How to Structure a Solvency II ORSA Narrative That Pre-empts PRA Capital Challenge

The ORSA is the single most revealing document an insurer sends the PRA. It tells supervisors whether the board actually owns its risk profile or whether the actuarial team is running the story alone. Weak ORSAs invite capital add-ons, s.166 reviews, and repeated information requests. Strong ORSAs close down challenge before it starts.

Key Executive Takeaways

• The PRA reads the ORSA for evidence of board ownership and forward-looking judgement, not model sophistication; structure the narrative around decisions taken, not calculations performed.
• Pre-empt capital challenge by explicitly addressing the four areas supervisors probe hardest: SCR appropriateness, stress severity, management action credibility, and reverse stress testing bite.
• Every material judgement should carry a visible audit trail linking risk identification, quantification, board debate, and resulting action or tolerance.

Start with the supervisory lens, not the template

The PRA reads your ORSA looking for specific things: whether the standard formula or internal model genuinely reflects your risk profile, whether stresses are severe enough, whether management actions are executable, and whether the board has actually engaged. Write the narrative to answer those questions directly. Do not bury the answers inside 200 pages of appendices.

The most common failure is treating the ORSA as a compliance artefact assembled bottom-up from risk function outputs. Good ORSAs are drafted top-down from a small number of board-level questions: Is our capital adequate under plausible severe conditions? Where is the business model most fragile? What would we actually do if things went wrong?

Address SCR appropriateness head-on

If you use the standard formula, do not simply assert appropriateness. Set out, risk module by risk module, where your profile diverges from the calibration assumptions and quantify the impact. Non-life insurers with concentrated exposures, life insurers with unusual guarantee structures, or firms with material operational risk profiles all attract PRA attention here. Show the analysis, name the deviations, and either add capital or explain rigorously why you have not.

Internal model firms face a different challenge: demonstrating that the model remains appropriate as the business evolves. Include a clear model change log, validation findings, and explicit commentary on limitations. The PRA is unimpressed by validation reports that find nothing.

Make stress and scenario testing bite

Stress severity is where most ORSAs fall short. A 1-in-200 stress that produces a 15% capital impact is almost always insufficiently severe. Calibrate against historical events, name them, and explain why your chosen severity is credible. Combine stresses rather than running them in isolation. Include at least one scenario that breaches the SCR and set out what happens next.

Reverse stress testing is where supervisors look for intellectual honesty. Identify the point at which the business model becomes unviable, not just where capital breaches occur. Vague reverse stresses signal a board that has not confronted its own fragility.

Management actions must be executable, not aspirational

The PRA has grown sceptical of management action credit. If your ORSA assumes reinsurance restructuring, capital raising, or dividend suspension under stress, document the specific triggers, decision-making authority, execution timeline, and any counterparty or market dependencies. Actions that require a functioning market during a market stress get discounted.

Good practice: table each material management action with trigger, owner, lead time, and evidence of feasibility (board minutes, standing facilities, precedent transactions).

Show the board actually engaged

Include a clear record of board challenge: which assumptions were questioned, what alternatives were considered, what changed as a result. An ORSA that emerges from the board unchanged is a red flag. Reference specific meetings, papers, and decisions. If the board increased a stress, tightened an appetite, or rejected a management action assumption, say so.

What good looks like

A supervisor should be able to read the executive summary and understand: the firm's material risks, the capital held against them, the severity of stresses considered, the actions available, and the board's conclusion on adequacy. Everything else supports that story.

Next step

Before the next ORSA cycle, run a red-team review against the four PRA probe areas above. If you cannot produce clean answers on any one of them, that is where your capital challenge will come from. Fix it in the drafting, not in the response letter.

Frequently Asked Questions

How long should the ORSA actually be?

There is no right length, but concise ORSAs (60 to 100 pages of core narrative plus technical appendices) tend to land better than sprawling documents. Length correlates weakly with quality and strongly with reader fatigue.

How much divergence from the standard formula triggers a capital add-on?

There is no threshold. The PRA looks at materiality in the context of the firm. A 5% understatement on a dominant risk module matters more than a 20% understatement on an immaterial one. Quantify honestly and either hold capital or defend the position.

Should we include scenarios the board finds uncomfortable?

Yes. Omitting obvious tail scenarios (pandemic recurrence, sovereign stress, cyber catastrophe, climate transition shocks relevant to your book) is more damaging than including them and showing you can survive.

How do we handle group ORSAs where the UK entity is one of several?

Produce a solo UK ORSA that stands alone on UK entity risks and capital. Cross-reference the group ORSA but do not rely on it. The PRA supervises the UK entity and wants to see the UK board's view.

What role should the CRO play versus the actuarial function?

The actuarial function produces the numbers. The CRO owns the risk narrative and the board engagement. If the ORSA reads like an actuarial report, the CRO has not done their job.