Haute Lumière
Commerce · VII.01 · MMXXVI · daylight
TEN CONCEPT BRIEFS · Chapter VII.01 — Planetary Boundaries as a Balance Sheet
One page each. A reader who reads only these ten pages has the chapter.
BRIEF 1 — The Control Variable
The idea. A planetary boundary is not a thing. It is a number attached to a measurable quantity, and the quantity is called the control variable. Everything about how much a boundary is worth follows from what its control variable is and how directly anybody can read it.
Climate change has two control variables: atmospheric carbon dioxide concentration, boundary 350 ppm, and radiative forcing, boundary 1.0 W/m2. Stratospheric ozone has one: the ozone column in Dobson units, boundary 275.5 DU — a 5 percent reduction from the pre-industrial 290 DU. Ocean acidification has one: the aragonite saturation state of surface seawater, boundary 2.75, which is 80 percent of the pre-industrial 3.44.
Worked example. Ask of each: could a satellite or an instrument read it today, without a model in between? For the ozone column, yes — several agencies do, continuously. For biosphere intactness, no: the index is modelled from land-use data and species-response curves, and the model is the measurement.
Why it matters. Two boundaries drawn as identical wedges can be a direct observation and an inference from an inference. The wedge does not say which.
You already know this because you already distinguish between a bank statement and a management estimate, and you would never let the two share a column.
BRIEF 2 — Threshold or Aggregate
The idea. This is the distinction the chapter is built on, and most popular accounts of the framework leave it out.
A threshold boundary sits in a globally well-mixed system where one number genuinely describes the state of the whole, and where crossing a level makes the system behave differently. Carbon dioxide mixes through the atmosphere in about a year; ozone chemistry is the same chemistry everywhere; surface ocean carbonate chemistry is set by a global partial pressure. Three of the nine are like this.
An aggregated boundary is a sum or an average of local conditions. Land-system change is a sum of cleared hectares. Nitrogen is a sum of fixed tonnes. Biodiversity loss is a sum of local extirpations. Brook, Ellis and colleagues asked directly whether the terrestrial biosphere has planetary tipping points and concluded that it very largely does not. A sum does not tip. It has an average.
Worked example. Halve the world's nitrogen use in the places that already use almost none and double it where it is already causing hypoxia. The global total can be unchanged. The damage is not.
Why it matters. A threshold is a level test. An aggregate is a running total. Managing one as though it were the other is the single most expensive mistake in this chapter, and it is mostly made by accident.
BRIEF 3 — The Uncertainty Band, Read as an Evidence Grade
The idea. Every boundary is published with a "zone of increasing risk" — a band running from the boundary to the point at which the authors think the risk becomes high. Divide the band's width by the boundary value and you have a like-for-like statement of how much the authors were willing to commit to.
ozone 5.3 %
aragonite saturation 12.4 %
forest cover 28.0 %
CO2 concentration 28.6 %
nitrogen 32.3 %
radiative forcing 50.0 %
phosphorus, regional 80.6 %
biosphere intactness 66.7 %
phosphorus, global 809.1 %
extinction rate 900.0 %The reading. The four climate-and-chemistry bands average 24.1 percent of their own boundary, median 20.5 percent. The two biosphere bands average 483.3 percent. That is 20.1 times on the means and 23.6 times on the medians — about twenty times, either way you compute it.
Why both are printed. The mean is carried by one band of 900.0 percent, and a ratio that depends on one row is a fact about that row. The median says the same thing without it. When two estimators disagree you print both; when they agree you have a finding.
You already know this because you already trust a builder's quotation given to the nearest thousand more than one given to the nearest order of magnitude, and you have never needed a statistician to tell you why.
BRIEF 4 — The Four Evidence Grades
The idea. Three tests, applied to each boundary, produce a grade that goes on the face of the statement in the same typeface as the number.
- Is the system well-mixed, so that one number describes the whole?
- Is the control variable directly measured, rather than modelled?
- Is the threshold demonstrated rather than inferred?
| Grade | Meaning | Boundaries |
|---|---|---|
| A | All three. A measured level in a global system with demonstrated threshold behaviour. | Climate, stratospheric ozone, ocean acidification |
| B | A genuine measured stock, aggregated, no demonstrated global tipping point. | Land-system change, freshwater change |
| C | A measured flow whose damage is local; the global total is a sum of incommensurables. | Biogeochemical flows, aerosols |
| D | Expert judgement; control variable contested, modelled or unquantified. | Biosphere integrity, novel entities |
The finding that grades the grader. Six of the nine — 66.7 percent — have had their boundary value or their control variable revised since 2009. The three that never have are climate, ozone and ocean acidification: exactly the Grade A set, with no exceptions in either direction.
Why it matters. The grade is auditable. Anybody can recompute it from the framework's own published tables, which means it is not an opinion about the science but a description of it.
BRIEF 5 — Covenant, Stock, Flow
The idea. Three accounting kinds, three loss functions, three places in the report.
| Kind | Loss function | Treatment |
|---|---|---|
| Covenant | Step. Inside or breached. | Covenant schedule: boundary, reading, headroom, trend, breach clause |
| Stock | Proportional. Depletion accumulates. | Asset register by location: gross, impairment, net (IAS 16, IAS 41) |
| Flow | Local and threshold-shaped. | Flow account by catchment or airshed, against that place's allowable input |
Worked example, and it is the whole argument for grading. Take a 5 percent overshoot on a $500,000,000 position. Under a covenant, a 5 percent breach makes $500,000,000 repayable on demand. Under a stock, a 5 percent impairment costs $25,000,000. The same 5 percent, twenty times apart.
Why it matters. Misclassify a covenant as a stock and you under-provide for a step. Misclassify a stock as a covenant and you raise an alarm that cannot be acted on, which is how a check gets switched off.
You already know this because you already know the difference between missing a payment and losing a little value, and you have never confused the two in your own affairs.
BRIEF 6 — The Aggregation Problem, in One Division
The idea. A global total is invariant to the distribution, and for a boundary whose damage is local and threshold-shaped, the distribution is the entire damage function.
global nitrogen fixed 190 Mt N/yr
cropland 1,600 Mha
-----------------------------------------
spread evenly 118.75 kg N/ha/yr
on a fifth of it 593.75 kg N/ha/yr 5.0 xBoth worlds report 190. One has anoxic estuaries, nitrate in the water table, and a continent that cannot buy fertiliser. The other has none of them.
The same error, one aggregate later. The familiar "1.6 Earths" figure is a composite of six components. Blomqvist and colleagues took it apart and found five of the six at or below unity, so the carbon component — a hypothetical forest area computed from emissions rather than an observed one — carries 37.5 percent of the aggregate and 100 percent of the overshoot. The headline is the climate problem restated in hectares.
Why it matters. Before trusting any aggregate, ask what it would look like if the components were rearranged and the total held. If the answer is "completely different", the total is not the management number.
BRIEF 7 — The Economic Critique, at Full Strength
The idea. The framework's economic critics are not denying the science. They are saying a limit is not a decision rule.
Nordhaus's objection. A boundary with no damage function and no discount rate tells you a place not to go and nothing about which of the many paths short of it to take. To Slow or Not to Slow (1991) and The Climate Casino (2013) are the tradition: optimise against damages, do not obey a line.
Sterner's objection, which is subtler and better. Sterner and Persson's An Even Sterner Review shows that as environmental goods become scarce their relative price rises, and that this one correction inside conventional machinery does much of the work a hard threshold reaches for — continuously, without a cliff, and with an allocation rule attached.
Nordhaus and Shellenberger's practical objection. If a boundary's number can be moved by a literature review, it is a research finding in a policy costume. The nitrogen boundary moved from 35 to 62 Mt N/yr between 2009 and 2015 — a revision of 77.1 percent — on the strength of de Vries and colleagues' reassessment. The nitrogen cycle did not change by 77.1 percent in six years.
The honest reply. For the Grade A three, the threshold is real and a price is not a substitute for a level test; you can price a covenant breach and you still cannot buy your way through it. For the other six, the critics are substantially right, and the fix is to stop presenting them as thresholds — which costs the framework nothing it should want to keep.
BRIEF 8 — Downscaling and Fair Shares
The idea. A global boundary that has not been divided into places is not yet an instrument.
Häyhä and colleagues set out the route: from a planetary boundary to a national share requires a biophysical step — how much of the global budget physically sits in this territory — and then a social and ethical step: on what principle is the remainder allocated. Equal per capita, grandfathered, ability-to-pay, or historical responsibility each give a different answer, and each is defensible.
Worked example. Nitrogen at 190 Mt N/yr against a boundary of 62 Mt N/yr is a global reduction requirement of two thirds. Applied as an equal proportional cut, a region using almost none cuts what it does not have, and a region at 593.75 kg N/ha remains far above every local threshold that matters. The global ratio of 3.065 does not say which.
The discipline. Label the ethical step as an ethical step, on the face of the allocation. Häyhä and colleagues do; most downscaling exercises quietly do not, and the quiet is what makes them contestable later.
Why it matters. Every allocation contains a value judgement. Naming it is what makes the arithmetic around it trustworthy.
BRIEF 9 — The Watch List, and Why Disclosure Is Not Demotion
The idea. Accounting already has a treatment for something real, material and not reliably measurable. It is called a contingent liability, it lives under IAS 37, and it is disclosed rather than recognised — precisely because it fails the measurement-reliability test.
Biosphere integrity and novel entities belong here. Intactness has a published band of 66.7 percent of its boundary; extinction rate a band of 900.0 percent. Novel entities has a boundary of 0 percent untested chemicals and no measured current value at all — the transgression is declared on the definitional ground that the share is certainly not zero, which may well be right and is not a measurement.
What disclosure is not. It is not dismissal. A contingent liability in a set of accounts is a live, legally consequential statement read by exactly the people who need to read it. Auditors ask about it. Lenders ask about it. It simply does not get added to a total.
The rule. Never sum across evidence grades. A total spanning a covenant and an expert judgement is not a total; it is the diagram again, with a decimal point.
You already know this because you already write "subject to survey" on an offer without meaning "ignore the roof."
BRIEF 10 — The Segmented Nature-Linked Facility
The idea. Price each segment the way its physics behaves.
Why the market's current answer does not work. The norm for a sustainability-linked loan is a margin ratchet of 25 basis points. On a drawn facility of $400,000,000 that is $1,000,000 a year, against a covenant whose breach makes $400,000,000 repayable. The ratchet is 0.25 percent of the consequence it claims to price — out by a factor of 400 times.
The structure.
| Segment | Metrics | Consequence |
|---|---|---|
| A | Grade A, level-tested quarterly | A real covenant with a breach clause and an equity cure |
| B | Grade B stocks by location | Borrowing-base adjustment, proportional to impairment |
| C | Grade C flows by catchment or airshed | Throughput covenant at the place of effect |
| D | Grade D | Disclosure only. No covenant on a band of 900.0 percent |
The number that decides it. Avoided impairment of $2,400,000 plus a margin benefit of $1,000,000, against verification at $340,000 and restoration at $1,250,000, returns 213.8 percent and a net $1,810,000 a year — and it clears before a single tonne of anything is restored, because most of what the spend buys is the documented right not to impair assets that were never impaired.
Why it matters. It converts a planetary argument into a loan term, and loan terms get negotiated by people who would never negotiate a planetary argument.
Every boundary value and reading in these briefs is quoted from Steffen et al. (2015) and Richardson et al. (2023) and labelled as an input in lib/verify/VII_01.py. Every ratio, band width and share is computed there.