the chapter risks legitimising the very hurdle that delays action on poorly quantified hazards; nitrate is its own example. - It mitigates this through lower-bound framing and by treating estimates as “a starting point for a broader discussion” (p. 577). - It never considers that demanding estimates can itself be a delaying tactic.

In its favour. - It is transparent about ranges: CAFE, Stern, and mercury with and without a threshold. - It concedes contested evidence and strong methodological critiques. - It admits the political limits of economic argument (the US ozone decision). - It disclaims precision (p. 564). - Treating economics as an ally of precaution makes it more useful to sceptical policy audiences than a purely normative case would be.

Notable quotes#

1. “the burden of proof is often distributed such that policymakers only respond to early warning signals from environmental hazards once the costs of inaction have been estimated” (p. 564)
2. “the ‘costs of inaction’ are simply analogous to the benefits that can be obtained with proper controls” (p. 565)
3. “Early warning signals are often located in this blurred area where linkages are not fully understood.” (p. 566)
4. “The precautionary perspective (TFEU art. 191) is really about acting in the absence of scientific proof” (p. 566)
5. “The main reason for abandoning a guide value was the absence of scientific proof to underpin it.” (p. 570)
6. “This tendency for cost-benefit analysis to lean toward conservative values further strengthens the case for adopting a precautionary perspective.” (p. 568)
7. “The acknowledgement that economics may not offer a mature consensus corroborates the role and significance of the precautionary principle.” (p. 571)
8. “‘The only thing that has been accumulated so far is a number of theories’” (Richard Scorer, 1975, quoted p. 575)
9. “‘million dollar problem with a billion dollar solution’” (what “some” called SO2 in the 1970s. It is cited to a 1977 Nature opinion piece whose title poses the phrase as a question; quoted p. 572.)
10. “Producing cost estimates should not be left to economists alone” (pp. 564, 577)

Open questions#

1. On balance, does requiring cost-of-inaction estimates speed up or slow down action? The chapter’s own nitrate and US ozone examples cut against its optimism. This could be tested across the LL1 and LL2 cases.
2. How often have early estimates of the costs of inaction, and of the costs of action, proved too low versus too high? “Often grossly underestimated” (p. 564) rests on two examples.
3. What did Heinzerling, Ackerman and Massey (2005) actually argue about CBA and lead? (Source to retrieve.)
4. Was the nitrate guide value ever reconsidered, and what have post-2001 cohorts found below 50 mg/l? [hindsight strand]
5. What does OECD (1977) actually say about Japan’s abatement investment as a share of investment and of GDP? (Source to retrieve.)
6. How should valuation handle distribution? This covers uniform mortality values across unequal incomes (p. 573), imported harms (p. 569) and valuing children’s cognition by future earnings (p. 567).
7. How should credit for outcomes be split between early warners and formal regulation (p. 576)?
8. When is linear scaling from occupational doses a reasonable default, and when does it mislead (p. 566)?
9. How have official social discount rates changed since 2013? [hindsight strand]
10. Do Chs 24–25 develop Part D’s “justice” and “innovation” framing? (Cross-reading.)

Post-2013 pointers for the hindsight strand [outside knowledge — all to verify]#

• Leaded petrol. UNEP announced the end of leaded petrol for road vehicles worldwide in 2021, when Algeria’s last stocks ran out (see p. 567).
• Lead.
• Later global modelling that includes adult cardiovascular deaths puts lead’s costs at several percent of global GDP (e.g. Larsen and Sánchez-Triana, Lancet Planetary Health 2023). This supports the chapter’s lower-bound warning (p. 568).
• The CDC’s blood lead reference value was lowered to 3.5 µg/dL in 2021.
• Mercury. The Minamata Convention was adopted in 2013 and entered into force in 2017. UNEP’s Global Mercury Assessment 2018 can be checked against the AMAP projection (p. 569).
• Nitrate.
• The IARC Group 2A classification (2010) predates the chapter.
• Danish registry cohorts (e.g. Schullehner et al. 2018) report colorectal cancer associations far below 50 mg/l.
• The recast Drinking Water Directive (EU) 2020/2184 kept 50 mg/l. Check whether any guide value returned.
• Air pollution.
• WHO’s 2021 guidelines lowered the PM2.5 guideline sharply.
• The EU revised its Ambient Air Quality Directive in 2024.
• The US set its ozone standard at 70 ppb in 2015, after the 2011 withdrawal (p. 571).
• Ozone.
• WMO/UNEP (2022) project recovery to 1980 values around 2040 (most latitudes), about 2045 (Arctic) and about 2066 (Antarctic), compared with 2050–2075 on p. 575.
• The 2016 Kigali Amendment extended the Montreal Protocol to HFCs for climate reasons (compare p. 576).
• Unexpected CFC-11 emissions were detected in 2018.
• Discounting and climate targets.
• US OMB Circular A-4 was revised in 2023 with a lower default rate, and later politically reversed [verify].
• The UK Climate Change Act target was raised to net zero in 2019 (compare p. 565).

Audit log#

Independent fact-check against the full text extract (PDF pp. 563–582). I re-rendered and checked PDF pp. 563, 566, 570, 572 and 575–578 by eye (the title page, chapter summary, Figure 23.1, the nitrate page’s italics, Box 23.2 with Figure 23.2, Table 23.1, and Box 23.3 with Figures 23.3–23.4). No web sources were used. The figure readings, the Table 23.1 values, the author line and the Part D front matter were all confirmed.

• Part D framing: “non-patentable substitutes” changed to the chapter’s wording, “not patent-protected”.
• Authors: Andersen is noted as co-author, not lead author, of Pizzol et al. (2010).
• Stance: “political fact” softened to “common political reality (‘often’)”.
• Stance: the list of disciplines corrected to the chapter’s own (health, ecology, demography, modelling, science).
• Stance: the CBA-as-obstacle contrast is flagged as my framing, not the chapter’s.
• Stance: “call to restore” the nitrate guide value corrected to “deserves reconsideration”.
• Chapter summary: added the second stated aim (warnings as a basis for estimates where the science is less consolidated).
• Chapter summary: added biodiversity omissions (p. 573) to the support for “often grossly underestimated”.
• 23.1: added the omitted phrase from the Late Lessons 1 quote, “posing particular problems of governance”.
• 23.2: “one open challenge” corrected to “one challenge” (the source uses the past tense).
• 23.2: flagged the loose wording on blood versus teeth measurements in the Needleman passage.
• 23.2: flagged the simplified gloss of TFEU art. 114(5) (outside knowledge — verify).
• Box 23.1: softened the CDC date/source mismatch, since a 2005 CDC statement may exist (verify).
• Box 23.1: noted that “Ackerman et al. 2005” is the same article as Heinzerling et al.
• Box 23.1: corrected “driven by one valuation choice”. The lead footnote compares two different studies, dates and currencies, so the gap is attributed to the discount rate but not solely caused by it.
• Box 23.1: noted that 1.4 % = Stern’s 0.1 % + 1.3 % growth.
• Box 23.1: flagged that EC (2008) likely recommends higher rates (verify).
• Box 23.1: added the missing Gould (2009) citation for the lower-bound point.
• Box 23.1 (mercury): added the chapter’s uncertainty caveat (“far from trivial”) and the “only harmful effects” point.
• Box 23.1 (mercury): noted that the EPA comparison uses a different methodology.
• 23.3: added the “double the amount of nitrates” point (p. 566).
• 23.3: flagged the tension between “no analysis” and the cited Van Grinsven “tentative cost-benefit assessment”.
• 23.3: corrected the italics note (only “potential” is italic).
• 23.3: noted that World Bank (2007) is a China study.
• 23.3: added “along with other types of studies” and “how significant are they?”.
• 23.3: noted that the EUR 2.6bn calculation is not shown.
• 23.3: restored the hedge “would suggest” in the p. 571 conclusion.
• 23.4: added that VSL “remains the conventional metric”; noted that the chapter itself raises the ethics of valuing late-life years.
• 23.4: noted that the US ozone withdrawal is uncited (the NYT 2011 citation belongs to the “senior death discount”).
• Box 23.2: added “some claimed not severe enough to warrant expensive action”.
• Box 23.2: flagged that the cited Nature piece’s title poses “million-dollar problem” as a question.
• Box 23.2: added that the CAFE figures cover combined particles from SO2, NOx and primary PM, not SO2 alone.
• 23.5: corrected the misattribution. The intergenerational-equity objection is attributed by the chapter to “many non-economists”, not asserted in its own voice.
• 23.5: Table 23.1 column header corrected to “Discounted costs of inaction”.
• Box 23.3: “immune suppression” reworded to the source’s “reduced resistance to infection”; added the chapter’s note that too little UV (vitamin D) is also a problem.
• Box 23.3: Figure 23.4 endpoint corrected from “above 30” to “roughly 30”.
• Box 23.3: EUR 2,150bn now described as the chapter frames it, a counterfactual offset cost had ODS been counted under Kyoto.
• Box 23.3: flagged that 215 Gt equals the whole counterfactual stream, against a commonly cited ~135 Gt (verify).
• Box 23.3: reworded the early-warner credit claim as the authors’ inference from total reductions since 1974, not an estimate of the scientists’ separate effect.
• References: added uncited entries (Morgenstern 1997, NASA 2010), the likely misdated IPCC entry (verify), the undated Reddy and Braun entry and the garbled Taylor page range.
• Timeline: Wallonia “1929 [sic]” now marked as outside knowledge (verify).
• Timeline: removed the wrong NYT 2011 attribution for the US ozone decision.
• Authors’ lessons 6: “well documented” replaced by a dated sequence plus the chapter’s own “ambiguous” caveat.
• Mechanism 1: removed the unsupported causal link “no CBA… so the guide value was removed”; the chapter gives absence of proof as the reason. US ozone wording softened.
• Mechanism 2: “preferring patentable lead” changed to the chapter’s “not patent-protected” alternatives.
• Mechanism 3: replaced a pseudo-quote (“is there scientific proof?”) with the verbatim wording.
• Mechanism 6: lead factor caveat added; equity objection attributed to “many non-economists”.
• Mechanism 6: “presented as technical” corrected, since the chapter explicitly raises the late-life valuation question.
• Mechanism 6: the labour-market reading is flagged as mine, and higher exposure added as a reason for Poland’s figure.
• Mechanism 7: “deregulatory administration” flagged as outside characterisation.
• Mechanism 9: “moved acidification from London” corrected. The chimneys relieved local smog and dispersed pollution, causing acidification in Scandinavia.
• Mechanism 12: “could not be patented” corrected to “were not patent-protected”.
• Mechanism 15: noted the question mark in the Nature title.
• Mechanism 16: corrected the overstatement that the box “credits early-warning scientists with 30–45 years”.
• Insight 1: noted that nitrate turned on proof rather than price, and that US ozone shows pricing is not sufficient.
• Insight 4: the pseudo-quote “lack of proof” replaced with the chapter’s “absence of scientific proof”.
• Insight 6: “one to two orders of magnitude” replaced with the actual spreads (1.5× to well over 10×, the largest confounded); Strong kept for sensitivity only.
• Insight 11: “history is contested” softened to “common account, details debated” (outside knowledge).
• Limitations: rebalanced the leaded-petrol motive point.
• Limitations: corrected “costs of action appear only where they are small” (PCB’s billions are mentioned, p. 566).
• Limitations: restored the “would suggest” hedge on nitrate.
• Limitations: added the Van Grinsven tension; softened “none independently corroborated”.
• Limitations: in the errors list, marked outside-knowledge items and added the 215 Gt, EC (2008) rate and uncited ozone items.
• Notable quote 9: re-attributed to “some” in the 1970s, as the chapter does, with the Nature title caveat.
• Digest: the lead per-gram range is now described as spanning different studies.
• Digest: nitrate now includes the “would suggest” hedge, the Van Grinsven tension and “calculation not shown”.
• Digest: CAFE figures are now described as combined pollutants.
• Digest: added the ODS 215 Gt caveat and the early-warner inference.
• Digest: “reinstating” changed to “reconsidering”.
• Digest: insight 5 no longer says “orders of magnitude”.
• Digest: “never counts the costs of action” changed to “weighs them only where small”.
• Digest: IARC wording made precise; the Wallonia date marked verify; the uncited ozone withdrawal added; the patent claim marked as single-source.