Haute Lumière

Commerce · VII.05 · MMXXVI · daylight

La Bourse  /  Volume VII  /  Nº VII.05

Bioregional Economics

Volume VII — Planetary and Cosmic

---

THE PLATE

---

THE LETTER

Somebody has handed you a map with a line on it that does not match any line in your accounts, and told you that the line is the real one.

It may have been a watershed. It may have been an ecoregion, a foodshed, a "fifty-mile radius", a county set drawn around a river system, or the boundary of a soil association that stops halfway through your second-largest supplier. Whoever handed it to you meant it kindly and probably meant it correctly. What they did not hand you was the arithmetic, and without the arithmetic the line is a sentiment with a shapefile.

This chapter supplies the arithmetic, and it does so in the order a finance function will want it. First: what a bioregional boundary actually is, who draws it, from what measurements, and how many of them there are — because the answer turns out to be a number, and the number changes the economics by an order of magnitude depending on which level of the hierarchy you pick. Second: the real self-sufficiency balance for a defined region, computed for a real city, by commodity, with the shortfall shown rather than summarised. Third: the trade question, done properly — comparative advantage stated at full strength, in Ricardo's own numbers, and then examined for the three things it leaves out.

You should know before you begin that the honest findings in this chapter are mixed, and that the mixture is the useful part. Most regions cannot feed themselves at the diet their people currently eat. Localisation raises the cost of most goods, and that cost lands first on the households with the least room for it. And a number of carefully conducted local-food studies have found the short chain carrying a larger transport footprint than the long one, for a reason that has nothing to do with distance.

None of that is an argument against bioregional economics. It is the argument for doing it where it wins, which is a specific and knowable list, with thresholds you can compute on the goods you actually buy. That list is the last third of this chapter and it is the part to take into a meeting.

Volume I, Chapter I.10 treats federation and scale — how many units a structure can hold before its coordination cost overtakes its shared benefit — and the ceiling it computes is the reason a bioregional body should be read as a tier in a federation rather than as a sovereign. Chapter VII.06 takes the watershed specifically, as a unit of account. This chapter takes the bioregion as an economic unit and tests it.

— The Editors

---

DISCOVERY

What is already working

The bioregion is usually introduced as a proposal. It is more useful, and more accurate, to introduce it as an instrument that already exists, is already drawn to a published method, and is already in daily administrative use.

The boundary is drawn, and it is drawn from measurements. James Omernik's method — adopted by the United States Environmental Protection Agency and then, with Canada and Mexico, by the Commission for Environmental Cooperation — does not draw a region from any single variable. It draws it from the place where several variables change together: geology, physiography, soils, potential natural vegetation, climate, hydrology, land use and wildlife. Where those patterns shift in concert, there is an edge; where they do not, there is not. The result is a nested hierarchy over North America of 15 Level I regions, 50 at Level II and 182 at Level III, each successive level dividing the one above it. At the global scale, Dinerstein and colleagues map 846 terrestrial ecoregions, of which 98 already exceed half-protected status and 313 more retain enough unaltered habitat to reach it. These are not literary boundaries. Water-quality criteria, biological reference conditions and habitat targets are set against them, by agencies, with budgets.

The idea was economic from the beginning. Peter Berg and Raymond Dasmann, writing in The Ecologist in 1977, defined a bioregion as both a geographical terrain and a terrain of consciousness — and the half of that definition usually quoted is the second half, which is a pity, because the first half is the one that has since been surveyed to a metre. Berg and Dasmann's claim was that the unit of livelihood and the unit of biology ought to coincide. The measurement community, working entirely independently and for entirely different reasons, went and drew the biological unit. It is sitting there, in a shapefile, waiting for somebody to put accounts on it.

Somebody has. Christian Peters and colleagues at Cornell built a spatial model of New York State's food production capacity and matched it against the food needs of its population centres. Their finding is the most quoted number in the foodshed literature and it deserves its position: given the diet they assumed, with all agricultural land in use and distribution optimised to minimise distance travelled, New York State could meet 34 percent of its total food needs from within the state, at an average distance of 49 kilometres. What is less quoted is the distribution underneath that average, and it is the appreciative finding of the whole study: the smallest population centres could meet virtually all of their food needs from a mean distance of about 25 kilometres, and urbanised areas of moderate size around 84 percent from about 51 kilometres. The state-level deficit is not general. It is specifically metropolitan, and it is concentrated in one city.

Inside a city, the capacity is larger than anyone expects. Kathryn Colasanti and Michael Hamm surveyed Detroit's publicly owned vacant parcels — 44,085 of them, 4,848 acres — and asked what could be grown there. Using high-yield biointensive methods and fewer than 300 acres, they found the city could supply 31 percent of the vegetables and 17 percent of the fruit seasonally consumed by its then 900,000 residents. Add season extension, hoop houses and storage and the figures rise above 75 percent of vegetables and 40 percent of fruit. Three hundred acres is a rounding error in a city of that size. The finding is not that Detroit should farm itself. It is that the marginal productivity of the first three hundred acres is extraordinarily high, and that nobody had costed it.

And at global scale the picture is better than the debate assumes. Pekka Kinnunen and colleagues, in Nature Food in 2020, built an optimisation model of minimum distance between production and consumption for six crop types worldwide. Their headline is usually read as a defeat for localisation — between 11 and 28 percent of the world's population can meet demand for a given crop within 100 kilometres, and between 26 and 64 percent need more than 1,000. Read it the other way and it is a map of where localisation is cheap: for some crops, in some regions, more than a quarter of humanity is already standing inside its own foodshed. The same paper identifies where closing yield gaps and cutting losses would move the line, and finds the largest movements in Africa and Asia.

Five findings, one shape. The boundary is drawn, the capacity has been measured, the measurement is uneven, and the unevenness is the information. Nothing above required a new institution. It required somebody to take a shapefile that already existed and put a quantity on it.

---

THE ARITHMETIC

What works, what does not, and where the line sits

Five calculations. The first says what a boundary costs you. The second computes a real foodshed for a real city and shows the shortfall commodity by commodity. The third is the honest negative and it has two halves. The fourth states comparative advantage at full strength and then examines it. The fifth is the practical payload: the goods localisation genuinely wins on, with the threshold.

One. The boundary is a resolution setting, and the setting decides the answer.

Ecoregions nest. Over North America the Commission for Environmental Cooperation publishes 15 at Level I, 50 at Level II and 182 at Level III; globally, Dinerstein and colleagues map 846. Divide and the grain becomes visible:

  North America land area          24.71 million km2
    Level I    15 regions           1,647,000 km2 each, on average
    Level III  182 regions            135,800 km2 each
  world land area                 148.94 million km2
    846 ecoregions                   176,100 km2 each

A Level I region is twelve times the area of a Level III region inside it. Since self-sufficiency is production over consumption and production scales with area while consumption scales with people, the same place is food-secure or food- dependent depending purely on which level you drew. That is not a defect in the classification. It is the reason the level must be stated.

The threshold it implies is worth memorising. At the diet Americans actually eat, Peters and colleagues' biophysical model puts the annual land requirement at 1.08 hectares per person — 0.18 ha of cultivated cropland, 0.16 ha of perennial cropland and 0.74 ha of grazing land. One square kilometre is 100 hectares. So:

  self-sufficiency requires      density  <=  100 x f / 1.08   persons per km2
       f = farmable fraction of the region's land

  f = 1.00  (every hectare farmed)      92.6 persons per km2
  f = 0.38  (the global agricultural share) 35.2 persons per km2

Ninety-three people per square kilometre, and that is the ceiling if you plough the streets. Most of western Europe sits above it, and every country there already knows so from its trade statistics rather than from its ecology.

Two. A real foodshed: New York City, by commodity.

New York City held 8,804,190 people at the 2020 Census. At 1.08 ha each it claims 9.51 million hectares of agricultural land — 23.5 million acres. New York State's entire land in farms, from the 2022 Census of Agriculture, is 6,502,286 acres. The city alone claims 3.6 times the whole state's farmland, and the state has 11.4 million other residents.

Now the same question asked a second way, sharing none of the first way's assumptions. Take national acreage from the 2022 Census, scale it by the city's share of the national population — 8,804,190 of 331,449,281, or 2.656 percent — and compare it with what New York State actually grows and keeps.

  commodity                 NYC's claim      New York State        covered
                                (acres)         (whole state)
  --------------------------------------------------------------------------
  corn for silage               158,800            501,938           316%
  milk cows (head)              247,300            631,199           255%
  forage, hay and haylage     1,438,600          1,655,850           115%
  vegetables, for sale          114,400            109,617            96%
  land in orchards              164,800            108,225            66%
  layers (head)              10,319,900          6,388,603            62%
  cattle and calves (head)    2,336,300          1,380,585            59%
  harvested cropland          8,004,100          3,563,036            45%
  corn for grain              2,140,900            603,642            28%
  soybeans                    2,247,200            341,509            15%
  wheat for grain               988,500            105,991            11%
  hogs and pigs (head)        1,960,800             42,889             2%
  rice                           60,600                  0             0%
  peanuts                        37,900                  0             0%
  sugarcane                      24,300                  0             0%
  --------------------------------------------------------------------------
  ALL land in farms          23,377,900          6,502,286            28%

Two independent routes, and they land in the same place. The biophysical diet model says the city needs 23.50 million acres; the Census-acreage route says 23.38 million. Half a percent apart, computed from different data by different methods. The 28 percent is real.

Read the table rather than the total, because the total is the number that hides everything. New York State runs a surplus in dairy, forage and silage — it could supply the city two and a half times over in milking cows. It is within four percent in vegetables. It is at two percent in pork, and it is at exactly zero in rice, peanuts and cane sugar, and will remain there for reasons of latitude that no policy will alter. Its entire citrus sector is one farm growing lemons, with the acreage withheld to protect a single operation.

A single self-sufficiency percentage averages a 316 percent surplus against a structural zero and returns a number that describes neither. This is why the published figures for the same state differ so widely and all of them are right: Peters and colleagues found 34 percent of New York's food needs meetable in-state at an average distance of 49 km, and, in a later paper working by food group, 69 percent on a fresh-weight basis at an average 238 km — and that second, larger figure is reached by supplying four food groups and not growing grain or meat in the state at all. Griffin and colleagues, measuring the Northeast as a region, found self-reliance running from 7 percent for pulses to 26 percent for vegetables, in a region holding about 6 percent of the nation's farmland and about 22 percent of its people.

And the geometry is unforgiving at the city scale. A circle of 100 miles around Manhattan encloses 81,367 km², or 8.14 million hectares. The city's claim is 9.51 million. Even if every square metre inside that circle — including the ocean, the estuary, the airports and the five boroughs themselves — were farmed at national average yields, it would supply 86 percent of one city's food.

Three. The honest negative, in two halves.

The first half: localisation raises the delivered cost of most goods, and the increase most likely lands first on the households with the least room for it. Fajgelbaum and Khandelwal, measuring the distribution of the gains from trade across consumers within 40 countries, find a pro-poor bias in every one. Closing a country off from trade entirely costs, on average across their sample, 63 percent of real income at the tenth percentile of the income distribution and 28 percent at the ninetieth. For the United States the gap is wider still: 69 percent at the tenth percentile against 4 percent at the ninetieth. The mechanism is basket composition — low-income households spend more of their money in the sectors that are actually traded. Their own figure: the average import share is 6.4 percent across service sectors against 20 percent for food and 48 percent for manufacturing.

And here the honest negative has to be honest about itself, because that finding is contested by good people with better data. Borusyak and Jaravel, measuring import shares directly in linked expenditure and customs microdata rather than inferring them from a demand system fitted to national aggregates, find import shares flat across the income distribution — hovering between 11.7 and 12.9 percent — and conclude that the purchasing-power gains from lower trade costs are distributionally neutral. They name Fajgelbaum and Khandelwal directly and argue that the demand system used there mechanically generates the pro-poor result.

Take both. What survives the disagreement is enough: a localisation premium is a consumption tax, its incidence is at best neutral and may be sharply regressive, and a programme that does not measure where it lands has chosen not to know.

The second half, and it is the one that surprises practitioners: several carefully conducted local-food studies find the short chain carrying the larger transport footprint. Coley, Howard and Winter compared a large organic vegetable-box operation — cold storage, packing, regional hub, doorstep delivery — against a customer driving to a farm shop, and found the crossover at a round trip of 6.7 kilometres. Beyond that, the short chain emits more. Nothing about distance causes this. Load factor causes it: a full lorry carries a kilogram at a small fraction of the cost of a car carrying ten.

Mundler and Rumpus found the same thing across a set of short chains in the Rhône-Alpes, measuring distribution energy in grams of oil equivalent per euro of product sold. Their range runs from 13.5 to 44.8 GOE/€: urban box schemes at 13.5, on-farm sales at 34.2, producer shops and collective sales points at 44.8 — against a best long-chain supermarket comparator that, on a basis where on-site storage is excluded from both sides, sits at 8.8 while the urban box scheme sits at 8.6. The box scheme wins by a whisker. The farm shop loses by a factor of five. In the periurban case, sixty percent of the total is the customer's own journey.

Weber and Matthews put the ceiling on the whole argument. Across the average US household's food footprint of 8.1 tonnes CO₂e a year, production accounts for 83 percent, all transport for 11 percent, and final delivery from producer to retail for 4 percent. Their measured freight intensities, per tonne-kilometre, are the reason:

  mode                          MJ/t-km     g CO2e/t-km
  --------------------------------------------------------
  international water, bulk         0.2            11
  international water, container    0.2            14
  rail                              0.3            18
  inland water                      0.3            21
  truck                             2.7           180
  air                              10.0           680

Truck is 12.9 times container ship per tonne-kilometre; air is 48.6 times. So the transport term is decided by mode, not by distance — and the maximum emissions saving available from perfect localisation is the 11 percent that transport occupies in the first place. If local production is more than about 11 percent less efficient than the distant producer's, localisation loses on carbon before the first lorry moves.

Now the cut. Take a car at 6.0 litres per 100 km; petrol carries 2.31 kg of CO₂ per litre, which is chemistry rather than policy, so the car emits 0.1386 kg per kilometre. Give the driver a ten-kilogram basket:

  0.1386 kg CO2 / km  /  10 kg  =  13.86 g CO2 per kg per km
  container ship                =   0.014 g CO2 per kg per km
  ----------------------------------------------------------
  one kilometre by car          =  990 kilometres by container ship

One kilometre in a car with the week's shopping costs what a thousand kilometres of ocean costs. Coley's 6.7 km round trip is 6,630 km of sea freight. A twenty-kilometre round trip to the farm shop is half the circumference of the Earth. The food mile was never the unit; the vehicle was, and the last mile is the long haul.

And the sharpest form of the negative: distant production is sometimes cleaner production. Saunders, Barber and Taylor measured New Zealand and United Kingdom production of four commodities, with shipping to the United Kingdom — 17,840 km — included on the New Zealand side:

  per tonne of product           New Zealand        United Kingdom
                                 incl. shipping
  ------------------------------------------------------------------
  lamb, carcass                  688.0 kg CO2        2,849.1 kg CO2
  dairy, milk solids           1,422.5 kg CO2        2,920.7 kg CO2
  apples                         185.0 kg CO2          271.8 kg CO2
  onions                         184.6 kg CO2          170.0 kg CO2

Lamb travelling half the world arrives at a quarter of the emissions of the lamb grown down the road, because New Zealand sheep stand outside on grass and British sheep need winter housing, fencing and supplementary feed. The shipping leg itself is trivial in the account: the report's own coefficient is 0.007 kg CO₂ per tonne-kilometre at sea against 0.027 for a lorry from Italy.

Read the fourth row before drawing a conclusion. Onions reverse it — the United Kingdom is lower per tonne despite using more energy, because of its electricity mix — and the report itself doubts whether British growers could actually cover the winter window it credits them with. Two further cautions belong with these numbers and the authors state the first one themselves: their New Zealand inventories are more complete than their United Kingdom ones, which biases the comparison against New Zealand; and the boundary stops at the United Kingdom border, so no domestic distribution, retail or consumer journey is counted on either side. The direction is solid. The magnitudes are not like-for-like, and the fourth row is the reason to say so.

Four. Comparative advantage, stated fairly and then examined.

State it at full strength, in Ricardo's own numbers from 1817. England needs the labour of 100 men for a year to make its cloth and 120 for its wine; Portugal needs 90 for cloth and 80 for wine. Portugal is better at both — and trade still pays, because the ratios differ:

  autarky      England 100 + 120  +  Portugal 90 + 80   =  390 man-years
  specialised  England 2 cloth = 200  Portugal 2 wine = 160  =  360
  ----------------------------------------------------------------------
  saving 30 man-years on the same four units            =    7.7%
  opportunity cost of wine:  England 1.20 cloth  ·  Portugal 0.89 cloth

That is the argument, and it is correct. Now examine it on three counts the model sets aside.

Transport. Ricardo's example carries no freight cost. Put one in. Two units cross the water; the gains vanish when transport reaches 15 man-years per traded unit — 15 percent of the English cloth's cost and 18.75 percent of the Portuguese wine's. Ocean freight on manufactured goods runs far below that, which is why Ricardo survives the sea comfortably. It does not survive the road for low-value goods, and that is the whole of the practical finding in part five.

Resilience. The model is a steady state with no variance term. It says nothing about what a specialised region does when its single crop fails, its single trading partner closes, or its single strait is blocked — and the correct response to that is not to abandon comparative advantage but to price the insurance separately, which is an instrument, not a sentiment.

The distribution of the gains. Autor, Dorn and Hanson, studying US local labour markets between 1990 and 2007, find that rising Chinese import competition explains one quarter of the aggregate decline in US manufacturing employment in that period, with unemployment, labour force participation and wages moving in the exposed markets and transfer payments for unemployment, disability, retirement and healthcare rising sharply there. The aggregate gain is real. The adjustment is local, it is slow, and the compensation the textbook assumes has never in practice been paid. Comparative advantage tells you the size of the pie. It is silent on the plate.

Five. Where localisation wins, and the threshold in each case.

This is the part to take into a meeting. Six categories, each with a test.

And the seventh case, which is not a category but a price: a good that is cheap to import and slow to substitute is a hedging problem, not a localisation problem. Pay for the option, not for the acreage.

---

DREAM

What becomes ordinary

In the version of this that has already happened, every organisation of any size knows which bioregion it sits in, at which level of the hierarchy, and can say so in one line without looking it up. The line is on the first page of the annual report, next to the registered office, and it is there for the same reason the registered office is there: it tells a reader which set of physical facts the enterprise is subject to.

Behind the line sits a regional balance, kept the way a treasury keeps a currency position. It has three columns — what the region supplies to itself, what it imports, what it exports — and it is kept by commodity rather than in aggregate, because the aggregate is the number that hides everything. Nobody finds this exotic. It is the same instrument a multinational already keeps for foreign-exchange exposure, pointed at physical goods instead of currencies, and the people who keep it were trained to keep the other one.

Procurement reads the balance before it tenders. Not sentimentally: the balance tells it which goods are cheap to source close and which are expensive, and the answers are frequently counter-intuitive, which is precisely why the report exists. Nobody in that organisation would propose localising a good that the arithmetic says to import, and nobody would import a good the arithmetic says is sitting in surplus forty kilometres away. The question should we buy local has been retired, because it was never a single question. It was between two and three hundred separate questions wearing one coat, and the balance answers them one at a time.

The trade that remains is deliberate and it is larger in value than the trade it replaced, because it is concentrated in the goods that genuinely travel well. The region exports what it is unusually good at, at a scale that would have been impossible when everything was exported a little. It imports what it cannot produce without heating a greenhouse in February, and it is not embarrassed about it, because the arithmetic on that greenhouse is public and anyone can check it.

The people who live there can see the balance. It is published, at the level of the ecoregion, once a year, in a form a person can read in ten minutes, and the schools use it. A child in that region can tell you which three things grow unusually well where they live and which four things arrive by ship, and can tell you why in terms of rainfall and soil rather than in terms of virtue.

And when something breaks somewhere else in the world — and something always does — the region knows within a day which of its lines are exposed, because the exposure was already written down. It does not have to discover its dependencies under pressure. It discovered them in a quiet year, on purpose, and priced the ones worth insuring.

None of this requires a bioregional government. It requires a boundary anybody can look up, a balance kept by commodity, and the discipline to let the arithmetic overrule the preference in both directions.

---

DESIGN

The structure that gets there

Four components, in order. Each is buildable by an existing organisation without anybody's permission.

One — declare the boundary, at a stated level, and stop arguing about it. The hierarchy is the gift here. Do not attempt to find the true bioregion; there is no such object, and the search for it has consumed more bioregional energy than any other single activity. Choose the published level whose grain matches the decision you are making, name it, and write down the level you chose. Water quality and habitat are Level III or IV questions. Labour markets and staple grains are Level II. Energy systems and long-distance freight are Level I or larger. The boundary is a resolution setting, not a discovery, and stating which setting you used is what makes your number comparable to anybody else's.

Two — build the balance by commodity, and never in aggregate. For each material line above a threshold of spend, three quantities: regional demand, regional production, and the delivered cost gap between the regional source and the incumbent. A single aggregate self-sufficiency percentage is worse than useless, because it averages a hundred-percent surplus against a structural zero and returns a number that describes neither. The balance is built once and refreshed annually; the first build takes a quarter and is mostly procurement data you already hold.

Three — sort the lines into the four categories and act differently on each.

The fourth row is the one most bioregional programmes miss, and it is where the real money is. A good that is cheap to import and catastrophic to lose is not a localisation candidate. It is a hedging candidate, and hedges are priced, contracted and audited, not aspired to.

Four — put the boundary inside a federation, not around a sovereign. Chapter I.10 computes the coordination ceiling for a federation of peers and finds it at twenty-four units for the optimum and eighty-seven for the point where a full mesh costs its members more than it returns. A bioregional body is a tier, and a tier is a single point of failure that must be capitalised and governed as one. Build it to hold the balance, the standards and the shared purchasing — and to hold nothing whose failure would take the members with it.

The sequence matters and it is the reverse of the usual one. Boundary, then balance, then policy. Most programmes begin with the policy, which is why most programmes discover in year two that they localised the one commodity where they had no advantage.

---

DESTINY

How it holds when nobody is pushing

A bioregional balance survives on the same three conditions as any other instrument: it is in the standing reporting pack, somebody's compensation moves with it, and it has a second owner. What is specific to this one is a fourth: it must be allowed to say no.

A balance that has never once recommended importing something is not a balance. It is an advocacy document with columns, and the finance function will read it as one within about two cycles. The single most durable thing you can do for a bioregional programme in its first year is to publish a line where the arithmetic said buy it from four thousand kilometres away, and act on it. That line is what makes every other line in the document credible.

Now the failure modes, named.

It fails when the level is allowed to float. A programme that quietly moves from Level III to Level II whenever the Level III answer is inconvenient is producing numbers that cannot be compared with each other, let alone with anybody else's. Fix the level in writing at the start, and change it only in a dated amendment that restates the prior figures.

It fails when the aggregate percentage escapes. One self-sufficiency number on a slide will outrun the commodity table every time, and it will be quoted back at you in a context where it is false. If you publish an aggregate, publish it as a range across food groups and never as a single figure.

It fails on the load factor. A regional supply chain assembled out of small vehicles running half empty does not merely fail to beat the long chain on emissions — it loses to it, measurably, and the published literature is unambiguous on this point. A programme that does not measure its own load factors is not measuring the thing it claims to be improving.

And it fails when the cost lands on the wrong households and nobody says so. The premium on a localised basket is a regressive tax unless it is deliberately offset. It can be offset — through the procurement of institutions rather than of individuals, which is Preston's whole method — but not by accident, and not in silence.

---

DELIGHT

What it feels like

There is a specific pleasure in learning the shape of the place you live, and it arrives the first time the balance tells you something you did not know. That the orchards two valleys over produce four times what the county eats. That the thing you assumed came from far away has been coming from thirty kilometres since before anyone measured it. That the soil changes at the lane end for a reason that is nine thousand years old.

It is the pleasure of a map resolving. You have lived somewhere for years as a set of addresses and journey times, and then one afternoon it becomes a system with a metabolism — rain arriving here, running that way, growing that, feeding these people, and leaving by that road. The place stops being scenery.

And there is a second pleasure, quieter, in being able to say plainly that the coffee comes by ship from eight thousand kilometres and that this is the right answer. A bioregional economics that can say that without flinching is one you can stay inside for a working lifetime, because it has stopped asking you to pretend. The relief is in the arithmetic, not in the loyalty — and relief, it turns out, is what makes a person keep the ledger.

---

OPERATIONALIZE THIS

At the level of finance

Here is the instrument, in the form a treasurer will recognise, and it is designed around the two findings that killed the previous generation of bioregional programmes: the premium is real, and the load factor decides the carbon.

The structure: a bioregional offtake facility with a load-factor covenant.

An anchor buyer — a hospital group, a university, a retailer, a food manufacturer, a construction contractor — enters a multi-year fixed-volume offtake with a producer cluster inside a named ecoregion at a stated level, covering only the commodity lines the balance placed in the regional by nature and regional in season categories. The producer cluster borrows working capital against the offtake. The premium the buyer pays is conditional on a measured delivery covenant.

The mechanics.

• Perimeter. The ecoregion, by published code and level, written into the contract. Not "local". Not "within fifty miles". A code, a level, a date of the boundary version used.
• Volume. A nominated annual tonnage over three to five years, with take-or-pay on around 70 percent of nomination and full flexibility above it. Three years is the minimum that changes a grower's planting decision, which is the entire point of the instrument.
• Price. A published regional cost-of-production index plus a fixed margin, collared against the incumbent's delivered landed cost so the buyer's exposure is bounded and boardable. The collar is what gets this past a CFO.
• The load-factor covenant — the clause that is new. The regional premium is payable only in periods where the weighted average vehicle load factor on the regional route exceeds a stated threshold, verified from weighbridge tickets or telematics. Below the threshold, price reverts to the incumbent's landed cost and the shortfall is recorded. This directly prices the failure mode the published load-factor studies found, and it turns the producer cluster's consolidation problem into the producer cluster's revenue problem, which is where it can actually be solved.
• Security. Assignment of the offtake receivables to the working-capital lender. Ordinary, well-precedented, and the reason a regional cluster can borrow at all.
• Counterparty. A cooperative or a single marketing agent. Never fifty separate farms — the administrative cost of a fifty-counterparty contract is itself a coordination cost of the kind Chapter I.10 computes, and it will eat the premium before the first delivery.

The balance-sheet treatment. For the buyer this is an executory purchase commitment: disclosed as an unconditional purchase obligation rather than recognised as a liability, under IAS 37 and, in US practice, ASC 440-10. Two things to settle with the auditor before signing. First, whether the take-or-pay tranche creates an onerous-contract provision if the collar binds — it can, and the collar width is where that is controlled. Second, whether any dedicated storage, plant or vehicle named in the agreement constitutes an identified asset and therefore a lease under IFRS 16; keep the asset unidentified unless you intend the lease. For the producer cluster, the offtake is not revenue until performance, but it is bankable, and that is its whole commercial function.

The number that decides it. One inequality, on the front page:

   regional premium per tonne   <=   c x (E_far - E_near)  +  p x L

   c        the firm's internal carbon price, currency per tCO2e
   E_far    delivered emissions of the incumbent, tCO2e per tonne
   E_near   delivered emissions of the regional route, tCO2e per tonne,
            measured at the covenanted load factor and not above it
   p        annual probability of a supply interruption on this line
   L        loss per tonne if that interruption occurs

If the inequality holds, the premium is not a preference. It is the cheapest available combination of abatement and insurance, and it is presented as such, in that order, with the carbon term and the risk term shown separately so that the board can disagree with one without rejecting both.

If it does not hold, the line does not belong in the facility, and saying so is what makes the lines that do belong credible.

The first ninety days.

Start with one commodity line and one counterparty. A facility covering one good that a grower can plant against is worth more than a framework covering forty that nobody has signed.

---

APPRECIATIVE QUESTIONS

Twelve, for a room

Discovery — what is already working

1. Which three things grow, run or get made unusually well where we are, better than they do a few hundred kilometres away — and what is it about the place that makes that true?
2. Which of our supply lines is already short without anyone having decided it should be? Who set it up, and what did they know?
3. When has a shortage somewhere else in the world left us untouched? What was it about how we were buying that protected us — and do we still buy that way?

Dream — what becomes possible

1. If we published an honest balance of what this region supplies to itself, what would we most want the first page to say in five years' time?
2. Imagine we knew, for every material line we buy, whether it was regional by nature or traded by nature. What would we stop arguing about?
3. If the people who live here could read the region's balance the way they read a weather forecast, what would they start asking us for?

Design — what we build

1. At what level of the hierarchy is the decision in front of us actually taken — and which level have we been arguing at instead?
2. Which single commodity line, if we contracted it for three years inside the boundary, would most change what a grower or a maker here is willing to invest in?
3. What would we have to measure about our own deliveries before we could claim, with a straight face, that the short chain is the cleaner one?

Destiny — how it holds

1. What would have to be true for this balance to still be published, by someone we have not met, ten years from now?
2. Where is the line we will publish that says buy this from four thousand kilometres away — and who needs to have agreed to it before it appears?
3. If the regional premium ends up landing on the households least able to carry it, who here would notice first, and what would we want them to do with that?

---

WORKS CITED

Twenty-six sources, in the order the argument needs them

Berg, P. and Dasmann, R. (1977). "Reinhabiting California." The Ecologist, 7(10), 399–401.

Omernik, J. M. (1987). "Ecoregions of the Conterminous United States." Annals of the Association of American Geographers, 77(1), 118–125.

Omernik, J. M. and Griffith, G. E. (2014). "Ecoregions of the Conterminous United States: Evolution of a Hierarchical Spatial Framework." Environmental Management, 54(6), 1249–1266.

Commission for Environmental Cooperation (1997, and subsequent revisions). Ecological Regions of North America: Toward a Common Perspective. Montreal. Level I, II and III counts as published by the CEC and the U.S. Environmental Protection Agency.

Dinerstein, E., Olson, D., Joshi, A. et al. (2017). "An Ecoregion-Based Approach to Protecting Half the Terrestrial Realm." BioScience, 67(6), 534–545. doi:10.1093/biosci/bix014.

Olson, D. M., Dinerstein, E., Wikramanayake, E. D. et al. (2001). "Terrestrial Ecoregions of the World: A New Map of Life on Earth." BioScience, 51(11), 933–938.

Peters, C. J., Wilkins, J. L. and Fick, G. W. (2007). "Testing a Complete-Diet Model for Estimating the Land Resource Requirements of Food Consumption and Agricultural Carrying Capacity: The New York State Example." Renewable Agriculture and Food Systems, 22(2), 145–153.

Peters, C. J., Bills, N. L., Lembo, A. J., Wilkins, J. L. and Fick, G. W. (2009). "Mapping Potential Foodsheds in New York State: A Spatial Model for Evaluating the Capacity to Localize Food Production." Renewable Agriculture and Food Systems, 24(1), 72–84. doi:10.1017/S1742170508002457.

Peters, C. J., Bills, N. L., Lembo, A. J., Wilkins, J. L. and Fick, G. W. (2012). "Mapping Potential Foodsheds in New York State by Food Group: An Approach for Prioritizing Which Foods to Grow Locally." Renewable Agriculture and Food Systems, 27(2), 125–137. doi:10.1017/S1742170511000196.

Peters, C. J., Picardy, J., Darrouzet-Nardi, A. F., Wilkins, J. L., Griffin, T. S. and Fick, G. W. (2016). "Carrying Capacity of U.S. Agricultural Land: Ten Diet Scenarios." Elementa: Science of the Anthropocene, 4, 000116. doi:10.12952/journal.elementa.000116.

Griffin, T. S., Conrad, Z., Peters, C. J., Ridberg, R. and Tyler, E. P. (2014). "Regional Self-Reliance of the Northeast Food System." Renewable Agriculture and Food Systems, 30(4), 349–363. doi:10.1017/S1742170514000027.

Conrad, Z., Blackstone, N. T., Peters, C. J. and Griffin, T. S. (2016). "The Role of Feed and Livestock in Regional Food System Self-Reliance." Renewable Agriculture and Food Systems, 32(2), 145–156. doi:10.1017/S1742170516000089.

Colasanti, K. J. A. and Hamm, M. W. (2010). "Assessing the Local Food Supply Capacity of Detroit, Michigan." Journal of Agriculture, Food Systems, and Community Development, 1(2), 41–58.

Kinnunen, P., Guillaume, J. H. A., Taka, M., D'Odorico, P., Siebert, S., Puma, M. J., Jalava, M. and Kummu, M. (2020). "Local Food Crop Production Can Fulfil Demand for Less Than One-Third of the Population." Nature Food, 1(4), 229–237. doi:10.1038/s43016-020-0060-7.

United States Department of Agriculture, National Agricultural Statistics Service (2024). 2022 Census of Agriculture, Volume 1, Chapter 1 (United States and State Level) and Chapter 2 (State Level), Table 1 and Tables 35–38.

United States Census Bureau (2021). 2020 Census Redistricting Data (P.L. 94-171) Summary File, Table P1, New York.

United States Department of Agriculture, Economic Research Service. Food Availability (Per Capita) Data System. Successive releases; commodity series terminate in different years, from 2010 for added fats and oils to 2023 for caloric sweeteners.

Weber, C. L. and Matthews, H. S. (2008). "Food-Miles and the Relative Climate Impacts of Food Choices in the United States." Environmental Science and Technology, 42(10), 3508–3513. Table 1 for modal energy and greenhouse gas intensities per tonne-kilometre.

Coley, D., Howard, M. and Winter, M. (2009). "Local Food, Food Miles and Carbon Emissions: A Comparison of Farm Shop and Mass Distribution Approaches." Food Policy, 34(2), 150–155.

Mundler, P. and Rumpus, L. (2012). "The Energy Efficiency of Local Food Systems: A Comparison Between Different Modes of Distribution." Food Policy, 37(6), 609–615. doi:10.1016/j.foodpol.2012.07.006. Figures for individual chains as reproduced, with page citations, by Maréchal, Plateau and Holzemer (2019), "La durabilité des circuits courts, une question d'échelle?", Économie Rurale, 367(1), 45–60.

Saunders, C., Barber, A. and Taylor, G. (2006). Food Miles — Comparative Energy/Emissions Performance of New Zealand's Agriculture Industry. Research Report No. 285, Agribusiness and Economics Research Unit, Lincoln University. Tables 6.13, 7.1, 7.3, 7.4 and 7.5.

Audsley, E., Brander, M., Chatterton, J., Murphy-Bokern, D., Webster, C. and Williams, A. (2010). How Low Can We Go? An Assessment of Greenhouse Gas Emissions from the UK Food System and the Scope for Reduction by 2050. Cranfield University, Ecometrica and Murphy-Bokern Konzepte for WWF-UK and the Food Climate Research Network. Table 13.

Williams, A. G., Audsley, E. and Sandars, D. L. (2006). Determining the Environmental Burdens and Resource Use in the Production of Agricultural and Horticultural Commodities. Defra Research Project IS0205. Cranfield University and Defra.

Ricardo, D. (1817). On the Principles of Political Economy and Taxation. John Murray, London. Chapter 7, "On Foreign Trade."

Fajgelbaum, P. D. and Khandelwal, A. K. (2016). "Measuring the Unequal Gains from Trade." Quarterly Journal of Economics, 131(3), 1113–1180. doi:10.1093/qje/qjw013.

Borusyak, K. and Jaravel, X. (2023). The Distributional Effects of Trade: Theory and Evidence from the United States. Centre for Economic Performance Discussion Paper No. 1953, London School of Economics; also NBER Working Paper 28957 (2021). A working paper, and cited here as one, because it is the strongest direct measurement contradicting the result above.

Autor, D. H., Dorn, D. and Hanson, G. H. (2013). "The China Syndrome: Local Labor Market Effects of Import Competition in the United States." American Economic Review, 103(6), 2121–2168. doi:10.1257/aer.103.6.2121.

Note on figures. Every figure in this chapter is computed in lib/verify/VII_05.py and printed with its inputs, its units and its source. The New York City foodshed is computed twice, by two routes sharing no assumptions — a biophysical complete-diet model and a per-capita share of national Census acreage — and the two agree to within half a percent. Where a figure is stated by a source rather than derived here, the module labels it as such and does not count it as computed. New York's citrus acreage is withheld by the Census to protect a single operation; withheld is not zero, and the module says so.