Waste as a Category Error

Volume IV — Production and Regeneration

Nine movements, one fence.

THE PLATE

Timber lodges at the foot of jagged mountains, their windows lit gold as the sun goes down over the sage. — Plate IV.06The Conveyor at Three Kilometres.The powder on that belt is a waste product at one end and a raw material at the other. Nothing happens to it in between. It crosses a property line.

THE LETTER

There is a word in your accounts that is doing more work than you have asked it to do, and the word is waste.

It appears as a disposal line, a gate fee, a scrap percentage, a yield loss, an obsolescence provision. In each place it looks like a description of a substance — this stuff here is waste, that stuff there is stock — and it is not. It is a description of a price. A flow is waste when the person holding it would pay to be rid of it. The same flow is a feedstock when somebody else would pay to have it. Nothing in the material decides which sentence is true. What decides it is where the fence runs, what a contract would cost to write, and what the law happens to call the stuff this year.

That sounds like a philosophical point and it is not. It is an operating instruction, and it produces a number.

This chapter does three things. It establishes that waste is an accounting category — a claim which is easy to make and easy to make badly. Then it tests the claim against the best industrial evidence there is: Kalundborg in Denmark, which has been running by-product exchanges since 1961 and publishes its volumes; the United Kingdom's National Industrial Symbiosis Programme, which ran the same idea at national scale for eight years and had its outputs evaluated; and the by-product synergy literature, which is honest enough to record where these things fail.

Then it computes the one number that explains most of the failures — the distance beyond which a symbiosis stops paying — and shows that the number is not really about distance at all.

You will leave with a threshold you can apply to any stream leaving your site by Friday, and with a clear view of the risk that comes attached, which is real and which has a price.

— The Editors

DISCOVERY

What is already working

Kalundborg, Zealand, Denmark. In 1961 an oil refinery on the Kalundborg fjord needed water and the groundwater would not carry it, so it laid a pipe to Lake Tissø. That is the whole of the founding event: a company solving its own problem, commercially, with no plan for an ecosystem. Sixty years on, sixteen partner organisations exchange steam, water, gas, gypsum, fly ash, sulphur, biomass residue and heat across a few square kilometres of industrial estate, and the Symbiosis publishes its annual figures: 635,000 tonnes of carbon dioxide avoided, 3,600,000 cubic metres of water saved, 100 GWh of energy, and 87,000 tonnes of material recycled, each year. Those boundaries are declared by the Symbiosis itself rather than by a public auditor, which is worth saying out loud and does not diminish them.

What matters more than the volumes is the character of the thing. Jørgen Christensen, who ran the refinery's technical function for much of the period, described it consistently as a set of bilateral commercial deals that happened to add up. Ehrenfeld and Gertler, in the study that introduced Kalundborg to the industrial ecology literature in 1997, found the same: no master plan, no subsidy, no central authority — a sequence of ordinary negotiations, each of which had to pay on its own.

And they did pay. The two accounts most often cited disagree with each other. One reports cumulative investment on the order of US$75 million against US$15 million a year of recurring savings. The other reports US$60 million invested against US$120 million of cumulative savings over five years. Take the first and the payback is five years. Take the second and it is two and a half. They disagree by a factor of two and the conclusion survives the disagreement: a payback somewhere between two and a half and five years sits comfortably inside an ordinary industrial capital hurdle. You do not need to resolve the dispute to make the decision.

Kwinana, Western Australia. An industrial area of alumina refineries, chemical plants, a power station and a port. Van Beers and colleagues documented it systematically and found forty-nine synergies in operation — thirty-two by-product exchanges and seventeen shared utilities. Again: no designer. The synergies were found by looking at what was already there.

Guitang, Guangxi, China. A state sugar group that built a paper mill on its bagasse, an alcohol plant on its molasses, a cement works on its filter mud and a fertiliser line on the residue of all three. The exchanges were not an environmental programme. They were how a low-margin sugar business stayed solvent, and Zhu and colleagues documented them as such.

NISP, the United Kingdom, 2005 to 2013. The idea taken national. Rather than building an eco-industrial park, NISP employed practitioners who knew regional industry, put companies in rooms together, and brokered the matches. Over the first five years the programme reported 7 million tonnes diverted from landfill, 6 million tonnes of carbon dioxide avoided, 11 million tonnes of virgin material saved, 11 million tonnes of water, 360,000 tonnes of hazardous waste eliminated, £1,000,000,000 in additional sales to industry, another £1,000,000,000 in cost savings, and around 10,000 jobs created or safeguarded.

Against public funding on the order of £25,000,000 for the period, that is 40:1 on cost savings alone and 80:1 counting the sales, and it works out at £4.17 per tonne of carbon dioxide abated. Move the funding figure to £15,000,000 and it is £2.50 per tonne; move it to £35,000,000 and it is £5.83. The conclusion is robust across the whole range: as an abatement measure this is roughly an order of magnitude cheaper than almost anything else a government buys. These are programme-reported outputs with evaluator validation, not national statistics, and the denominator is self-declared. Say so, and the figures still stand up.

Five cases, four continents, one shape. In every one, somebody found a flow that was already leaving a building and changed its destination rather than its quantity. Nobody invented a material. Nobody built a technology. They changed who owned what, and when.

THE ARITHMETIC

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

First, the claim, stated so it can be tested.

Take the global material flow. Around 100.6 gigatonnes of materials are extracted into the world economy each year. Of that, something between 7.2 and 8.6 percent is cycled back, depending on the year and the accounting; the European Union's own circular material use rate stands at about 11.5 percent. Which means 91.9 gigatonnes a year are not cycled.

Now ask the physical question. How much of that 91.9 gigatonnes is destroyed?

Zero. Mass is conserved. Every tonne is still here — dispersed, oxidised, buried, diluted, alloyed into something nobody can separate, but present. What changed when we called it waste was not its existence. It was its sign: the flow moved from the asset side of somebody's accounts to the cost side, and the moment it did, the rational thing to do with it became to minimise contact with it as cheaply as possible.

That is the category error, and it is genuinely an error of category rather than of judgement. Waste is a term in the accounts that has been mistaken for a term in the inventory.

Second, the test. If the claim is true, then moving the fence should move the category — and it should do so at a computable price.

Here is the computation, and it is the spine of the chapter.

A by-product exchange pays when the margin per tonne exceeds the cost of moving the tonne. Start with the cost of moving, derived rather than quoted:

  HGV all-in operating cost     1.40 EUR/km   (driver, fuel, vehicle, overhead)
  bulk payload                  25.0 t
  empty-return factor           2              (the truck comes back empty)
  ------------------------------------------------------------------
  c  =  2 x 1.40 / 25.0   =     0.112 EUR per tonne-kilometre

At the low end of haulage cost that is 0.096; at the high end, 0.128. Use your own if you have one — the method is what matters.

Now the margin. For any stream, the value of an exchange to the two parties together is the virgin input the receiver no longer buys, plus the disposal the sender no longer pays for, minus what it costs to handle and qualify the material:

  m  =  avoided virgin input  +  avoided disposal  -  handling

Four real streams, priced at European order of magnitude:

  stream                              virgin  disposal  handling      m
  ----------------------------------------------------------------------
  FGD gypsum -> plasterboard           30.00     15.00      8.00   37.00
  bagasse -> pulp                      45.00      6.00     12.00   39.00
  fermentation residue -> farmland      8.00     20.00     10.00   18.00
  spent foundry sand -> fill            6.00      9.00     12.00    3.00

And the threshold distance is simply d_max = m / c:

  bagasse -> pulp                    39.00 / 0.112  =   348.2 km
  FGD gypsum -> plasterboard         37.00 / 0.112  =   330.4 km
  fermentation residue -> farmland   18.00 / 0.112  =   160.7 km
  spent foundry sand -> fill          3.00 / 0.112  =    26.8 km

A factor of 12.3 between the best stream and the worst, and nothing in that spread is about the technology. It is about the price of the thing the material displaces.

The same arithmetic in capital form explains why heat never travels. A 50 MW recoverable flow at 4,000 hours a year is 200,000 MWh, worth about 5,000,000 euros at 25 euros a megawatt-hour. A buried transmission main costs on the order of 1,500 euros a metre, which at a 7 percent discount over thirty years is a capital recovery factor of 0.0806, or 120,880 euros per kilometre per year. So d_max = 5,000,000 / 120,880 = 41.4 kilometres — and European district heating transmission mains, in practice, rarely run beyond twenty or thirty. The threshold predicted the observed world before we looked at it.

This is the point at which the chapter can be checked rather than believed. Jensen and colleagues measured the distances that NISP synergies actually travelled and found them concentrated in the tens of kilometres with a long, thin tail. That distribution is exactly what these four numbers generate from cost alone. A model that reproduces an observed distribution it was not fitted to has earned a little trust.

Third — and this is the cut — the distance is not the constraint.

Watch what happens when the flow gets small. Writing the contract is a one-off cost: legal drafting, a waste-classification dossier, a permit variation, sampling and due diligence. Call it 40,000 euros over a five-year contract. Per tonne, that term is T / (Q · L), and it behaves violently:

  Q (t/yr)    transaction    net margin    d_max
  ---------------------------------------------
   170,000        0.05 EUR       36.95    329.9 km
    20,000        0.40           36.60    326.8 km
     5,000        1.60           35.40    316.1 km
     1,000        8.00           29.00    258.9 km
       500       16.00           21.00    187.5 km
       300       26.67           10.33     92.3 km
       250       32.00            5.00     44.6 km

And the flow at which it goes to nothing:

  Q_zero  =  T / (L · m)  =  40,000 / (5 x 37.00)  =  216.2 t/yr

Below roughly 216 tonnes a year, the contract costs more than the material is worth at zero distance. Two plants sharing a wall, with a perfectly good exchange available, and it does not pay — not because of the truck, because of the lawyer.

Which means the threshold that governs industrial symbiosis is not a distance. It is the cost of crossing an ownership boundary, and distance is only one of the things you pay for crossing it. Oliver Williamson gave the general case its name half a century ago: site specificity, asset specificity, and the transaction costs that attach to both. Ronald Coase had already explained, in 1937, why a firm exists at all — because some exchanges are cheaper inside a boundary than across one. Industrial symbiosis is that theory in a pipe.

And the legal term is not a metaphor. Under the European Waste Framework Directive, Article 5 sets out when a production residue is a by-product and Article 6 when waste ceases to be waste. Moving a stream from one side of that line to the other changes no molecule, no distance and no truck. It changes T — sometimes by an order of magnitude. "Zero waste" is therefore a statement about contracts, not about matter. The matter was never going anywhere.

Fourth, the honest negative, and it is a real one.

A symbiosis is a mutual dependency, and a dependency has a price that nobody puts on the page at the time.

Kalundborg's own flagship exchange is the case. Asnæs power station burned coal; desulphurising the flue gas produced synthetic gypsum; roughly 170,000 tonnes a year of it crossed 3 kilometres of harbour to a plasterboard works that had been sited there in the 1970s precisely because of it. It was, for decades, one of the best by-product exchanges in the world.

In 2019 Asnæs stopped burning coal and moved to wood chips. No sulphur, no gypsum. The receiving plant inherited a building designed around a conveyor.

Price it. Natural gypsum delivered from the Spanish trade — some 2,500 kilometres by sea — against a fenceline stream that arrived for near the cost of handling it. At an assumed delta of 25 euros a tonne, that is 4,250,000 euros a year; at 15 euros, 2,550,000; at 35, 5,950,000. The exact contracted price is commercial and is not knowable from outside, which is why the figure is given as a range and labelled as an assumption rather than dressed as a measurement.

The carbon runs the same way. Sea freight at 0.008 kilograms of carbon dioxide per tonne-kilometre over 2,500 kilometres is 3,400 tonnes a year; the fenceline conveyor equivalent is 63. The substitution carries 3,337 tonnes a year, which is 0.53 percent of the Symbiosis's whole reported annual avoidance — one broken link giving back half a percent of sixty years of work.

So price the dependency in advance. The expected annual cost of a partner leaving is p · C / Q — the probability their process changes, times the cost of reconfiguring, spread over the flow:

  stream          m      p         C          Q        p·C/Q    m_net   share
  --------------------------------------------------------------------------
  large fence   25.00   3%   8,000,000    170,000       1.41    23.59    5.6%
  mid           18.00   3%   2,000,000     20,000       3.00    15.00   16.7%
  small         18.00   3%   2,000,000      5,000      12.00     6.00   66.7%

The small stream loses two-thirds of its margin to a risk nobody priced. And notice that it is the same term, 1/Q, that killed it on transaction cost. A small flow across a fence is expensive twice, for two unrelated reasons, and the two multiply.

The threshold, complete. A by-product exchange pays when:

      m  -  T/(Q·L)  -  p·C/Q   >   c · d

Everything in this chapter is that inequality, and every failure in the literature is one of its terms.

DREAM

What becomes ordinary

In the economy that has taken this seriously, a material flow leaving a site has a counterparty the way a product has a customer, and it is somebody's job.

The disposal line in the management accounts has been split in two. One part is genuine loss — dispersed, contaminated, not worth recovering, and named as such without embarrassment. The other part is called what it is: inventory held at a negative price because no contract exists yet. The second number is reviewed monthly, and it falls.

Contracting has become cheap, which is the quiet revolution. A standard by-product offtake agreement exists the way a standard lease exists — a form that two mid-sized companies can execute in a fortnight without either of them paying for original drafting. Because T fell, Q_zero fell with it, and streams of two or three hundred tonnes a year that were never worth a contract are now worth a contract. The number of viable exchanges in an economy is a function of its legal stationery, and everybody knows it.

Regulators have separated the two questions they used to ask as one: is this material dangerous, and is this material unwanted. The first is answered with tests. The second is answered with a signature. A residue whose composition is known and stable receives by-product status as a matter of routine, and the dossier is reusable across a sector rather than rewritten per company.

Industrial estates are laid out by flow, not by parcel. A planner siting a new plant asks what leaves the neighbours and at what temperature, and the answer changes the price of the land. Sites with a hot, steady, well-characterised stream next door command a premium, openly, in the way a site near a motorway junction does now.

And the dependency is written down. Every significant fenceline exchange carries a notice period matched to the redesign time of the plant that depends on it, and a reserve that accrues out of the savings. When a partner does change fuel — and partners change fuel — the other one has thirty-six months and a funded account, instead of a surprise and a conveyor.

None of that requires a new technology. It requires a form, a register, a default clause, and somebody whose job it is.

DESIGN

The structure that gets there

Start from what is already leaving. Pull three years of disposal, scrap, write-off and gate-fee lines, sorted by tonnage and by gate fee per tonne. The second sort is the one that finds money: a high gate fee means someone is already paying to make a material disappear, and the whole of that payment is available as margin.

Apply the three-two heuristic. Marian Chertow's working definition of a symbiosis — as opposed to a bilateral deal or a recycling contract — is at least three different entities exchanging at least two different resources, none of them primarily in the recycling business. It is a useful filter because it separates a system, which develops its own resilience, from a single dependency, which does not.

Then run the inequality, in this order.

m — the margin. Avoided virgin input plus avoided disposal minus handling. If m is negative, stop; no distance saves it.
Q_zero — the contract floor. T / (L · m). If your flow is below it, the answer is not a contract. It is an aggregator, a sector-standard agreement, or a co-location.
d_max — the distance. (m − T/(Q·L)) / c. Compare with the real road distance, not the map distance.
p · C / Q — the dependency. What does the receiver have to rebuild if the sender changes process, and how likely is that over the term?

Broker it, do not plan it. This is the clearest empirical finding in the field and it deserves its weight. The planned eco-industrial parks of the 1990s largely did not achieve the exchanges they were designed around; Heeres and colleagues, and Gibbs and Deutz, both report it, and both find that company-led arrangements did better. NISP's model was not a park. It was practitioners who knew an industry standing in a room with people who had materials, and the brokering was the product. No single clean completion rate is published across that literature — which is worth saying plainly rather than estimating into a table — but the direction is not in dispute.

Govern it as a commons, because it is one. Ostrom's design principles apply almost line for line: clear boundaries about who is in the exchange, rules that fit local conditions, the people affected able to change the rules, monitoring by the participants themselves, graduated sanctions, and cheap conflict resolution. Kalundborg's participants meet, and have met for decades. That is not collegiality. It is principle four.

Write the dependency in at the start. Notice period matched to redesign time. A substitution covenant naming the qualified alternate. A concentration limit — no single fenceline source above a stated share of feedstock without a reserve. And the reserve itself, funded from the margin the exchange creates.

DESTINY

How it holds when nobody is pushing

It holds because each link is a commercial contract that pays on its own terms. Kalundborg has survived sixty years, three ownership changes among its partners, and a complete change of fuel at its largest node, because no link in it was ever load-bearing on goodwill. A symbiosis held together by shared values is a club. A symbiosis held together by invoices is infrastructure.

It fails in four ways, and all four are visible in the inequality.

It fails when Q is small — the contract and the dependency both scale with 1/Q, so small streams are penalised twice and die of paperwork.

It fails when T is set by a regulator who will not distinguish a residue from a refuse. This is the most fixable failure in the whole field and it is fixed with a form, not a technology.

It fails when p was never estimated. The plasterboard works at Kalundborg is the honest example: a superb exchange for thirty years, and then a fuel change three kilometres away and several million euros a year of inherited cost.

And it fails when somebody builds the park before finding the flows — designing a symbiosis into a masterplan and then recruiting tenants to fit it. The evidence on that is consistent enough to treat as settled: find the flows, then draw the site.

DELIGHT

What it feels like

There is a specific pleasure in the moment a disposal line turns into a revenue line, and it is not the money. It is the feeling of having been wrong in a way that is entirely to your advantage — of discovering that something you had filed as a cost of doing business was a category mistake you inherited and never examined.

People who work in symbioses talk about the neighbours. Not in a sentimental way: they know what temperature the plant next door runs at, what its shutdown weeks are, what it needs in February. An industrial estate stops being a set of parcels that happen to share a postcode and becomes a place where the people know each other's processes. That is a nicer place to work and everyone who has done it says so.

And there is the plain sensory fact of it. A covered belt moving pale powder three kilometres across a harbour, in weather, on a Tuesday, worth several million euros a year to two companies and costing almost nothing to run. It is the least dramatic infrastructure imaginable and it is doing more than most things that photograph better.

OPERATIONALIZE THIS

At the level of finance

The instrument: a by-product offtake agreement with a dependency reserve.

This is a supply contract, not a sustainability programme, and it belongs in the same drawer as your other supply contracts. What follows is the structure, in the form a treasurer and a general counsel will both recognise.

The structure.

Term. Ten years, matched to the receiver's plant investment cycle. Shorter terms do not justify plant designed around the input, which is the whole value.
Volume. Nominal annual tonnage with a take-or-pay floor at 60 percent. On a 170,000-tonne nominal that is 102,000 tonnes a year the sender can count on. The floor is what makes the sender's own capital case work.
Price. Indexed to the delivered price of the virgin substitute, less a stated discount. Never a fixed price: a fixed price on a ten-year by-product contract converts a shared benefit into a bet on a commodity index, and whichever party loses that bet will stop cooperating.
Specification and rejection. Named test method, named sampling frequency, named tolerance, and a rejection route that does not require litigation.
Notice. Thirty-six months, matched to the receiver's reconfiguration time. This single clause is the difference between an orderly transition and a stranded plant.
The dependency reserve. An escrow accruing at 15 percent of verified annual savings — on a 4,250,000-euro saving that is 637,500 euros a year. Over ten years at 7 percent that accrues to 8,807,986 euros against an assumed 8,000,000-euro reconfiguration cost: a cover ratio of 1.10. It releases to the receiver if supply ends and reverts to the operating units if it does not.

The balance-sheet treatment. For the sender, a stream that moves from disposal cost to revenue removes a provision and adds a receivable — talk to your auditor about whether the historic disposal provision can be released, because that is frequently a larger one-off than the first year of margin. For the receiver, the offtake is an executory contract; the take-or-pay floor is a commitment to disclose. Where the exchange lets you build a plant you could not otherwise justify, the contract is the asset's economic life, so depreciate to the contract, not past it.

The counterparty. Your neighbour. Start inside the fence if you can: a transfer between two business units of the same firm needs no waste classification, no permit variation and no lawyer, which sets T near zero and Q_zero near zero with it. Prove it internally, then take the proven structure across the boundary.

The first ninety days.

Day	Action	Artifact
1–15	Rank every outbound stream by gate fee per tonne	The disposal register, re-sorted
16–30	Compute `m`, `Q_zero` and `d_max` for the top five	One page of arithmetic
31–45	Find the counterparty for the best one	A named company and a named person
46–60	Classification: by-product or end-of-waste dossier	The regulator's answer, in writing
61–75	Draft offtake, floor, notice period and reserve	Term sheet
76–90	Signature and first load	A weighbridge ticket

The number that decides it. Two, and they take one line each.

  d_max  =  ( m  -  T/(Q·L)  -  p·C/Q )  /  c      must exceed the real haul

  concentration  =  fenceline tonnes / total feedstock tonnes

If the concentration ratio is above 50 percent and the notice period is shorter than the receiver's redesign time, the reserve is mandatory, not optional. At Kalundborg the ratio was 85 percent — 170,000 tonnes out of 200,000 — and the reserve did not exist. That is the whole lesson, in one fraction.

APPRECIATIVE QUESTIONS

Twelve, for a room

Discovery — what is already working

Which material leaves this site that somebody already pays us for, or would if we asked — and who found that out, and how?
Think of a time we turned a disposal cost into something useful. What made it possible, and who had to agree?
Of everything we send out of here, which stream do we know the most about — composition, moisture, consistency, timing — and why do we know it so well?

Dream — what becomes possible

If every outbound flow had a named counterparty the way every product has a customer, what would change about who we talk to in a week?
Imagine the disposal line split in two on next year's management accounts — genuine loss, and inventory awaiting a contract. What would we want the second number to be?
If contracting with a neighbour took a fortnight and a standard form, which three streams would we do first?

Design — what we build

What is the largest gate fee per tonne we pay anywhere in this business, and who is nearest to us that could use the material?
Which of our inputs comes from furthest away, and is there anybody within a hundred kilometres making it as a residue?
What would a good notice period be for us — how long would we actually need to reconfigure if a fenceline supply stopped?

Destiny — how it holds

Which of our exchanges would survive the person who set it up leaving, and what makes the difference between those and the others?
If our largest supplier of a by-product changed their process next year, what would we wish we had signed today?
Who else is in this exchange besides the two of us — and what would it take to make it three parties and two materials rather than one deal?

WORKS CITED

Chertow, M. R. (2000). "Industrial Symbiosis: Literature and Taxonomy." Annual Review of Energy and the Environment, 25, 313–337.

Chertow, M. R. (2007). "'Uncovering' Industrial Symbiosis." Journal of Industrial Ecology, 11(1), 11–30.

Chertow, M. R. and Lombardi, D. R. (2005). "Quantifying Economic and Environmental Benefits of Co-Located Firms." Environmental Science & Technology, 39(17), 6535–6541.

Circle Economy (2020). The Circularity Gap Report 2020. Amsterdam.

Circle Economy (2023). The Circularity Gap Report 2023. Amsterdam.

Coase, R. H. (1937). "The Nature of the Firm." Economica, 4(16), 386–405.

Desrochers, P. (2004). "Industrial Symbiosis: The Case for Market Coordination." Journal of Cleaner Production, 12(8–10), 1099–1110.

Ehrenfeld, J. and Gertler, N. (1997). "Industrial Ecology in Practice: The Evolution of Interdependence at Kalundborg." Journal of Industrial Ecology, 1(1), 67–79.

European Union (2008). Directive 2008/98/EC on waste (Waste Framework Directive), Articles 5 and 6.

Eurostat. Circular material use rate. Annual series.

Frosch, R. A. and Gallopoulos, N. E. (1989). "Strategies for Manufacturing." Scientific American, 261(3), 144–152.

Gibbs, D. and Deutz, P. (2007). "Reflections on Implementing Industrial Ecology through Eco-Industrial Park Development." Journal of Cleaner Production, 15(17), 1683–1695.

Heeres, R. R., Vermeulen, W. J. V. and de Walle, F. B. (2004). "Eco-Industrial Park Initiatives in the USA and the Netherlands: First Lessons." Journal of Cleaner Production, 12(8–10), 985–995.

Jacobsen, N. B. (2006). "Industrial Symbiosis in Kalundborg, Denmark: A Quantitative Assessment of Economic and Environmental Aspects." Journal of Industrial Ecology, 10(1–2), 239–255.

Jensen, P. D., Basson, L., Hellawell, E. E., Bailey, M. R. and Leach, M. (2011). "Quantifying 'Geographic Proximity': Experiences from the United Kingdom's National Industrial Symbiosis Programme." Resources, Conservation and Recycling, 55(7), 703–712.

Kalundborg Symbiosis. Published annual figures and exchange map. Kalundborg.

Lombardi, D. R. and Laybourn, P. (2012). "Redefining Industrial Symbiosis: Crossing Academic–Practitioner Boundaries." Journal of Industrial Ecology, 16(1), 28–37.

Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press.

van Beers, D., Corder, G., Bossilkov, A. and van Berkel, R. (2007). "Industrial Symbiosis in the Australian Minerals Industry: The Cases of Kwinana and Gladstone." Journal of Industrial Ecology, 11(1), 55–72.

Williamson, O. E. (1985). The Economic Institutions of Capitalism. Free Press.

Zhu, Q., Lowe, E. A., Wei, Y. and Barnes, D. (2007). "Industrial Symbiosis in China: A Case Study of the Guitang Group." Journal of Industrial Ecology, 11(1), 31–42.

Note on figures. Every figure in this chapter is computed in lib/verify/IV_06.py and printed with its inputs, its units and its source. Prices, haulage costs, reconfiguration costs and exit probabilities are stated assumptions at European order of magnitude, computed at low, central and high values, and labelled as assumptions rather than measurements. Kalundborg and NISP outputs are as reported by those programmes, with self-declared boundaries. The completion rate of identified-to-implemented synergies is not published cleanly across the literature and is named as unknown rather than estimated.