For the Corporation

Volume VI — Governance and the Commons · Extension II of III Nine movements, one register.

THE PLATE

A woman in linen standing at a window beside a tall vase of dried grasses, looking out into the light. — Plate VI.E2The Item That Waited Four Weeks.A firm counts the cost of every decision it takes. There is no line anywhere in its accounts for the cost of the ones it is still taking.

THE LETTER

You have a governance system already. It is not a preference, a philosophy or a culture: it is a schedule of authorities, a committee calendar, a delegation matrix and a set of thresholds, most of them set once, by somebody who has left, for a reason nobody in the building can now reconstruct. It is running, it is expensive, and it has never been priced.

Volume VI is unusually useful to you for a reason that has nothing to do with cooperatives. Four of its eleven chapters are arithmetic about the exact machinery a firm already operates — what a decision rule costs (VI.01), what an escalation rule costs (VI.03), at what level a function belongs (VI.08), and how to read the record you already keep to find out whether any of it is working (VI.10). Two more are about the thing your firm calls shared services and does not govern at all: the internal commons.

This chapter re-cuts those for a board, a group finance function and an executive committee. Four instruments, one register, and a facility that pays for the audit. It also names what to stop, with figures beside each item, because a governance chapter that only adds is a chapter that will be read once and filed.

There is a result in here that will not go away once you have seen it, and it is about the one number in your governance that has never appeared in a paper: what it costs the firm while a decision is still being taken. Put that on the page beside the harm of deciding wrongly — which is the only harm your current rules price — and the rule that comes out is not the rule you are running. On the worked board it is not close.

— The Editors

DISCOVERY

What is already working, inside firms, at scale

Start with the evidence from organisations that had every option and chose structure deliberately, because the strongest cases here are not cooperatives.

Buurtzorg. Nursing teams capped at twelve, holding their own intake, rostering, hiring and clinical judgement, with no team leader. At roughly ten thousand nurses the coordinating establishment is about one per cent of the front line, against a sector overhead near a quarter of revenue, and the KPMG evaluation found care delivered in about forty per cent fewer hours than comparable providers (VI.03). Not a philosophy. A span, a cap, and an escalation rule.

Morning Star. Roughly four hundred colleagues, no managers at all, and in place of a job description the Colleague Letter of Understanding: each person writes their own commercial mission, the activities they will perform and the metrics they will be judged on, and negotiates it directly with the ten or so people whose work touches theirs (VI.03). The CLOU is the invention worth stealing. It converts who is in charge of this into who did I agree this with, and it does it in a document a third party can read.

Haier. The RenDanHeYi transition removed on the order of ten thousand middle-management roles and replaced them with several thousand microenterprises that contract with each other on negotiated terms, hold their own notional P&L and may raise outside capital (VI.03). Whatever one thinks of the ideology, the mechanism is an internal transfer-pricing regime with decision rights attached, and it survived being run through an acquired American business.

The German cooperative banks. A joint protection scheme in continuous operation since 1934 and through a ninety per cent consolidation of its own members, with no depositor loss (VI.08). The members kept the lending decision, the customer and the board, and assigned exactly one function upward — the risk pool. One function, correctly placed, held for ninety years while everything around it changed shape.

FINRA. Rules written and enforced by the industry for roughly three thousand three hundred broker-dealers, funded by a levy on its own members, with every rule filed with the Securities and Exchange Commission, which may approve it and may abrogate it (VI.11). Self-governance with reserved powers above it, in a market that moves money in microseconds. If a firm wants a template for how to delegate to an internal body without losing control of it, this is the one with the longest record.

And a board that audited itself in public. After its crisis, the Co-operative Group commissioned an independent governance review, published it whole, and the members then voted to change the rules (VI.10). A published finding that the structure was not functioning as written, followed by a vote to change the structure. That sequence is rare and it is the one this chapter is trying to make ordinary.

Six organisations, one pattern: each named a function, priced it, and placed it once — rather than deciding, in the abstract, how centralised to be. That decision does not exist. There is only a list of functions, and every function has its own answer.

THE ARITHMETIC

The number your rules have never carried

First, re-derive the board, so the change is visible.

VI.01's worked board: nine directors, each accepting a given proposal with probability 0.60, three hours of every member's time per round at a loaded £120 an hour — £3,240 a round — twelve contested decisions a year, and a loss of £30,000 borne by a member left outside the winning coalition.

   k    P(X>=k)   rounds   decision £/yr   external £/yr     total £/yr
   4     0.9006     1.11          43,169         200,000        243,169
   5     0.7334     1.36          53,011         160,000        213,011
   6     0.4826     2.07          80,562         120,000        200,562
   7     0.2318     4.31         167,740          80,000        247,740
   8     0.0705    14.18         551,146          40,000        591,146
   9     0.0101    99.23       3,858,025               0      3,858,025

The minimum sits at k = 6 of nine — 66.7 per cent — at £200,562 a year. Two-thirds is not a tradition in that table; it is the bottom of a sum.

Now the cut, and it is one term.

Every entry in that table prices two things: the time it takes to assemble agreement, and the harm done to whoever the decision goes against. Neither of them is the harm of not deciding. A firm that spends four rounds reaching a pricing decision has not merely spent four rounds of directors' time — it has spent four rounds of not having the price, and that is a real quantity with a real owner.

Call it D: the value the firm loses per round while a decision is still open. Add it to the cost of a round and re-minimise:

  D = £0 / round          k* = 6   (66.7 %)    total   £200,562 / yr
  D = £5,000 / round      k* = 5   (55.6 %)    total   £294,818 / yr
  D = £20,000 / round     k* = 4   (44.4 %)    total   £509,644 / yr
  D = £50,000 / round     k* = 3   (33.3 %)    total   £895,285 / yr

Solve for the exact point at which two-thirds stops being optimal. A move from six to five buys one seat of external cost — £40,000 a year — and costs 0.7086 extra rounds per decision. They balance at:

  D  =  £1,464 per round   =   12.2 hours of loaded time

Twelve hours. If leaving a decision open for one more round costs the firm more than about twelve hours of one person's loaded time — and for anything touching a price, a hire, a customer commitment or a competitor's move, it costs far more than that — the same board, on the same stake, with the same people, should not be voting two-thirds. At a delay cost of £20,000 a round, the optimum is a bare majority of four, and holding two-thirds instead costs £188,214 a year on twelve decisions.

May's theorem (1952) is the other half of this and it has been sitting in plain sight since before most delegation matrices were written: simple majority is the only rule that treats the proposal and the status quo alike. Every other rule is a deliberate thumb on one side of the scale. A supermajority is a subsidy to the status quo, it is priced in rounds, and no firm has ever put it on the P&L. That is not an argument against supermajorities. It is an argument for knowing what one costs before granting it — because on the worked board, the supermajority everybody assumes is prudence is, at any realistic delay cost, the most expensive rule available.

Second, the escalation rule, recomputed at your size.

VI.03's finding is the one to carry into any reorganisation: a nested structure and a hierarchy are the same tree, at the same span and the same depth, differing in one rule — whether a decision rises when it meets authority or only when it cannot be contained. VI.03 computed the break-even containment rate for ten thousand people and got 37.8 per cent. Recompute it for a firm of 600:

  teams of 12                                          50
  span 5, depth = ceil(log_5 50) + 1                    4 levels
  L at c = 0.55 (authority rule)                   1.7436 levels
  L at c = 0.85 (containment rule)                 1.1759 levels
  ------------------------------------------------------------
                                HIERARCHY          NESTED
  coordination posts                   63               9
  posts                        £4,410,000        £630,000
  inside-team coordination       £297,000      £1,783,200
  escalation                   £2,891,214        £683,802
  ------------------------------------------------------------
  total                        £7,598,214      £3,097,002
  saving                       £4,501,212   =   £7,502 per person
  break-even containment  c*        35.4 %

35.4 per cent — lower than the ten-thousand-person firm's 37.8. A six-hundred-person company has to close only about a third of its decisions inside the unit to beat the hierarchy it is currently running, against a modelled 0.85, a margin of 49.6 points. Nesting is not a large-organisation technology; VI.03's own finding is that most of what it can do per decision, it has done by about fifteen hundred people.

And price the emergency honestly, because it is the clause every firm skips. At containment zero — which is what an emergency is, an event that spans units — a decision climbs the whole tree by consent: 4.0 levels, 20.0 working days against 5.0 with a pre-named commander, a factor of 4.0. That is what the exception clause buys, and it is why the clause is written on day one with an automatic expiry rather than after the first crisis without one.

Third, where each function belongs.

VI.08's rule needs two estimates per function and no money at all: s, the share of the benefit that lands outside the deciding unit, and h, how unlike the units' own answers would be. They balance at s* = h / (1 + h).

One of those estimates can be derived rather than guessed. For anything invested in a person, s is the share of the benefit that leaves with them. At 13.0 per cent voluntary attrition over a five-year payback, the expected years retained are 3.3567, so s = 0.3287:

  function                         s       h      s*     down      up   verdict
  procurement of common inputs   0.45    0.15   0.130   0.669   0.022   UP
  safety standards               0.60    0.10   0.091   1.000   0.010   UP
  the brand                      0.70    0.25   0.200   1.000   0.062   UP
  transferable training          0.33    0.20   0.167   0.240   0.040   UP
  hiring into a team             0.10    0.40   0.286   0.012   0.160   local
  local working practice         0.05    0.45   0.310   0.003   0.203   local

Two of the "down" losses are shown at 1.000 because the formula's shortfall exceeds the efficient level entirely and the loss is capped; read those rows as decisively up rather than as a computed magnitude. Procurement is the row worth arguing in a board paper: the loss of holding it locally is 29.75 times the loss of holding it centrally, which is a sentence, not a sentiment.

Fourth, the internal commons nobody chartered.

Your platform team, your shared data estate, your internal tooling group and your central capital pool are commons in the strict sense — a shared resource, many appropriators, and a provision side funded by somebody else. VI.02's arithmetic says when a shared thing needs a written rulebook rather than a handful of bilateral understandings, and it is a count. Price the firm's version: a bilateral service agreement between two units costs about twenty hours a year of two managers' time — £4,800 — and a written charter with a chargeback formula, an escalation route and a published service level costs about £96,000 a year to run:

  n = 6     15 pairs      £72,000 / yr   — bilateral still wins
  n = 7     21 pairs     £100,800 / yr   — the charter wins
  crossover, exactly                       6.84 units

Seven units. VI.02's consortium crossover, on an entirely different price list, is eight. Below seven consuming units a platform team should be running service agreements and no rulebook; above it, the rulebook is the cheaper instrument and the argument for it is arithmetic rather than a plea for governance.

And the fault to look for first is VI.02's principle 2B, congruence between appropriation and provision. In most firms a platform team's appropriation is free at the point of use and its provision is a single central budget line — which is congruence of exactly zero, the same structural condition that makes a knowledge commons thin on the provision side while every other indicator reads healthy.

Fifth, the room.

VI.10's effective number — one over the sum of squared decision weights — is the cheapest governance measurement in this volume, and it can be run on an executive committee in ten minutes. Weight an eleven-person committee the way it actually decides and:

  people at the table                                11
  sum of squared shares                          0.2109
  N_eff                                            4.74 voices
  as a share of the headcount                      43.1 %

Eleven people, four and three-quarter voices. Nobody has to be accused of anything; the number comes out of the weights the committee itself would recognise.

Now the honest negative, and it is aimed at everything above.

Every threshold in this chapter is downstream of one column, and that column is the easiest number in the building to inflate. Hold the board's parameters fixed and move only the loss-if-wrong:

  loss if wrong £5,000        k* = 4   (44.4 %)
  loss if wrong £10,000       k* = 5   (55.6 %)
  loss if wrong £30,000       k* = 6   (66.7 %)
  loss if wrong £100,000      k* = 7   (77.8 %)
  loss if wrong £300,000      k* = 8   (88.9 %)

A tenfold inflation of a single cell moves the rule two seats — and the person most motivated to inflate it is the person who wants the decision reserved. The same applies to s and h, to the divergence rate, and to D itself, which an advocate of speed will overstate exactly as reliably.

And the firm cannot currently check any of it, because it does not keep the record. A six-hundred-person firm makes on the order of 14,400 coordination decisions a year. A year of board and committee minutes records perhaps 120 of them — 0.83 per cent. Every number above is computed from a register that does not yet exist.

Three exits, and the first is nearly free.

Build the register and the precision arrives inside a year. VI.10's sampling arithmetic says a proportion to ±5 points needs 787.50 decisions once the clustering inside meetings is allowed for. A register capturing 1,000 decisions a year reaches that in 0.79 years. Unlike the small group in VI.E1, which needs the better part of a decade, a firm is large enough to measure its own governance within one financial year — it simply has never written the rows down.

Fill the loss column from outside the room. The person who will not be in the decision estimates the loss; the person who will, states the delay. Two names on two cells, and both are challengeable.

Publish the sensitivity beside the answer. Never a single k*. The rule, and the band of losses over which it is stable. A threshold whose neighbourhood is printed cannot be quietly moved by a spreadsheet.

DREAM

What becomes ordinary

In the firm that has done this work, nobody argues about whether something needs to go to the committee. They look it up, and the lookup takes eleven seconds.

There is a register. Every recurring decision is a row: the class, the rule, the body, the delegated limit, the two parameters that placed it there, the date they were last estimated, and the name of the person who estimated them. It sits where the delegation matrix used to sit, and it is the most-read internal document in the company because it answers the question everybody actually has.

Every proposal template has two fields at the top that did not used to be there: what this costs if it is wrong, and what it costs the firm per week while it is open. The second field is the one that changed the culture, and it changed it quietly. A committee looking at a paper that says the delay is worth forty thousand pounds a week does not ask for another round of analysis in order to look rigorous.

Thresholds move, in both directions, at a scheduled annual review, and the review is a computation rather than a negotiation. Two or three classes drift across a line every year. Nobody takes it personally, because the rule was never about trust.

The platform teams have charters. Each names its resource, its member classes, its contribution formula as an equation, who monitors what and to whom they answer, a published ladder, a conflict route with a service level, and — the clause everybody skips — what the centre may and may not do to it, on what notice, with what written reason. The teams inside them notice that their autonomy went up when the charter was written, which surprises everybody the first time and nobody the second.

And once a year the governance numbers go to the audit committee on two pages, computed from the register by somebody the register does not report to, with the boundary statement printed underneath: what was looked at, what was not, and how far the sample carries.

DESIGN

What to start, and — as plainly — what to stop

Start: five moves, in this order.

One, the decision register. One row per recurring decision, extracted from twelve months of minutes and purchase orders. This is an afternoon of work, not a system, and everything else is downstream of it.

Two, two columns on every proposal: loss-if-wrong and delay cost. The second is new and it is the one that pays. Estimate D in the crudest available way — margin at risk per week, or the cost of the next-best week of the same team — and print it.

Three, price each class and let the rule fall out. Three or four bands by loss-if-wrong, each with its own k* computed at your own n, p, hourly cost and delay. Most firms find the same two things: their highest band is under-protected, and their lowest band is carrying a rule that costs more than the decisions are worth.

Four, the assignment register. s and h for every function, each to one decimal place, each by a named person in ten minutes. Move the clear rows. Mark the near-misses contested and leave them exactly where they are — a firm that moves its close calls spends all its political capital where the answer barely matters.

Five, charter the internal commons above seven units. Below seven, service agreements. Above it, the twelve covenants, with the chargeback formula written as an equation and the centre's reserved powers written as four: approval, abrogation, substitution with a sunset, and standing for whoever the boundary excludes (VI.11).

Stop: seven things, each with its price.

Stop running one decision rule for the whole company. On the worked board, a two-thirds rule held against a £20,000 delay cost burns £188,214 a year. The rule belongs to the decision, not to the body.

Stop approving what you have already delegated. An authority limit that is routinely re-confirmed above the line is not a limit; it is a queue. Set it at the eighty-fifth percentile of the unit's own historical decisions and then let it stand.

Stop calling a platform team a service and charging it like a tax. Free appropriation against centrally funded provision is congruence of zero, and it is the most reliable way to hollow out a shared capability while every dashboard still reads green.

Stop reporting attendance as participation. Eleven at the table and 4.74 voices is a fact about weights, and it is available from the minutes today.

Stop the emergency reorganisation that has no expiry. Substitution without an automatic sunset is not an emergency power, it is a transfer — and VI.03's own record is that a structure which takes command in a crisis and does not hand it back has become a hierarchy that once had a good reason.

Stop renegotiating governance targets annually. A target renegotiated every year is a target nobody has to meet (VI.10). Reset on evidence, on a three-year term.

And stop granting rights you have not funded anybody to exercise. VI.07's finding is the sharpest transferable result in this volume: of ten enacted rights-of-nature instruments, eight named a guardian with no money, and the one with an endowment has outlived the unfunded ones by a factor of thirty. Inside a firm the same rule governs the ombudsman, the safety escalation, the ethics line and the works council. Endowment times draw rate over annual cost of the office, greater than or equal to one, or it is a declaration rather than a right.

DESTINY

How it holds when nobody is pushing

Three things keep a priced governance system honest, and each is cheap.

The register is reviewed on a date, not on a trigger. Anything reviewed by exception is reviewed when somebody is angry, and an angry review moves thresholds for reasons the arithmetic cannot see.

Every parameter carries a name and a date. An estimate with nobody's name on it cannot be challenged, so it hardens into a fact. An estimate with a name on it gets corrected by the person who knows better, which is the only mechanism that keeps a register true.

The measurement is computed by someone who is not measured by it. This is the feature most often dropped when the practice moves from a statistical office into a company, and it is the only one that cannot be replaced by good intentions.

Now the failures, plainly.

The delay cost becomes a rhetorical device. D is as inflatable as loss-if-wrong and it is inflatable by the faster-moving party. The counter is symmetry: both cells are filled by people who are not the proposer, and both are printed with their sensitivity band.

The coach re-acquires authority. A nested structure decays into a hierarchy one helpful intervention at a time. The counter is structural — the integrative role holds no budget, signs nothing, and is evaluated on the containment rate of the units it serves, which falls every time it decides something.

Teams drift past the cap. The mesh cost is quadratic and invisible; VI.03's own numbers put the cap around twelve and the point at which a managed hub becomes cheaper at about nineteen. The counter is a pre-written split rule and a standing report of team sizes.

And the register is summed into a score. VI.10's warning, from Baggio and colleagues: no single design principle is necessary or sufficient, so a total adds terms that do not add, and a score invites a target. Six numbers with their history and a boundary statement. Never one index.

DELIGHT

What it feels like

The pleasure is specific and every executive who has felt it describes it the same way: the paper that does not need to be written. There is a particular expenditure of spirit that goes into building a case for something you already know is right, for a committee that will approve it, and getting that expenditure back is not subtle. You get the week, and you get the part of yourself that was composing the argument on the train.

Then a quieter one, on the other side of the table. A director who has voted against something and lost, under a rule that was computed before anybody knew the question, does not carry it. The most common source of boardroom bitterness is not losing a vote; it is discovering afterwards that the rule was chosen, quietly, by people who could see which way it would fall. A rule set before the question is a rule you can lose to without injury.

And the best hour of the year is the one where somebody reads the register out loud and two rows are obviously wrong — a threshold that has been costing a fortune, a function sitting two levels from where it belongs — and moving them takes ten minutes because the argument is about two estimates rather than about who is to be trusted.

OPERATIONALIZE THIS

At the level of finance

The instrument: a Decision Charter, priced on both sides, financed by a governance-linked facility.

One document and one facility. The document consolidates four things a firm already half-has — a schedule of reserved matters, an assignment register, a containment covenant and an internal-commons charter. The facility is what makes the first audit free.

The mechanics.

Three classes in the articles or the shareholders' agreement. Class A, ordinary operating decisions at a simple majority of the operating team. Class B, above the computed threshold, board majority plus a stated consent. Class C, constitutional, at a special resolution — in England and Wales, seventy-five per cent under the Companies Act 2006 — which hands you a statutory Class C threshold you do not have to invent.
Each class's rule computed on both costs. k* minimises the sum of decision cost, delay cost and external cost. Print D beside the loss. A rule computed on one side only is the rule this chapter exists to retire.
The veto test, per reserved matter. A consent right earns its place only where the external cost it prevents exceeds the expected cost of the delay it creates. Run it for every item and delete the ones that fail: a schedule of reserved matters that has never had an item deleted has never been priced.
The assignment register, s and h per function, reviewed annually, near-misses marked and left alone.
The containment covenant. The Authority Schedule sets each unit's limit at the eighty-fifth percentile of its own decisions; the unit reports its measured containment rate quarterly against its own c* — 35.4 per cent on the worked firm.
The internal-commons charter, above 7 consuming units, with the four reserved powers and a contribution formula that makes appropriation proportional to provision.

The balance-sheet treatment, and it is the part most teams miss. Under IFRS 10 control turns on power over the relevant activities of an investee, and a minority holder with substantive participating rights over those activities can prevent the majority holder from consolidating, while merely protective rights do not. The drafting of your reserved-matters schedule therefore decides whether a subsidiary or joint venture appears line by line or as one equity-accounted investment — a governance decision with a direct effect on reported gearing and covenant headroom. Bring the auditors into the drafting rather than the review; it is a conversation about the substance of rights, which they have every year.

The facility. VI.10's covenant, pointed at decision data. Three targets — measured containment, non-implementation rate and rule–practice divergence — computed from the register by a verifier the lender engages and pays, with one published holdout excluded from the ratchet in writing so both sides can see the gaming component.

  facility                                £50,000,000
  margin step                                   25 bp
  annual margin saving                       £125,000 / yr
  register audit and verification             £60,000 / yr
  ------------------------------------------------------------
  cover ratio                                    2.08 x
  break-even facility at 25 bp            £24,000,000
  break-even step at £50m                        12.0 bp

Above a facility of £24,000,000 the audit is paid for by the margin. Below it, run the audit anyway and drop the ratchet — the register earns £188,214 a year on one rule and £4,501,212 on one structure, which is not a close decision at a cost of £60,000.

The counterparty. Internal first: group finance to one business unit, one schedule, two quarters. The external version is available once you hold two completed cycles and a measured series, at which point you are presenting evidence rather than intent.

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

    decision cost saved  +  delay cost saved
   ------------------------------------------   >   1
    divergence rate x decisions x loss if wrong

Above one, delegate and do not reserve. Below one, reserve it and accept the delay — having priced the delay, which is the term the same inequality in VI.01 did not carry and the reason this chapter exists.

The first ninety days.

Day	Action	Artifact
1–15	Extract twelve months of real decisions from minutes and orders	The decision register
16–30	Add two columns: loss-if-wrong, and delay cost per week	The two-sided register
31–45	Band by loss; compute `k*` on both costs; print the sensitivity	The priced schedule
46–60	Estimate `s` and `h` per function; move only the clear rows	The assignment register
61–75	Count consuming units; charter the commons or write SLAs	Charter or agreements
76–90	Take the covenant to one lender; name the holdout in writing	Term sheet and holdout

APPRECIATIVE QUESTIONS

Twelve, for a room

Discovery — what is already working

Name a decision this firm made in the last two years that everybody, including the people it went against, still thinks was decided well. What was the rule, and who was in the room?
Which unit here already behaves like a whole business — knows its own numbers, holds its own edges? What was it given to get there?
Where do we already decide quickly and well with nobody checking a policy, and what is it about those decisions that makes that safe?

Dream — what becomes possible

If every paper arrived with the cost of waiting printed next to the cost of being wrong, what would our meetings be about instead?
Imagine the register a year from now, with two rows in different places than they are today. Which two, and what changed about the week?
If our most junior manager could read one page and know exactly what they are entitled to decide, what would they do first?

Design — what we build

What is the single decision here with the largest loss if it is wrong, and what rule is it currently made under?
What is the best proxy we already collect for how much of a benefit leaves the unit — something somebody could read this week?
How many units consume our largest shared service, and does that number put us above or below seven?

Destiny — how it holds

Who fills in the loss column, who fills in the delay column, and what would make us confident either number was honest?
What would be the first sign that an emergency structure had quietly become permanent, and who in this room would notice it first?
What would have to be true for this register to still be in use, and still accurate, when everyone here has moved on?

WORKS CITED

Baggio, J. A., Barnett, A. J., Perez-Ibarra, I., Brady, U., Ratajczyk, E., Rollins, N., Rubiños, C., Shin, H. C., Yu, D. J., Aggarwal, R., Anderies, J. M. and Janssen, M. A. (2016). "Explaining Success and Failure in the Commons: The Configural Nature of Ostrom's Institutional Design Principles." International Journal of the Commons, 10(2), 417–439.

Buchanan, J. M. and Tullock, G. (1962). The Calculus of Consent: Logical Foundations of Constitutional Democracy. University of Michigan Press.

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

Hamel, G. (2011). "First, Let's Fire All the Managers." Harvard Business Review, December.

Hansmann, H. (1996). The Ownership of Enterprise. Harvard University Press.

International Accounting Standards Board (2011). IFRS 10 Consolidated Financial Statements. IFRS Foundation.

Jensen, M. C. and Meckling, W. H. (1976). "Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure." Journal of Financial Economics, 3(4), 305–360.

KPMG (2015). The Best of Both Worlds: Buurtzorg. KPMG International.

Laakso, M. and Taagepera, R. (1979). "'Effective' Number of Parties: A Measure with Application to West Europe." Comparative Political Studies, 12(1), 3–27.

Loan Market Association (2023). Sustainability-Linked Loan Principles. LMA, APLMA and LSTA.

May, K. O. (1952). "A Set of Independent Necessary and Sufficient Conditions for Simple Majority Decision." Econometrica, 20(4), 680–684.

Myners, P. (2014). The Co-operative Group: Report of the Independent Governance Review. The Co-operative Group.

Oates, W. E. (1972). Fiscal Federalism. Harcourt Brace Jovanovich.

Olson, M. (1969). "The Principle of 'Fiscal Equivalence': The Division of Responsibilities among Different Levels of Government." American Economic Review, 59(2), 479–487.

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

Simon, H. A. (1962). "The Architecture of Complexity." Proceedings of the American Philosophical Society, 106(6), 467–482.

United Kingdom. Companies Act 2006, section 283.

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

Note on figures. Every figure in this chapter is computed in lib/verify/VI_E2.py and reproducible with python3 lib/verify.py VI.E2, which prints its inputs and its intermediate terms before its results. VI.01's board is re-derived rather than quoted, so that a reader can see the unchanged case land where VI.01 left it before the delay term is added. Scenario inputs — the loaded hourly rate, the acceptance probability, the delay cost, the attrition rate, the spillover and heterogeneity estimates, the facility and the audit fee — are labelled ASSUMED in that module. Two rows of the assignment table show a capped loss of 1.000 and are read as decisively up rather than as a magnitude. The curvature term rho is set to 1 as in VI.08, and every threshold moves if it moves.