If you are on the boardEasy

Non-executives, trustees and audit committee members have to ask the right question about an instrument they will never trade, in a meeting that has already decided how long the item takes.

6 min read · 1 060 words

A board does not run anything. It reads a paper somebody else wrote, listens to somebody else explain it, and decides whether the answer is good enough — usually without the specialist knowledge that produced it, and always in less time than the subject deserves. This page adds no new subject matter: it is the shortest route through what is already here, arranged around the questions that seat is actually accountable for.

The position you are in, stated plainly

• You cannot verify the number and you are responsible for it. That is not a failure of preparation; it is the structure of the role, and pretending otherwise produces worse questions rather than better ones.
• Everything reaching you has been through somebody's judgement. Not dishonestly — a paper is a summary, and every summary is a set of decisions about what to leave out. The useful skill is knowing which decisions those usually are.
• A paper is written to be agreed with. Its author knows what the meeting is for. This is ordinary and it means the objection is rarely in the paper, so it has to come from somewhere else.
• And the thing that will hurt is structural, not a number being slightly wrong. Every case study on this site is a structure that worked until an assumption stopped holding, and in almost none of them was the arithmetic incorrect.

Four questions that do most of the work

They are general on purpose: they work on a treasury paper, a valuation paper, a new product approval and an outsourcing decision alike.

• Which of these numbers was observed, and which was produced? A trade that happened, a model's output, and the figure a statement had to print are three different things — price, value and mark is why the difference decides how much weight a figure can carry.
• What would have to be true for this to be wrong? Not "is it right". A paper defends its conclusion; almost none of them states the assumption that would overturn it, and asking for that assumption by name is the single most productive thing a board member does.
• Who can do something here without asking us? Which document governs sets out the four places discretion is written down — valuation, dealing, determination and substitution — and every one of them is a power somebody holds that a board paper rarely mentions.
• What happens if we cannot wait? Almost every failure on this site is a holder who was right and could not hold on. What a number is measuring is why an exposure and a funding obligation are different questions, and only one of them has a date.

Read these four before the next pack

• Which risk decides — five failure modes, and which one drives the outcome for each instrument. It is a two-minute exercise and it is the vocabulary a risk paper assumes you already have.
• Stress testing — a conditional question with no probability attached, deliberately. The section on reverse stress testing is the one to read twice: it is the only version that searches where the scenario library does not.
• Model risk — calibration reproduces today's market, estimation describes the world, and the dangerous model is the one that has been right long enough that its assumptions stopped being read as assumptions.
• How a fund is built — six parties and why the separation between them is the whole of the protection. The general principle behind every governance question: the party making the decisions should not be the party holding the money or the one checking it.

If the institution is a bank

• Bank capital — the misunderstanding first, because it is nearly universal: capital is a claim nobody has to be repaid, not a reserve in a vault, which is why "hold more capital" never means "lend less". Then the risk-weighted denominator, the leverage ratio as the check on it, and the buffer tension no page can resolve.
• Reading a capital disclosure — the document the bank publishes about itself, and which parts of it repay attention.
• How a position hits the books — the same holding reported three ways depending on a classification made when it was bought, which is how a capital ratio and a fair-value note can tell two stories about one balance sheet. SVB in 2023 is that mechanism and a funding event in one episode.

If it is a pension scheme or an insurer

• Defined benefit against defined contribution — who carries investment, longevity, inflation and sponsor risk, and the last row that is the whole page: each structure hides exactly the thing the other shows.
• LDI in 2022 — a hedge that was correct on exposure and fatal on funding, which is the shape of the trustee question that matters: not whether the hedge works, but whether it can be paid for on the day it does.
• The pensions seat and the insurance investment seat — what the people writing your papers are measured on, which is usually the fastest explanation of why a paper is shaped the way it is.

What this page will not do

• Make you a specialist. Four hours of reading does not replace the person who does this daily, and a board member who starts second-guessing the arithmetic has taken over a job that was not theirs and stopped doing their own.
• Tell you what to decide. Nothing on this site is advice, and a page that ranked options for a board would be exactly that.
• Give you a checklist to work through in the meeting. The four questions above are habits rather than an agenda item, and the one that does the most work — what would have to be true for this to be wrong — is the one no template can ask for you.

The one habit

Ask where the number came from before asking whether it is good. Almost every question worth asking in that room is a provenance question wearing something else's clothes, and it is the one question a paper's author can always answer and rarely volunteers.