Prepared by the ENSI Foresight Division. Built on the 183-document primary library assembled for this programme, with no new sourcing. Forty-eight candidate arguments were generated and scored on evidence, independence and decision-relevance; the sixteen below are what survived, and the other thirty-two are folded into them rather than discarded. Compiled 24 October 2026.
The argument, before the list
Most cases for a new economic arrangement fail in the same place. They assemble reasons to believe and stop — leaving the reader convinced and with nothing to do on Monday. The companion forecast in this series found that the published framework literature has the same disease at scale: of eight major architectures for a better economy, exactly one names a funding source for its own programme. A reason to believe without a move attached is a preference wearing an argument’s clothes.
So this report pairs each argument with one move. The structure is deliberate: the argument is why it works, the move is what makes it work, and where the move is unknown, that is stated rather than finessed.
There is one more reason to do it this way, and it comes from the evidence rather than from style. The thing that distinguishes the arrangements in this library that worked from the ones that merely persuaded is almost never the quality of the idea. It is whether somebody had to do something specific, and whether anybody’s interests were engaged in making them keep doing it. Italy’s worker-buyout law works — 257 buyouts, roughly 9,300 jobs, survival at 13.12 years against a national average of 13.5 from a distressed starting point — because it specifies that workers may capitalise three years of advance unemployment benefit against a matched fund built from 3% of the cooperative sector’s annual surplus. The 2019 stakeholder-capitalism statement did nothing, and we know this precisely: of all 128 signatories, no bylaw changed and none conceded that signing required any change. Same sentiment, opposite outcomes, and the difference is a mechanism.
The sixteen arguments divide into three groups, and the division carries information.
Six are mechanisms — reasons the arrangement works that do not depend on anybody choosing to make it work. These are the load-bearing ones, because they operate whether or not the design is good.
Six are design conditions — things that must be set correctly, where the default setting is wrong and the correct setting is known.
Four are missing institutions — things that do not exist and have to be built, where the evidence tells us what is needed but not yet that it works.
That ratio is itself an argument. A forecast resting entirely on institutions that do not exist is a wish. A forecast with six mechanisms underneath it has a floor: even if nobody builds anything, the first six still operate, and the arrangement partially emerges in a worse form than it could have. The question is not whether this happens. It is which version.
One caution carried over from the forecast, because it governs everything below. This arrives as a hiring freeze, not a firing wave — the OECD found no measured employment or wage effect from AI adoption, with adjustment through slowed hiring, and the ILO finds only clerical work highly exposed at 24% of its tasks against 1–4% elsewhere. So the population this serves is not displaced mid-career specialists with a client list. It is entrants who never got the client list, and several of the moves below change substantially once that is taken seriously.
The main points, in brief
The ceiling on specialisation was never cognitive — it was that a narrow specialist could not operate alone. Agents supply the adjacent competencies, so the minimum viable scope of a specialist collapses and depth can exceed anything a firm structure permitted.
Judgement is the residual because Hayek’s objection was never answered. Computation was solved decades ago; the acquisition of local, tacit, fast-changing knowledge was not, and that gap does not close as models improve.
Small owned units already outperform their reputation. Italian cooperatives have a median employment of 153 against 72 for conventional firms, and on separately estimated production functions conventional firms would produce more using the cooperative technology.
The defect that kept them marginal has a named fix. A fully cooperative economy sits at 73% of capitalist output entirely through under-sizing — and if individuals own capital while units rent it, “the cooperative economy becomes isomorphic to the capitalist one.”
Self-governance beats imposed management on measured outcomes. Subjects choosing their own sanctioning system reached 90% of optimal returns; an externally imposed limit with weak inspection increased defection relative to talking alone.
Process ownership is a free performance gain and a pay illusion. The clean German discontinuity shows no wage effect at all, and simultaneously 2–8% higher value added per worker with more capital per worker.
Exit capability is the whole distributional question. Low asset specificity buys ownership; high specificity buys a dashboard — and Mondragón’s cooperative share of employment fell from 86% to 29.5% because members do not admit members.
Duration, not generosity, produces enterprise formation. Kenyan recipients invested because income was guaranteed for twelve years — then lost the entire gain to one shock, which is why the floor needs insurance attached.
Buyer concentration decides whether this is liberation or subcontracting. Labour markets already run a concentration index of 3,157 with a 17% wage penalty across the distribution, and the best-protected market in the OECD covers none of the self-employed.
The entry ladder is the thing being quietly destroyed, because the measurable interface-clean work that packages most easily is also how people have always learned the trade.
How to read the sixteen
Each point has the same two parts. The argument — why this supports the arrangement, with the evidence that carries it and, where relevant, the counter-evidence that limits it. The move — the single most consequential thing to do about it, stated concretely enough to be assigned to someone.
Where an argument absorbs others from the original forty-eight, they appear inside it rather than as separate claims, because several of the discarded thirty-two were not weak arguments but the same argument from a different angle.
Three of the sixteen have a move that nobody currently knows how to make. Those are flagged as open, and they are the honest frontier of this forecast rather than a gap to be covered with confident prose.
Part One — The six mechanisms
These operate whether or not anybody designs for them. They are the reason this is a forecast rather than a proposal.
1. The ceiling on specialisation was never cognitive, and it has just been removed
The argument. The division of labour is the oldest known productivity mechanism in economics, and for two centuries it ran into a constraint that is almost never named: a narrow specialist cannot operate alone. Someone has to supply the adjacent competencies — the administration, the client acquisition, the quality control, the handful of things that are not your specialism but without which your specialism cannot be delivered. The firm existed partly to supply them, and in exchange it set the minimum viable scope at which a person could function economically. That scope is why a brilliant narrow talent had to become a mediocre generalist to survive independently.
Agents supply the adjacent competencies directly. The minimum viable scope collapses, and specialisation can therefore go deeper than any firm structure previously permitted. This is not the same claim as “agents make people more productive,” which is a claim about rate. It is a claim about the shape of a viable economic unit, and it has a second-order effect worth as much as the first: time scales sub-linearly with volume for the operator, so the unit does not hit the ceiling that converts a specialist into a manager of other specialists.
The reallocation effect matters as much as the depth effect. If detail-oriented execution is absorbed, the people best at framing problems stop spending their hours on work that does not require framing. Judgement is scarce and has historically been spent on tasks that did not need it — which means the gain is not only that more gets done but that the scarcest input gets pointed at the thing only it can do. The firm, correspondingly, keeps the functions that cannot be packaged and becomes more concentrated in what it irreducibly is.
The honest limit: depth without a second client is fragility, not strength. A unit narrow enough to be specific and not deep enough to be irreplaceable has traded breadth for exposure.
The move. Define the specialism by the problem it owns, not the task it performs — and verify that the specialism survives the disappearance of the task. A unit defined as “we produce the onboarding content” dies when content production is commoditised. A unit defined as “we own time-to-productivity for new joiners” survives, because the problem persists when the method changes. For a firm, the equivalent move is to scope every unit to an outcome with a named owner rather than to a deliverable, since the first is a relationship and the second is a purchase order.
2. Judgement is the residual, because Hayek’s objection was never answered
The argument. The strongest theoretical support for a person in front of an agent stack is not that the person is better. It is an eighty-year-old argument that nobody has refuted, and it is worth separating from the argument that was refuted.
The computational objection to central allocation is dead. Exploiting the sparsity of a technical-coefficient matrix reduces a ten-million-product plan from roughly 10²¹ operations — about 1.5 million years — to approximately seventeen minutes, and the only implemented and timed model of it scales polynomially at O(n^3.3). Arithmetic was never the binding constraint, and the strong impossibility claim was always refuted by the fact that planned economies ran for decades.
What was never answered is Hayek’s actual objection, which concerned the acquisition of dispersed, local, tacit, fast-changing knowledge rather than the computation performed on it once acquired. The cybersocialists concede it explicitly: their argument “says nothing about the task of gathering the vast amount of data required.” And the catastrophic demonstration is the Great Leap Forward, which killed 17 to 30 million people with aggregate food production at least 205% above survival needs, because quotas set from stale data systematically over-procured the most productive regions — a result the governing model generates even under a utilitarian planner with no regional favouritism. Not malice, not insufficient compute: an information architecture that could not see the present.
Transposed to a firm: a system acts on the world as recorded; an operator acts on the world as encountered. The deal going sideways for reasons nobody logged, the candidate whose hesitation means something, the exception that is signal rather than noise. The operator’s advantage is that they hold coefficients the system does not have — and because the gap is about acquisition rather than processing, it does not close as models improve.
The limit, stated plainly: this predicts operators beat systems where local knowledge is decisive, and predicts nothing elsewhere. In high-volume, low-variance, fully instrumented work, the system should win.
The move. Measure the judgement share of any function before packaging it — the fraction of its value that comes from tacit, local, fast-changing knowledge rather than instrumented throughput. Build units where that share is high; automate where it is low; and stop defending work whose judgement share has already collapsed, which is a losing position held expensively.
3. Small owned units already outperform their reputation
The argument. The textbook picture of worker-owned enterprise — undercapitalised, fragile, unproductive — is simply wrong on the measured evidence, and this matters because it is the nearest existing analogue to a performance-valued unit.
On size, they are larger: in a comparison of 2,000 Italian cooperatives against 150,000 conventional firms, average employment was 284 against 228 and median employment 153 against 72. On survival, French birth and death rates over two decades were 0.16 and 0.11 for cooperatives against 0.12 and 0.11 for all firms with employees. On productivity the result is stronger than parity: estimating separate production functions for each organisational form and applying each to the other’s inputs, conventional firms would on average produce more using the cooperative technology. The predicted underinvestment does not appear — investment was equal in three and faster in four of seven French industry groups, and cooperatives retain more profit than statute requires.
Two further properties transfer directly. They adjust pay rather than employment, confirmed by four independent studies across three countries, which makes them counter-cyclical stabilisers — fewer jobs created in booms, more preserved in slumps. And the wealth evidence on the employee-ownership variant is strong: employee-owners aged 28–34 show 92% higher median household wealth, 33% higher median wages and 53% longer median tenure, with the honest caveat, supplied by the study itself, that this is association and such firms may simply hire better people.
The move. Stop arguing the viability question and start arguing the scaling question, because the evidence has already settled the first and not the second. For anyone building a unit: the performance case is made, so spend the argument on exit capability and client concentration instead, which is where the actual risk is.
4. The defect that kept them marginal has a named fix
The argument. If small owned units perform this well, why are worker-members only about 11.1 million people — roughly 0.38% of world employment? The answer is specific, measured, and almost nobody knows it.
Brzustowski and Caselli’s calibration puts a fully cooperative economy at 73% of capitalist output — a 28% welfare loss — and attributes the entire gap to static inefficiency from the units being too small, since their saving rate is in fact slightly higher. Not productivity. Not capital access in the naive sense. Size. The same mechanism is independently identified as the cause of Yugoslavia’s growth collapse, where labour-managed firms restricted entry, job-finding and separation rates collapsed after 1965, and 10% of the labour force emigrated. Three separate literatures converge on under-sizing as the systematic defect.
And then the sentence this whole report turns on. The authors name the fix and leave it to future work: the static inefficiency exists because the units take their capital stock as given, and if you instead let individuals own capital and the units rent it, “the cooperative economy becomes isomorphic to the capitalist one.”
That is precisely the architecture of a performance-valued unit operating inside or alongside a large firm. The unit owns its method, agent stack, relationships and accountability. It rents the firm’s capital, platform, data, compliance perimeter and distribution. Ownership of the process is separated from ownership of the capital the process runs on — which is the named escape from the defect that has held this organisational form at a third of a percent of world employment for a century.
The limits, because this is the strongest claim in the report and deserves them. It is a result from a calibrated general-equilibrium model, not an observed outcome. It says nothing about agents. And the authors are explicit that their 27% gap is an upper bound, since “market power and concentration, environmental and other types of externalities, asymmetries in bargaining power vis-à-vis workers and labour-market monopsony… are omitted from the model, and so the deck is stacked against the cooperative economy.”
The move. Write the rental agreement as the founding document, not the services contract. It should specify what the unit owns outright — method, tooling, performance record, client relationships — and what it rents, on what terms, with what notice. A unit that cannot name what it owns does not own anything, and compliance-perimeter rental deserves explicit treatment, since it is what lets a one-person unit operate in a regulated activity at all.
5. Firms already do this, partially, and already own the instrumentation
The argument. The transition requires less new machinery than it appears, because two of its preconditions are already in place in most large organisations.
The attribution infrastructure exists. Pipeline systems, applicant tracking, ticketing platforms, advertising consoles and close-management tooling already produce dated, named, countable objects for most of the functions that package cleanly. The firm does not need to invent measurement for tier-one work; it needs to decide to pay against measurements it already collects.
And the precedent exists. Localisation, payroll, benefits administration and performance marketing are already routinely external in large parts of the market — which is the proof that the model works rather than a prediction that it might. The constraint on payroll was never measurability; it was accuracy. Falling transaction costs of contracting, escrow and payment rails have meanwhile lowered the price of the arrangement itself.
The honest counterweight: the existing external market for these functions is mostly agencies and business-process outsourcers, not owner-operated units — which means the precedent demonstrates the contracting, not the ownership. The thing that is new is who holds the method, and that part has no precedent at scale.
The move. Run the instrumentation audit before the reorganisation. For each candidate function, ask whether the firm already possesses a dated, attributable, auditable measure of its output — and treat the absence of one as a reason to defer, not as a reason to invent a proxy. Functions whose measurement has to be built specially are functions whose attribution will be contested forever.
6. Self-governance measurably outperforms imposed management
The argument. This is the most under-used body of evidence in the whole debate, and it is unusually hard.
The field results first. Of 47 irrigation systems studied, 12 were government-managed, of which only 40% achieved high performance; of 25 farmer-managed systems, over 70% did. In Nepal, farmer-governed systems beat government systems on all three performance measures — physical condition, tail-end water availability, agricultural productivity — controlling for environment, confirmed on an expanded database of 229 systems. Across 178 forest user groups, monitoring intensity correlated strongly with condition, robust to controls for formal organisation, resource dependence and social capital. And monitor accountability was the strongest of eleven design principles tested across 91 studies, with a supportive-to-unsupportive ratio of 11.7.
Then the laboratory result, which is the one to remember. Anonymous non-communicating subjects overharvest exactly as theory predicts. Allowing discussion reduces it. And subjects who chose their own sanctioning system after communication achieved 90% of optimal net returns — the best outcome in any common-pool-resource experiment. Meanwhile an externally imposed harvest limit with a one-in-sixteen inspection probability increased withdrawal relative to the communication-only condition.
Imposed measurement with weak enforcement produces worse cooperation than no measurement at all. That finding should be printed at the top of every performance-management redesign, because it inverts the instinct: the problem with imposed targets is not that they are insufficiently strict but that they displace something that was working.
The limit, from the same author, and it cuts against enthusiasts: “we cannot be overly optimistic and presume that dilemmas will always be solved by those involved. Many groups have struggled and failed.” Property regime is not the explanatory variable; fit to local conditions is.
The move. Make the units author their own measures and sanctions, with the firm holding a veto rather than the pen — and make whoever monitors them accountable to them in some concrete way, such as publishing the monitor’s own error rate. A scoring system the scored had no hand in writing will be gamed, and the evidence says it may produce less cooperation than having no scoring system at all.
Part Two — The six design conditions
These have a correct setting and a default setting, and the two are different. Each is a place where the arrangement becomes either a redistribution of agency or a redistribution of risk.
7. Attribution quality is the constitution, not the reporting layer
The argument. If standing derives from attributable contribution, then the measurement regime is the political system, and its quality determines whether the arrangement is a meritocracy or a lottery with good branding.
The founding measurement text states the mechanism — “what we measure affects what we do; and if our measurements are flawed, decisions may be distorted” — and then specifies a discipline that the entire field subsequently abandoned: a dashboard of indicators interpretable as variations in underlying stocks, with physically distinct things kept physically separate, and explicitly no composite index, because combining current performance with sustainability into one number is like a car meter “that added up in one single number the current speed of the vehicle and the remaining level of gasoline.” Within fifteen years the dashboards had been replaced by exactly the composite indices the Commission warned against. A procurement process wants a single score; that is the pressure, and resisting it is the discipline.
Two specific failure modes follow. Attribution gets confused with measurement, constantly — plenty of functions emit abundant numbers that cannot be causally assigned to the unit producing them, and the corporate-power literature supplies the warning at scale: the headline markup series everyone cited implied roughly 70% of GDP as pure economic profit and a capital output elasticity of −0.25, with an independent reconstruction finding no rise at all. Confident numbers, wrong object.
And the unmeasurable gets systematically underpaid, and it is not the low-value work. Strategy, culture, governance, relationships and crisis judgement fail attribution completely. In a system allocating reward by attributable contribution, that work either degrades or gets protected by people powerful enough to exempt themselves — which produces the structural irony that the people with the most power will be those whose contribution the system cannot measure.
The move. Publish the scoring function and make it contestable by the scored, with a standing route to challenge it. The warning comes from the data-dividend modelling: identical data produces near-equal payouts or a Gini of around 80 with a 2.5-to-1 median gender gap, depending entirely on a weighting transform nobody voted on. A scoring function is a distributive institution dressed as a measurement one, and it should be governed like the former. Keep a dashboard, refuse the composite, and maintain an explicit protected list of work that is paid for without being scored.
8. Exit capability is what converts a dashboard into ownership
The argument. This is the distributional question, and it resolves on leverage rather than merit. A unit that can serve another client will be offered ownership. A unit that cannot will be offered a dashboard and a bonus.
The historical record on whether incumbents concede ownership voluntarily is unambiguous and discouraging. Mondragón’s cooperative share of employment fell from 86% in 1991 to 29.5% in 2007 — not because cooperatives failed but because the group grew by acquiring subsidiaries staffed with non-members. The detail is damning: at one plant workers bought in at €13,000 each under what a union delegate called blackmail, almost 80% joined for job security, and only 7% could state the group’s cooperative principles a year later. Basque members openly refused to cooperativise the Chinese and Slovak plants because their own jobs would be at risk. The generalisable mechanism: incumbent members have a rational interest in not admitting new ones, and the people who would approve the dilution are the people diluted.
Corroboration from the corporate side is a census rather than a sample. Of all 128 Business Roundtable signatories to the 2019 stakeholder statement, no bylaw changed, none conceded that signing required any change, and of the seventy with published director-incentive principles, every one aligned directors with shareholders and none with stakeholders. Voluntary stakeholderism is a substitute for structural change, not a weak version of it.
So exit capability cannot be requested. It has to be constructed, and it has four components: standardised interfaces so a second client is cheap to serve; output and data portability, so the unit takes its work product and its performance record; compliance-perimeter rental on terms that survive leaving; and switching-cost symmetry, meaning the firm’s cost of replacing the unit is comparable to the unit’s cost of losing the firm. Where that symmetry is absent, every other term is negotiated under duress.
The move. Copy Legge Marcora. Workers facing a failing or restructuring employer may capitalise up to three years of advance unemployment benefit, matched one-to-one from a fund built on 3% of the cooperative sector’s annual surplus. The record: 257 buyouts, roughly 9,300 jobs, the state vehicle involved in 77.4% of them, survival at 13.12 years against a national firm average of 13.5 from a distressed starting point, and 78 new buyouts against 25 closures since 2009. A state that wants specialists to be able to buy their own process out of a restructuring firm does not need to invent an instrument; it needs to translate this one. For a firm, the equivalent move is to price the buy-out option into the unit’s founding agreement, because an option that has to be negotiated under duress is not an option.
9. A durable floor plus insurance is the precondition for anyone trying
The argument. The mechanism that produces enterprise formation is credibility over time, not generosity — and this is the most counterintuitive finding in the transfer literature.
In the Kenyan long-term trial, recipients invested in new enterprises because the income was guaranteed for twelve years. The duration did the work. The three-year US trial at $1,000 a month found entrepreneurial intention up and entrepreneurial activity flat, with participation down 4.1 points, household earnings down 28 cents per dollar transferred, and the freed hours going to leisure rather than to learning, childcare or exercise. A short, generous transfer buys rest. A long, credible, modest floor buys risk-taking.
And the Kenyan gain was entirely lost to a single shock, which is the other half of the design requirement. A floor that survives a political cycle but not a bad quarter produces nothing.
The universality finding is the encouraging footnote and the open research question. Forty years of a universal dividend at roughly $2,000 per person per year show no employment effect at all and a 1.8-point rise in part-time work, with effects in tradables and none in non-tradables — consistent with universal payments generating the demand that buys back the labour they withdraw. The labour-supply penalty measured so precisely in the individually randomised trial may be an artefact of paying a few people inside a large economy rather than a property of unconditional income.
The move. Legislate the horizon, not the amount — a floor with a statutory duration long enough that someone can plan a business against it, paired with shock insurance so one bad event does not erase the accumulated gain. For a firm, the translatable version is a minimum guaranteed engagement period for new units, because a unit on thirty-day terms cannot invest in its own method and will behave accordingly.
10. Process ownership is a free performance gain — and not a pay substitute
The argument. There is a clean natural experiment on whether giving people formal standing in how work is governed changes anything, and it answers two questions in opposite directions.
Germany abolished mandatory one-third board representation for newly incorporated stock corporations in 1994 while grandfathering it for older cohorts, tested across the universe of German social security records. On pay: nothing. Point estimates of +0.4% to +2.3%, confidence intervals ruling out anything above 4%, no change in the within-firm wage structure, an insignificant labour-share change of −0.03, and identical rent-sharing elasticities of about 0.07 in both groups. Formal governance standing did not move pay at all.
On performance: the same study refutes the forty-year-old objection that worker involvement destroys capital. Codetermined firms hold significantly more fixed assets per worker and produce 2 to 8% more value added per worker, with no revenue, employment or profitability penalty.
Read across to process ownership, the lesson is precise and unwelcome to both camps. Ownership of a process reliably improves how the work is done, and does not by itself change what the owner is paid. Firms offering “ownership” instead of compensation are making an offer that is empty on the pay dimension and real on the performance dimension — which is exactly why they will make it.
The move. Separate the two negotiations explicitly. Ownership of the process is agreed on its own terms as an operating design; compensation is agreed against the market for the unit’s output. A specialist who accepts ownership as compensation has, on this evidence, accepted something worth between nothing and 4% — and the firm has acquired a 2–8% productivity gain for it. Price the exception; assume the default.
11. Buyer concentration decides whether this is liberation or subcontracting
The argument. This is the failure mode that would make the whole arrangement a downgrade, and it is the default rather than a tail risk.
Concentration is already severe. The average US labour-market Herfindahl index is 3,157 — well into the range that would block a merger in a product market — and the instrumented wage-concentration elasticity is −0.127, implying a 17% wage penalty moving from the 25th to the 75th percentile of concentration. This is the part of the corporate-power case that survives scrutiny, unlike the markup series.
Convert workers into suppliers and that concentration becomes procurement power over micro-firms that have none of what the employee had: no unfair-dismissal claim, no bargaining coverage, no notice period, no employer social contribution.
The operational evidence is explicit, from a market projected to employ over 23.5 million workers by 2029–30. Workers experience order distribution, pay and incentives through an application while believing humans decide — naming “supervisor,” “team leader,” “a team in Bangalore” — producing a situation in which the least advantaged party lacks the information and agency to contest decisions determining their livelihood. Verification misfires, including one worker penalised for not wearing a company shirt while wearing it. Drivers are deactivated after customer complaints about accidents they did not cause. Chat support is the most-used feature and fails precisely on penalties, deactivations and payouts. This is the dark mirror of the whole thesis: not people directing agents, but people managed by agents with no appeal.
The move. Three things, in order of tractability. Payment-terms regulation, because micro-firms die of working-capital timing long before they die of pricing. Explicit liability allocation in every unit agreement, so the unit knows which failures it carries. And buyer-concentration screening — applying the monopsony framework already used in merger review to supplier markets, since the analytical tools exist and the legal hook does too. For a unit, the move is simpler and non-negotiable: no client above a fixed share of revenue, set before the first contract rather than after the third.
12. Units can only invest if contracts give them horizons
The argument. This is the least evidenced of the sixteen and included because its absence voids several of the others. A unit that owns its method has to invest in that method — tooling, agent infrastructure, domain depth — and investment requires a horizon over which to amortise it. Thirty-day terms produce units that behave like temporary labour, because that is what they are.
The system-level reason to expect this to be workable rather than utopian comes from the observation that none of it requires a change of economic system. All real economies are hybrids: calling an economy capitalist is shorthand for a hybrid “within which capitalist relations are dominant.” A supplier economy with long horizons, rented capital and portable reputation is a reconfiguration of components that already exist, not a replacement of them — which is why it can emerge incrementally and why the companion reports rank incremental reconfigurations above wholesale alternatives on evidence.
The move. Make the default engagement a multi-period commitment with a performance exit, rather than a short term with a renewal option. The two look similar and behave oppositely: the first lets a unit invest and gives the firm a quality remedy, while the second transfers all planning risk to the party least able to carry it.
Part Three — The four missing institutions
These do not exist. The evidence tells us what is needed and does not yet tell us that it works, so three of the four moves below are marked open — a frontier rather than a plan.
13. A portable, verifiable, contestable record of delivery
The argument. If the unit of value is an owned process with a named owner, then the scarce asset is a record of what that owner has actually delivered — the thing that replaces the employer’s name on a CV as the signal of competence. Today that record lives in private references, platform-specific ratings that do not travel, and self-assertion.
The structural reason this is buildable rather than wishful is the strongest joint result in the framework literature, reached independently by three separate traditions: market outcomes are constituted by law and are therefore available for redesign. Credentials, professional licensure, limited liability and the enforceability of a reference are all constructed, and all currently presuppose an employment relationship. A supplier economy needs the equivalents and does not have them.
Two cautions. A reputation layer distributes however its scoring function says it distributes — the data-dividend modelling produces either near-equal payouts or a Gini around 80 from identical inputs. And it hits the complementarity constraint: institutional practices cluster non-randomly, and importing one component into a system lacking its complements buys the cost without the return. A verifiable reputation layer without portable benefits, supplier bargaining and an insurance floor is not a partial version of a functioning supplier economy. It is a more efficient market in people who have no protection.
The move — open. Nobody has built a delivery record that is simultaneously portable across clients, verifiable by the counterparty, resistant to gaming, and contestable by the person it describes. The four properties are in tension: verification wants a closed system, portability wants an open one. This is the most commercially obvious gap in the whole forecast and the one most likely to be solved badly first — which argues for the governance question being settled before the technical one, since whoever sets the weighting sets the distribution.
14. Supplier-side bargaining, portable benefits and a right of appeal
The argument. Every protective institution we have was built for the employment relationship, and the shift moves people out from under all of them at once.
The evidence that this is not a gap at the margin: Denmark — 67% union density, 82% bargaining coverage against the UK’s 27%, on-site representatives in one firm in two, 1.2% of GDP on active labour market policy against an OECD average of 0.4% — leaves the solo self-employed and zero-hours workers outside the model entirely. One in three Danish workers is now in non-standard employment, marginal part-time has more than doubled since 2000, and 12% of wage earners have bought private unemployment top-up insurance. If the best-protected labour market in the OECD covers none of this population, none currently does. And the European pattern identified in the comparative literature is dualisation — protecting insiders at outsiders’ expense — rather than convergence on any model.
The genuinely hopeful finding sits in the same evidence and deserves weight. During a Delhi quick-commerce dispute, workers switched off their IDs over payout cuts while new hires earned more for identical work; some whose own pay was unchanged joined in solidarity; the company raised payouts and the strike ended. Collective leverage exists among atomised, nominally independent suppliers. The organisational form has to be invented; the power does not.
The move. One part is known and two are open. The known part: a time-bound right of escalation to a human on any decision that terminates income — this is the single most implementable protection in the report, it costs almost nothing, and the gig evidence shows precisely where its absence does damage. The open parts: bargaining rights that attach to a supplier rather than an employee without colliding with competition law, which is a genuine legal problem rather than a drafting one; and benefits that follow the person across clients, which requires a contribution mechanism nobody has settled.
15. The entry ladder has to be rebuilt deliberately
The argument. This is the quiet casualty and the one that compounds.
The work that packages most cleanly — measurable, interface-clean, bounded error — is also, almost exactly, the work through which people have always learned the trade. Lead generation, sourcing, bookkeeping, support, QA, first-pass review: these are the tier-one functions and they are the junior roles. Remove them and you do not primarily remove jobs from the people currently holding them, because adjustment comes through slowed hiring rather than dismissal. You remove the route by which anyone becomes senior.
And the thing that becomes most valuable is precisely the thing the removed work produced. Judgement is the residual because it rests on holding local, tacit, fast-changing coefficients — and those coefficients were acquired by doing the detailed work for several years. The forecast therefore contains a paradox it cannot resolve internally: it makes tacit judgement the scarcest input while dismantling the only known production route for it.
There is a design precedent in the predistribution literature, and it is small but real: procurement conditioned on apprenticeship training. A firm buying from external units can require that the units train, in the same way states have required it of contractors.
The move — open in substance, available in form. The form is procurement conditionality: make training a term of the contract rather than a hope about the market. The substance is unresolved, and it is the subject of the companion report in this series: if the detail work is gone, where do the coefficients come from? Nobody knows, and the answers currently on offer — simulation, apprenticeship to an operator, deliberately retained manual work — are untested.
16. The coordination capability nobody budgets for
The argument. A firm running thirty external performance-valued units has replaced management with procurement, and procurement at that granularity is itself a substantial function — specifying, contracting, measuring, arbitrating and renewing across many small suppliers. This cost is real, it is routinely omitted from the business case, and when it surfaces it gets read as evidence that the model does not work rather than as evidence that it was under-resourced.
Two pieces of evidence bear on it. The commons literature makes monitoring by monitors accountable to the users a load-bearing design requirement rather than an overhead, and the experimental finding is that imposed rules with weak inspection probability produced more defection than communication alone — so under-resourced monitoring is worse than none. And the comparative-institutions result is that institutional quality can substitute for social trust: above roughly German levels of regulatory quality, trust’s marginal effect on growth becomes statistically indistinguishable from zero. That is the hopeful reading — a low-trust organisation is not condemned to fail at this, but it has to build the institution rather than wait for the culture.
The move. Budget the coordination function explicitly, at a named headcount, before the first unit is created — and give it the job of arbitration, not just contracting. Most of what kills multi-supplier arrangements is not price but unresolved disputes about attribution, and nobody owns that.
What the sixteen add up to
Count the structure. Six mechanisms operate whether or not anyone designs for them. Six design conditions have a known correct setting and a wrong default. Four institutions do not exist, and three of those have no known move.
That distribution is the actual finding of this report, and it supports a more specific conclusion than either enthusiasm or scepticism.
The arrangement will emerge. Not because it is desirable but because the mechanisms do not require anyone’s consent: the minimum viable scope of a specialist has collapsed, judgement is the residual for reasons that do not weaken as models improve, the organisational form already outperforms its reputation, the defect that kept it marginal has a named fix, the instrumentation is already installed, and self-governance beats imposed management on measured outcomes. Six independent pushes in the same direction.
The version that emerges by default will be the bad one. On current settings, attribution is poor and composite, exit capability is absent, the floor is short and discretionary, ownership is granted in vocabulary rather than in equity, buyer concentration is already at 3,157 with a 17% wage penalty, and horizons are thirty days. Each of those defaults is a choice nobody made, which is what makes them defaults.
And the gap between the two versions is almost entirely about power, not capability. Of the sixteen arguments, two concern what people and agents can do. The other fourteen concern who measures, who can leave, who carries the downside, and who has standing to contest a number. That is the same ratio the companion forecast found in its parameter set, and it arrived there independently.
The three moves that do the most work, if only three get made:
Write the rental agreement. The unit owns method, agent stack, relationships and accountability; it rents capital, platform, data, compliance perimeter and distribution. This is the one structural change with a general-equilibrium result behind it, and it is the difference between a unit and a cost centre with a dashboard.
Legislate the horizon, not the amount. Duration is what produced enterprise formation in the only long trial we have, and the gain was lost to a single shock, so insurance is part of the instrument rather than an addition to it.
Give a time-bound right of appeal to a human on any decision that terminates income. It is the cheapest protection in the report, the gig evidence shows exactly where its absence does damage, and it is the one thing on this list that could be implemented by a single firm next quarter without anyone’s permission.
What remains unresolved should be stated as such rather than smoothed. Nobody knows how to build a delivery record that is portable, verifiable, gaming-resistant and contestable at once, and the four properties are in genuine tension. Nobody has solved supplier bargaining without colliding with competition law. And nobody has a replacement for the entry ladder — which matters most of all, because the arrangement makes tacit judgement the scarcest thing in the economy while removing the only route by which anyone has ever acquired it.