Outgrace, home
Governance & RiskHow-to guide

How to write a board risk appetite statement directors can apply

Set a measure, a tolerance, an owner and a review date for each major risk, and keep the whole statement to one page a director will actually read.

A risk appetite statement that cannot be measured is a mission statement with a risk heading. Most boards have one. Few can say, in the meeting where a decision is due, whether the company is inside it. The fault is rarely in the wording. It lies in the absence of a number that someone is responsible for reading every month. The Bank of England's Prudential Regulation Authority puts the principle plainly. Its rules, it says, require financial firms to maintain sufficient capital and have adequate risk controls, and it supervises firms so that it can step in if they are not run in a safe and sound way. Regulated firms have a legal floor to meet. Everyone else can set a higher standard for themselves, and most of the work in this guide applies to unregulated companies as much as to banks.

Before you start

Five things need to be true before anyone drafts a line. Skip any of them and the statement will read well and do nothing. The first is an owner. The chair and the chief executive should agree, in writing, who proposes the statement to the board. In most companies that is the chief risk officer or, where there is none, the chief financial officer. The second is a list of risks the board actually reviews, taken from the last twelve months of board packs and minutes. The third is data. Every category needs one reliable source, and if the figure takes three weeks to assemble it cannot be a tolerance, however sensible it sounds. The fourth is a current position. Write down today's value for each measure, not the figure from last year's plan. The fifth is agreement on purpose: the statement is a tool for decisions, not a file for auditors.

  • A named proposer, agreed by the chair and the chief executive.
  • The risks taken from last year's board packs, not from a template.
  • One data source per category, with a known refresh date.
  • Today's value for every proposed measure.
  • A shared view that the statement exists to change decisions.
  1. Step 1

    Start with the decisions

    Ask each director to name the three decisions from the past year that would have come out differently with a clear limit in place. Write your categories from those answers. The usual mistake is to begin with a generic list of twenty risk types and ask the board to rate each one, which produces a heat map nobody discusses.

  2. Step 2

    Choose the categories the board owns

    Keep the list to the categories the board can actually move. In our experience, six to eight is the most a board tracks with any care. Where two categories overlap, merge them. Operational risk and cyber risk, for instance, often share a measure and a owner, and listing them separately doubles the reporting without doubling the insight.

  3. Step 3

    Write one measure per category

    Each category needs a single number or a dated test that a director can read in the room and judge as inside or outside the limit. Ask whether a non-specialist could tell from the figure alone that the company had a problem. If the answer is no, pick another measure. The common failure is the qualitative tolerance, such as low, moderate or acceptable, which every director agrees with and nobody can check.

  4. Step 4

    Set appetite, tolerance and trigger

    Set three separate figures for each measure. Appetite is where management aims to run. Tolerance is the limit the board will accept before it demands action. The trigger is the point at which the breach comes to the board, usually earlier than the tolerance itself. Do not set the tolerance at today's level, because a limit the company already sits on will never breach and will teach directors to ignore it.

  5. Step 5

    Assign owners across three lines

    A common way to divide the work follows the IIA's Three Lines Model. The first line runs the activity and reports the number. The second line proposes and challenges the measure and the limit. The third line, internal audit, tests whether the number is right. Write the three names against each measure. The usual mistake is one owner for everything, or a second-line team that both reports the figure and marks its own work.

  6. Step 6

    Fix the review cycle

    Set three review points and put them in the board calendar before the statement is approved. Rewrite the whole statement once a year. Check the numbers every quarter. Review any category the same week it breaches its tolerance. An annual review that simply re-approves last year's text is the most common way a statement goes stale, and a quarterly check nobody attends is the second.

  7. Step 7

    Test it against a bad quarter

    Choose one plausible bad scenario and walk the statement through it. A large customer fails to pay, or a cyber incident takes a core system offline for a day. Note which measures breach, when the trigger fires, and who acts. The European Central Bank's analysis points to cybersecurity threats and non-bank vulnerabilities among the risks it monitors in a highly uncertain environment, which is a reasonable place to look for scenarios. Most statements are tested only against a good quarter, which t

  8. Step 8

    Put it on one page

    Lay the statement out as a single table: category, measure, appetite, tolerance, trigger, owner and next review date. Anything that does not fit is either supporting detail for the appendix or a sign that the category should be cut. A forty-page document will be approved and then filed, and no director will read it twice.

Where boards go wrong

The most common failure is a tolerance that cannot fail. It looks cautious on paper because it sits at a sensible-sounding level, but it was set after the current position was measured, so it sits safely beside it. Directors then discover, in a bad month, that the company has been running inside a limit that never constrained anything. The fix is mechanical. Put the current value next to each proposed tolerance and ask the proposer to explain the gap. If there is no gap, the tolerance is a description, not a limit.

The second failure is a measure that arrives too late. A tolerance on a monthly figure that takes six weeks to close will be breached before the board hears of it. Check the lag for every measure, and where the lag is longer than the time the board needs to act, either find a faster proxy or move the trigger earlier. The third failure is disconnection from capital and funding. The Financial Stability Board estimates private credit assets at $1.5–2 trillion, and says the asset class supports financing for mid-sized companies while creating potential financial stability risks. A company that borrows from private credit lenders should expect those lenders to ask about funding and liquidity measures, and its own statement should contain a funding tolerance that the finance team can defend in the same language.

Regulated firms face a further point. The Bank of England's Prudential Regulation Authority opened a consultation on the Senior Managers and Certification Regime, Phase 2, as CP14/26, dated 22 October 2026, and the consultation is open. If your firm is in scope of that regime, ask counsel whether accountability for the statement, and for each measure's owner, will change once the rules settle. We do not have the detail to say so here.

Our rules require financial firms to maintain sufficient capital and have adequate risk controls in place.

Bank of England, PRA policy page, bankofengland.co.uk
Does every risk category need a number?

No, but every category needs a test that could fail. Where a sound number exists, use it. Where none exists, such as for reputation or a change in regulation, use a dated, observable event: a complaint volume that crosses a set figure, or a required filing that is not made by a set date. The test should let a director say yes or no at the meeting.

Who should approve the statement?

The board approves it. The chief executive or chief risk officer proposes it, and the audit or risk committee reviews the measures before they reach the full board. Record the approval date and the version number so that a later breach can be traced to the limit in force at the time.

How often should we change the limits?

Change them when the strategy, the capital plan or a major exposure changes, and otherwise at the annual rewrite. Frequent changes to make a breach disappear are a warning sign, and the board should see every change with the reason written beside it.

What if our data is not good enough yet?

Start with the measure you have and write down its weakness. Then set a dated plan to improve the data. A tolerance you cannot yet measure is an aspiration, and the board should label it as one rather than pretend the limit is in force.

Are we legally required to have one?

Regulated firms must meet the PRA's rules on capital and risk controls, and the detail of any statement they need should come from their own rulebook and their regulator. The sources we checked set out that principle, not a template. Ask counsel or your regulatory adviser before you treat any part of this guide as a compliance requirement.

This week, take the three decisions from last year that a clear limit would have changed, and write them on one sheet of paper. Ask the finance lead and the head of risk to attach a measure to each, with today's value beside it. Then put the three names against the three lines, and book the first quarterly check into the board calendar before the next meeting. A director can judge a statement on that sheet within a single meeting. If the numbers are missing, you have found the first piece of work, and the board can see it.

Newsletter

The Briefing

One email a week: the leaders we profiled, the ideas worth stealing, and nothing else.

Weekly, free, one click to leave. Read our privacy policy.