Testing Board Assumptions Against External Reality Before Regulatory Review

Start with the gap that actually matters

Boards rarely fail regulatory reviews because they misunderstand the rules. They fail because they misjudge what external decision-makers care about most, and in what order. The technical submission is fine. The tone, the sequencing of concessions, the framing of risk appetite: these are where assumptions diverge from reality.

The question to answer is not 'what does the regulator want' in the abstract. It is: on this specific review, with these specific reviewers, at this specific moment in the political and supervisory cycle, what will they actually push on?

Extract the board's assumptions before you test them

Before you can check the board against reality, you need the board's assumptions in written form. Most boards have never articulated them explicitly.

Run a short structured exercise with the chair, SID, risk committee chair and CEO separately. Ask each: name the top three concerns you expect the regulator to raise, rank the stakeholders by influence over the outcome, and state what you think 'good' looks like in the closing letter. Do not do this in a group. You want the divergence visible.

What you will typically find: the executives focus on remediation progress, the NEDs focus on culture and tone, and no one has a clear view of which second-order stakeholders (rating agencies, buy-side analysts, consumer bodies, Treasury) will shape the regulator's stance behind the scenes.

Test against three external reference points, not one

A single source of external input, usually a former regulator on retainer, gives you a confident but narrow read. Triangulate across three:

Recent supervisory output. Read the last 18 months of Dear CEO letters, speeches, enforcement notices and portfolio letters in your subsector. Note the verbs and the recurring examples. Supervisors telegraph priorities more than firms credit them for.

Peer intelligence. Talk to two or three peers who have been through similar reviews in the last year. Ask what surprised them, not what they prepared for. The surprises are the assumption gaps.

Adjacent voices. Consumer groups, trade press, sell-side analysts and select politicians shape the environment the supervisor operates in. If they are loud on a topic your board considers minor, that is a signal to reweight.

Score the divergence, do not just describe it

Put the board's ranked assumptions and the external-derived ranking side by side. For each item, mark: aligned, minor gap, or material gap. Then ask two questions on every material gap:

1. What is the consequence if we go into the review holding the board's view and the external view is right?
2. What would we need to see, or stop seeing, to change our mind?

The second question is the important one. Assumptions that cannot be falsified are not assumptions, they are beliefs, and beliefs do not survive contact with a skeptical supervisor.

What usually goes wrong

Three recurring failures:

• Confirmation from friendly sources. Advisers who depend on the relationship soften their read. Insist on written pre-read from at least one source with no ongoing commercial interest.
• Testing too late. Once the submission is drafted, the board is anchored. Do the assumption test before the submission outline is agreed, not after.
• Treating the regulator as monolithic. The case officer, the head of department and the executive committee often have different concerns. Map them separately.

What good looks like

A one-page document, signed off by the chair, that states: here are our five working assumptions about what will drive the review outcome, here is the evidence base for each, here is what would cause us to revise, and here is who owns the watch on each assumption between now and the review. Revisit it monthly.

This is a mundane artefact. It is also the single most useful thing a board can produce in the six months before a major review.

Your next move

Before your next board or risk committee meeting, ask each member privately for their top three predicted regulator concerns. If you get more than two different top-threes across the group, you have an alignment problem to solve before you have an external validation problem to solve. Fix the internal picture first, then test it.