Haute Lumière

Commerce · VII.04 · MMXXVI · daylight

La Bourse  /  Volume VII  /  Nº VII.04  /  Workbook — the executive

A woman on a veranda holding a cup of tea, a book open beside her, a lotus garden and pavilion beyond.
Plate VII.04 · Workbook — the executiveThe Certificate and the Box.Two great commons, one atmosphere and one ocean, and only one of them has ever been governed successfully. The difference was never the size of the thing. It was whether the rule could be tested on an object that had to pass through somebody's hands.

WORKBOOK — THE CORPORATE EXECUTIVE

Chapter VII.04 · The Ocean and Atmospheric Commons

For the person with a P&L, a supply chain, a board and a signature authority. You do not govern a commons. You buy from one, or you are exposed to one, and this chapter is the arithmetic that tells you which of your suppliers' assurances are worth what they cost you.

THE PREMISE, STATED COMMERCIALLY

Every assurance you buy about conduct in your supply chain is a claim about a probability. Somebody is telling you that a supplier, a vessel, a mill or a processor is complying with a rule. That claim is worth exactly p · F to the party doing the complying — the chance they are caught and sanctioned, times what it costs them.

You are almost certainly paying for audits where that product is close to zero. Which means you are paying for a document and receiving a probability you have never computed.

This workbook is not about the ocean. It is about the fact that the ocean has the cleanest available worked example of a compliance regime whose economics can be computed all the way through, and once you have done it there you can do it on any tier-two supplier you have.

Three things follow directly into your numbers.

One. Your assurance spend is misallocated by construction. Audit and certification budgets are sized by risk perception and by what vendors sell. They are almost never sized by G — the gain to the supplier from non-compliance. If G is $180,000 and your audit raises the supplier's expected cost by $2,500, you have bought a document.

Two. Coverage bought before consequence is bought is wasted. The chapter's central result is that at the penalties actually in force in high-seas fisheries, the required detection probability exceeds one. The same structure appears in supplier contracts where the only sanction is "corrective action plan".

Three. Collateral beats judgment, and it is cheap. A performance bond costs one to three per cent of face. That is a known, budgetable, contractible number that changes the supplier's arithmetic by a factor you can compute in advance.

PART ONE — DISCOVERY

Days 1–30

Exercise 1.1 — The assurance inventory (one week, with procurement)

List every assurance instrument you currently buy: certifications, third-party audits, self-declarations, chain-of-custody schemes, attestations, supplier code sign-offs. For each, four columns:

InstrumentAnnual cost to usWhat it would cost the supplier to fail itWhose money is at risk if it is wrong?

The fourth column is the one that matters, and in most organisations it is blank for most rows. An assurance where nobody's money is at risk if the assurance is wrong is a document, not a control. Name the rows where the answer is "nobody" before doing anything else.

Exercise 1.2 — Find where it already works (half a day)

Now the appreciative half, and do not skip it. Somewhere in your supply chain there is a control that genuinely works — a test at goods-in, a payment gate, a system field that cannot be bypassed, a licence whose loss would end a relationship. Find three.

For each, write what makes it hold. You will find the same pattern the chapter finds in every case that ever worked: a physical or transactional object had to pass through a named place, and a binary test was run on it there.

Those three are your model. Everything in Part Three is an attempt to make the other controls look more like them.

Exercise 1.3 — The board conversation, framed appreciatively (one session)

Do not open with exposure. Open with this: "We already run three controls that genuinely work, and here is precisely what makes them work. I would like to make four more of our controls look like those three, and I can show you the arithmetic."

You are asking for a design change, not a budget increase. In most cases the budget goes down.

PART TWO — THE ARITHMETIC

Days 31–45

Exercise 2.1 — Compute G for your three largest exposures (one week)

For each of your three most material compliance exposures, compute what non-compliance is actually worth to the counterparty.

  G  =  (gain from the non-compliant act)
        −  (marginal cost of performing it)

Be concrete. Substituting a cheaper input. Running a line past its qualified hours. Under-declaring a volume. Sourcing outside an approved list. For each, the number is usually obtainable from your own cost model, because you know what the compliant version costs you.

This is the single most useful figure nobody in your organisation has. It is also the figure that tells you what a control has to be worth in order to function.

Exercise 2.2 — Compute p honestly (2 days)

p is a joint probability and the second half is where it dies:

  p  =  P(detected)  ×  P(a consequence actually follows)

For each exposure: what fraction of activity does your assurance actually observe? And in the last three years, how many detected non-conformances resulted in a financial consequence to the supplier, as opposed to a plan?

Most organisations, doing this for the first time, find P(consequence) between zero and 0.2. If it is zero, p is zero, and every pound of coverage you buy is multiplied by nothing.

Exercise 2.3 — Run the frontier, and take the result (half a day)

  F*  =  G / p         the consequence required at today's coverage
  p*  =  G / F         the coverage required at today's consequence

If p exceeds 1, no amount of auditing can work and you should stop buying more of it this year. If F exceeds what you could ever contractually impose, the control needs a different structure — which is Part Three.

Worked, from the chapter, so the shape is visible:

  G = $180,000 · p = 0.025 · F = $100,000
  expected consequence   $2,500        the violation pays 72 ×
  F* = $7,200,000        4.8 × the counterparty's annual revenue
  p* = 1.80              impossible

Exercise 2.4 — The sensitivity that protects you in the room (2 hours)

Before you present any of this, vary every input across its plausible range and find which conclusions survive. In the chapter's case, the specific 72× does not survive — it ranges from 9x to 144x — but the conclusion that consequence and not coverage is binding survives every cell.

Present the surviving conclusion and the range, never the point estimate alone. A finance director who finds one number soft will not check the second one, and they are right to work that way.

PART THREE — DESIGN

Days 46–60: the instrument

Exercise 3.1 — Convert one judgment into collateral (one week, with legal and treasury)

Take the exposure with the worst ratio and restructure its consequence as collateral rather than as a claim.

The options, in ascending order of friction:

Size the face on the frontier, not by negotiation:

  face  =  G / p        at the coverage you will actually have

The chapter's worked case: at 100 per cent electronic monitoring and 50 per cent follow-through, p = 0.50 and the face is $360,000 — rounded to $400,000, which is 26.7 per cent of the counterparty's annual revenue. Premium at two per cent: $8,000 a year. Total cost to the counterparty, with monitoring: $17,400, or 1.16 per cent of gross revenue.

Exercise 3.2 — The balance-sheet conversation, early (one meeting)

Have this with your controller and your auditors before you draft anything.

The sentence for the board paper: the P&L moves by the premium while the expected consequence moves by the frontier ratio. In the chapter's case $17,400 against an eighty-fold change.

Exercise 3.3 — Move the test to the chokepoint (one week)

Map where product physically or transactionally enters your control: goods-in, customs clearance, the first payment, the system that raises the purchase order.

Then make the assurance a required field at that point rather than a periodic review. The chapter's ratio is the argument: 3,620 moving vessels against roughly twenty fixed ports, a factor of 181, and the ports do not move.

A control that runs where the thing must pass anyway costs almost nothing to operate and cannot be skipped by a busy person.

Exercise 3.4 — Run the Montreal checklist on your own transition (half a day)

Whatever material transition your firm is committed to — a substitution, a phase-out, a sourcing shift — score it against the four conditions.

  1. Does a substitute exist, and do we or our suppliers own it?
  2. How many points must change? Count them.
  3. Is the success signal visible inside the tenure of the people who must act?
  4. Is there a chokepoint where a binary test can be run?

Where a condition fails, the design job is to manufacture it. Condition 3 fails most often, and the fix is a leading indicator published monthly. That is a reporting change, not a strategy change, and it is usually the difference between a programme that survives a management change and one that does not.

PART FOUR — DESTINY AND DELIGHT

Days 61–90

Exercise 4.1 — Into the standing pack (one conversation)

One line in the monthly pack: the expected-consequence ratio for your three material exposures. p · F / G, which must exceed 1.00. Anything reviewed monthly persists.

Exercise 4.2 — Give the underwriter the data (one cycle)

Once a bond exists, the underwriter wants evidence and will price against it. Give them the monitoring data directly. Two things follow: your premium falls as the record accumulates, and you acquire a second reader of your own assurance data who loses money if they read it wrong. That is a free, permanent, commercially motivated auditor, and you did not hire them.

Exercise 4.3 — Publish one ratio (this quarter)

Publish, externally, one number about your own conduct that nobody required you to publish. Not a target — a measurement, with its denominator. The chapter's Destiny movement is blunt about why: visibility is the mechanism that corrects without anybody pushing, and the first organisation in a sector to publish a particular ratio sets the shape of it for everyone who follows.

Exercise 4.4 — Name the crowding risk, in writing (1 hour)

Before you roll any of this out, write half a page on where you currently get compliance for free because people believe in the thing. Then say specifically how the new instrument avoids converting that into a price. Aim the bond at the tier where the calculation is actually being made — which is a small, known set — and not at the whole supply base.

This is not a caveat. It is a design requirement, and the firms that have got this wrong lost something they had not costed.

Exercise 4.5 — Delight, for a firm (ongoing)

There is a real and underrated pleasure in a control that nobody experiences as suspicion. The field is filled, the load clears, the payment runs. Nobody was accused of anything. Your honest suppliers — who are most of them — have stopped competing against people who are not, and they will tell you so.

THE FAILURE MODES, NAMED

It fails when coverage is bought before consequence is fixed. The most likely failure, because coverage is procurable and consequences are negotiated.

It fails when the bond face is set by negotiation with the supplier. You end up with a licence fee. Set it from the frontier or do not set it.

It fails when the instrument is applied uniformly. A bond is regressive: it is cheapest for the best-capitalised supplier. Applied flat, it consolidates your supply base toward large incumbents, which may be exactly what you did not want. Tier it, or fund it for smaller suppliers as the Multilateral Fund did — more than $4 billion, which is the least-discussed reason Montreal has universal ratification.

It fails when nobody publishes the ratio, and the whole thing quietly reverts to a document.

BOARD PAPER TEMPLATE

Subject: Restructuring assurance on [exposure] from audit to collateral

What we buy today. £X a year of audit and certification covering Y per cent of activity. Detected non-conformances in the last three years: N. Financial consequences imposed: M.

What non-compliance is worth to the counterparty. G = £Z, computed from our own cost model. Range £Z₁–£Z₂.

What our current control is worth to them. p · F = £W. Ratio to G: [number]. This is the control's actual strength, and it is the figure this paper exists to change.

The frontier. At today's coverage, the consequence required is £F. At today's consequence, the coverage required is p. [If p > 1: no achievable level of audit can make this control work.*]

Proposal. Replace [audit scope] with [retention / bond] of face £[from the frontier], premium [1–3] per cent, triggered on [defined evidence], released annually on a clean record.

P&L effect. £[premium] operating expense. No balance-sheet charge; collateral is restricted cash where posted.

The number that decides it. p · F / G moves from [a] to [b]. It must exceed 1.00.

What we are not claiming. The range on G; the inputs we could not verify; and the compliance we currently receive without paying for it, which this design deliberately does not disturb.

APPRECIATIVE QUESTIONS FOR YOUR LEADERSHIP TEAM

  1. Which of our controls genuinely works, and what specifically makes it work?
  2. Where do we already have a chokepoint that could carry one more test?
  3. If an underwriter priced our conduct, what would they want to see — and what would we be glad to show them?
  4. Where are we currently receiving compliance for free, and what would we be careful not to break?
  5. What is the one ratio about ourselves we could publish this year that nobody in our sector publishes?
  6. If this programme were quietly decaying in three years, what is the first number that would show it, and who would be looking at it that week?