Haute Lumière

Commerce · IV.03 · MMXXVI · daylight

La Bourse  /  Volume IV  /  Nº IV.03  /  Ten concept briefs

TEN CONCEPT BRIEFS · Chapter IV.03 — Agriculture as the Model Case

One page each. A reader who reads only these ten pages has the chapter.

---

BRIEF 1 — Soil Organic Carbon as a Stock

The idea. Soil carbon is a stock with an inflow and an outflow, and almost every mistake made about it comes from treating it as a flow.

Carbon enters soil as plant residue, root exudate and manure. It leaves as CO₂ when microbes respire it. The stock is whatever the difference has left behind, and it is large: about 700 Gt C in the world's top 30 cm and about 1,500 Gt C in the top metre — more than the atmosphere and all vegetation combined. Human land use has removed roughly 133 Gt of it over twelve thousand years.

Worked example. A typical arable topsoil carries about 50 t C/ha in the top 30 cm. A good management change adds about 0.30 t C/ha/yr. That is a 0.6 percent annual change against a stock you cannot see, which is the entire reason this subject is hard. It is not that the effect is not there. It is that the effect is small relative to the thing it sits inside.

The unit discipline. Always carry tonnes of carbon per hectare, and convert to CO₂ only at the point of sale, by 44/12 = 3.6667. Mixing the two is the most common arithmetic error in the field, and it inflates every figure by a factor of 3.67.

You already know this because you have looked at a bank balance where a steady salary and a steady spend nearly cancel, and you know that the balance tells you almost nothing about either flow — you need the statement, not the figure at the bottom.

---

BRIEF 2 — The 4-per-1000 Arithmetic

The idea. The claim that raising soil carbon by 0.4 percent a year would offset human emissions is arithmetically sound and agronomically out of reach, and both halves matter.

  0.004 × 1,500 Gt C (top metre)    =  6.0  Gt C/yr
  0.004 ×   700 Gt C (top 30 cm)    =  2.8  Gt C/yr
  atmospheric growth 2.4 ppm × 2.124 =  5.10 Gt C/yr

So the top-30-cm version covers 55 percent of the annual atmospheric increase. That is the true half.

Now the reachable half. Measured rates on cropland cluster between 0.2 and 0.5 t C/ha/yr (Minasny et al.) with a wider outer range of 0.05 to 0.76 (Lal). Take 0.30:

  1.6 Gha × 0.30 t C/ha/yr = 0.48 Gt C/yr
    →  17 % of the 2.8 Gt target
    →  9.4 % of atmospheric growth
    →  1.76 Gt CO₂/yr = 4.8 % of global fossil CO₂

Why it matters. 4.8 percent is a real and valuable contribution that no serious person should discard — and it is not a climate solution. A practitioner who says the first sentence without the second will be corrected in public by someone holding this brief, and everything else they said will be discarded with it.

You already know this because you have watched a total addressable market quoted at the size of the whole category, when the reachable share was a fraction of it, and you knew the difference mattered before anyone said so.

---

BRIEF 3 — Saturation

The idea. A soil carbon sink does not stay open. It fills, and the rate of filling decays toward zero.

The standard model is first-order approach to an equilibrium:

  C(t) = C∞ − (C∞ − C₀) e^(−kt)
  rate(t) = k · (C∞ − C(t))

The rate is proportional to the remaining gap. As the gap closes, the rate falls. There is no management that repeals this.

Worked example — Broadbalk, Rothamsted, 1843 to now. Thirty-five tonnes of farmyard manure a hectare, every year, for a hundred and eighty years. Topsoil carbon went from about 28 to about 85 t C/ha. Fit the model with an assumed equilibrium of 90 and k = 0.01399/yr:

Half-life on the remaining gap: 49.6 years. Time to 95 percent of the total gain: 214 years.

The figure to carry. 4 per 1000 of Broadbalk's current stock would require 0.34 t C/ha/yr. The plot delivers 0.070 — 21 percent of it, on the most heavily manured arable soil on record.

Why it matters. Saturation means soil carbon is a transition subsidy, not an annuity. It pays for the change in practice, once, and then stops. Build a business model on it lasting forever and the model fails in year thirty with no one at fault.

You already know this because you have refilled something — a reservoir, a pipeline, a hiring backlog — and watched the last ten percent take longer than the first sixty.

---

BRIEF 4 — Reversibility and Permanence

The idea. Soil carbon leaves about four times faster than it arrives, so a soil carbon credit is a tenancy and must be priced as one.

The measurement. Davidson and Ackerman's synthesis of cultivation studies found 20 to 40 percent of soil organic carbon lost when previously untilled soils are brought into cultivation — centred near 30 percent, most of it within about twenty years.

Worked example. Broadbalk holds 85 t C/ha after 180 years of manuring, of which 57 t C/ha was built by the experiment.

  loss at 30 %      85.0 × 0.30 = 25.5 t C/ha
  as a share of everything built   25.5 / 57 = 45 %
  loss rate   25.5 / 20  = 1.275 t C/ha/yr
  build rate    57 / 180 = 0.317 t C/ha/yr
  ratio                  = 4.0 ×

Pricing the hazard. Over a hundred-year horizon:

A ten-year vintage at 1 percent carries a 9.6 percent reversal probability, which is why a serious facility holds a buffer above that — 22 percent in this chapter's instrument.

The design consequence, and it is not obvious. Never write a clawback against the farmer. A farmer facing a personal clawback in a bad year will till, and you will lose the carbon and the member. Cancel from the buffer instead. The buffer exists to protect the relationship, not only the ledger.

You already know this because you have seen a penalty clause produce exactly the behaviour it was written to prevent.

---

BRIEF 5 — The Minimum Detectable Difference

The idea. Whether you can afford to measure soil carbon is a statistics question with a dollar answer, and you can compute it before you spend anything.

  n  =  2 (z₀.₉₇₅ + z₀.₈₀)² σ² / Δ²      (unpaired, two groups)
  n  =    (z₀.₉₇₅ + z₀.₈₀)² σ_d² / Δ²    (paired, same points re-sampled)

Worked example. Stock 50 t C/ha. Within-field spatial CV 15 percent, so σ = 7.50. Four years at 0.30 t C/ha/yr, so Δ = 1.20. (z + z)² = 7.8490.

  unpaired  n = 2 × 7.8490 × 7.50² / 1.20²  =  614 cores per group
  paired    n =     7.8490 × 3.00² / 1.20²  =   50 points

At $25 a sample, taken twice: $30,700 unpaired against $2,500 paired — twelve times cheaper for the same confidence. The paired design works because the variance that matters becomes the variance of the change at a fixed point (CV near 6 percent), not the variance of the stock across the field.

The three things that move the answer. Reduce σ by stratifying and pairing. Increase Δ by contracting over more years or choosing a higher-rate practice. Accept less power — and say so out loud, because an undeclared drop from 80 percent power is how a project quietly stops being able to detect its own effect.

Why it matters. A project that cannot detect its own result will produce a number anyway, and that number will be noise wearing a decimal point.

You already know this because you have seen an A/B test called early on a sample that could never have resolved the difference being claimed.

---

BRIEF 6 — The Aggregation Threshold

The idea — and this is the chapter's one inversion. The cost of verifying soil carbon scales with the number of soil types, not the number of hectares. So the same carbon is worthless at farm scale and valuable at co-operative scale, and the missing ingredient is not a technology but a counterparty.

Worked example. Four years at 0.30 t C/ha/yr is 1.20 t C/ha, which is 4.40 t CO₂e/ha, worth $132/ha at $30 a tonne.

On a 65-hectare field the unpaired design costs $107.34 per tonne verified against a $30 price — the measurement costs three and a half times the carbon. Across a 3,000-hectare co-operative with five soil-series strata, paired:

  MRV  5 × $2,500 = $12,500  =  $4.17/ha
  net  $132.00 − $4.17 = $127.83/ha over four years
  total across the membership  =  $383,500

Per tonne: $0.95 of measurement cost, and $22.45 net after a 22 percent buffer.

Why it matters. Every debate about whether soil carbon markets "work" is really a debate about scale, conducted by people who have not computed the threshold. Below it the answer is no and above it the answer is yes, and the threshold is a number you can work out in an afternoon.

You already know this because you have seen a fixed cost — a certification, a laboratory account, a compliance officer — destroy a small operator and cost a large one nothing, for exactly the same activity.

---

BRIEF 7 — The Yield Gap, and the Land Claim Inside It

The idea. Organic systems yield less on average, the gap is well measured, and a yield gap is always also a land-use claim.

The three meta-analyses, built independently.

Mean of the four central estimates: 21.1 percent. Ponisio's 95 percent interval: 11.9 to 26.5 percent.

The land arithmetic, which must not be dodged.

  land for equal output = 1 / (1 − gap)
    gap 25.0 %  →  1.333  →  33.3 % more land
    gap 19.2 %  →  1.238  →  23.8 % more land
    gap  9.0 %  →  1.099  →   9.9 % more land

What the DOK trial adds. A 20 percent yield loss bought a 34 to 53 percent cut in fertiliser and energy input and a 97 percent cut in pesticide. Output per unit of input therefore rose by 21 to 70 percent. The gap is real and the efficiency gain is also real; a serious economics states both.

You already know this because you have compared two suppliers on unit price and then discovered that the cheaper unit price was buying a much larger total bill.

---

BRIEF 8 — Substitution Is Not Ecology

The idea. Most of the measured organic yield gap comes from taking an input out without putting a biological process back. Where the process goes back, most of the gap closes — and no certificate is required.

Worked example — Marsden Farm, Iowa State. Three systems: the standard two-year corn–soybean rotation, a three-year rotation, and a four-year rotation with a small grain and a legume forage.

Profitability held. There is no organic premium in this result, no label, and no counterparty. The mechanism is time: a longer rotation breaks pest and weed cycles and fixes nitrogen biologically, so the purchased inputs are not replacing anything — they were replacing a rotation that had been shortened.

Read alongside Ponisio, whose 19.2 percent gap falls to 8 or 9 percent precisely where rotations and multi-cropping are present, and the two results say the same thing from opposite directions.

Why it matters. It relocates the argument. The question stops being organic versus conventional — a question about certificates — and becomes how much time is in the rotation, which is a question about agronomy that any farm can answer and any lender can covenant.

You already know this because you have watched a team lose a capability, replace it with a contractor, and then discover that the contractor was never the expensive part.

---

BRIEF 9 — The Transition Hole

The idea. The barrier to converting a farm is not belief and it is not agronomy. It is three years of working capital, and it is computable in advance.

Under the US National Organic Program, land must be free of prohibited substances for 36 months before an organic crop can be harvested as organic. So for three years you take organic yields at conventional prices. That is the whole problem in one sentence.

Worked example, per hectare, on a Midwest corn–soybean farm.

  conventional:  revenue 1,600 − cost 1,150            =  margin 450
  transition y1: 1,600 × 0.70 = 1,120 − 980 − 40 = 100 →  short 350
  transition y2: 1,600 × 0.75 = 1,200 − 980 − 40 = 180 →  short 270
  transition y3: 1,600 × 0.80 = 1,280 − 980 − 40 = 260 →  short 190
                                          the hole  =  810 USD/ha
  on 400 ha  =  324,000 USD of working capital

Then the recovery. Mature organic at Ponisio's 0.808 yield factor:

The break-even premium for this cost structure is 13.7 percent. Crowder and Reganold measured actual premiums of 29 to 32 percent and a break-even nearer 5 to 7 percent for a steady-state comparison that carries no transition cost. Two routes, two figures, and the gap between them is the transition.

Why it matters. "Three years of lower income" is a fear. "$810 a hectare, $324,000, repaid from year eight" is a facility. Only one of those can be financed.

You already know this because you have seen a capital project that was obviously right get refused, not on the return, but on the cash profile.

---

BRIEF 10 — Pay for the Difference, Never the Stock

The idea. A carbon contract may only honestly pay for the carbon that would not have been there anyway — the difference between the practice and its counterfactual, not the stock under the field.

Worked example — the Rodale Farming Systems Trial over 22 years. Soil carbon rose 27.9 percent in the organic-animal system, 15.1 percent in the organic-legume system, and 8.6 percent in the conventional system. Against a 50 t C/ha stock:

  organic-animal   0.634 t C/ha/yr
  organic-legume   0.343 t C/ha/yr
  conventional     0.195 t C/ha/yr
  the difference   0.634 − 0.195  =  0.439 t C/ha/yr

0.439, not 0.634. Sell the larger figure and you have sold 0.195 t C/ha/yr that the conventional control was building anyway — about 31 percent of the contract, and a fraud whether or not anyone intended one.

The three tests a project must pass. Additionality: would it have happened anyway? Permanence: what holds it in place, and for how long — see Brief 4. Leakage: if this field produces less, does someone else plough more elsewhere — see Brief 7's land arithmetic.

And the register to say it in. At $30 a tonne, net carbon income is about $24.70/ha/yr, which covers $74.09 of the $810/ha transition hole — 9.1 percent. Carbon would need to reach $315.90 a tonne to fund a transition by itself. So say the true thing: the premium and the input bill finance the transition, and the carbon finances the measurement that makes the premium defensible.

You already know this because you have watched a business attribute all of a good quarter to its new initiative, without ever asking what the rest of the market did that quarter.

---

Every figure in these briefs is computed in lib/verify/IV_03.py, labelled [src], [asm] or [calc], and sourced in the chapter's Works Cited.