The IFRS 17 accounting standard has improved technical discipline across the insurance industry. New models, new systems, new processes have made results more consistent, more auditable, and more defensible.
But improved compliance has not always translated into improved understanding.
At the board and executive level, a familiar frustration remains: when reserves move, explanations often sound confident — but are not always convincing.
This is not a technical failure.
It is a governance one.
Across the industry, the shift from compliance to comprehension is exposing a new kind of risk: not whether organizations can produce explanations, but whether those explanations are robust enough to govern decisions.
Two pressures, one growing gap
Senior leaders now sit between two powerful pressures.
- The first is IFRS 17 itself. While the standard does not explicitly require causal attribution, it does demand coherent explanations of movements over time. Results must be traceable, narratives consistent, and changes defensible across reporting periods.
- The second pressure comes from boards, auditors, regulators, and investors. They are not interested in model mechanics. They want to understand whether changes are structural or transitory, whether action is required, and whether the organization is truly in control.
Between these two pressures lies a widening gap: the distance between what leaders are expected to explain and what the data can reliably support.
That gap is where governance risk now lives.
1. Accurate forecasts are not the same as credible explanations
Many executive discussions implicitly assume that if the numbers are right, the explanation must be right, as well.
That assumption is wrong.
In complex systems — insurance included — prediction and explanation are distinct tasks. A model can forecast outcomes accurately while offering little reliable insight into why those outcomes changed.
Under IFRS 17, leaders are increasingly asked not just whether results are reasonable, but what actually drove them. Treating prediction success as evidence of causal understanding is the first and most common executive blind spot.
This distinction is well known in public health and medico-legal risk assessment, where prediction, explanation, and attribution have long been treated as separate governance questions.
2. IFRS 17 creates accountability for explanation, not just compliance
Much of the early IFRS 17 effort focused on implementation: systems, calculations, disclosures.
But as the standard matures, the real challenge has shifted. IFRS 17 has made explanation unavoidable. Executives are now accountable not only for reported numbers but for the stories attached to them — and for their consistency over time.
This accountability is new. And many organizations are still responding with tools designed for forecasting, not for governing explanation.
3. Over-confidence in attribution narratives is a strategic risk
When data does not support stable attribution, organizations rarely say "we don't know." Instead, they fill the gap with confident narratives.
This is understandable — and dangerous.
Over-confident explanations drive decisions: pricing changes, underwriting shifts, capital reallocation. When those explanations are fragile, the organization risks acting decisively on the wrong signals.
The danger is not getting attribution slightly wrong.
The danger is not knowing when attribution is unreliable — and acting anyway.
4. More sophisticated models often increase confidence, not insight
A common response to weak explanations is to add complexity: more variables, more refinement, more layers of modeling.
This often worsens the governance problem.
Complex models can improve forecasts while simultaneously reducing transparency for decision-makers. They can make explanations sound more authoritative, even as the underlying signal becomes harder to assess.
From a board perspective, this creates a false sense of control: explanations feel stronger precisely when they should be treated with greater caution.
5. Explainability is not the same as storytelling
Clear charts, fluent commentary, and alignment across functions are often taken as signs of understanding. They are not.
In many real-world portfolios, different drivers — frequency, severity, inflation — move together or are structurally entangled. In such cases, attribution can look plausible without being reliable.
Across multiple reserving datasets we have examined, some portfolios contain enough independent signal to support stable attribution, while others do not — regardless of model sophistication. In the latter case, attribution stories can still be produced, but they are not stable. The limitation lies in the data structure, not in the models.
Crucially, failed attribution does not imply failed governance — if it is recognized early. When attribution feasibility is assessed upfront, models can be reframed to restore interpretability, allowing leaders to distinguish between what happened and what can safely guide decisions.
When this step is skipped, organizations tend to act on confident explanations that the data cannot sustain. For boards, the distinction between explainability and post-hoc storytelling is critical — and rarely made explicit.
6. The most important question executives should ask
The executive implication is simple: attribution is not a yes-or-no capability. It is a property of the data and the portfolio.
Governance improves not by forcing explanations everywhere, but by knowing where attribution is reliable, where it is uncertain, and where it should not be used to drive decisions at all.
This is why attribution under IFRS 17 should start as a diagnostic question, not a reporting requirement.
When presented with an attribution analysis, the instinctive executive question is: "Which driver mattered most?"
Under IFRS 17, that is often the wrong question.
The more important one is: "How reliable is this explanation — and what decisions can safely be based on it?"
This reframing shifts the conversation from debating drivers to governing uncertainty — from defending narratives to assessing their limits.
What this means for boards and senior leaders
IFRS 17 did not just introduce new accounting mechanics: it introduced a new responsibility for explanation.
Attribution has become a board-level issue not because it improves models, but because it determines whether strategic decisions, capital allocation, and external communications are grounded in evidence — or in confident storytelling.
In the IFRS 17 era, good governance is no longer about having the most persuasive explanation.
It is about knowing when explanations deserve confidence — and when they deserve restraint.
Ultimately, IFRS 17 is forcing leaders to confront a deeper question: not how much certainty they can present, but how much uncertainty their governance can safely carry.
